Az angol nyelv kiejtése

The Pronunciation
of English
Balogné Bérces Katalin
Szentgyörgyi Szilárd
2006
Bölcsész
Konzorcium
043-cimlap.indd 1 2006.07.31. 10:39:37
Kiadta a Bölcsész Konzorcium
A Konzorcium tagjai:
Eötvös Loránd Tudományegyetem
Pécsi Tudományegyetem
Szegedi Tudományegyetem
Debreceni Egyetem
Pázmány Péter Katolikus Egyetem
Berzsenyi Dániel Főiskola
Eszterházy Károly Főiskola
Károli Gáspár Református Egyetem
Miskolci Egyetem
Nyíregyházi Főiskola
Pannon Egyetem
Kodolányi János Főiskola
Szent István Egyetem
Szerkesztette: Balogné Bérces Katalin
Lektor: Dávid Gyula
Siptár Péter
A kötet megjelenése az Európai Unió támogatásával,
a Nemzeti Fejlesztési Terv keretében valósult meg:
A felsőoktatás szerkezeti és tartalmi fejlesztése – HEFOP-3.3.1-P.-2004-09-0134/1.0













ISBN 963 9704 34 2
© Bölcsész Konzorcium. Minden jog fenntartva!
Bölcsész Konzorcium HEFOP Iroda
H-1088 Budapest, Múzeum krt. 4/A.
tel.: (+36 1) 485-5200/5772 – dekanbtk@ludens.elte.hu
Contents
Preface…………………………………………………………………... vi
1. English pronunciation: phonetics and phonology……………………. 1
2. The phonology of English consonants: an introduction……………… 13
3. The phonology of English vowels: an introduction………………….. 30
4. R-influence on vowels……………………………………………….. 45
5. The English syllable………………………………………………….. 61
6. Laryngeal features……………………………………………………. 78
7. Connected speech…………………………………………………….. 91
8. Word stress – Part 1: The degrees of stress………………………….. 108
9. Word stress – Part 2: Primary stress…………………………………. 123
10. Sentence stress and intonation……………………………………… 138
11. Letter-to-sound rules – Part 1: Consonants…………………………. 152
12. Letter-to-sound rules – Part 2: Vowels……………………………... 176
Bibliography……………………………………………………………. 191
Subject index…………………………………………………………… 193
v
Preface
This book contains twelve chapters introducing the basic characteristics of
the pronunciation of standard English, and it is designed for a one-term
course with twelve weekly topics elaborated on in ten to twenty pages on
average. It is the authors' intention to keep both the amount of material
covered and the students' reading load to the absolute minimum: the book
leaves, on purpose, considerable time and space for practice, revision and
assessment as well as for the inclusion of the personal preferences of the
instructor teaching the course.
Primarily for pedagogical reasons, but also out of space limitations,
the book describes the pronunciation of the standard dialects of English only,
though reference is made, whenever relevant to the topic, to non-standard
regional varieties, too. As usual in similar textbooks on English used in
Hungarian higher education, the description focuses on standard British
English pronunciation ("RP"), as this is the accent which most Hungarian
students of English appear to be familiar with. However, an attempt is made
to include the characterization of standard American English ("GA") as well,
especially where the two reference accents significantly differ, both because
neither of the standards should be considered inferior to the other and
neglected, and because Hungarian students seem to be exposed to American
English at least as much as, if not more than, to British English. In addition,
the authors believe that all students holding a degree in English are required
to be aware of what pronunciation differences are to be expected between
native speakers of different linguistic backgrounds. Once the major varieties
are introduced, it is generally accepted that the two together present a good
starting point for the discussion of other dialects.
vi
Throughout the book, a certain amount of preliminary knowledge of
the basic notions of linguistics is assumed since most BA programmes
contain an introductory tier with at least one course in elementary linguistics.
Nevertheless, all the discussions aim to be as self-explanatory and self-
contained as possible. At the end of this section you will find a list of the
terms which are considered to be elementary.
The book is accompanied by a digital material, which contains
exercises to practice and revise the topics covered by the readings. These
exercises are primarily meant to be used in their interactive form as home
practice for the weekly readings. Alternatively, their static version can be
handed out in class – this is totally at the instructor's discretion.
As it has been mentioned several times, the book grants considerable
freedom to the instructor, and ample opportunities for extension. Also, it
covers so little phonological theory that it leaves the way open for advanced
studies in both BA and MA programmes to elaborate on all the topics in more
detail.
Before you study this book, check whether you are familiar with the
following terms: adjective, adverb, allomorph, allophone, articulation,
assimilation, bound morpheme, complementary distribution, compound,
conjunction, consonant, demonstrative pronoun, diacritic, diphthong,
distribution, free morpheme, free variation, function word, homophone,
idiomatic, interrogative pronoun, lexical content word, loanword, manners
of articulation, minimal pair, monomorphemic, monophthong,
monosyllabic, morpheme, noun, phonetics, places of articulation, root,
schwa, segment, speech organs, spelling, stem, stress, suffix, syllable,
triphthong, verb, voicing, vowel, Wh-question, Yes/No question
vii
1. English pronunciation: phonetics and phonology
This is a book on the pronunciation of English. No matter how obvious our
topic might seem, it needs considerable clarification. Namely, we have to
explain what we mean by "English" on the one hand, and "pronunciation" on
the other.
English has as many as 400 million native speakers in the British
Isles, North America, Australia and New Zealand as well as parts of Africa
and Asia. It is the most popular language learnt and used as a second or
foreign language. As you know, it is a member of the Indo-European family
of languages, and as such, is genetically related to a number of tongues
spoken all over Europe and Asia: from the Indian subcontinent to Western
and Southern Europe. In contrast, Hungarian is of Uralic (more precisely,
Finno-Ugric) origin, cognate to, among others, Finnish, Estonian, Lapp, and
the Samoyed languages. Therefore, from a historical point of view, English
and Hungarian could not be farther from each other. This results in numerous
linguistic differences between the two languages, which is why the
Hungarian student of English (as well as the English student of Hungarian) is
faced with so many difficulties.
However, languages can not only be related genetically – English is
not only related to the other Indo-European languages (and most closely, of
course, to West Germanic German, Flemish, Dutch, Afrikaans, Frisian, and
Yiddish). Compare English to German, for example: apart from the core of
the word stock, they exhibit very few of the similarities one would expect
from two languages that have evolved from a common ancestor. As far as
linguistic structure is concerned, English shows more resemblances to
Chinese (with its comparative lack of different word endings) than to any of
Chapter 1
the other members of the Indo-European family. Languages, then, can also be
related to each other according to what linguistic type they belong to, that is,
typologically.
Finally, all languages are naturally related to others culturally
through the contacts they come into. Therefore, English is so related to North
American Indian languages: although they share neither early history nor
(much) linguistic structure, they have borrowed a number of place names and
terms from each other, in both directions. This is the way in which English
can be considered as related to Hungarian, too. Even though the amount of
English influence on Hungarian vocabulary is obviously larger, we are able to
identify a handful of English words as words of Hungarian origin, including
the well-known loanwords coach (from kocsi 'carriage') and biro ('ball-point
pen', from the name of the inventor, László Bíró).
English does not only have contacts with non-Germanic languages
outside the British Isles; even in its homeland English lives side by side with
a couple of Celtic languages, which belong to another branch of the Indo-
European family: Welsh (spoken in Wales), Irish (or Irish Gaelic, still spoken
in parts of Ireland), and Scottish Gaelic (in the north-west of Scotland,
especially the Hebrides Islands). It should be kept in mind that these are only
distantly related to English, and are languages in their own right.
(Unfortunately, all the Celtic languages formerly spoken on the European
continent are now extinct, e.g., Gaulish, cf. Asterix and company).
Besides its intricate pattern of connections to other languages and its
dominant status on the linguistic map of the world, English is very special in
at least one more respect. Due to a series of historical events, a discussion of
which is beyond the present purposes, English has developed two standard
varieties, that is, two forms, both of which are equally accepted by the
societies of their respective countries. One is Standard British English in
2
English pronunciation: phonetics and phonology
England, the other is Standard American English in the USA. As this book is
exclusively concerned with pronunciation, henceforth we will concentrate on
the pronunciation varieties (called accents
1
) of the two standards. The
standard accent of England is traditionally referred to as Received
Pronunciation (where received means 'accepted'), abbreviated to RP,
whereas that of the USA is often referred to as General American, or GA for
short. The two accents differ in many ways, most of which concern vowels.
On the one hand, a number of systematic sound correspondences can be
identified, e.g., whenever an RP speaker uses the vowel /co/ as in know, go,
boat, a GA speaker pronounces /oo/; RP /o/ in lot and dog corresponds to a
somewhat longer /o/ in GA. On the other hand, there are differences which
are not as general as that but only affect certain individual words. For
example, a couple of words pronounced with /o:/ in RP, e.g., after, ask, bath,
can't, chance, class, dance, glass, grass, half, last, pass, past, path, rather,
staff, have /æ/ in GA. Further examples are given in the chart below:
typical RP typical GA
address /c'orcs/ /'æorcs/
advertisement /co'va:tismcnt/ /'æovcrtaizmcnt/
ate /ct/ /cit/
clerk /kIo:k/ /kIark/
figure /'Iigc/ /'Iigjcr/
inquiry /in'kwaicri/ /'inkwcri/
laboratory /Ic'borctri/ /'Iæbrct5:ri/
1
Notice that this sense of the word accent is much wider than in everyday use, where it
basically coincides with what linguists refer to as a foreign accent. Here, in contrast, it is a
general expression to refer to the pronounced form of any variety of any language, that is, the
standard accent (standard English pronunciation, standard Hungarian pronunciation, etc.) is
just another accent in the same way as geographically or otherwise localizable forms (e.g.,
Australian English, working class London English (called Cockney), Black English (that is,
the African American vernacular), or the Szeged dialect of Hungarian). Every speaker has an
accent.
3
Chapter 1
leisure /'Ic¸c/ /'Ii:¸cr/
lieutenant /IcI'tcncnt/ /Iu:'tcncnt/
(n)either /'(n)aiðc/ /'(n)i:ðcr/
schedule /'jcoju:I/ /'skco¸u:I/
shone /jon/ /joon/
tomato /tc'mo:tco/ /tc'mcitoo/
vase /vo:z/ /vcis/
Z /zco/ /zi:/
Although the two varieties of English mostly differ in pronunciation, there
exists a certain amount of vocabulary, spelling and grammatical
differences as well. Since these are irrelevant to the discussion in the rest of
this book, we will mention just a few. The following table lists some of the
notions that American English (AmE) and British English (BrE) use different
words for.
AmE BrE AmE BrE
apartment flat faucet tap
baggage luggage french fries (potato) chips
bill (bank) note garbage rubbish
cab taxi gasoline petrol
candy sweets hood (of a car) bonnet
closet wardrobe line queue
cookie biscuit (potato) chips (potato) crisps
corn maize sidewalk pavement
diaper nappy the first floor the ground floor
elevator lift truck lorry
eraser rubber trunk (of a car) boot
fall autumn vacation holiday
In contrast to pronunciation and vocabulary differences, the two systems of
spelling and grammar do not deviate considerably. As to differences in
spelling, there are two types again: some are systematic (e.g., words ending
in -our, -ise and -re in British English end in -or, -ize and -er in American
4
English pronunciation: phonetics and phonology
English, e.g., colour/color, realise/realize
2
, centre/center, theatre/theater; in
AmE final -l is not usually doubled, e.g., AmE traveler, leveling – BrE
traveller, levelling), some characterize individual words only, e.g., BrE
cheque, gaol, plough, programme, pyjamas, tyre correspond to AmE check,
jail, plow, program, pajamas, tire. Grammatical differences do not abound,
either; perhaps the most conspicuous concerns the usage of have, as in AmE
Do you have a problem? vs. typical BrE Have you got a problem? In
addition, American English uses simple past tense in some cases where
British English has present perfect, e.g., AmE He just went home. As to verb
forms, in British English the past tense and past participle of burn, dream,
lean, learn, smell, spell, spill and spoil are typically irregular while in
American English they are regular; fit, quit and wet are regular in British
English but irregular in American English (all three forms being the same);
dive is regular in British English but irregular in American English
(dive/dove/dived); and the past participle of get is gotten in American
English, got in British English. Finally, there are small differences in the use
of prepositions, e.g., AmE meet with sy – BrE meet sy, AmE stay home – BrE
stay at home, AmE Monday through Friday – BrE Monday to Friday.
Turning back to pronunciation, in the rest of the book the main emphasis falls
on RP since it is the pronunciation most students of English as a foreign
language are familiar with all over the world, but its occasional differences
from GA (and some other accents) are not left unmentioned, either.
But what elements is pronunciation composed of? Let us first take a
look at the basic mechanism that is used to produce speech sounds in English
and in most Indo-European languages and also in Hungarian. The first phase
2
In this book, we follow the British conventions for spelling. However, in words with
alternative -ise/ize, -ize is used henceforth, as this form is getting so widespread that even
major British publishers recommend it to their authors.
5
Chapter 1
in the process of articulation (speech production) is called a pulmonic
egressive airstream mechanism, meaning that the source of the air to be
used in speech is the lungs ("pulmonic") and that the direction of the airflow
is outward ("egressive"). It is important to note as other languages might use
other kinds of airstream mechanisms to produce certain speech sounds – e.g.,
implosives, ejectives and clicks, which are not found in Indo-European
languages or in Hungarian.
As the air leaves the lungs it continues upward in the windpipe
(trachea), and enters the so-called vocal tract, where it is modified in various
ways by the movements of the speech organs called articulators. All speech
sounds can be classified according to where in the vocal tract this
modification takes place (the so-called place of articulation) and how
exactly this modification is carried out (the manner of articulation). As a
third term in the description of speech sounds, we can specify how active the
vocal cords are: whether they vibrate (in voiced sounds) or not (in voiceless
sounds). A more detailed discussion of the articulation of English consonants
is found in Chapter 2, of the articulation of English (more precisely, RP)
vowels in Chapter 3, and of voicing and related phenomena in Chapter 6.
The features mentioned above characterize individual segments, that
is, speech sounds – manner, place and voicing are the so-called segmental
features of speech. However, larger chunks of pronunciation also have
characteristics of their own – these are the so-called suprasegmental
features. They are named so because in some sense of the word they are
situated "above" segments, they affect elements which are higher up in the
hierarchy of linguistic units: syllables, phrases, sentences. The two most
significant suprasegmental features are stress (discussed in Chapters 7, 8, and
9) and intonation (Chapter 10). Notice that it is never a single consonant or
vowel which is stressed, but the combination of consonants and vowels called
6
English pronunciation: phonetics and phonology
syllable; it is never a single consonant or vowel which has a characteristic
intonational contour, but a whole phrase or sentence.
The scientific study of the segmental and suprasegmental features of
speech is called phonetics. Although this is not clear from the above
discussion, phonetics does not only deal with the process of articulation, that
is, speech production, but it is concerned with acoustics (the way speech
travels in the air in the form of sound waves) and speech perception
(sometimes referred to as auditory phonetics), too. All the three aspects of
pronunciation are of equal importance, nevertheless, in the rest of the book
we will concentrate on articulation as it comprises a minimally necessary
element of the physical properties of speech, which is at the same time
sufficient for the present purposes.
A different point of view is taken by the branch of linguistics called
phonology. It also deals with speech and sounds, and it borrows the terms
and notions of phonetics, but it only uses them as tools to achieve the
ultimate objective: describe the functions the segments have in speech, the
relationships they contract with each other, and the various systems and
patterns they constitute. For example, the same sound, i.e., the same phonetic
object, may serve as an independent unit (a phoneme) in one language but
only as a form, a positional variant (an allophone) of a phoneme in another.
Two phonemes always enter into such a relation that they contrast and
distinguish words; allophones never do so but are predictable instead. For
example, a plain [k] sound and its aspirated version [kº] (with an extra puff
of air following the consonant – see Chapter 6) are separate phonemes in,
e.g., Hindi, where a lot of word pairs (called minimal pairs) like /kan/ and
/kºan/ are distinguished by this very feature – and it does matter which word
you mean as the former means 'ear' while the latter 'mime'. The same is not
true for these sounds in English: in skin it is plain but in kin it is aspirated,
7
Chapter 1
however, this is totally predictable as all word-initial k's are aspirated unless
they are preceded by a s, in which case they are always plain. Because of this
it is impossible for them to appear in identical phonological positions, they
mutually exclude each other, that is, are in complementary distribution.
Phonology also attempts to handle cases when a sound appears in
different forms in different environments, i.e., when phonemes or allophones
alternate. For example, the two types of English k above can be argued to
stand in such a correlation: a common underlyer k is realized as aspirated at
the beginning of words, as plain after an s. Such alternations in linguistics
are commonly referred to as rules. A note of warning is in order here, though.
The word rule should not be taken here in the same sense as in the case of,
e.g., the rules of the Highway Code, or the rules of etiquette. Instead, the
rules of language are more like the rules (or "laws") of physics or football: it
is the rules which constitute the system, which cannot exist without these
rules. There is no physical world without, say, gravitation, and a ball game in
which the players are allowed to catch the ball cannot be football. In contrast,
traffic does exist without the Highway Code (in fact, there used to be a time
when cars were already used but no traffic signs had been invented yet; and
we also know how often drivers and pedestrians break these rules without
traffic as such coming to an end); and it is possible to show (some kind of)
human behaviour without respect to the rules of politeness (and how many
people do so at least in certain situations!). The rules of a language are not
like that. Languages do not exist without their rules – in fact, the rules define
the languages. A system in which all k's are plain cannot be (native) English;
pronunciations like that of kin with a plain k are ill-formed (or,
ungrammatical), at least in standard English, and will henceforth be
indicated with an asterisk (*), e.g., *[kin]. The word rule therefore denotes
8
English pronunciation: phonetics and phonology
the observation of some systematic regularity rather than a regulation which
must be obeyed by all good citizens.
To summarize the discussion so far, we can state that phonetics treats
speech sounds from the viewpoint of their physical properties, while
phonology is concerned with their function and patterning within a linguistic
system. Sometimes these two viewpoints arrive at totally different
conclusions. For example, compare two vowels of English, the one at the
beginning of about (called schwa) and the vowel of bird. Phonetically, they
are almost identical: in both cases the airstream entering the vocal tract is
only slightly modified, with the tongue resting in its neutral position.
Phonologically, however, they are each other's opposites: the former can only
occur in weak, unstressed syllables, whereas the latter can only occur in
strong, stressed syllables. That is why no minimal pairs exist for these two
sounds: they mutually exclude each other. Therefore, two phonetically nearly
identical objects are evaluated by phonology as two distinct elements. A great
number of further examples illustrate that the phonetic and the phonological
classes of sounds do not necessarily coincide.
The two different points of view may also influence the notational
conventions analysts use. No matter to what extent the transcription system
of the International Phonetic Alphabet (IPA for short) is based on universal
agreement, phonologists vary as to which symbols to use in their description.
For instance, someone primarily governed by phonetic criteria will choose
the same symbol for about and bird, e.g., /c/; while phonology-oriented
researchers will transcribe only about with a schwa (/c/) and mark the bird-
sound differently, e.g., with /a/. Actually, both types exist in Englsh: the first
solution characterizes the so-called Jonesian notation (found in, e.g., the old
bilingual Országh-dictionaries) whereas the second one is utilized in the so-
called Gimsonian system (e.g., the latest editions of Oxford Advanced
9
Chapter 1
Learner's Dictionary). (You may have already guessed that both are named
after the person who devised them.) The two transcription systems vary in
several other respects as well; this book favours Gimson's version of the IPA.
The differences of phonetic detail versus phonological analysis in
transcription can be highlighted by enclosing the two types in different
brackets: usually square brackets [ ] denote a phonetic transcription while
slant brackets / / stand for a phonological one. E.g., English kin [kºin] can be
phonologically transcribed as /kin/ since the aspirated k is simply a phonetic
variant (allophone) of the phoneme /k/.
A crucial difference between phonology and phonetics lies in their
status concerning native speakers: namely, the former is, but the latter is not,
part of linguistic knowledge. Native speakers know when to aspirate their k's
because they are native speakers; they learn this when they acquire their
mother tongue, and anybody who is ignorant of aspiration must be a non-
native speaker. They also know, subconsciously of course, that the first vowel
in police is not the same as the one in pearly: only the former is weak enough
to be dropped, for instance, so that /pc'Ii:s/ frequently becomes /'pIi:s/ (cf.
Chapter 5), but the same is impossible with the latter: /'pa(r)Ii/ is never */'pIi/.
As opposed to this, native speakers are typically unaware of phonetic facts
like the near physical identity of the two vowels, or the exact articulatory
gestures involved in their production (not to mention their acoustics).
Not only the phonetic aspect of a language is considered to lie outside
the linguistic knowledge of native speakers – so is the history of the language
(would you as a native speaker of Hungarian know that your mother tongue
is a Finno-Ugric language had you not been taught this at school?) as well as
the written form, the spelling.
10
English pronunciation: phonetics and phonology
The relationship between English pronunciation and spelling (also
called orthography, and commonly given between angle brackets, < >) is
worthy of interest in this book, for at least two reasons. On the one hand, the
spelling system of English is a mixture of several different traditions and no
major reforms have affected it for centuries – as a result, there are not too
many one-to-one spelling-to-pronunciation correspondences, or letter-to-
sound rules. The same sound, say /e/, can be spelt, with single letters, as
<e> (in bed), <a> (in many), <u> (in bury), or with digraphs (combinations
of two graphic symbols) like <ea> (in head) (for more detail on letter-to-
sound rules, see Chapters 11-12). On the other hand, being non-native
learners of English, we are very often first confronted with an unknown word
in its spelt form and consequently we tend to overestimate its role: pronounce
long consonants (called geminates) – erroneously – for double consonant
letters (e.g., Emma), or pronounce silent letters (e.g., iron, Wednesday).
Notice however, that human language is primarily spoken: children learn to
speak first and are explicitly taught to read and write somewhat later (in fact,
some native speakers never acquire the spelt form of their language and
remain illiterate); and speech comes first in the history of language itself, too
(writing systems have emerged to represent already existing spoken
languages, and many cultures have never employed an orthography). Thus
writing must be conceived of as a derived version of the spoken language,
and not the other way round.
That component of linguistic knowledge that we are concerned with
here is, therefore, phonology. Nevertheless, we are forced to make constant
reference to other components as well, since phonology seems to be heavily
influenced by the rules of word formation (morphology) and, to a more
limited extent, the rules of sentence formation (syntax). In addition, most
phonological processes have exceptions, which cannot be accounted for by
11
Chapter 1
the rules but must be stored in the speakers' memory (called the lexicon). For
example, the fact that stress falls on the first syllable in personal but on the
second in personify is due to the morphological difference between these
words (viz., they are derived from the same stem, person, using suffixes of
different types); the fact that stress falls on the first syllable in blackbird
('feketerigó') but on the second in black bird ('fekete madár') is due to the
syntactic difference between a single (compound) noun and a phrase
composed of an adjective and a noun. Finally, the fact that stress falls on the
first syllable in the noun present but on the second in event is an example of a
mere idiosyncracy: the stress pattern of event must be memorized in the
lexicon as irregular. (On stress placement, see Chapters 8 and 9.)
In the remaining eleven chapters follows a description of the main
phonetic and phonological features of standard English pronunciation (RP
and GA), together with all the morphological, syntactic and lexical
conditions, which every student of English in higher education is expected to
be aware of, be able to recognize in native speech, and consciously use in
order to improve their pronunciation.
12
2. The phonology of English consonants: an introduction
Before you study this chapter, check whether you are familiar with the
following terms: allophone, compound, deletion, free variation, GA,
homophone, larynx, morpheme, nasal cavity, oral cavity, orthography, RP,
schwa, soft palate (velum), stress, suffix, syllable, vocal cords
In this chapter we take a brief look at what phonological processes affect the
consonants of English, especially those of RP and GA, and which aspect of
sound structure determines their behaviour. Since we assume a knowledge of
the basics of phonetics, the articulatory classification of consonants will only
be touched upon and not discussed in great detail.
According to their articulation, the consonants of English can be
classified along three terms: voicing, the place of articulation, and the manner
of articulation. Voicing results from vocal cord vibration, and it is dealt with
in considerable detail in Chapter 6. For the present purposes it is sufficient to
say that when the vocal cords vibrate, we get voiced sounds, and when they
do not, we get voiceless sounds. The place and manner of articulation refer to
where and how the airflow is obstructed during the production of the
consonant. There are four major places of articulation: labial (involving a
lip or both lips), coronal (involving the blade of the tongue called the
corona), velar (the back of the tongue moving towards the soft palate or
velum), and glottal (involving some kind of manipulation of the opening
between the vocal cords in the larynx called glottis), which are illustrated in
the following diagram showing the cross-section of the head.
Chapter 2

VELAR
CORONAL
LABIAL
GLOTTAL
These places of articulation can be further divided into subcases, e.g.,
whether a labial makes use of both lips (bilabial) or just the lower one plus
the upper teeth (labiodental). These are included in the consonant chart
below.
According to the manner of articulation, there are several possible
divisions of consonants. For example, some consonants are oral (the air
escapes through the oral cavity and the mouth), others are nasal (the air
escapes through the nasal cavity, i.e., the nose). If the articulation involves a
total obstruction of the air in the larynx or the oral cavity, that is, if the air is
stopped for a short period, we get stops, or, in other words, noncontinuant
sounds. Otherwise the sound is continuant. One consonant, /I/, is special:
although the tip of the tongue touches the alveolar ridge, which normally
results in noncontinuant articulation, in this case the air is able to escape
along the sides of the tongue (the name of this manner is lateral), and
therefore it is generally assumed to belong to the continuants.
14
The phonology of English consonants: an introduction
The major subclasses of consonants, however, stem from a further
aspect of the manner of articulation: the degree of openness of the vocal tract,
with the resulting relative loudness of sounds called sonority. The more open
the vocal tract (that is, the smaller the degree of obstruction), the more
sonorous the sound is. Accordingly, the following classes can be set up.
degree of sonority
——————————————————————————————→
oral stops (plosives) and affricates – fricatives – nasal stops – liquids – glides (semivowels) (– vowels)
As you can see, sonority increases from left to right, and this order of sound
classes constitutes a sonority scale. That is, oral stops and affricates are the
least sonorous as their production involves complete obstruction to the
airflow. The most sonorous consonants are the glides, but the most sonorous
sounds are the vowels. From plosives up to (and including) fricatives the
obstruction is considerable; these consonants are called obstruents. The
remaining classes (nasals, liquids, glides) are the sonorant consonants
because they are dominated by sonority.
The following chart includes all places and manners of articulation
relevant to the description of English consonants. When consonants appear in
pairs, the one on the right is voiced, the one on the left is voiceless. Unpaired
obstruents are voiceless; all sonorants are voiced.
15
Chapter 2
16
The phonology of English consonants: an introduction
The obstruents in the shaded cells (s z j ¸ tj o¸) are the hissing and
hushing sounds called sibilants. The consonant /w/ is produced with two
important articulatory gestures, and consequently appears in the chart twice:
it involves considerable lip rounding on the one hand, and a velar gesture on
the other. For this reason, it is sometimes termed labio-velar.
After this brief introduction to the phonetics of consonants, we turn to
their phonology, that is, what processes they undergo and trigger. First, let us
see one of the most salient differences between RP and GA: the
pronunciation of orthographic <r>. Consider a pair of rhymes in the song An
Englishman in New York by British pop musician Sting, with the rhyming
words underlined:
You can hear it in my accent when I talk
I'm an Englishman in New York
The two words, talk and York, rhyme because Sting (along with millions of
English speakers) does not pronounce the <r> in York, which is not true for,
e.g., the speakers of GA. (Notice that the <l> in talk is not pronounced in any
accent of English, therefore this word contains three sound segments only.)
The result is /t5:k/ for talk and, more importantly, /j5:k/ for York. The process
whereby certain spelt <r>'s are not pronounced, that is, are deleted in certain
accents of English is traditionally called R-dropping. Those accents of
English whose speakers pronounce all orthographic <r>'s are referred to as
rhotic accents (after the name of the corresponding letter in the Greek
alphabet, rho). These include GA, Canadian English, British English in North
and South-West England, Scottish English, and Irish English. In contrast, RP,
Southern British English (including Cockney and London English), Welsh
English, Australian English, New Zealand English, South African English,
and American English in New England and parts of Southern US are non-
rhotic, that is, their speakers drop the <r> when it is followed by a consonant
17
Chapter 2
or a pause (i.e., nothing in speech), and they only pronounce it when it is
followed by a vowel. The following chart illustrates this with a few
examples.
no /r/ pronounced /r/
before a consonant before a pause before a vowel
(a) (b) (c) (d) (e) (f) (g)
York
party
bird
allergy
leopard
particular
bears
tired
iron
aren't
feared
retirement
fires
rarely
your
car
err
refer
teacher
particular
bear
tire
bore
care
lyre
restore
more
centre
ring
routine
rhyme
refer
restore
retirement
rarely
crow
pray
tribute
shrimp
Africa
poetry
arrive
tiring
boring
error
referee
fiery
furry
rarest
In non-rhotic accents like RP, then, no /r/ is pronounced before a consonant,
as in the examples in column (a). Sometimes the consonant is a suffix, e.g.,
bears. Very often the letter <r> is immediately followed by a vowel letter
which is not pronounced, as in column (b) – in such cases the <r> really is
before a consonant and behaves accordingly. The most frequent silent letter is
<e>, so it can be misleading when some other vowel letter, e.g., the <o> in
iron, is unpronounced. (You may be able to recall a Bob Marley song entitled
Iron Lion Zion, which makes good use of those three rhyming words. Notice
that Jamaican English is also non-rhotic.) The <r>'s in (c) and (d) are in final
position – again, silent letters (in (d)) do not count.
When the <r> is followed by a vowel, however, it is pronounced even
in non-rhotic accents, be it at the beginning (column (e)) or the middle
(column (f)) of the word. In the word rhyme in (e), the letter <h> is not
pronounced, so the /r/ is followed by a vowel, /ai/. In (g), the following
vowel, which enables the <r> to be pronounced, is part of a suffix. Notice
18
The phonology of English consonants: an introduction
that word-final <r>'s are like ghosts: sometimes you see them, sometimes you
do not. You "see" them when they are followed by a vowel-initial morpheme
like -ing, -er/-or, -ee, -y, etc. (those /r/'s are often called Linking-R), but they
disappear when they are final or when they are followed by a consonant-
initial morpheme like -(e)d (the <e> is mostly silent), -ment, -ly, etc., as in
retirement and rarely in column (e) above. Hence the difference between tire/
tired vs tiring, bore vs boring, err vs error, refer vs referee, and fur vs furry.
Note that it never matters whether the /r/ is spelt as a single letter <r>
or double <rr>. Also, remember that non-rhotic accents are named so not
because they do not pronounce any /r/ sounds, but because they do not
pronounce certain orthographic <r>'s, as opposed to rhotic accents, which
pronounce all <r>'s that are present in spelling. In fact, there are some cases
when rhotic speakers do not pronounce an /r/, and there is no <r> in spelling,
however, most non-rhotic speakers (perhaps except only for conservative RP,
i.e., RP spoken by older generations) pronounce one, e.g., in sawing /'s5:rin/,
gnawing /'n5:rin/, rumbaing /'r\mbcrin/, subpoenaing /sc'pi:ncrin/
'summoning sy to appear in a lawcourt', guffawing /gc'I5:rin/ 'giving a noisy
laugh', baahing /'bo:rin/ (of sheep), blaher /'bIo:rc/ 'more mediocre'. This is
called Intrusive-R. Intrusive-R is only found in non-rhotic accents, and it
only appears at (certain) morpheme boundaries, after certain vowels. We
learn more about this in Chapter 7.
An interesting consequence of R-dropping in non-rhotic accents is
that new homophones emerge, so sore sounds the same as saw, pour sounds
the same as paw, aren't sounds the same as aunt. Further examples include
farther – father, fort – fought, source – sauce, more – maw, tuner – tuna,
sort – sought, court – caught, spar – spa, career – Korea.
19
Chapter 2
When a word ends in an <r>, it can not only "escape" being dropped
when a vowel-initial suffix is attached to the word as in the examples in
column (g) in the chart above, but in fact any vowel-initial morpheme
following the word is able to produce the same effect, even a following
vowel-initial word. Therefore, Linking-R is heard in phrases like more
exciting, your eyes, (to) err is (human), care about, centre of, tire us, etc. This
can even happen between two sentences, e.g., He doesn't care. I do or There's
a spider. I'm scared. Similarly, under the same conditions as between a word
and a suffix, Intrusive-R appears between the two words in visa application,
(the) idea is, (the) Shah of (Persia), schwa insertion, law and (order), Gloria
Estefan, (cats) claw at (the furniture), (the giant) panda is (an endangered
species), etc., and between the two sentences in Try that sofa. It's softer or
Call Maria. I need her. Further homophones arise, e.g., vanilla ice – vanilla
rice, Amanda Avon – Amanda Raven, the spa is broken – the spar is broken,
put the tuna in the box – put the tuner in the box. Such homophones are only
possible in the non-rhotic accents of English exhibiting Intrusive-R.
Now we turn to another consonant, /I/. As exemplified by the word talk
mentioned above, the letter <l> is also sometimes not pronounced, but this
time we are only concerned with the sound /I/, that is, when the <l> is
pronounced. In several dialects of English, it has two possible
pronunciations, i.e., two allophones. In certain positions, the /l/ is the same
alveolar lateral as in Hungarian; the traditional name of this sound is clear-L
(or light-L). In other positions, however, it becomes velarized, that is, its
articulation involves the movement of the tongue towards the soft palate
(velum); this is usually called dark-L. Roughly, in RP the /l/ is only
pronounced clear when followed by a vowel, and it becomes dark (the IPA
20
The phonology of English consonants: an introduction
symbol of dark-L is [I]) when a consonant or pause follows. This is the rule
of L-darkening. The following chart shows the details.
dark-L clear-L
bef. a cons. bef. a pause before a vowel before /j/
(a) (b) (c) (d) (e)
spilt
belch
Albert
else
killed
tallness
pill
bell
rebel
stale
kill
tall
lip
look
Linda
lateral
libido
lullaby
slip
splendid
Ashley
colon
killing
taller
value
cellular
million
evaluate
volume
schoolyard
As you can see above, the /l/ is dark when it is followed by a consonant
sound (column (a)) or when final in the word (column (b)). Spelling does not
count, so <l> and <ll> behave identically since they are both pronounced in
the same way: short /l/. Do not let silent letters (e.g., killed, stale) mislead
you. In the three columns on the right, the /l/ is clear because it is prevocalic,
either word-initially (column (c)) or medially (column (d)), or it precedes a /j/
sound (column (e)), which may be part of the complex vowel /ju:/ or its
reduced counterparts /jo, jc/, or may result from the reduction of an
unstressed /i/ (e.g., million /'miIjcn/). If we compare the examples killed –
kill – killing and tallness – tall – taller, we find that word-final /l/ is only dark
when followed by a consonant or a pause, but it becomes clear when a
vowel-initial suffix follows. In schoolyard, the /l/ is at the end of the word
school, but it is followed by a /j/ in the compound, and therefore it is clear.
Within and across sentences, the pronunciation of word-final /l/ is
determined by the following segment in the same way. While it is dark in feel
and feel me, it is clear in feel at home; dark in spell and spell this word but
21
Chapter 2
clear in spell it. Compare kill and kill Bill with kill you, smile and smile back
with smile at me. There is one type of word-final /l/, however, which is
always dark and which no following vowel can "rescue". In words like cycle
['saikI], martial ['mo:jI], or channel ['tjænI], the /l/ is found in a syllable that
lacks a vowel: notice that in the second syllables of these examples, the /l/
constitutes the syllable along with another consonant. Since in such cases the
/l/ is considered to take up the role of the vowel and constitute the syllable, it
is usually referred to as a syllabic /l/, and is transcribed [I ] (syllabicity is
indicated in the IPA by a short vertical line under the main symbol). For some
reason, syllabic /l/ is always dark, even if it is followed by a vowel sound in
the next syllable, so it is also dark in cycling ['saikI in], martial arts
['mo:jI 'o:ts], or Channel Islands ['tjænI 'aiIcnoz].
Let us emphasize at this point that the discussion of L-darkening
refers to RP only. In other accents of English, the distinction between clear
and dark-L may not be present at all (as in GA, where /l/ is usually dark in all
positions), or may have slightly different conditions. In several non-standard
varieties of English, dark-L is often articulated as an [o]-like vowel, e.g., milk
[miok], shelf [jcoI], feel [Ii:o]. This is called L-vocalization, as the
consonant /l/ is replaced by a vowel.
Finally, we take a look at plosives in English, especially voiceless plosives,
since they exhibit a wide range of allophones. Word-initially and before a
stressed vowel, the voiceless plosives (= /p t k/) are followed by a short [h]-
like sound. This phenomenon is referred to as aspiration, and a detailed
description of its articulatory basis is given in Chapter 6. Phonetically, i.e., in
physical reality, there are various degrees of aspiration, with word-initial pre-
stress plosives having the strongest possible aspiration, word-medial pre-
22
The phonology of English consonants: an introduction
stress plosives and unstressed initial plosives having somewhat less, and
other word-internal ones and those in word-final position having even less. In
one case, however, they are definitely unaspirated: after an /s/. This is
illustrated in the chart below. The accents on top of vowel letters in the
example words denote stress (the acute accent <´> means stronger stress than
the grave accent <`> – see Chapter 8 for the degrees of stress in English).
aspirated unaspirated 1 unaspirated 2
——————————————————————————————→
(a) (b) (c) (d) (e) (f)
pát
póker
tén
tíger
kíll
cút
repéat
suppórt
retúrn
detér
índicàte
raccóon
potáto
políce
tomáto
todáy
cajóle
collápse
léopard
clípper
tomáto
vánity
quáker
póker
ráp
gállop
cút
suppórt
póke
láck
spíll
wásp
stóp
stándard
scúll
skín
The columns (a)-(f) give examples of the degrees of aspiration from strong in
(a) to zero in (f). The chart also shows that on the basis of the behaviour, i.e.,
the possible pronunciations, of the plosives, we can identify a phonologically
relevant binary distinction between aspirated and unaspirated. In (a)-(c), only
aspirated plosives are pronounced by speakers of English, and it is just the
degree of the aspiration which distinguishes the subcases. The other columns,
however, differ in that aspiration is either optional ("unaspirated 1" in (d)-(e))
or impossible ("unaspirated 2" in (f)). The optionality of aspiration in (d)-(e)
means that in the positions in question other allophones can also appear.
Besides a plain unaspirated plosive, in GA or informal-colloquial British
English /t/ is frequently realized as a so-called tap (or flap) in the cases in (d)
(i.e., before an unstressed vowel), the IPA symbol of which is [r], and the
23
Chapter 2
process is referred to as tapping (or flapping). (In fact, this is also true for
/d/, which is, of course, not a voiceless plosive but a voiced one, so it does
not undergo aspiration.) Examples include tomato [tºc'mcirco], vanity
['væniri], matter ['mærc(r)], butterfly ['b\rcIIai], nobody ['ncobcri], little
['IirI ]. In (e), that is, word-finally, the voiceless plosives /p t k/ (and also /tj/)
are usually unreleased and/or are accompanied by a short closure of the vocal
cords called the glottal stop (symbolized by [1] in the IPA), and the
phenomenon is accordingly dubbed glottalization. E.g., sleep [sIi1p], match
[mæ1tj], not [no1t]. (See Chapter 6.)
Turning back to aspiration, it actually has two realizations. One is the
short [º]-like sound following the plosive, mentioned above. The other
manifestation of aspiration is the devoicing of a following sonorant
consonant. In pray [pr ci], plug [pI\g], simplicity, attract, queen [kwi:n], cube
[kju:b], liqueur, twist, the underlined sonorant consonants are voiceless. In
/tr/ sequences, the /t/ is aspirated and therefore the /r/ is devoiced, and the
resulting [tr ] sounds very much like a /tj/, as if it was an affricate. Notice how
minimal the difference is between train and chain.
If we take consonant sequences (called clusters) under closer
scrutiny, we make an interesting observation. In words like attractive, betray,
the /t/ is aspirated and thus the /r/ is devoiced. In contrast, in examples such
as Atlantic, Scotland, the /t/ is glottalized and the /l/ is fully voiced. In both
cases the voiceless plosive is followed by a sonorant consonant. Where can
the difference come from? Notice that, crucially, thousands of English words
start with a /tr/ cluster (tray, trip, trombone, etc.), but there are no such
examples with /tl/. This cannot be an accident. In addition, most native
speakers of English would insist that the /t/ in attractive belongs to the
24
The phonology of English consonants: an introduction
second syllable of the word along with the /r/, whereas in Atlantic there is a
syllable division (or syllable boundary, customarily indicated by a dot)
between the /t/ and the /l/. This yields a.ttrac.tive (the double <t> in spelling
stands for a single /t/ sound) but At.lan.tic. In conclusion, /tr/ is a frequent
syllable-initial cluster, while in /tl/ the /t/ is at the end of one syllable and the
/l/ is at the beginning of the next one. This also explains why /tr/ is possible at
the beginning of words (after all, the beginning of the word is, at the same
time, the beginning of the (first) syllable), and why /tl/ never occurs in that
position (the /t/ is at the end of a syllable and not at the beginning). Now, it
appears that the pronunciation of the /t/ (and the other voiceless plosives as
well) depends on its position within the syllable: it is aspirated when syllable-
initial but glottalized when syllable-final. If we (or rather: native speakers of
English) syllabify the example words given above and repeated here for
convenience (only the relevant syllable divisions are indicated), we identify a
syllable boundary to the left of all the "aspirated" cases: in (a) and (c), the
plosive is at the beginning of the word, while in (b) there is a unanimous
agreement among speakers as to the syllable boundary. In (e) and (f), this
clearly does not hold: the plosive is either word- (that is, syllable-) final or
medial but certainly not initial.
aspirated unaspirated 1 unaspirated 2
——————————————————————————————→
(a) (b) (c) (d) (e) (f)
pát
póker
tén
tíger
kíll
cút
re.péat
su.ppórt
re.túrn
de.tér
índi.càte
ra.ccóon
potáto
políce
tomáto
todáy
cajóle
collápse
léo.p.ard
clí.pp.er
tomá.t.o
váni.t.y
quá.k.er
pó.k.er
ráp
gállop
cút
suppórt
póke
láck
spíll
wásp
stóp
stándard
scúll
skín
25
Chapter 2
Column (d) is, however, problematic. When a single consonant is followed
by an unstressed vowel, native intuition fails to make unambiguous
judgments: some speakers would opt for leop.ard, others for leo.pard, yet
others "feel" as if the /p/ belonged to both syllables – i.e., the syllabification
of such consonants is ambivalent. (In the chart, we have indicated this
hesitation by dots assigned to both possible locations.) For this reason,
phonologists often refer to such a situation as ambisyllabicity, and to such
consonants as ambisyllabic. We can conclude, then, that ambisyllabic
voiceless plosives can be plain or weakly aspirated, ambisyllabic /t/ and /d/
may even be tapped. Our findings are summarized in the following table.
Syllabic position Pronunciation
initial strongly aspirated
ambisyllabic weakly aspirated or tapped
final unaspirated glottalized
For the sake of experiment, let us revisit the allophonies discussed earlier,
and investigate into syllable boundaries in the examples. The chart
illustrating R-dropping is repeated presently.
no /r/ pronounced /r/
before a consonant before a pause before a vowel
(a) (b) (c) (d) (e) (f) (g)
York
par.ty
bird
aller.gy
leopard
par.ticular
bears
tired
iron
aren't
feared
retire.ment
fires
rare.ly
your
car
err
refer
teacher
particular
bear
tire
bore
care
lyre
restore
more
centre
ring
routine
rhyme
refer
restore
retirement
rarely
crow
pray
tribute
shrimp
A.frica
poe.try
a.rrive
ti.r.ing
bo.r.ing
e.rr.or
refe.ree
fie.r.y
fu.rr.y
ra.r.est
26
The phonology of English consonants: an introduction
The position of the relevant syllable divisions reveal that R-dropping affects
syllable-final /r/, which can be absolute final (as in, e.g., party in (a), and all
the example words in (c)-(d)) or part of a final cluster (as in, e.g., leopard).
Syllable-initial /r/, whether absolute initial (in (e) and in arrive and referee)
or part of an initial cluster (in (f)), escapes being dropped, and so does
ambisyllabic /r/ (in (g)).
Syllabic position Pronunciation
initial pronounced
ambisyllabic pronounced
final dropped
Notice that initial and ambisyllabic consonants have something in common:
they (can) occupy the beginning of the syllable.
Now here is the chart for L-darkening.
dark-L clear-L
bef. a cons. bef. a pause before a vowel before /j/
(a) (b) (c) (d) (e)
spilt
belch
Al.bert
else
killed
ill.ness
pill
bell
cancel
stale
kill
ill
lip
look
Linda
lateral
libido
lullaby
slip
splendid
A.shley
co.l.on
ki.ll.ing
i.ll.er
va.l.ue
ce.ll.ular
mi.ll.ion
eva.l.uate
vo.l.ume
schoo.l.yard
In the first two columns, the /l/ is always at the end of a syllable: it is either
absolute final (as in, e.g., Albert, illness, and the examples in column (b)) or
part of a final cluster (as in the rest of the words in (a)). Notice that killed
goes under exactly the same rubric as spilt, belch or else: the <e> in the -ed
suffix is silent, so the /l/ is immediately followed by the final consonant /d/.
27
Chapter 2
In the remaining three columns, however, the /l/ is syllable-initial (including
cases when it is part of an initial cluster) or ambisyllabic. In column (e), it is
always ambisyllabic since it may end the syllable to the left, but it may as
well form an initial cluster with the following /j/ – words like lucid /'Iju:sio/,
lucrative /'Iju:krctiv/, ludicrous /'Iju:oikrcs/ exemplify /Ij/, at least in one
possible pronunciation (the other alternative does not contain the /j/ – see
Yod-dropping in Chapter 5). Therefore, we can conclude the following.
Syllabic position Pronunciation
initial clear
ambisyllabic clear
final dark
Careful readers must have noticed that in L-darkening, just like in the
allophony of plosives and in R-dropping, the ambisyllabic situation patterns
with the initial rather than the final position. The only exception is tapping,
which is possible when the /t/ or /d/ is ambisyllabic but impossible when it is
clearly initial. Apparently, /l/ is clear and /r/ is pronounced in at least partial
initial position, whereas for a plosive to be (considerably) aspirated it must
exclusively occupy the beginning of the syllable.
When we look at the "fate" of word-final consonants (e.g., the /r/ in
more, the /l/ in feel, or the /t/ in get) in phrases and sentences, things become
particularly exciting. In isolation or at the end of an utterance (i.e., a stretch
of speech uttered without a pause), they are, obviously, syllable-final. But
how are strings like more exciting, your eyes, feel at home, spell it and get up
syllabified? As we have just stated, the underlined consonants can be
syllable-final, but notice that they can also start the following syllable – after
all, there are a whole lot of English words beginning with /r/, /l/ and /t/.
Consequently, the syllabification of such consonants is not unambiguous –
28
The phonology of English consonants: an introduction
they are ambisyllabic. (This is also reflected by the fact that phrases like a
nice cream and an ice-cream are indistinguishable. In fact, this phenomenon
is not only observable in English but in other languages, e.g., Hungarian, as
well: the native intuition concerning the position of, e.g., the /z/ in az 'the' is
readily illustrated by the title of a Miklós Vámos book, Zenga zének 'the song
resounds', or the Hungarian translation of Heffalump, Zelefánt, in the cartoon
Micimackó és a Zelefánt, Pooh's Heffalump Movie). Therefore, the relevant
syllable boundaries can be located as mo.re. exciting, you.r. eyes, fee.l. at
home, spe.ll. it. (Remember to ignore the silent <e> at the end of more.) This
straightforwardly explains why such /r/'s are pronounced and such /l/'s are
clear. The same is not true, however, of phrases like more beautiful and feel
me, where ambisyllabicity is ruled out, and as a result, such /r/'s are dropped
and such /l/'s are dark. As far as syllabic consonants are concerned, they are
necessarily always syllable-final, since no English syllable can start with one
(that is why no English word starts with one). It follows, then, that syllabic
/l/, as in Channel Islands, is always dark.
In light of the above discussion, you could have guessed by now what
happens to a final /t/ in a phrase like get up: since it is ambisyllabic, it can
only be moderately aspirated, what is more, it must be tapped in the so-called
tapping dialects like GA or informal-colloquial British English. This
prediction is supported by the facts: in GA, for example, the underlined /t/ in
right away, not a joke, get up, at all is usually pronounced as a tap.
In sum, this chapter has shown what major phonological processes
affect the consonants of English, and how these processes are driven by the
position the consonants occupy within the syllable. The next two chapters
introduce the phonology of English vowels, but in Chapter 5 we take up the
discussion of syllable structure again, and discover some of its further
aspects.
29
3. The phonology of English vowels: an introduction
Before you study this chapter, check whether you are familiar with the
following terms: back, central, close, consonant cluster, diphthong, front,
full vowel, half-close, half-open, high, lax, low, manner of articulation,
mid, monophthong, open, place of articulation, rounded, schwa, stress,
suffix, syllable, tense, triphthong, unrounded, vowel shift, weak vowel
In this chapter we take a look at English vowel sounds and their possible
classifications, compare them with the Hungarian vowel system and see what
typical vowel alternations occur in English.
Vowels differ from consonants in two very important ways: they are
articulated without any kind of obstruction in the oral cavity – i.e., the
articulators do not form a complete or partial closure or a narrowed passage
in the way of the exhaled air. On the other hand, vowels differ from
consonants in their behaviour, too: while consonants typically occur in
syllable marginal positions – they appear at the peripheries of the syllable –,
vowels form the very core of the syllable and occur in syllable central
position.
As suggested in Chapter 1, vowel sounds may be classified according
to two types of factors: phonetic and phonological. In the first case,
classification is based on some articulatory characteristics while in the second
it is some aspect of vowel behaviour that serves as the basis for classification.
Let us first examine what phonetic classes may be defined in the
English vowel system. In some vowels the position of the tongue is relatively
stable during articulation; such vowels are called monophthongs. In other
vowels, though, the position that the tongue occupies at the beginning of the
vowel differs significantly from what it occupies at the end of the vowel; i.e.,
The phonology of English vowels: an introduction
some tongue movement is involved. Such vowels are referred to as
diphthongs (and triphthongs). We may also think of this difference as a
difference in how many vowels are found within one syllable: in
monophthongs there is one – e.g., /i, e, o, 5:, o:/ –, in diphthongs there are
two – e.g., /ai, cc, co/ – while in triphthongs there are three – e.g., /aic, aoc/.
Note though that triphthongs are not found in all dialects of English: those
dialects that pronounce all underlying /r/'s – the so-called rhotic dialects, cf.
Chapter 2 – typically lack triphthongs – and even some of the diphthongs as
we will show in Chapter 4.
On the other hand, vowels may be short – e.g. /i, e, o/ – or long – e.g.,
/5:, o:, cc, co, aic/ – depending on their duration: long vowels are
approximately twice as long as short ones. Note that diphthongs and
triphthongs are just as long as long monophthongs. Whenever we refer to
long vowels, we always mean long monophthongs, diphthongs and
triphthongs together. Note that length in English varies depending on the
environment – i.e., length is not a stable property. For more on length
alternations, see Chapter 6. The vowels of RP are the following:
Short vowels Long vowels
i, o, c, o, \, æ o:, i:, u:, 5:, a: ci, ai, 5i, ao, co, ic, cc, oc aic, aoc
Monophthongs Diphthongs Triphthongs

To further demonstrate that length is not a purely phonetic property of
English vowels, we may refer to the controversy of length marking: the
vowel length of monophthongal – or pure – vowels is indicated with a colon.
However, one of the so-called short monophthongs, the vowel /æ/ is just as
long in actual pronunciation as any of the long monophthongs or diphthongs
and it even undergoes the very same shortening process as long vowels do
31
Chapter 3
(see Chapter 6). However, its length is not indicated in transcription with the
colon. The vowel /æ/ is categorized as a short vowel because it behaves like
other short vowels do. The phonetic length of /æ/ may be due to the fact that
during its production the lower jaw and the tongue are in their most open
position, a gesture which might take long enough to cause a perceivable
length difference.
Another important note concerning vowel length is due here: while
most Hungarian short-long vowel pairs consist of vowels of more or less the
same quality with just a length difference (e.g., /y/-/y:/ tüze 'his/her/its fire'
vs. tűz-e 'does he/she/it stitch?', /o/-/o:/ kör 'circle' vs. kőr 'hearts (in cards)',
/i/-/i:/ Sirok (a placename) vs. sírok 'I cry'), English short-long vowel pairs
always involve a quality difference, that is, there is no English short-long
vowel pair in which the qualities of the two vowels are the same. This is also
reflected in the phonetic symbols used to indicate them. Consequently, while
there is a short /i/ and a long /i:/, there is no /i:/; similarly, while there is a
short /o/ and a long /u:/, there is no /o:/. The only exception to this rule is the
vowel pair /c/-/a:/, mentioned in Chapter 1, where the quality of the vowels is
the same. However, in this case it is the full vowel-weak vowel distinction, to
be discussed presently, that justifies the use of the different symbols.
As far as phonological classifications of vowels are concerned, the
two major phonological classes are based on the type of syllable the vowel
appears in. English behaves quite differently from Hungarian as far as
stressed and unstressed syllables are concerned. On the one hand, while it is
always the first syllable of the word that carries the main stress in Hungarian,
it may be the first, second, third, etc. syllable of an English word that carries
primary stress.
1
On the other hand, English unstressed syllables have reduced
1
For the degrees of stress, see Chapter 8.
32
The phonology of English vowels: an introduction
vowels only, in the sense that these vowels are shorter, weaker in energy and
closer to schwa /c/ in place of articulation. Thus, in unstressed syllables only
weak vowels – /c/, /i/ and /o/ – may be found while in stressed syllables we
may only find so-called full vowels – i.e., all the other vowels of English,
also including /i/ and /o/, which, besides occurring in unstressed syllables,
may also function as full vowels.
Within the class of full vowels we may distinguish two subclasses:
tense and lax vowels. One has to be very careful when using these two terms
as they are often used as phonetic labels, too. In a phonetic sense, these terms
refer to the muscle bundles located at the backmost part of the tongue, against
the back wall of the pharynx (throat). Whenever these muscles are tense, the
vowel is tense (in a phonetic sense); when such muscle tenseness is not
present, the vowel is lax (phonetically). However, we will use these terms in
a purely phonological sense, i.e., to refer to a certain kind of vowel
behaviour. (We might just as well call the two types of vowel Type1 and
Type2 was it not for our wish to follow the tradition.) As we will show below,
tense and lax vowels (in a phonological sense) occur in different types of
environment.
Tense Lax
Monophthongs i:, u:, 5:
3
i, e, æ, \, o, o, o:, a:, 5:
1
, 5:
2
Diphthongs and
triphthongs
ai, ci, 5i, ao, co, ic, cc, oc,
aic, aoc
There are a few generalizations to be drawn on the basis of the above table:
all short vowels are lax and all diphthongs and triphthongs are tense while
long monophthongs are divided between the two classes. Non-high long
monophthongs – that is /o:/, /a:/ and /5:/ – are lax, except in the case of /5:/
3
.
33
Chapter 3
The behaviour of /5:/ is twofold: sometimes it is tense, in other cases
it is lax. There are two general types of spellings that indicate a lax /5:/ while
a further set of spellings represents the tense variant.
/5:/
1
/5:/
2
Lax
No <r> in spelling
ball, called, saw, bought,
broad, Shaw, clause, fault
<ar> or <or> in spelling in
word-final position or followed
by a consonant letter
for, horde, morning,
gorgeous, cord, north, war,
dwarf, quarter, horn, sport
/5:/
3
Tense
<or> in spelling followed by a
(pronounced or silent) vowel
letter
before, historian, store,
more, adore, bored, shore
<oar>, <oor>, <our>, <aur> in
spelling
soar, boar, roar, door, floor,
four, pour, aura, Laura

The behaviour and alternations of tense and lax vowels are discussed below
and also in Chapter 4, where we take a look at their behaviour before <r>.
The following table sums up what we have discussed about the manner of
articulation and the behaviour of vowels so far.
Full vowel
lax tense
short long long
Weak
vowel
Monophthong i, e, æ, \, o, o o:, a:, 5: i:, u:, 5: i, o, c
Diphthong - - ai, ci, 5i, ao, co, ic, cc, oc -
Triphthong - - aic, aoc -
Let us now turn to the places of articulation of vowels. Before we actually
discuss these we have to point out that places of articulation are not as
clearcut for vowels as for consonants for the very simple reason that while in
consonants the place of articulation refers to the articulators producing some
34
The phonology of English vowels: an introduction
degree of obstruction, in vowels it is simply inapplicable as they do not
involve any kind of obstruction. Instead of referring to obstruction sites, we
will use three criteria to classify vowels according to horizontal tongue
position, vertical tongue position and lip rounding.
The places of articulation of the monophthongs of RP are the
following:
Front Central Back
unrounded unrounded unrounded rounded
Close /i:/ beat - - /u:/ boot
Half-close /i/ bit
Half-open /e/ bet
/c/ ago
/a:/ burn
- /o/ put
- /5:/ bought
Open /æ/ bat /\/ but /o:/ bar /o/ Bob
We have to note that besides the terms used in the table above, close vowels
are often referred to as high, open vowels as low, while the ones inbetween
as mid. As it can be seen from the above table the following generalizations
may be drawn: front and central vowels are unrounded while back vowels are
rounded, except for /o:/.
The places of articulation of the diphthongs of RP are the
following:
Front Central Back
unrounded unrounded unrounded rounded
Close
Half-close
Half-open ei
co
5i
Open ai ao
/ci/ bay
/ai/ bye
/5i/ boy
/ao/ bound
/co/ boat
35
Chapter 3
Front Central Back
unrounded unrounded unrounded rounded
Close
Half-close ic
Half-open ec
oc
Open
/ic/ beer
/cc/ bear
/oc/ boorish
Diphthongs may be classified according to several factors. On the one hand,
we may distinguish them according to their second component: if it is a
schwa /c/, then we talk about centring diphthongs. In all other diphthongs
the second component is more close than the first, and these are thus called
closing diphthongs; those that end in /i/ are fronting (and closing) while
those ending in /o/ are backing (and closing). On the other hand, closing
diphthongs may be classified according to the articulatory distance between
the two components: the diphthongs /ci/, /co/ are narrow (and closing), while
the rest, /ai/, /ao/, /5i/ are the so-called low-starting or wide diphthongs.
This is summarized in the table below:
Centring Closing
Fronting Backing
Narrow ic, cc, oc ci co
Wide - ai, 5i ao
As shown in the following table, unlike English, Hungarian also has front
rounded vowels. In addition, Hungarian back vowels are all rounded – note
that /a:/ is a central vowel. Attention must also be paid to the fact that the
traditional Hungarian terminology might be misleading: the so-called
"magas (hangrendű)" vowels are actually front – and not high –, while "mély
(hangrendű)" vowels are central or back – and not low –, i.e., magas and
36
The phonology of English vowels: an introduction
mély do not refer to tongue height, but are metaphors for the acoustic effect
made by the vowel.
The places of articulation of Hungarian vowels are the following:
Front Central Back
unrounded rounded unrounded rounded
Close /i:/ hív
/i/ ki
/y:/ tűz
/y/ üt
- /u:/ út
/u/ kulcs
Half-close /e:/ kér /o:/ nő
/o/ kör
- /o:/ tó
/o/ hoz
Half-open /r/ kert - - -
Open - - /a:/ ház /o/ kar
Let us now turn back to phonology and the discussion of tense and lax
vowels. One of the differences between them is in what positions they may
appear in a word. In English, unlike in Hungarian, when a word is suffixed,
often it is the pronunciation of the word stem that changes and not that of the
suffix. One such alternation involves the change of an original tense vowel
into a lax one. The phenomenon is called vowel shift, a historical version of
which – the Great Vowel Shift – applied to English long vowels around the
15th century.
The vowel shift is thus a case of tense-lax alternations. Tense vowels
of word stems become lax in certain environments. The tense-lax vowel pairs
are as follows:
37
Chapter 3
Regular
type
Vowel
letter
Before R
1. /ci/-/æ/ grade-gradual
sane-sanity
vane-vanity
A
compare-comparison
prepare-preparatory
barbarian-barbaric
1. /cc/-/æ/
2. /i:/-/c/ meter-metric
secret-secretary
keep-kept
E
imperial-imperative
severe-severity
hero-heroine
2. /ic/-/c/
3. /ai/-/i/ final-finish
decide-decision
Bible-biblical
I or Y
satire-satirical
tyrant-tyranny
lyre-lyrical
3. /aic/-/i/
4. /co/-/o/ holy-holiday
know-knowledge
sole-solitude
O
historian-historical
explore-exploratory
flora-florist
4. /5:/-/o/
There are two types of word pair: one in which the stressed vowel is followed
by the letter <r> and one in which it is not. As vowels may be influenced by a
following <r> – Pre-R Breaking for tense vowels and Pre-R Broadening for
lax vowels, for details see Chapter 4 – we have to consider pre-R cases
separately. Note that in the examples relevant to the present discussion, Pre-R
Breaking does apply for tense vowels (that is, tense vowels differ according
to what follows them) but Pre-R Broadening does not (that is, the same lax
vowels appear in both the first and the last columns of the table).
It is also clear from the table that some of the tense vowels, namely
/(j)u:/, /5i/, and /ao/ have no lax counterparts and as a result do not participate
in the alternation (they are non-laxable). There are a few untypical pairings
that may occur: /ao/-/\/ pronounce-pronunciation, /5i/-/\/ join-juncture, /u:/-
/\/ do-does, etc. Also, there are a few cases that involve some alternation but
38
The phonology of English vowels: an introduction
it is either not one of the regular vowel pairs above – e.g., /ic/-/æ/ clear-
clarity, /ci/-/c/ break-breakfast – or they involve lax-lax or tense-tense
alternations – /o:/-/c/ example-exemplify, /aic/-/ic/ empire-imperial. Let us
now turn to the environments in which vowel shift may occur.
Probably the most influential such laxing process is Trisyllabic
Laxness, in which a stressed vowel in (at least) the third-last syllable must be
lax – e.g., sane-sanity, grade-gradual, compare-comparison. As we have
noted above, this rule has regular exceptions: the tense vowels /u:/, /ju:/ and
their variants /oc/, /joc/ are regular exceptions, i.e., they freely occur in
trisyllabic environments, e.g., unity, purify, stupefy. Besides these, there are
irregular exceptions, too. In a few cases other tense vowels may also occur in
trisyllabic environments, e.g., nightingale, Abraham, notify, isolate. What
makes this rule problematic is that there is a great number of exceptions, both
regular and irregular. Also, the rule is sensitive to the morphological structure
of the word: it applies if certain suffixes are attached to the stem but not if
others are added. That is, it seems that the syllables of certain suffixes are
counted when we count the three syllables from the end of the word while
others are not. Whether to count the syllables of the suffix or not depends on
whether the suffix is a regular, productive suffix, which can be added to
almost all members of a category (noun, verb, adjective, etc.) to produce a
large number of words, or a non-productive suffix which is only added to
certain stems of a class and therefore has fewer examples. Some of the
typical examples of the two suffix classes are shown in the following table.
39
Chapter 3
Productive suffixes
not counted in Trisyllabic Laxness
Non-productive suffixes
counted in Trisyllabic Laxness
-ness lazy-laziness, tidy-tidiness -ity grave-gravity, sane-sanity
-ly total-totally, lazy-lazily -al crime-criminal
-ary/-ery/-ory advise-advisory -ative provoke-provocative
-ing pilot-piloting -ible divide-divisible, eat-edible
We must mention here that Trisyllabic Laxness is not just an active
phonological rule that applies to certain roots if they are followed by certain
suffixes but also a so-called morpheme structure condition, a passive
constraint that requires that a stressed vowel which is in at least the third
syllable from the end of the word must be lax – even if no suffix is added to
it.
/æ/ /c/ /i/ /o/
animal, stamina,
fantasy,
cannibal, janitor,
character
penetrate,
separate,
demonstrate,
several, decorate,
intelligent
irritate, miracle,
similar, limerick,
stimulate,
frivolous
opera, positive,
sonorant,
homonym,
dominate, oracle
/\/ /a:/ /o:/ /5:/
company,
gullible,
succulent,
Gulliver
terminal,
permanent,
pertinent,
courtesy
participle,
harmony,
carnival,
parsimony
orthodox,
auditor, audible,
autism
Short /o/ is missing from the charts above simply because it is so rare in
present-day standard English that it is almost impossible to find relevant
examples, e.g., bulletin.
40
The phonology of English vowels: an introduction
Another laxing process applies if a so-called laxing ending is added
to the word stem: a stressed syllable followed by one of the laxing endings
must be lax. Examples include monosyllabic suffixes typically spelled with
<i> or <e>: e.g., -ic, -ish (n/v), -id, -it, -et, -el, as in metre-metric, final-finish,
satire-satirical, etc. As indicated in brackets, -ish (n/v) is a laxing ending
only if the word ending in -ish is a noun or a verb. However, if it is an
adjective, the ending is non-laxing and the preceding stressed vowel may
remain tense. Compare the sample words finish (n/v), vanish (v) and greenish
(adj), Swedish (adj). The first two examples are nouns and verbs and thus the
stressed vowel must be lax as opposed to the other two examples which are
both adjectives and, as a result, the suffix does not influence the
pronunciation of the stressed vowel.
Just as in the case of trisyllabic laxness, there are exceptions to this
laxing rule, too. On the one hand, the vowels /u:/, /ju:/ and their variants /oc/,
/joc/ are regular exceptions; stressed /u:/ and /ju:/ vowels are not affected by
this laxing process: cube-cubic, stupe-stupid, Cupid. There are irregular
exceptions as well; some roots resist laxing, e.g., base-basic. It is important
that this rule does not only apply if the endings are separate morphemes
attached to a root but also if they are just part of the root. For instance, the
ending -ic causes the laxness of the stressed vowels in the names Eric,
Patrick although the very same stressed vowels would be tense where they
are followed by some other kind of ending, e.g., era ['icrc], patron ['pcitrcn].
It is clear then that this rule is not just an active phonological rule but also a
letter-to-sound rule that determines how letters must be pronounced
depending on the environment.
The third relevant laxing rule is triggered by the presence of a
consonant cluster – a sequence of at least two consonants – immediately after
the stressed vowel, and thus a stressed vowel followed by at least two
41
Chapter 3
consonants must be lax: e.g., intervene- intervention, receive-reception, etc.
This regularity is sometimes dubbed Pre-cluster Laxness.
There are two more rules that may cause the laxness of a vowel but
they are clearly not phonological rules but letter-to-sound rules, that is they
tell us how to pronounce vowel letters in certain environments in spelling.
For this reason we just mention them here very briefly and they will be
discussed in detail in Chapters 11-12, where we discuss letter-to-sound rules
exhaustively.
The first such rule is Laxing by free U, which requires that if the
stressed syllable is followed by a free U – roughly, a letter U followed by a
vowel letter (as in venue, statue) – then the stressed vowel must be
pronounced lax, e.g., grade-gradual, rite-ritual, etc. Just like in all other
laxing rules, the vowels /(j)u:/ and /(j)oc/ are regular exceptions, they stay
tense before a free U, e.g., use-usual.
The other letter-to-sound rule causing laxness is the so-called CiV
Laxing rule, which forces a stressed vowel letter – spelled with <i> or <y> –
to be pronounced lax /i/ when followed by a consonant letter, another letter
<i> and one more vowel letter. That is, the stressed vowel letter <i/y> is
followed by the CiV configuration in spelling, hence the name of the rule,
e.g., decide-decision, revise-revision, idiot, familiar, Syria, etc. It is important
to note that all the other vowel letters undergo CiV Tensing in the same
environment, i.e., other vowel letters must be pronounced with a tense vowel,
e.g., manic-mania, Albania, Celia, Gloria, senior, radio, etc. Interestingly,
this rule is able to block the application of the laxing rules. In all the sample
words above the stressed vowel is the third-last vowel from the end of the
word, still, Trisyllabic Laxness does not apply and make them lax. The reason
for this is that CiV tensing is more powerful and robust than the laxing rules,
and thus it can override their effect. Of course, there are exceptions to the
42
The phonology of English vowels: an introduction
CiV tensing rule as well, in which the stressed vowel is lax even though it is
followed by CiV, e.g., national, special, Italian, Daniel, etc.
Besides CiV Tenseness, there is another regularity in English which
requires a vowel in a certain position to be tense. It is called Prevocalic
Tenseness, as its effect is to ensure that all stressed vowels preceding other
vowels are tense. The situation when two separate vowels (the centres of their
respective syllables) are adjacent is generally referred to as hiatus, e.g., the
underlined portions of Noam, Leo, hiatus. Prevocalic Tenseness does nothing
but describe the observation that in English the first member of a hiatus, if
stressed, is always tense, namely /co i: ai/ in the examples above. Note that it
does not apply to unstressed vowels, e.g., react, which are of course reduced.
In addition, it is only relevant to pronunciation: compare Leo to mean or
people – the underlined vowel letter is pronounced as a separate sound in Leo
only, in the others it combines with the following vowel letter to represent a
single sound. That is, in mean and people there is no hiatus, and consequently
Prevocalic Tenseness is not applicable.
Similarly to CiV Tenseness, Prevocalic Tenseness is also stronger than
the laxness rules: in variety, for instance, either Trisyllabic Laxness or
Prevocalic Tenseness could in principle take effect, but it is the latter that
"wins".
In this chapter we saw that, although sometimes English chooses a
tense vowel systematically, in many situations tense vowels are replaced by
their lax counterparts. There are numerous examples where the originally
tense stressed vowel becomes lax although none of the above environments
may be blamed for the change. In such cases we may only say that these are
unexplained, idiosyncratic cases of vowel shift, the surviving effects of older
rules which are no longer active in the language, e.g., read (present)-read
(past), life-live (v), shade-shadow, mead-meadow.
43
Chapter 3
Although the above discussion of the phonetic and phonological
classification of vowels concentrated on RP, most of it is valid in the case of
GA, too. The tense-lax distinction applies to GA in the same way, together
with the tenseness and laxness rules, with just a handful of examples where
the two dialects diverge, e.g., apricot, pronounced (irregularly) with /ci/ in
RP but very often (conforming to Trisyllabic Laxness) with /æ/ in GA. There
are only a few minor differences in the vowel inventories, e.g., recall from
Chapter 1 that all RP /co/'s correspond to /oo/ in GA. Some of these also
affect the classification of vowels, e.g., RP /o/ in lot, odd, wash is usually
long and unrounded /o:/ in GA, still, the vowel behaves as lax in the same
way in the two accents: cf. tone – tonic RP /tcon/ – /'tonik/, GA /toon/ –
/'to:nik/, etc. As we will see in the next chapter, the rest of the dialectal
deviations are caused by the differing distribution of /r/.
44
4. R-influence on vowels
Before you study this chapter, check whether you are familiar with the
following terms: allophone, centring diphthong, complementary
distribution, diphthong, distribution, foreignism, fricative, full vowel, GA,
hiatus, homophone, Intrusive-R, labial, lax, letter-to-sound rule, Linking-R,
low-starting diphthong, minimal pair, monophthong, morpheme, nasal,
non-productive suffix, non-rhotic accent, phoneme, productive suffix,
rhotic accent, R-dropping, RP, tense, triphthong
This chapter mainly focuses on the behaviour of full vowels before an /r/, the
phonological and letter-to-sound rules related to this behaviour and some
further phenomena concerning vowels. As it is demonstrated in Chapter 2 the
two main accent types of English, rhotic and non-rhotic accents, are most
easily distinguished by whether an /r/ is pronounced in all positions or not. In
General American, a rhotic accent, all /r/'s are pronounced while in Received
Pronunciation, a non-rhotic variant, only prevocalic ones are. Besides this,
these – and other – dialects may also be distinguished by the behaviour of
stressed vowels before an /r/, briefly mentioned in the previous chapter.
To remind the reader of the most important vowel classes that will be
referred to we repeat one of the tables from Chapter 3 for convenience.
Tense Lax
Monophthongs i:, u:, 5:
3
i, e, æ, \, o, o, o:, a:, 5:
1
, 5:
2
Diphthongs and
triphthongs
ai, ci, 5i, ao, co, ic, cc, oc,
aic, aoc
Chapter 4
Recall that we have come up with a few generalizations in Chapter 3, namely
that all short vowels are lax, all diphthongs and triphthongs are tense, non-
high long monophthongs are lax, except for /5:/, which behaves in an
ambiguous way: sometimes it is tense, in other cases it is lax. For the details
of this controversy, see Chapter 3.
Let us first consider the behaviour of tense vowels and the rule called
Pre-R Breaking. Tense vowels may be further classified into two subgroups
on the basis of their distribution, i.e., the environments in which they may
occur.
Non-low-starting Low-starting
Plain-Tense i: (j)u: ci co ai ao 5i
Broken-Tense ic (j)oc cc 5:
3
aic aoc 5ic
The rule of Pre-R Breaking seems to be a very simple allophonic rule at first
sight: the members of the Plain-Tense – Broken-Tense vowel pairs appear to
occur in complementary distribution: Broken-Tense vowels only appear
before r within the same word while Plain-Tense vowels occur everywhere
else but never before r within the word.
46
R-influence on vowels
Plain-Tense Broken-Tense Plain-Tense Broken-Tense
bead [i:]
tea [i:]
cohesion [i:]
cute [ju:]
futile [ju:]
unity [ju:]
baby [ci]
staple [ci]
Rumanian [ci]
stone [co]
cloakroom [co]
broken [co]
beard [ic]
tear (n) [ic]
adherence [ic]
curious [joc]
furious [joc]
Europe [joc]
bare [cc]
staring [cc]
Hungarian [cc]
story [5:]
3
roaring [5:]
3
glorious [5:]
3
fight [ai]
tonight [ai]
pine [ai]
pint [ai]
town [ao]
cloud [ao]
Downing [ao]
moist [5i]
fire [aic]
admire [aic]
pirate [aic] or [ai]
iron
1
[aic]
hour [aoc]
flour [aoc]
dowry [aoc] or [ao]
Moira [5ic] or [5i]
Since the members of the pairs are in complementary distribution and are
phonetically quite similar to each other, we may just as well assume that they
are variants, allophones of the same phoneme.
/i:/ /u:/ /ci/ /co/ /ai/ /ao/ /5i/
[i:] [ic] [u:] [oc] [ci] [cc] [co] [5:] [ai] [aic] [ao] [aoc] [5i] [5ic]
1
Be careful with the word iron since its second vowel letter, <o> is silent, and the
pronunciation of the <r> is determined accordingly: dropped in RP /'aicn/, but not in GA
/'ai(c)rn/.
47
Chapter 4
The vowel phonemes in the upper row of the chart on p.46 are divided into
two major classes: the last three, the so-called low-starting diphthongs - /ai/,
/ao/, /5i/, appearing in lighter shaded cells in the table, and the rest of the
vowels - /i:/, /u:/, /ci/, /co/. The differences between these two groups are
twofold: on the one hand, in non-low-starting tense vowels the broken tense
variant is typically a centring diphthong except for /5:/, in the darker shaded
cell in the table. This tense /5:/
3
variant historically derives from the centring
diphthong */5c/ now always pronounced as /5:/. Also, in these four vowel
phonemes the second half of the vowel is changed into /c/, if we think of a
long monophthong as consisting of two identical short components (as
opposed to diphthongs whose two components are different). In low-starting
diphthongs the broken tense variant contains an extra element, /c/, that is, it is
always a triphthong.
Also, there is a difference between the nature of Breaking in the two
vowel groups. While in non-low-starting tense vowels it is always obligatory,
that is, whenever a tense vowel from this group is followed by an r in the
same word it is always replaced by its Broken-Tense counterpart, in low-
starting diphthongs it is not always so: in low-starting diphthongs Breaking is
only obligatory if the r is at the end of a word or followed by a productive
suffix (cf. Chapter 3). For instance, in words like fire /Iaic(r)/ the r is word-
final; in fired /Iaico/ and firing /'Iaicrin/ it is followed by a productive suffix
(-ed and -ing) and as a result the stressed vowel always has to be realized by
a Broken-Tense vowel, [aic]. On the other hand, if the low-starting diphthong
is followed by an r which is morpheme-internal or followed by a non-
productive suffix, then Breaking is optional, and the vowel may be Plain or
Broken-Tense, e.g., pirate /'paicrct/ or /'pairct/, biro /'baicrco/ or /'bairco/.
48
R-influence on vowels
A process that is closely related to Pre-R Breaking is the
simplification of Broken-Tense vowels in fast casual speech, Smoothing, and
its extreme form, the complete monophthongization of diphthongs or
triphthongs. Smoothing influences the triphthongs resulting from the above-
mentioned mechanism of obligatory or optional Breaking of low-starting
diphthongs. Typically the middle component, [i] or [o], of the triphthong is
dropped in casual speech; in faster speech even the last component, schwa [c]
may be dropped: this process is known as monophthongization. To make up
for the loss of the second and third components of the triphthong, the first
part is lengthened, a process often referred to as compensatory lengthening
(for more detail, see below).
Triphthong Middle component
dropped
Last component dropped +
first component lengthened
fire ['Iaic(r)]
tired ['taico]
hours ['aocz]
dowry ['oaocri]
['Iac(r)]
['taco]
['acz]
['oacri]
['Ia:(r)]
['ta:d]
['a:z]
['oa:ri]
It is interesting to note that in many dialects of English, for instance in
Southern dialects of American English, the low-starting diphthongs /ai/ and
/ao/ may also be simplified, i.e., replaced by a long monophthong, in a non-
pre-R environment:
49
Chapter 4
Diphthong Second component dropped
why [wai]
I'm [aim]
wow [wao]
about [c'baot]
[wa:]
[a:m]
[wa:]
[c'ba:t]
Another similar process by which certain diphthongs become simplified
concerns the pronunciation of the diphthongs /(j)oc/ and /cc/. The tendency
especially in the speech of younger speakers of RP is to pronounce /(j)oc/ as
/(j)5:/ and /cc/ as a long half-open unrounded front /c:/. In some dialects, like
Australian English for instance, /ci/ is also being replaced by [c:], that is, the
tendency does not only influence the Broken-Tense but also the Plain-Tense
variant of the vowel.
/(j)oc/ [5:] /cc/ [c:]
poor [pº5:(r)]
purify ['pj 5:riIai]
Europe ['j5:rcp]
rural ['r5:rI]
tourist ['tº5:rist]
bureau ['bj5:rco]
stairs [stc:z]
parent ['pºc:rcnt]
hairy ['hc:ri]
repair [ri'pºc:(r)]
Hungarian [h\n'gc:ricn]
fairness ['Ic:nis]
Note, however, that this monophthongization only affects those /(j)oc/'s
which are the result of Breaking; the same sequence arising from hiatus, as in
fuel or ritual, is left uninfluenced.
50
R-influence on vowels
Pre-R Breaking, then, is one of the most salient allophonic rules
affecting RP vowels. Some might argue, on the basis of minimal pairs like
bee /bi:/ vs. beer /bic/, bead /bi:o/ vs. beard /bico/, that the plain and broken
vowels are independent phonemes, at least in non-rhotic accents like RP.
However, notice that the spelling of Broken-Tense vowels always involves an
<r> (cf. beer, beard), which means two things. On the one hand, Pre-R
Breaking also qualifies as a letter-to-sound rule: whenever a tense vowel is
followed by the letter <r> within the word, it is broken. On the other hand, it
is possible to analyse all Broken-Tense vowels as the outputs of R-influence,
in such a way that the trigger itself (the /r/) is subsequently deleted if the
conditions of R-dropping are met. All in all, the status of Pre-R Breaking in
English phonology is not straightforward, therefore we will simply follow the
traditional practice of indicating Plain-Tense and Broken-Tense vowels
separately in phonological transcriptions, that is, beer /bic/, beard /bico/, etc.
As regards GA, the lack of the rule of R-dropping results in the
absence of apparent minimal pairs like bee and beer, GA /bi:/ and /bi(c)r/,
respectively. It is also shown in the transcriptions that consequently, Pre-R
Breaking is never obligatory in GA, not even in the case of non-low-starting
tense vowels (except for the /oo/-/5:/ pair, which behaves in the same way as
in RP, cf. stone – story GA /stoon/ – /'st5:ri/), and it practically never occurs
before a syllable-initial /r/ (e.g., hairy /'hcri/). As a further result, descriptions
of GA do not normally consider Pre-R Breaking as either a phonological rule
or a letter-to-sound regularity – the occasional appearance of the schwa is
usually taken to be the result of an optional schwa-insertion rule taking place
before syllable-final /r/. A consequence of this is that the GA inventory of
diphthongs is much smaller than that of RP (no centring diphthongs) and
triphthongs are missing altogether. It also follows that smoothing and
51
Chapter 4
monophthongization are not as extensive in GA: fire is always /Iai(c)r/, sure
and poor are usually /jo(c)r/ and /po(c)r/, respectively, and in stairs and
hairy the monophthong is automatically created if the schwa is not inserted
(cf. /stc(c)rz/, /'hcri/).
Let us now turn our attention to the other major group of full vowels, and
their behaviour before r. Lax vowels may also be divided into two major
groups: Plain-Lax vowels and Broad-Lax vowels. Short (lax) vowels all
belong to the former group while the three long lax vowels all fall into the
latter as indicated in the following table:
Plain-Lax æ o c i \ o
Broad-Lax o: 5:
2
a:
The rule of Pre-R Broadening seems to be very similar to Pre-R Breaking as
Broad-Lax vowels will replace their Plain Lax counterparts before r. Note,
however, that this is not so as it will be clear from the discussion below.
Instead, it will turn out that Pre-R Broadening is a practical rule concerning
the relationship between the spelling and pronunciation of vowel letters
before r. Also, from the table above it is obvious that four of the Plain Lax
vowels, namely /c, i, \, o/ share a Broad Lax counterpart, /a:/, which also
makes Pre-R Broadening different from Pre-R Breaking as in the latter all
Plain-Tense vowels had a Broken-Tense counterpart of their own.
On the other hand, Pre-R Broadening, unlike Pre-R Breaking, cannot
be considered an allophonic rule as the distribution of Plain-Lax and Broad-
Lax vowels is not complementary, i.e., the two types of vowel do appear in
the same environment – with certain limitations. Since these Plain-Lax –
Broad-Lax vowel pairs do not occur in complementary distribution, the
52
R-influence on vowels
sample word pairs have been set up on the basis of spelling: the Plain-Lax –
Broad-Lax vowel pairs are represented by the very same vowel letter in the
pairs.
Plain-Lax Broad-Lax Plain-Lax Broad-Lax
cat [æ]
fan [æ]
bad [æ]
fond [o]
bond [o]
clock [o]
stem [c]
send [c]
head [c]
car [o:]
far [o:]
bar [o:]
for [5:]
2
abort [5:]
2
lord [5:]
2
stern [a:]
serve [a:]
heard [a:]
fit [i]
bingo [i]
stick [i]
hut [\]
cutlery [\]
spun [\]
put [o]
bush [o]
buffet [o]
firm [a:]
bird [a:]
stir [a:]
hurt [a:]
curl [a:]
spur [a:]
purr [a:]
burst [a:]
burp [a:]
Having taken a look at the examples containing a Broad-Lax vowel, we may
notice that although there is always an r in spelling in these words, it is not
pronounced in non-rhotic accents like RP, either because it is followed by a
consonant – e.g., abort, stern, bird, burp – or it is word-final and is followed
by a pause – e.g., car, for, stir, purr.
Thus we can conclude that Pre-R Broadening does apply if the r after
the Lax vowel is silent, i.e., it is dropped because of the R-Dropping rule
(Chapter 2). As a result of this one might easily find a very attractive
explanation for the lengthening component of broadening: since the r is
dropped in these environments, its now empty position becomes available for
the vowel before it. That is, the vowel lengthens to make up for the loss of
53
Chapter 4
the r in the word – the kind of process referred to above as compensatory
lengthening.
X X X X → X X X X X X X X → X X X X
             
s t o r → s t o: h o r o → h o: o
The diagrams above demonstrate compensatory lengthening: the X's stand for
timing units. If a sound segment is linked to one timing unit, it is short while
if it is linked to two, it takes twice the time to pronounce, i.e., it is long. Both
star and hard originally contain four short segments. When the r in the words
is dropped, its "place" is preserved, and the preceding lax vowel lengthens by
becoming linked to this empty timing unit, as the broken lines indicate. The
process is very similar to what frequently happens in certain non-standard
varieties of Hungarian, where the other liquid, l, can be dropped before a
consonant.
X X X X → X X X X X X X X → X X X X
             
b o I t → b o: t z o I o → z o: o
bolt 'shop' zöld 'green'
In such Hungarian examples the l is deleted but its timing unit is retained,
and as a result the preceding vowel is lengthened. Notice that the effect of
compensatory lengthening is very similar to the so-called law of mass
preservation in the physical world: we have the same amount of material –
that is, the same number of timing units – on both sides of the equation.
Pre-R Broadening, then, can be accounted for with reference to
compensatory lengthening. Nevertheless, note that however attractive this
54
R-influence on vowels
explanation may be, it cannot be true in all cases. While there is evidence that
there is an /r/ phoneme in words like star as the word final /r/ is often realized
as a Linking-R (see Chapter 2), words like hard pose a problem for babies
aquiring a non-rhotic dialect like RP. They will always hear such words
pronounced without r as there is no environment in which the r of hard
would be present in actual pronunciation, and as a result they will have to
assume that these words do not contain an /r/ phoneme. However, if there is
no /r/, then it is actually not dropped and thus the vowel is not lengthened
because the r has been dropped.
While it seems that we have to give up our idea of compensatory
lengthening as a motivation for Pre-R Broadening, in many cases it can be
shown to be a component of this rule, and it still is a useful kind of
explanation when teaching pronunciation. All the more so as a major
difference between RP and GA can only be accounted for if we separate
Broadening proper (influencing the quality of the target vowel) and
compensatory lengthening (responsible for vowel quantity). As GA is a rhotic
accent, no r's are dropped; consequently, compensatory lengthening is
impossible. Therefore in GA we find the same vowels in car, lord, stern,
firm, hurt as in RP, only they are short: /kor/, /I5ro/, /starn/, /Iarm/, /hart/ –
Broadening, but not compensatory lengthening, has taken place.
The next question that we turn to is what happens if the r following
the lax vowel is realized in pronunciation. Let us take a look at some sample
words containing such a sequence:
55
Chapter 4
[æ] [o] [e] [i] [\]
barrier
carrot
chariot
Harry
marriage
narrow
wheelbarrow
borrow
corridor
Morris
sorrow
sorry
tomorrow
torrent
berry
bury
Jerry
merit
serendipity
terrible
terror
irritate
lyrical
miracle
mirror
pirouette
pyramid
spirit
burrow
courage
current
curry
furrier (n)
furrow
hurry
There are two possible conclusions that can be drawn on the basis of the data
in the table: on the one hand, it does not contain sample words containing an
[o] before a pronounced [r] as this vowel does not regularly appear in such a
position with a few exceptions like courier /'koric(r)/; on the other hand, it is
also clear that Pre-R Broadening does not apply in any of the above words.
Thus we may claim that Pre-R Broadening only applies if the r is dropped,
i.e., the r is syllable-final (cf. Chapter 2). As the absence of Broadening is
typical in words like carrot, where the r is followed by a pronounced vowel,
this regular absence of Broadening is usually referred to as the Carrot-Rule.
The Carrot-Rule is often indicated in spelling by the doubling of the r, e.g.,
marriage, borrow, Jerry, mirror, curry, although it does not always happen,
e.g., bury, miracle, courage.
We have seen so far that normally it is Broad-Lax vowels that occur
before an r. If the /r/ is not silent, however, the lax vowel before it will be a
Plain-Lax vowel as the Carrot-Rule will block the application of Broadening,
i.e., it results in a group of regular exceptions. The Carrot-Rule itself is not
without exceptions, either. In GA, for instance, although in most cases it
applies in the same way as in RP, a few irregular words are exempt from it –
that is, Broadening does take place even though the following /r/ is not
56
R-influence on vowels
syllable-final. E.g., courage, currency, current, curry, hurry, Murray,
occurrence, turret, worry, all with /\/ in RP but /a/ in GA, and squirrel RP /i/
vs. GA /a/.
In certain cases when the r is pronounced, i.e., the Carrot-Rule should
block Broadening, it does not do so and as a result Broadening will apply
resulting in a Broad-Lax vowel before a pronounced r. This is the case when
the r is followed by a vowel-initial productive suffix (again! – cf. the
discussion of Pre-R Breaking above) or a vowel-initial word. Non-productive
suffixes, on the other hand, behave as if they were not separate morphemes
and the word was morphologically simple. For example, Broadening affects
both occur (with a syllable-final /r/) and occurring (with productive -ing), in
contrast to occurrence (with non-productive -ence), which exhibits the same
pattern as, say, current.
2
syllable-final followed by non-productive suffix
occur
bar
err
clergy
/c'ka:(r)/
/bo:(r)/
/a:(r)/
/'kIa:o¸i/
occurr+ence
barr+en
err+or
cler+ical
/c'k\rcns/
/'bærcn/
/'crc(r)/
/'kIcrikI/
2
Note, however, the differences between RP and GA, discussed above, concerning words
like occurrence and current.
57
Chapter 4
syllable-final followed by productive suffix
occur
blur
refer
fur
bar
star
/c'ka:(r)/
/bla:(r)/
/ri'Ia:(r)/
/Ia:(r)/
/bo:(r)/
/sto:(r)/
occurring
blurring
referring
furry
barring
starring
/c'ka:rin/
/'bIa:rin/
/ri'Ia:rin/
/'Ia:ri/
/'bo:rin/
/'sto:rin/
It is clear from the tables above that if a non-productive suffix follows – e.g.,
-ence, -ical –, then the Carrot-Rule will block the application of Broadening
as expected. However, if the r precedes a productive suffix – e.g., -ing, -y –,
then the Carrot-Rule will not be able to block Broadening, which will hence
normally apply to the vowel making it Broad-Lax.
Thus it seems that one half of the original suggestion concerning
Broadening has already been borne out: before an r lax vowels are not always
Broad-Lax as in some cases, as a result of the Carrot-Rule, they will remain
Plain-Lax. Let us now take a look at the other half of the story and see some
examples in which Broad-Lax vowels appear in environments other than
before r, i.e., cases of Broadness without r.
3
3
Most of these examples are repeated in Chapter 12 as groups of deviating words.
58
R-influence on vowels
/o:/
1. Foreignisms imitating the original Greek, French or Italian pronunciation
(so called DRAMA-words): bourgeois /'boc¸wo:/, bra /bro:/, drama /'oro:mc/,
gratis /'gro:tis/, Shah /jo:/, sonata /sc'no:tc/, spa /spo:/.
2. The vowel letter <a> followed by a voiceless fricative or a
nasal+consonant cluster (so called ASK-words): ask /o:sk/, aunt /o:nt/, bath
/bo:0/, brass /bro:s/, can’t /ko:nt/, class /kIo:s/, dance /oo:ns/, last /Io:st/,
laugh /Io:I/, pass /po:s/, path /po:0/, task /to:sk/.
4
3. The vowel letter <a> followed by a silent <l>+labial consonant cluster (so
called CALM-words): almond /'o:mcno/, balm /bo:m/, calm /ko:m/, palm
/po:m/, calf /ko:I/, half /ho:I/, halve /ho:v/, psalm /so:m/.
5
4. Irregular cases: father /'Io:ðc(r)/, lather /'Io:ðc(r)/, rather /'ro:ðc(r)/.
6
/5:/
1. The vowel letter a followed by a pronounced l+consonant or nothing or by
silent l+k (so called CALL-words): bald /b5:Io/, ball /b5:I/, calling /'k5:Iin/,
fallen /'I5:Icn/, stalk /st5:k/, talk /t5:k/, tall /t5:I/, walk /w5:k/, wall /w5:I/.
2. -ough or -augh (so-called THOUGHT-words): bought /b5:t/, caught /k5:t/,
fought /I5:t/, sought /s5:t/, thought /05:t/.
3. -au or -aw word-finally, before a voiceless consonant or a nasal (so-called
SAUCE-words): author /'5:0c(r)/, claw /kI5:/, dawn /o5:n/, law /I5:/, lawn /I5:n/,
raw /r5:/, sauce /s5:s/, saw /s5:/.
4. Irregular cases: abroad /c'br5:o/, broad /br5:o/, water /'w5:tc(r)/.
4
Recall from Chapter 1 that all these words are pronounced with /æ/ in GA.
5
Some of the examples, e.g., calf, half, halve have /æ/ in GA. Note irregular salmon /'sæmcn
/, too.
6
In GA, lather and rather contain /æ/.
59
Chapter 4
/a:/
1. Only in one word: colonel /'ka:nI/.
7
These examples of Broadness without r illustrate that, on the one hand,
Broadening is not always predictable (sometimes it takes place without a
potential trigger being present in the word), and on the other hand, numerous
pairs of homophones exist, although in non-rhotic English only, with an r in
one member but with no r in the other. For example, words like roar and raw,
pore and paw, spar and spa, baa 'make the bleat of a sheep' and bar are
totally indistinguishable for a non-rhotic speaker – a fact which contributes to
the emergence of the so-called Intrusive-R, mentioned in Chapter 2 but
treated in detail in Chapter 7.
7
Although this word is generally considered to be an exception, that is, one with a Broad
vowel without an /r/, the corresponding rhotic pronunciation, /'karnI/, shows that in fact the
<l> after the stressed vowel represents an /r/, and the second <o> is silent. Consequently, the
/r/ is in syllable-final position, and dropped in non-rhotic accents like RP. This means that
this word actually falls under the same rubric as, say, kernel.
60
5. The English syllable
Before you study this chapter, check whether you are familiar with the
following terms: coronal, distribution, fricative, glide, homophone, liquid,
monosyllabic, morpheme, nasal, obstruent, plosive, rhotic accent, suffix,
velar
In this chapter, we take a closer look at the structure of English syllables. In
Chapter 2 it was demonstrated that the syllable plays a significant role in
defining what positions host the targets of phonological processes like
aspiration or R-dropping. However, this is not the only way it affects the
patterning of speech sounds; as it is shown below, the syllable is one of the
major factors determining the restrictions on sound sequences.
You may have already noticed that in languages in general only a very
small portion of theoretically possible sound sequences is used as actual
words. On the one hand, there are always thousands of combinations whose
absence cannot be accounted for: they are potential words but have no
meaning. Such "nonsense words" are sometimes referred to as accidental
gaps since they are gaps (that is, missing items) in the vocabulary by accident
only and may gain some meaning later on. As an example, let us cite the first
stanza of Lewis Carroll's Jabberwocky, a nonsense poem in his book entitled
Through the Looking Glass, together with the Hungarian translation (by
István Tótfalusi).
Chapter 5
JABBERWOCKY A GRUFFACSÓR
'Twas brillig, and the slithy toves
Did gyre and gimble in the wabe:
All mimsy were the borogoves,
And the mome raths outgrabe.
Nézsonra járt, nyalkás brigyók
turboltak, purrtak a zepén,
nyamlongott mind a pirityók,
bröftyent a mamsi plény.
If you do not understand the italicized words (that is, virtually the whole
text), do not panic – however well-formed English words brillig and mimsy
and gimble look, they are nonexistent, just like the quasi-Hungarian words
nézson and brigyó and plény. What is crucial is the fact that there is no
principled reason for their nonexistence; they really sound like English and
Hungarian words, therefore it would not be impossible to imagine them as,
say, dialectal forms of existing words. In fact, certain accidental gaps do
become part of the language with time, e.g., a nonsense trademark can start a
life of its own, as it happened in the case of spam (once a trademark for a
canned meat product, it appeared in a skit on the British television series
Monty Python's Flying Circus; now it is generally accepted as a term to refer
to unsolicited, usually commercial e-mail sent to a large number of addresses,
and it is even used as a verb).
Not all gaps are accidental, though. In a great many cases, a sound
sequence is not a potential word as it contains some combination which is
systematically rejected by the language. For example, while brillig and plény
are acceptable as words of English and Hungarian, respectively, neither
rbillig nor lpény would be, although they contain exactly the same segments.
No English or Hungarian words start with /rb/ or /Ip/, and it is completely
unlikely that some ever will, not even as trademarks or internet terminology.
Notice, however, that word-finally you observe just the opposite: /rb/ or /Ip/
is possible (cf. kerb /karb/ in the rhotic accents of English, or Hungarian talp
62 62
The English syllable
'sole') but /br/ or /pI/ is not. In sum, both languages seem to impose strict
restrictions on what sounds can appear in what order in what position. These
restrictions are called phonotactics in phonology.
In what follows, we discover the major phonotactic restrictions in
English. The chart below illustrates some of the most frequent two-member
combinations of sounds on either edge of English monosyllabic morphemes
(O = obstruent, N = nasal, L = liquid, G = glide, V = vowel, F = fricative, P =
plosive). As we will see below, all single consonants except for /n/ can start
such a morpheme (e.g., pit, heart, lie), so there are almost as many
consonant+vowel sequences as the number of consonants multiplied by the
number of vowels – therefore they are not included in the chart. Bear in mind
that we are talking about sounds here, not letters, and English spelling can
sometimes be misleading. For example, <kn->, <ps->, <gn->, or <wh-> in
spelling never stand for clusters because one or the other letter remains silent,
cf. knife (cf. Hungarian knédli 'steamed dumpling'), psychology (cf.
Hungarian pszichológia), gnome (cf. Hungarian gnóm), who or which. Such
sequences of letters are not taken into consideration either.
O+O O+N O+L O+G V+G
1
G+L /r/+/l/ L+N N+F F+P
stop
Spain
ski
sphere
snake
snore
shmuck
schnook
slay
shrimp
plead
trap
%suit
%tune
queen
swing
eye
tow
%fire
%hour
file
owl
%earl
%girl
%earn
%harm
elm
kiln
ounce
nymph
east
raft
clasp
ask
1
Diphthongs can be analysed as vowel-glide sequences. Notice that the second members of
closing diphthongs, viz. /i/ and /o/, are phonetically so close to the glides /j/ and /w/,
respectively, that some transcription systems denote them with the symbols of the glides,
e.g., /aj/ for /ai/, or /ow/ for /oo/. No wonder glides are also called semivowels! The intuition
that there is no clear dividing line between vowels and glides is also reflected in the choice of
the ancient Roman alphabet to represent both with the same symbol. Thus you can find
inscriptions like GAIVS IVLIVS CAESAR for Gaius Julius Caesar.
63 63
Chapter 5
The percentage mark (%) customarily indicates that the given example only
applies to certain speakers – in most cases it shows dialectal variation. In the
chart above it mainly refers to /r/, which is only pronounced in rhotic accents,
e.g., earl GA /arI/ (vs. RP /a:I/). In words like tune the yod (/j/) is not
pronounced in GA (/tu:n/) so the example is only relevant to RP (/tju:n/),
while in words like suit the yod is only pronounced by conservative (that is,
older) speakers of RP (/sju:t/) but not by younger speakers or speakers of GA
in general (/su:t/). This phenomenon, the absence of a yod in certain
positions, is called yod-dropping, and it is elaborated on below and in
Chapter 11.
On the basis of the examples, we arrive at the following order in
which sound segments are usually organized in the syllable:
obstruents - nasals/liquids/glides - vowels - glides - /r/ - /I/ - nasals - fricatives - plosives
The careful reader may have noticed that this list is more or less
symmetrically organized, having similar groups of consonants on either end
(namely, obstruents), vowels in the centre, and sonorants inbetween.
Moreover, it bears a spooky resemblance to the sonority scale discussed in
Chapter 2 and repeated here for convenience.
degree of sonority
——————————————————————————————→
oral stops (plosives) and affricates – fricatives – nasal stops – liquids – glides (semivowels) (– vowels)
Therefore we can make the following generalization: within syllables,
sonority increases towards the vowel, which forms a sonority peak, and then
sonority decreases; or, on both sides of syllables, sonority increases towards
64 64
The English syllable
the vowel. Henceforth we will call this the Sonority Principle. Let us
illustrate with a few examples how the Sonority Principle describes the
structure of well-formed syllables. The word tramp /træmp/, for example,
starts with a plosive, then comes a liquid, then the vowel, a nasal, and another
plosive at the end. This can be schematically represented as follows.
V æ
G
L r
N m
F
P t p
tramp
Further examples:
V a a 5
G i w w
L I r r
N n m
F s
P b o k k
blind GA quirk GA swarm
Notice what happens in words like tender or button: since there are two
vowels, there are two sonority peaks, that is, two syllables! Even if the /n/ is
syllabic in button, the number of sonority peaks, that is, the number of
syllables is unchanged. The difference between the schwa-ful and the schwa-
less pronunciations is that in the latter case the second sonority peak is not a
vowel but a consonant (the /n/).
65 65
Chapter 5
V c c \ c \
G
L
N n n n
F
P t o b t b t
tender button button
A simple definition of syllabic consonants ensues: they are consonants
functioning as the sonority peak in a syllable. It also follows that not only
vowels can occupy the sonority peak, thus the Sonority Principle needs
reformulating: on both sides of syllables, sonority increases towards the peak,
which is a vowel or a syllabic consonant. The conditions on syllabic
consonant formation in English are discussed towards the end of this chapter.
It is very interesting that the above definition of the Sonority Principle
can be turned inside out and translated as the definition of the syllable: it is a
phonological unit which contains a sonority peak. What we have seen above,
then, directly follows: in a word, there are as many syllables as sonority
peaks. The English word rhythm, for instance, can only be pronounced with
two syllables as it contains two such peaks (a vowel and a (syllabic) /m/). If
we shuffle the segments in a well-formed syllable, e.g., tramp /træmp/
(mentioned above), resulting in /træpm/ or /rtæmp/, we arrive at the same
conclusion: these must be disyllabic words.
66 66
The English syllable
V i æ æ
G
L r r r
N m m m
F ð
P t p t p
rhythm /træpm/ /rtæmp/
The difference between rhythm and (hypothetical) /'træpm/ is that the latter is
simply non-existent, i.e., an accidental gap. The difference between
(hypothetical) /'træpm/ and (hypothetical) /rtæmp/, however, is much graver:
while /'træpm/ is a possible (disyllabic) word of English, the same is not true
for the other: /rtæmp/ starts with a syllabic /r/, and for independent reasons
English words never start with a syllabic consonant. The Sonority Principle
is, therefore, one of the major factors determining and explaining what
qualifies as a well-formed English syllable.
However, there exist a number of examples where the Sonority
Principle fails. Consider the following words: they all contain two sonority
peaks, still, all speakers of English insist that they are monosyllabic.
V o i: æ o
G
L
N
F s s s I s
P t p k p k
stop ski apse fox
In addition, in a few cases segments of equal sonority follow each other
within the syllable, and consequently sonority neither rises nor falls.
67 67
Chapter 5
V i æ
G
L
N n
F s I s
P k k t
sphinx act
Notice that the word sphinx, for example, is doubly problematic: on the one
hand, it starts with flat sonority rather than the expected rise; on the other, it
ends in a sonority rise rather than the expected fall. The following chart
summarizes the possible exceptions to the Sonority Principle and gives a
couple of examples.
initial final rise flat sonority flat sonority finally
fall simple complex initially simple complex
stop
ski
Spain
fox
apse
axe
hits
lads
eighth
sphinx
sphere
svelte
act
adopt
corrupt
fifth
ached
robbed
As you can see, the word-final examples fall into two categories: they are
either monomorphemic (i.e., morphologically simple), e.g., fox or act, or the
problematic segments straddle a morpheme boundary (i.e., the word is
morphologically complex), e.g., hit-s or fif-th. Although words like these
contradict the Sonority Principle, we can still conceive of it as a
generalization describing the majority of the data, and treat stop and the like
as exceptions. It is intriguing, however, that even these exceptions are
constrained: the problems are caused by obstruents, in most cases fricatives,
more specifically /s/. We will see below that /s/ takes part in the construction
of syllables in a special way in several further respects.
68 68
The English syllable
In sum, the Sonority Principle serves us with a considerably reliable
definition of the syllable, although sometimes it is overridden by native
intuition. Compare rhythm and fox, discussed above: both contain two
sonority peaks but only one of them (rhythm) is judged by speakers of
English to be disyllabic.
From a cross-linguistic perspective, the role of the Sonority Principle
is far from uniform. In most languages, there are strict conditions on sound
sequences. In certain languages each syllable must start with a consonant,
that is, with a sonority rise; in others there cannot be more than a single
consonant at the beginning. Yet others (like English or Hungarian) allow for
clusters syllable-initially but only certain types, usually with a strict
adherence to the Sonority Principle. And finally, there are a few languages
(e.g., the Slavic languages like Russian) where (almost) any combination of
their consonants is possible, and numerous violations of the Sonority
Principle are attested. As to the syllable-final position, some languages permit
no consonants and therefore all syllables end in a vowel (such syllables are
called open); others (like Italian or Japanese) differentiate between word-
internal and final syllables, and only have syllable-final consonants in one of
the two types. In languages like English, it is possible to find consonants at
the end of any syllable (making it a so-called closed syllable), but there is
always a limit on the maximal number of consonants. In English, this is four,
which means that some syllables contain such "monster clusters" as /ksts/ or
/ks0s/ in texts or sixths, for example. (You may have noticed that the clusters
at the end of texts or sixths are not only too long, but they also contain
violations of the Sonority Principle.) Nevertheless, there is one syllable type
which is universal, i.e., which is possible in all the languages of the world:
one starting with a single consonant and ending with a vowel. Besides being
universal, this very simple configuration is also the first to emerge during the
69 69
Chapter 5
process of language acquisition, that is, when babies learn their mother
tongue. Just list the words Hungarian kids learn first, and you will see.
Let us now turn back to the discussion of the English syllable. In what
follows, we provide a brief description of what language-specific
phonotactic restrictions accompany the Sonority Principle. As we have
seen, the centre of the syllable is the sonority peak, which is usually a vowel,
and in fact in English (and Hungarian) this peak is the only obligatory
constituent – that is, there are syllables with a single vowel and no
consonants (e.g., English I/eye /ai/ or Hungarian ő '(s)he'), but there are no
syllables without a peak (in Hungarian, without a vowel). The English peak
can be preceded by zero to three consonants and followed by zero to four.
If there is a single consonant before the peak, it can be any consonant
except /n/. Certain consonants like /¸/ and /ð/ are relatively infrequent in this
position. Two-member clusters usually consist of an obstruent and an
approximant, since these obey the Sonority Principle (e.g., twin, trip, tube,
play, pray, puke, quick, cry, clean, cube, fry, fling, dry, Gwen, etc.). One
consonant, /s/, can be combined with any of the others except for voiced
obstruents and /r/ (e.g., snip, slip, swim, sport, skirt, stink, sphere, etc.), and
this very often leads to the sonority sequencing violations mentioned above.
Recall that it is usually /s/ that is to blame!
A few rising-sonority clusters, however, are ruled out, e.g., *pn, *ps,
*gn and *kn. The warning is still in effect that you should not let words like
pneumonia, psycho, gnu and knight mislead you – they only start with a
consonant cluster in spelling. Similarly, the letter <x> at the beginning of
words like Xerox, xylophone and Xanadu does not denote a /ks/ sequence but
a single /z/. The spelling of the words pterodactyl and mnemonics suggest
initial clusters of flat sonority, but in pronunciation they are simplified, and
only a single consonant is pronounced. (Such spelling-to-pronunciation
70 70
The English syllable
regularities are discussed in detail in Chapter 11.) The nonexistence of these
clusters of rising or flat sonority is curious because apparently they are
completely acceptable in Hungarian (cf. the Hungarian equivalents of the
above words, or the examples given earlier), although their foreign origin is
evident.
There is another set of rising-sonority clusters which is unattested in
English, but this time the same holds for Hungarian, and in fact, we will be
able to find a principled explanation for why they are so unpopular. These
include, e.g., *tl, *tn, *pw, *fw – no English (or Hungarian) examples are
available for them. What these clusters have in common is that they are
homorganic, i.e., their members share the place of articulation. Both /t/ and
/l/ are coronal, and so is /n/; both /p/ and /w/ are labial, and so is /f/. Although
a whole lot of other homorganic clusters exist, e.g., /tr, or, 0r, jr, jn/ plus the
/s/+coronal clusters (recall that /s/ can combine with almost all other
consonants), there is a clear dispreference for clustering consonants to share
the place, one manifestation of which is a phenomenon referred to above,
Yod-dropping. There is an absolute ban on /j/, the coronal glide, to appear
after coronal /j, ¸, tj, o¸, r/. There are no English syllables beginning with
/jj, ¸j, tjj, o¸j, rj/. After coronal /l/, it is again impossible to find a yod if the
/l/ is preceded by another consonant, that is, when it is part of a cluster: e.g.,
*/bIj-/. Following a single /l/, the yod can "survive" dropping but only in
conservative RP, cf. lucid /'Iju:sio/, lucrative /'Iju:krctiv/, ludicrous
/'Iju:oikrcs/. Even in RP, however, the pronunciations without the yod, i.e.,
/'Iu:sio/, /'Iu:krctiv/, /'Iu:oikrcs/, are more frequent, and the same applies to
/sj, zj/ in words like suit /sju:t~su:t/, super /'sju:pc~'su:pc/, Zeus /zju:s~zu:s/,
presume /pri'zju:m~pri'zu:m/. In GA, this tendency to drop the yod has
71 71
Chapter 5
become generalized to take place after all coronals – not only /I, s, z/ but /0, t,
o, n/ too. That is why new is /nju:/ in RP but /nu:/ in GA, tuna is /'tju:nc/ in
RP but /'tu:nc/ in GA, dude is /'oju:o/ in RP but /'ou:o/ in GA. It is only in GA
that the title Looney Tunes can refer to lunatic toons (cartoon characters)
since both are pronounced /tu:nz/, as opposed to RP, where tune is /tju:n/. In
contrast, the yod is rather stable in both dialects in unstressed syllables, e.g.,
after a lone /l/ in value /'væIju:/, after an /s/ in capsule /'kæpsju:I/, although
after an /n/ as in avenue both options are available in GA /'ævcnu:~'ævcnju:/.
All in all, /j/ is gradually disappearing after the other coronals, which can be
considered as another illustration of the dispreference of homorganic
syllable-initial clusters.
2
As it has been mentioned above, the maximal number of syllable-
initial consonants in English is three. The three-member sequences are,
however, heavily constrained: they always begin with /s/ (again, it is /s/!),
which is followed by a legitimate two-member cluster (strength, spring,
square, splash, %stew RP /stju:/, etc.). Since all such syllables contain the
/s/+(voiceless) plosive+approximant sequence, they always violate the
Sonority Principle.
Turning to the syllable-final position, we can state that any single
consonant except for /h/ can occupy it. In addition, in non-rhotic accents like
RP, /r/ is also banned at the end of syllables, as it was discovered in Chapter 2
– therefore the rule of R-dropping can be treated as a phonotactic restriction
characterizing non-rhotic accents only. In two-member clusters after the peak,
we usually find nasal/liquid+consonant sequences, which exhibit falling
sonority, e.g., lamp, month, land, mince, help, bulb, elf, %carp, %herb,
2
You find further examples of Yod-dropping in Chapter 11, where it is discussed again from
a slightly different point of view: as a letter-to-sound rule. In addition, it is argued there that
the yod is in fact part of a complex vowel /ju:/.
72 72
The English syllable
%search, film, %harm, %curl, etc. Notice that within the class of liquids /r/
systematically "pretends" to be more sonorous than /l/: -rl is possible (at least
in rhotic accents) but -lr is not.
When two obstruents compose a syllable-final cluster, one of them is
usually /s/ (again!): /s/+obstruent in grasp, last, risk, etc., obstruent+/s/ in
lapse, axe, etc. Flat sonority contours are also attested (apt, act, etc.) but the
second consonant is always a coronal. In three-member strings (prompt,
against, next, etc.) the third member is always a coronal obstruent, and in
morphologically complex words additional combinations yielding the
"monster clusters" with four consonants in a row can also be formed (ending
in -ed, -s, -th – all coronals).
The examples of final clusters we have seen up to this point also
appear word-medially, e.g., /mp/ is found in both lamp and campaign, /lm/ in
film and helmet, /st/ in last and asterisk, and /pt/ in apt and chapter. There
are, however, certain word-internal consonant clusters which are impossible
word-finally. In such cases, the consonant cluster suggested by the spelling
undergoes simplification, and remains simple even if a suffix is attached to
the word. For instance, /gn/ is well-formed within words like cognate,
dignity, magnet, signature, resignation, but the /g/ is deleted in sign and
resign as well as in signing and resigning. The same goes for /mn/: it is
acceptable in alumnus, amnesty, chimney, insomnia, damnation, hymnal,
autumnal, but simplified (with the /n/ lost) in damn and damning, hymn,
autumn. Homorganic nasal+voiced plosive sequences are also highly
restricted unless the consonants are coronal: /no/ is unmodified irrespective
of its position (cf. lend, bind, wound, and candle, tender, boundary) whereas
/mb/ and /ng/ only survive word-internally (amber, ambulance, bombard;
finger, anger) but not finally (bomb, bomber, bombing; long, strong, sing,
73 73
Chapter 5
bang, singer, singing, banger). The distribution of the velar nasal is
particularly intriguing: it does not normally appear between vowels in
morphologically simplex forms like finger or anger (*fi[n]er, *a[n]er) – with
just a few exceptions such as hangar. Next to a morphological boundary,
however, it is rather frequent in such position, as we have seen above (singer,
singing, banger, etc.). In this respect, what happens in the comparative and
superlative forms of adjectives is surprising: the simplified cluster of the
positive forms long, strong, young is "regained" in longer, stronger, younger;
longest, strongest, youngest (all with /ng/).
3
Besides the restrictions on syllable-initial and -final consonant
sequences, there is an additional type of phonotactic constraint, namely, one
which applies to the vowel and the following consonant(s) together. Since in
poetry this part of the syllable determines whether two words rhyme,
phonologists conventionally refer to it as the syllable rhyme. There are
several restrictions on the English rhyme, e.g., /ao/ can only be followed by
coronal consonants (shout, crowd, south, town, etc.); /5i/ can only be
followed by alveolars (exploit, void, voice, noise, coin, coil, moist, point); a
long vowel is only possible before a consonant cluster if the cluster is made
up of coronals (mind, boost, faint, etc.); and in word-final open syllables (i.e.,
without a closing consonant) the vowel has to be either long (monophthong
or diphthong, e.g., taboo, array, RP near) or unstressed (happy, comma, etc.).
Before discussing the restrictions concerning the peak, let us take
another look at syllable and word edges, and the asymmetry between them.
On the one hand, in word-final clusters more consonants are possible than in
word-medial ones; what is more, they frequently violate the Sonority
Principle (cf. sixth, text), which also holds for word-initial clusters (cf. stop,
Spain, screw, strip). On the other hand, it is a well-known fact that not all
3
Examples like these are repeated in Chapter 11, in the discussion of silent letters.
74 74
The English syllable
combinations of well-formed syllables yield a well-formed word, so the
attempt at joining the apparently well-formed right edge /kst/ of a syllable
like text with the apparently well-formed left edge /str/ of a syllable like strip
will result in the string /kststr/ unattested word-internally. It seems impossible
to talk about phonotactic restrictions without making reference to the position
of the syllable within the word.
Finally, let us see some of the phonotactic constraints on the syllable
peak. In most cases it is occupied by a vowel, either monophthong or
diphthong. As far as diphthongs go, we find that they are heavily restricted:
not all the possible combinations of the vocalic segments of English exist.
Moreover, their second members can only be one of three vowels, /i o c/ –
this number is radically smaller than the number of English monophthongs.
Besides vowels, certain consonants can also constitute the peak of the
syllable, in which case they are syllabic consonants. Recall from Chapter 2
that syllabic consonants are indicated with a subscript [ ], e.g., table [-bI],
button [-tn], faculty [-kIti], finally [-nI i], national [-jn I]. In RP, syllabic
consonant formation (SCF) is only possible in unstressed syllables, where
an alternative pronunciation (mainly used in slow, careful speech) contains a
schwa followed by a non-syllabic version of the consonant. For instance, the
word table has two possible pronunciations, one with a schwa ['tºcibcI] and
one without, in which case the final consonant is syllabic ['tºcibI ]. Basically,
what happens is that the schwa drops out but the number of syllables is
preserved since the following consonant steps up to act as a peak instead. The
process, however, has a number of conditions. First, the consonant following
the schwa must be a sonorant; in most cases, it is /n/ or /l/. Second, the
consonant following the schwa must be more sonorous than the one
75 75
Chapter 5
preceding it. In camel, e.g., SCF is possible, yielding ['kºæmI], because /l/ is
more sonorous than /m/; compare this with column /'koIcm/, where it is not.
The same applies to kennel (['kºcncI] or ['kºcnI]) versus melon ['mcIcn] and
not *['mcIn]. This sonority condition does not hold if the first consonant is
/r/: barrel ['bærI] is well-formed although both consonants are liquids; barren
['bærn] is possible alongside examples like banner GA ['bænr ]. In non-rhotic
English (including RP) /r/ can only become syllabic word-internally, e.g.,
natural ['nætjr I]; but in rhotic English (especially GA) /r/ can also become
syllabic word-finally (e.g., better ['bctr ] or ['bcrr ]) or even in stressed
syllables (e.g., bird [br o]).
SCF is not the only form of schwa deletion, though. Schwa can also
drop out in such a way that the number of syllables is NOT preserved – a
vowel is lost, consequently a peak is lost, consequently a syllable is lost.
Such straightforward examples of vowel loss are traditionally referred to as
syncope. Intriguingly, the conditions on syncope are more strict after a
stressed vowel than before it. For post-stress syncope to take place, the
consonant following the schwa must be a sonorant, and it must be more
sonorous than the one preceding it. In addition, the following vowel must be
unstressed, that is, weak (cf. Chapter 3). That is how the underlined vowels in
camera, family, different, separate (adj) can be deleted, yielding disyllabic
/'kæmrc/, /'IæmIi/, /'oiIrcnt/, /'scprct/, but not in vanity (the /t/ is not a
sonorant), felony (nasals are less sonorous than liquids) or separate (v) (the
third syllable contains a full vowel).
Pre-stress syncope, however, is not as restricted: although it always
occurs in initial syllables, the consonants surrounding the target schwa do not
necessarily obey the sonority constraint. The underlined vowel can not only
76 76
The English syllable
be omitted from words like terrain or parade but also in suppose, suffice,
potato, etc. Interesting new homophones emerge: terrain may sound the same
as train, parade as prayed, Sapir as spear, support as sport, and police may
only differ in the final consonant from please.
The difference between syncope and SCF, then, is that the number of
syllables in the word is affected in the former but not in the latter. Both are,
however, in most cases (except for pre-stress syncope) governed by some
kind of sonority condition, similarly to the overall structure of the syllable.
77 77
6. Laryngeal features
Before you study this chapter, check whether you are familiar with the
following terms: allomorph, allophone, aspiration, devoicing, frequency,
glottalization, glottal stop, glottis, hiatus, larynx, organs of speech,
phoneme, pulmonic egressive airstream, root, sibilant, suffix, syllabic
consonant, utterance, vocal cords/folds, voice assimilation, voiced,
voiceless, voicing
In this chapter we take a look at the articulatory role of the glottis, the vocal
cords/folds and all the different phenomena that are related to the operation
of the larynx. This includes voicing and voicelessness, (a comparison of
English and Hungarian) voice assimilation, devoicing, aspiration and
glottalization, and the effect of voicelessness on preceding vowels, Pre-fortis
Clipping.
Recall from Chapter 1 that the basic mechanism that is used to
produce speech sounds in English and Hungarian is a pulmonic egressive
airstream mechanism. Having left the lungs, the air continues upward in the
windpipe up to the larynx – the front, shield-like part of which is called
Adam's apple – then into the pharynx and the supraglottal cavities, the oral
and nasal cavities. In the larynx it has to pass between the vocal cords/folds,
two bundles of muscle, which may produce numerous different effects.
Before discussing these, let us take a look at the structure and parts of the
larynx and their different configurations.
As it can be seen in the diagram, the larynx consists of some
cartilages – one fixed in the front and two movable ones in the back, plus
one more on the top not shown in the diagram, the epiglottis, which can close
the windpipe, and a circular one serving as the frame for the larynx –, the
Laryngeal features
vocal cords connecting the cartilages, and the opening between them, the
glottis. Depending on how tense the muscles of the vocal cords are, the two
cartilages in the back will move closer together or further apart. This way the
vocal cords will close or open the glottis to different degrees.
When the vocal cords are wide apart then silent breathing is produced. When
the vocal cords are slightly pulled together but still do not close the glottis
and no vibration is produced, voiceless consonants are articulated. If the
vocal cords are pulled a bit closer together than in the case of voiceless
consonants, a voiceless glottal [h] sound is produced. In another
configuration the elastic vocal cords are pulled together and the outflowing
air pushes them apart and then, because of their elastic nature, they return
into their original position. This is repeated periodically at a rate of about 120
times in average in males and about 220 times in average in females (that is,
the basic frequency in males is about 120 Hz and about 220 Hz in females),
this way producing vibration, that is, voiced sounds. This mechanism is very
similar to when children put a blade of grass or a piece of a leaf between their
two thumbs and then blow it producing a high pitch trumpet-like noise. In
79
Chapter 6
both cases it is a flexible, elastic string – the vocal cords or the blade of grass
– that is forced to move in a fast, periodic way.
Finally, it is possible to produce a total closure, a complete
obstruction to stop the air in the larynx. This way a glottal stop is produced,
the sound often heard in the pronunciation of words like bottle ['bo1I] or
['bo
1
tI] in British dialects (mentioned in Chapter 2 and discussed in more
detail below), and in Hungarian as an extralinguistic device to express
surprise in [o1'o:] or to optionally break up a hiatus – the sequence of two
vowels – in words like kiiktat ['ki1iktot].
Let us now take a look at how English makes use of the qualities
voiced and voiceless in the different classes of speech sounds. To start with
the easy part, consider vowels first: as all vowels are always produced as
voiced, we can say that voicing is not a distinctive quality in vowels, i.e., it
does not distinguish vowels from each other. Voiceless vowels are only used
when whispering, partially devoiced vowels – vowels which have lost part of
their voicing, the very beginning, the first few milliseconds of the vowel
being voiceless – occur after aspirated stops, a topic already discussed in
Chapter 2 and to be discussed later in this chapter. Sonorant consonants
behave in a very similar way to vowels: they are always voiced by default
and they only become partially devoiced after aspirated stops.
This leaves us with obstruents: stops, fricatives and affricates. If one
takes a look at the table of manners and places of articulation in Chapter 2,
then it is easy to see that obstruents occur in voiced-voiceless pairs or to put
it very simply: in the obstruent part of the table there are always voiced-
voiceless pairs of stops, fricatives and affricates at each place of articulation.
The only exceptions seem to be the glottal stop and /h/. The glottal fricative
does not have a voiced counterpart in English – as opposed to Hungarian,
80
Laryngeal features
where the phoneme /h/ does have a voiced allophone [n] occurring between
sonorants, e.g., konyha ['konno], but not a separate voiced glottal /n/
phoneme.
The most interesting aspect of the voicing of obstruents is the stability
of voicelessness and the relative instability of voicedness in English. The so-
called voiced obstruents of English are very often realized by a partially or
fully devoiced allophone – this devoicing is represented in transcription by a
small circle below the symbol of the sound, e.g., [z]. As English voiced
obstruents seem to be voiced only phonologically (i.e., they only behave as if
they were voiced) in many cases, two other terms have been introduced
instead of voiced and voiceless: lenis (Latin for 'weak') and fortis (Latin for
'strong'). Lenis obstruents are weak and often lose their underlying voiced
quality; they are phonologically voiced and may be realized by voiced or
voiceless speech sounds in actual pronunciation depending on the
environment. Fortis obstruents, on the other hand, are strong, and are thus
always realized by voiceless speech sounds.
The devoicing process affecting lenis consonants typically applies in
utterance-initial, utterance-final positions and next to fortis obstruents.
81
Chapter 6
Utterance-initial Utterance-final Next to a fortis sound
(a) (b) (c) (d)
Bravo! ['bro:vco]
Good! ['goo]
Zany! ['zcini]
Damn! ['oæm]
Very much! ['v cri]
Mad! ['mæo]
Go ahead! [c'hco]
Think big! ['b ig ]
Bob! ['b ob]
Leave! ['Ii:v ]
obtain [cb 'tºcin]
cheesecake ['tji:zkºcik]
bigfoot ['b ig Iot]
egghead ['cghco]
roadster ['rcoostc(r)]
matchbox ['mætjboks]
baseball ['b cisb 5:I]
cookbook ['kºokbok]
life gear ['IaiIgic(r)]
Shoot back! ['ju:t 'b æk]
In (a) and (b), in utterance-initial and -final position, i.e., before or after a
pause, lenis obstruents often devoice at least partially: in initial position it is
typically the beginning of the obstruent that is voiceless while in final
position it is the end. This is probably the consequence of the fact that there is
a timing difference between the beginning or end of vocal cord vibration
(voicing) and the beginning or end of the closure. In initial position closure is
produced first and voicing starts only a few milliseconds later, while in final
position voicing stops first and only after that is the stop released. Note that
Hungarian is different in this respect as there is no such devoicing in initial or
final position. The voicing of obstruents normally starts before the closure is
produced and voicing only ends after the closure ceases – Hungarian is said
to have pre- and postvoicing in obstruents.
In column (c) all the sample words contain a lenis obstruent followed
by a fortis one. As a result of the influence of the fortis (voiceless) sounds,
the preceding obstruents become devoiced, partially or fully voiceless. It is
similar to what happens in identical Hungarian clusters.
82
Laryngeal features
English Hungarian
(c) (d) (e) (f)
obtain [cb'tºcin]
cheesecake ['tji:z kºcik]
bigfoot ['bigIot]
egghead ['cghco]
roadster ['rcoostc(r)]
matchbox ['mætjboks]
baseball ['bcisb 5:I]
cookbook ['kºokbok]
life gear ['IaiIgic(r)]
Shoot back! ['ju:t 'bæk]
rabtól ['ropto:I]
rézkarc ['rc:skorts]
hangfal ['honkIoI]
éghez ['c:khrz]
roadster ['ro:tstrr]
matchbox ['mro¸boks]
baseball ['bc:zbo:I]
tökből ['togbo:I]
afgán ['ovga:n]
kertből ['krrobo:I]
As it can be seen, whenever a fortis and a lenis obstruent of English occur
adjacently, one of them changes its voice quality. Similarly, whenever a
Hungarian voiced and voiceless obstruent occur adjacently, one of them has
to change its voice value. We may call this a case of voice assimilation by
which the voicing of one sound becomes identical to that of a neigbouring
one. It is called regressive if the sound that changes precedes the sound that
influences it, and it is called progressive in the opposite case.
The difference between the lenis+fortis case in the English and
Hungarian examples – columns (c) and (e) – is manyfold: devoicing is not
necessarily complete in English, but it is in Hungarian as indicated by the
difference in the symbols; devoicing is not obligatory in English but it is in
Hungarian – but assimilation is regressive in both languages. As for the
difference between the fortis+lenis cases – columns (d) and (f) – the
difference is even bigger. It is not just a matter of degree – partial or complete
– and nature – optional or obligatory – but also a matter of value: in English
the lenis obstruent assimilates to the fortis one – devoicing applies –, in
Hungarian it is the fortis obstruent that assimilates to the lenis one – voicing
applies. Consequently, in English we see progressive devoicing while in
Hungarian we see regressive voicing. That is, it seems that in English it is the
result of assimilation that is fixed – it is always voicelessness –, while in
83
Chapter 6
Hungarian it is the direction – it is always regressive. The following table
sums up the differences between voice assimilation and devoicing in English
and Hungarian.
English Hungarian
► optional
► partial or complete
► its result is always devoicing
► may be regressive or progressive
► initial or final devoicing may
apply
► obligatory
► complete
► its result may be devoicing or
voicing
► always regressive
► no initial or final devoicing
There is one more special area of English voice assimilation that we have to
mention here: the assimilation of the suffixes -s 'plural', '3rdSg' or 'possessive'
and -ed 'past tense' or 'past participle'. Normally it is assumed that the basic
forms of these suffixes are /z/ and /d/, respectively, as these are the ones that
appear after vowel-final roots. These forms then assimilate to the root-final
consonants.
/z/ /s/ /iz/
legs /'Icgz/
tabs /'tæbz/
heads /'hcoz/
means /'mi:nz/
girls /'ga:Iz/
ways /'wciz/
shows /'jcoz/
kicks /'kiks/
blokes /'bIcoks/
taps /'tæps/
turnips /'ta:nips/
hats /'hæts/
laughs /'Io:Is/
baths /'bo:0s/
churches /'tja:tjiz/
judges /'o¸\o¸iz/
bushes /'bojiz/
garages /gc'ro:¸iz/
kisses /'kisiz/
buzzes /'b\ziz/
stretches /'strctjiz/
84
Laryngeal features
As can be seen in the last column of the table, /iz/ is used after the sounds
/tj/, /o¸/, /j/, /¸/, /s/, and /z/, that is, after sibilant consonants, as it would be
difficult to pronounce the sibilant /z/ of the suffix after the root-final sibilants.
The first column of the table shows that the basic form /z/ is used after all
non-sibilant voiced sounds – both consonants and vowels – while the second
column demonstrates that /s/ occurs after voiceless non-sibilants. All in all,
we can say that the suffix consonant progressively assimilates to the root-
final consonant. We have to remember, though, that this voice assimilation is
different from the devoicing cases in that it is always complete, progressive
and obligatory.
The suffix -ed behaves in a very similar way to -s presented above.
/o/ /t/ /io/
begged /'bcgo/
robbed /'robo/
advised /co'vaizo/
depraved /oi'prcivo/
damaged /'oæmio¸o/
contained /kcn'tcino/
filled /'IiIo/
followed /'IoIcoo/
clicked /'kIikt/
ripped /'ript/
laughed /'Io:It/
passed /'po:st/
kissed /'kist/
hushed /'h\jt/
stretched /'strctjt/
attached /c'tætjt/
wanted /'wontio/
mended /'mcnoio/
protected /prc'tcktio/
beheaded /bi'hcoio/
located /Ic'kcitio/
paraded /pc'rcioio/
navigated /'nævigcitio/
vaccinated /'væksincitio/
In the last column, after root-final alveolar stops /t/ and /o/, the /io/ allomorph
is used, as an /i/ is inserted between the two alveolar stops. After all other
voiced root-final phonemes the basic variant /o/ is used while after all other
85
Chapter 6
voiceless root-final phonemes a /t/ allomorph occurs. This assimilation
process is also always complete, obligatory and progressive.
Now we turn to laryngeal processes other than voicing. In Chapter 2,
aspiration is mentioned as one of the processes affecting the voiceless
plosives /p t k/, but it is left unexplained what exactly aspiration is
phonetically. We know that it has two forms. One is the short [º]-like sound
following the plosive, which is in fact not a separate sound segment but
merely the acoustic impression that we get due to the first half of the
following vowel being devoiced. As you know, all vowels are normally
voiced, i.e., their articulation involves vocal cord vibration. In words like pay
[pºci], the /p/ is voiceless, and its voicelessness spreads onto the vowel, as a
result of which the vocal cord vibration characteristic of all vowels starts
much later than the release of the plosive and the onset of the vowel.
Therefore, what is heard right after the burst of the plosive is a short period
with a voiceless vowel (= open vocal cords plus no considerable obstruction
to the airflow in the oral cavity), which is, phonetically, identical with /h/ (=
open vocal cords plus no considerable obstruction to the airflow in the oral
cavity).
The other manifestation of aspiration is the devoicing of a following
sonorant consonant in words like play [pI ci], true [tr u:], quick [kwik]. Of
course, the two forms of aspiration are not unrelated, on the contrary, they are
the same: the voicelessness of the plosive spreads onto the following
segment. Whether it is a vowel or a consonant, its vocal cord vibration lags
well behind the release of the plosive. This is schematically represented in the
diagram below. The two parallel lines symbolize the vocal cords; when they
are straight, there is no vibration, when they zigzag, there is. In bay, we have
a voiced plosive; in spay and spray, a voiceless unaspirated one owing to the
86
Laryngeal features
preceding /s/; in pay and pray, a voiceless aspirated /p/, which devoices the
following vowel and sonorant, respectively.
1
Another process caused by laryngeal activity is glottalization. As it is
explained in Chapter 2, the voiceless plosives /p t k/ (and also /tj/) are in
certain positions accompanied by a short closure of the vocal cords, i.e., by
the so-called glottal stop ([1]), e.g., bat [b æ
1
t], actor ['æ
1
ktc(r)], teacher
['tºi
1
tjc(r)]. This is called glottal reinforcement. Sometimes, especially
before a syllabic /n/, a /t/ can be completely replaced by it, that is, glottal
replacement can take place, e.g., button ['b \1n]; in several non-standard
varieties of English, especially London English, this can even happen in
words like let [Ic1], butter ['b \1c] (or ['b o1c]) or city ['si1i]. What connects
these examples to the previous discussion is the fact that the production of
the glottal stop involves the movement of the vocal cords only, and no
gesture above the larynx.
1
We ignore the slight devoicing of /b/ at the beginning of bay.
87
Chapter 6
A final rule that belongs to laryngeal processes – although not a
strictly laryngeal one – is the way fortis consonants influence the phonetic
length of the preceding vowel. If a long vowel – monophthong, diphthong or
triphthong – is immediately followed by a fortis consonant or a nasal and a
fortis consonant within the word, then the vowel will be shortened or clipped,
hence the name of the process, Pre-fortis Clipping. Long vowels become
approximately as short as real, phonologically short vowels but it is important
to remember that there is no change in the quality of the vowels at all.
Remember (from Chapter 3) that the phonologically short but phonetically
long vowel /æ/ behaves identically in this respect, i.e., it patterns exactly like
phonologically long vowels do. The change in vowel length is indicated in
narrow, phonetic transcription with the symbol [·] instead of [:] after the
vowel.
long monophthong shortened long
monophthong
long diphthong shortened
diphthong
balloon [bc'Iu:n]
believe [bi'Ii:v]
store ['st5:(r)]
star ['sto:(r)]
stir ['sta:(r)]
span ['spæn]
carnival ['kºo:nivI ]
spawn ['sp5:n]
boot ['b u·t]
speak ['spi·k]
sports ['sp5·ts]
start ['sto·t]
shirt ['ja·t]
attacked [c'tºækt]
dance ['oo·ns]
daunting ['o5·ntin]
obey [c'bci]
decide [oi'saio]
allowing [c'Iaoin]
towed ['tºcoo ]
cleared ['kIico]
destroy [oi'str5i]
staring ['stccrin]
curious ['kj ocrics]
mistake [mi'stcik]
tonight [tºc'nait]
bounce ['baons]
boat ['bcot]
fierce ['Iics]
catering ['kºcitcrin]
spouse ['spaos]
biting ['baitin]
Note that in the case of the vowel /æ/ we cannot indicate shortening as the
vowel is classified as a phonologically short vowel and, as a result, the
phonetic length of the vowel is not indicated with the colon originally. The
case of diphthongs is similar: their length is encoded in the combination of
88
Laryngeal features
two symbols rather than a length mark – since neither of the two elements is
lost through clipping, we are again unable to show this phonetic shortening in
our transcriptions.
What is intriguing about Pre-fortis Clipping is that it is clearly
conditioned by the fortis character of the following consonant, rather than its
voicelessness. Recall that lenis obstruents systematically become
devoiced/voiceless in, for instance, utterance-final position, e.g., said [sco].
Still, a preceding long vowel remains long, e.g., seed [si:o ], as opposed to
words like seat [si·
1
t], where the consonant following the vowel is not only
voiceless but also fortis. Similarly, the /ai/ is much shorter in write [rait] than
in ride [raio ], and the /æ/ is much shorter in atom ['ætcm] than in Adam
['æocm]. In the so-called tapping dialects of English, mentioned in Chapter 2,
the distinction between /t/ and /d/ may be lost in certain positions, both being
replaced by a tap, but the length of the vowel is still there to show the
fortis/lenis character of the original consonant: short in atom ['ærcm] but
long in Adam ['ærcm]; short in writing ['rairin] but long in riding ['rairin];
short in seater ['sirc(r)] but long in seeder ['si:rc(r)]. The duration of the
vowel ultimately becomes the indirect indicator of the nature of the following
consonant.
To sum up the discussion of laryngeal processes, we can state that the
vocal cords play a crucial role in the articulatory process not only in
determining the voicedness of speech sounds, but also in producing
individual segments like /h/ or the glottal stop. In addition, they are
responsible for certain phenomena, e.g., devoicing, aspiration and
glottalization, which constitute some of the most significant allophonic rules
that English consonants undergo. We have also seen that although both
Hungarian and English obstruents take part in voicing assimilation, there is a
89
Chapter 6
huge difference between the two languages: in Hungarian, the direction of the
assimilation is fixed (namely, it always proceeds from right to left), whereas
in English the output of the process is fixed (namely, it always results in
voicelessness). Finally, it has been demonstrated how complex an interaction
exists between the vocal cord activity characterising a consonant and the
phonetic length of the preceding vowels.
90
7. Connected speech
Before you study this chapter, check whether you are familiar with the
following terms: accent, alveolar, ambisyllabic, aspirated, assimilation,
allophone, clear-L, compound, dark-L, dental, devoicing, diacritic, full
vowel, function word, glottal stop, glottalized, hiatus, high vowel,
idiomatic, intrusive-R, linking-R, morpheme, nasal, non-rhotic, palatal,
plosive, R-dropping, reduced vowel, rhotic, segment, stressed/unstressed,
suffix, suprasegmental features, syncope, tapping/flapping, utterance, velar,
yod
This chapter deals with the phenomena that characterize connected speech,
that is, combinations of words rather than individual words uttered in
isolation. These cross-word processes are of crucial importance since rarely
do we pronounce a single word only – normally we use phrases and
sentences, very often several sentences one after the other. We have already
mentioned a number of such processes, especially in Chapter 2, which are
recapitulated below. Recall, also from Chapter 2, that word-final consonants
are always ambisyllabic when followed by a vowel in the next word, and
choose their pronunciation variant accordingly.
First, in the discussion of L-darkening in RP, we found that syllable-
final /l/ is dark, but it is clear elsewhere. Elsewhere includes the ambisyllabic
position, too. Therefore, the /l/ at the end of mill, for example, is dark when
the word is final in the utterance (e.g., Where's the mill?) or when the next
segment is a consonant (e.g., There are two mills here or The mill was
closed). However, when the following morpheme – suffix or word – starts
with a vowel, it is clear (e.g., It's Mr. Miller or The mill is closed). Bear in
Chapter 7
mind that syllabic /l/ is dark, whatever it may be followed by; just as dark in
cancel a meeting as in cancel the meeting. This is a logical consequence of
the fact that peaks are never ambisyllabic – after all, they define syllables (cf.
Chapter 5), therefore their affiliation cannot be ambiguous.
Second, recall that /t/ has several allophones in the dialects of English,
e.g., aspirated, glottalized, tapped/flapped. It is tapping in GA or informal-
colloquial British English that we are concerned with here most, since this is
the process which clearly affects ambisyllabic consonants only. While within
words a consonant must be followed by an unstressed vowel to be
ambisyllabic (the /t/ is tapped in átom but not in atómic), across words this
stress-sensitivity ceases to exist, and all word-final consonants followed by
(any!) vowel undergo the process; not only do we find tapping in get alóng,
where the next vowel is unstressed, but in get úp, too. In sum, /t/ has the
following major allophones: plain [t] (e.g., after an /s/), aspirated [tº]
(syllable-initially), glottalized [
1
t] or replaced by a glottal stop [1] (cf.
Chapter 6) (syllable-finally), and replaced by the tap/flap [r] (in ambisyllabic
position). Word-initial /t/ is always aspirated and never tapped, as in a tease;
word-final /t/ is never aspirated but may be tapped, as in at ease. The same
contrast is found in the pair might I (tap) vs. my tie (aspiration). The
expression at all is pronounced differently in the two standard varieties of
English: the expected unaspirated pronunciation of the /t/ is only found in GA
(of course, with tapping: [c'r5:I]), whereas in RP the string is, rather
exceptionally, treated as if it was a single word, just like retúrn, and
consequently aspiration appears: [c'tº5:I]. Let us emphasize that this is an
isolated, irregular example, and word-final plosives in general do not
normally become aspirated, cf. plum pie (aspirated) vs. plump eye
(unaspirated). Similarly, the aspiration-killing effect of a preceding /s/ can
92
Connected speech
only be exerted if the /s/ is in the same syllable as the following plosive: the
/t/ is plain in both stake and mistake but aspirated in miss Tom.
Besides L-darkening and tapping, there is a third rule which applies
across words in the same fashion as word-internally, R-dropping. You may
be able to recall from Chapter 2 the phenomenon called Linking-R, a word-
final <r> which does not undergo R-dropping because the next morpheme
starts with a vowel, which "saves" it. We have also seen that sometimes a
"historically unmotivated" /r/ shows up between two morphemes, an /r/
which is absent from spelling and from the rhotic accents of English. This is
called Intrusive-R. We observe a few interesting facts when we compare
Linking-R and Intrusive-R:
(i) They are phonetically identical.
(ii) Both of them characterize the non-rhotic accents of English only –
linking and intrusion go hand in hand with R-dropping.
(iii) Since a word-final <r> can only be preceded by a broken tense
vowel, a broad lax vowel, or, in unstressed final syllables, a schwa (as the
discussion on the R-influence affecting preceding vowels in Chapter 4
shows), it follows that Linking-R always follows one of /o: 5: a: c/, that is, a
non-high vowel.
(iv) It is a general feature of Intrusive-R in all the non-rhotic accents
exhibiting it that it does not appear in a random fashion, but after certain
vowels only, namely /o: 5: a: c/, that is, after a non-high vowel.
(v) Both Linking-R and Intrusive-R are always sandwiched between
two vowels: they are preceded by a (non-high) vowel and followed by
another vowel in the next morpheme. That is, both always pop up between
vowels in a hiatus (cf. Chapter 3); in fact, they break up, i.e., destroy, the
hiatus.
93
Chapter 7
How can all these five observations be accounted for in the simplest
way? On the one hand, it should be clear that Linking-R and Intrusive-R are
virtually the same: they appear in the same position (intervocalically, after a
non-high vowel), and have the same function (to fill a hiatus). On the other
hand, it should also be clear that the existence of both crucially depends on
the presence of the R-dropping rule. Let us illustrate how Intrusive-R must
have come into being.
Suppose you are a speaker of non-rhotic English. For you, words like
paw and pore, spa and spar, manna and manner, are homophones: /p5:, spo:,
'mænc/, respectively. (You may only face the fact that they are spelt
differently when you start learning to read and write at school. Doesn't this
remind you of the sufferings you underwent in primary school while trying to
memorize that gólya 'stork' is written with <ly> but bója 'buoy' with a <j>,
although both are pronounced with the same sound, /j/?) You also notice that
when words like pore, spar, manner are followed by a vowel-initial element,
an /r/ suddenly appears between them: …pore is… /'p5:riz/, …spar is…
/'spo:riz/, …manner is… /'mæncriz/. You conclude that whenever a word
ends in /o: 5: a: c/, and the next morpheme begins with a vowel, an /r/ is
inserted inbetween. You start treating paw, spa, manna analogously to pore,
spar, manner.
1
pore paw spar spa manner manna
before a pause /p5:/ /p5:/ /spo:/ /spo:/ /'mænc/ /'mænc/
before a cons. /p5:/ /p5:/ /spo:/ /spo:/ /'mænc/ /'mænc/
before a vowel /p5:r/ /p5:r/ /spo:r/ /spo:r/ /'mæncr/ /'mæncr/
1
Recall the discussion of Broadness without r in Chapter 4.
94
Connected speech
Therefore, Linking-R and Intrusive-R are both the manifestations of the same
process of hiatus-filling after word-final non-high vowels, facilitating the
smooth transition between the vowels. Such phenomena are frequently
referred to as liaison, a French noun meaning 'connection, link'.
The question arises what happens in hiatuses when the first member is
not a non-high vowel. Can they be similarly broken up by a hiatus-filler
consonant? The answer is yes, although in such cases it is not a /r/ which is
inserted but a semivowel. If the first vowel is high and front, e.g., /i:/, it is the
yod, e.g. me and you /'mi:'cn'ju:/. If the first vowel is high and back, e.g., /u:/,
it is /w/, e.g., you and me /'ju:"cn'mi:/. Footnote 1 in Chapter 5 mentioned the
close connection between the high vowels and the glides, and now we are
faced with a further example illustrating it. Notice that the choice of the glide
is not random, either: /j/ is coronal, that is, produced by the front surface of
the tongue (just like /i:/), while /w/ is formed in the back of the oral cavity,
being velar (just like /u:/). The major difference between the hiatus-filling
glides and /r/ is that the latter only has this function in non-rhotic accents,
whereas the former characterize all the dialects of English.
After the discussion of what processes affect vowels meeting across
morpheme boundaries in connected speech, let us turn our attention to what
happens to consonants in such situations. There are two basic phenomena
which need mentioning: assimilation and deletion.
Assimilation processes are of several different types. First, as it is
described in Chapter 6, a form of voice assimilation is observable in English,
although it is not obligatory, it is not always complete, and it is more limited
than what we find in Hungarian. The output of voice assimilation in English
is always devoicing, that is, a voiceless consonant affects a voiced one,
irrespective of the relative order of the two. As a result, the direction of the
95
Chapter 7
assimilation is not fixed – it can be either regressive or progressive.
Regressive voice assimilation, whose directionality coincides with
Hungarian, is most likely for fricatives and affricates (therefore it is also
referred to by certain authors as "Fricative Devoicing"), as in his tie [z t],
live show [v j], have to [v t] (or even [I t]). Progressive devoicing is what
poses particular difficulties for Hungarian speakers of English since it
subsumes cases where the opposite happens in Hungarian. Thus in the oft-
cited example matchbox, the /b/ assimilates to the voiceless affricate, yielding
['mætjb oks], rather than the other way round, yielding *['mæo¸boks] or
*['mro¸boks] (this latter is in fact the standard Hungarian pronunciation of
the word). This does not only apply in words within a compound but also
across words within the utterance, e.g., catch Bill is ['kºætj 'b iI] rather than
*['kºæo¸ 'biI]; what’s this is ['wots 'ð is] rather than *['wooz 'ðis]; missed
Jane is ['mist 'o ¸cin] rather than *['mizo 'o¸cin].
Besides voice assimilations, English exhibits a variety of regressive
place assimilations, including the dentalization of alveolar /t d n l/ when
they are followed by dental /0/ or /ð/ (this is indicated by the diacritic [ ],
e.g., Matt thanked [t 0]), the labial assimilation of /t d/ (e.g., eight pence
[p p], blood pudding [b p]), the velar assimilation of /t d/ (e.g., it could [k k],
bad company [g k]), and nasal place assimilation (Green Peace [m p], in
question [n k]). These cases are not problematic to Hungarian speakers as
such processes automatically take place in Hungarian, too. However, there is
a phenomenon which is unattested in Hungarian: (occasional) cross-word
palatalization. This is brought about by a /j/ that follows either an alveolar
fricative (/s/ or /z/) turning it into its postalveolar equivalent (/j/ or /¸/,
respectively), or an alveolar plosive (/t/ or /d/) turning it into a postalveolar
96
Connected speech
affricate (/tj/ or /o¸/, respectively). The expression "occasional" refers to the
fact that this assimilation is optional (characterizing faster, colloquial speech
rather than slow and careful pronunciation) and only applies on certain
restricted occasions. Namely, it produces palato-alveolar /j ¸ tj o¸/ only if a
word ending in one of the alveolar obstruents /s z t o/ and a function word
beginning with /j/ (e.g., you, your, yet, plus a few other common words
including year and usual) are combined. A few examples: this year ['ðij jic]
or ['ðijic], ease your pain ['i:¸ jc 'pºcin] or ['i:¸c 'pºcin], why don't you love
me ['wai 'ocontjo 'I\v mi:], mind your head ['maino¸c 'hco], could you see
['kºoo¸c 'si:]. (For word-internal palatalization, see Chapter 11.)
Assimilations, however, are not the only processes affecting
consonants across word boundaries. Optional consonant deletion is just as
frequent, especially when more than two consonants "pile up". You may have
noticed that such "congestions" get simplified word-internally, as in
Wednesday, handkerchief (cf. the diminutive form hankie), Christmas,
exactly and grandmother (cf. gran, granny) for instance, where the
underlined consonant letters are normally unpronounced. The same happens
across words to a /t/, e.g., Saint Paul, first knight, next day, I don't know, etc.,
or a /d/ between an /n/ and another consonant, e.g., send Jim, rock and roll,
Guns and Roses, find me, etc. In addition, a /h/ is very often silent in the
function words he-him-his, her, have-has-had and so is the voiced dental in
them (for the so-called weak forms of function words, see below), which is
sometimes reflected by spellings like 'im, 'er, 'em, too. Thus the underlined
consonants in I met him, We told her, Who is he?, That'll teach them may
remain unpronounced. This is called Aitch-Dropping, for obvious reasons.
2
2
In fact, aitches are not only dropped at the beginning of unstressed function words but also
word-internally before an unstressed vowel, cf. vehícular with a pronounced /h/ vs. véhicle
without one, or herd with an /h/ vs. shépherd without one. In addition, it is a wide-spread
97
Chapter 7
Finally, it is worthy of mention that the final /v/ of words like give or leave
can also be deleted if they are followed by an unstressed function word (e.g.,
leave me alone; cf. contracted gimme from give me).
Curiously enough, some of the processes mentioned above only apply
when the second of the two words juxtaposed is a function word:
palatalization is possible in miss you but not in, say, miss Yolanda; /v/-
deletion is possible in leave me alone but not in leave Maureen alone.
After the discussion of the cross-word phenomena affecting individual
speech sounds (segments) – that is, segmental phenomena – let us scrutinize
the suprasegmental features of connected speech, i.e., those that
characterize larger strings like syllables or utterances.
One of the most important of such features, intonation, will be
devoted a whole chapter to later (Chapter 10), so here we can concentrate on
the other one: stress. The way stress is placed in English words is dealt with
in Chapters 8-9 – for the time being, suffice it to say that all non-function
words (that is, nouns, verbs, adjectives and adverbs) contain at least one
syllable that constitutes a rhythmic beat (called major stress), which makes
it stronger, more prominent, than the neighbouring syllables. When words are
combined into phrases, usually the (final) major stress of the final element is
even stronger than the others, and when phrases are combined into sentences,
the strongest major stress of the final phrase in the string receives the greatest
emphasis. For example, in the sentence Colourless green ideas sleep
furiously, green is stronger than the first syllable of colourless, the second
syllable of ideas is stronger than green, and the major stress of furiously is
stronger than that of either sleep or ideas. In sum, the strongest phrasal
stress normally falls on the final element. Of course, this generalization can
feature of non-standard pronunciation in all dialects (but, perhaps, more extensively in
England than the US) to "drop one's aitches" at the beginning of content words as well,
yielding 'ouse instead of standard house, for example.
98
Connected speech
be overridden if the speaker wishes to put extra emphasis on another word in
the phrase or sentence for some reason, so colourless may as well become the
most prominent in, e.g., Colourless green ideas sleep furiously and not
coloured ones!
As opposed to phrasal stress, compound stress, i.e., maximal
prominence within a compound word, is normally assigned to the first term,
thus producing a stress pattern which is the mirror image of the usual phrasal
stress pattern. For example, a big wig is simply a wig which is big (it is a
phrase consisting of an adjective and a noun), while a bigwig is an important
person (a compound with its idiomatic meaning). Similarly, a red skin is a
skin which is red, but a redskin is a North American Indian. (Notice how
much this resembles Hungarian examples like eladó lány 'marriageable girl',
which is a phrase, versus eladólány 'salesgirl', which is a compound.) The
rule applies in longer compounds as well, so pet is the strongest term in both
pet shop and pet shop boys. Further details on compound stress are adduced
in the next chapter, when the various degrees of stress are discussed.
The alternation of rhythmic beats and weak (unstressed) syllables
produces the rhythm of speech. A major difference between English and
Hungarian lies in the type of rhythm they exhibit. While in Hungarian each
syllable is pronounced in about the same time, and therefore the basic unit of
speech rhythm is the syllable (this is called syllable-timed rhythm), in
English it is the sound string stretching from one major stressed syllable up to
the next one (the so-called foot), and consequently the time elapsing between
two major stresses is approximately the same. This is traditionally referred to
as stress-timed rhythm. What follows from this is the fact that in English
(and similar languages, but not in Hungarian) rhythmic beats occur at more
or less equal intervals: the greater the number of following unstressed
syllables is, the shorter the stressed vowel and the more compressed the
99
Chapter 7
unstressed syllables become. For illustration, consider the following
sentence:
He | gave a | digital | camera to | George | Clooney for his | birthday
The rhythmic beats are indicated by underscores, and the vertical lines denote
foot boundaries. Due to the stress-timed rhythm of English, the strings
between any two such boundaries are much the same in duration, from which
a number of consequences ensue, e.g., George is pronounced considerably
long, whereas camera tends to get compressed to disyllabic cam'ra (cf.
syncope in Chapter 5) so that camera to fits into more or less the same time
span as, e.g., digital.
Perhaps the most effective strategy whereby syllables can be
"squeezed" is vowel reduction, that is, the replacement of full vowels with
the weak (reduced) vowels /c i o/ (see Chapter 3). In connected speech, this
reduction process characteristically manifests itself in the reduction of
unstressed function words. In the example sentence above none of the
function words (he, a, to, for, his - /hi, c, tc, Ic(r), (h)iz/) contains anything
other than those three vowels. Of course, as we have already seen above, any
word can in principle be stressed in an utterance for special emphasis, and
under such circumstances these function words may contain unreduced
vowels (/'hi:, 'ci, 'tu:, 'I5:, 'hiz/). Let us see the details.
There are roughly forty words in English that have two basic forms:
one which is the usual, unstressed pronunciation (called the weak form –
very often, the same word exhibits several different weak forms), and
another, stressed pronunciation (called the strong form or full form), which
is only used in certain specific situations (see below). The list of the most
common such words is given in tabular form in the Appendix at the end of
100
Connected speech
the chapter. Most of them belong to the closed class of function words
(determiners and pronouns [1-11 in the table in the Appendix], prepositions
[12-17], conjunctions [17-22] and auxiliaries [23-30]), although certain
highly frequent major category words (e.g., the noun saint when part of
compound proper names – [+1]) also show this kind of dual behaviour. It is
possible to use only strong forms in speaking, and some foreigners (including
the typical Hungarian speaker of English) do this, but native speakers find
such "all-strong-form" pronunciations unnatural and foreign-sounding;
moreover, the unnecessary lack of reduction creates the impression of
emphasis, which may even lead to misunderstanding. In addition, it is crucial
for learners of English to be familiar with the use of weak forms or else they
are likely to have difficulty comprehending (native) speakers who do use
them (statistically, as many as 95% of the occurrences of a function word in
native speech are weak).
3
The unstressed, weak forms are normally used sentence-medially
(e.g., It's time to /tc/ go on), and, with the exception of auxiliaries [23-30],
sentence-initially as well (e.g., To /to/ err is human), whereas the strong
forms occur at the end of the sentence (e.g., I can do it if you want me to
/tu:/). It has already been repeated several times that even otherwise
unstressed words can become prominent for purposes of emphasis or
contrast, for example – accordingly, the strong form is chosen when the word
is contrasted or co-ordinated with another one (e.g., Both of them can /'kæn/,
but only Jack will /'wiI/, answer this question or It's at /'æt/ the corner, not on
/'on/ the corner), when it is cited or quoted (e.g., Don't say "but"! /'b\t/), or it
is simply emphasized (You must /'m\st/ hold on! or He does /'o\z/ do the
homework regularly!). In addition, when a preposition is followed by a
3
One exception is singing, in which strong forms are often used in normally unstressed positions, although articles
are generally weak even there.
101
Chapter 7
pronoun at the end of a sentence, usually the strong form of the preposition
(and, of course, also of the pronoun) is used (e.g., I'm looking at you /æt ju:/;
cf. It's at /ct/ the corner).
These general rules, however, have a number of exceptions. First,
object pronouns [7-10] are not normally full even sentence-finally (e.g., Have
you seen them? /ðm/). Second, auxiliary verbs never have the weak
pronunciation in their negative form (i.e., combined with not) – the very
nature of negation involves emphasis (e.g., I can't /'ko:nt/ (or cannot
/'kænot/) dance) and, as it has been mentioned above, usually, though not
always, they have the strong pronunciation at the beginning of the sentence
(Can /kæn/ you dance? as opposed to John can /kn/ dance the tango).
Finally, there are a few function words that have a strong form only, e.g.,
auxiliaries (did, may, might, need), prepositions (in, off, on, up), conjunctions
(though, when), pronouns (that, these, those, who
4
), and the negative particle
not (but it shortens to n't when contracted with certain auxiliaries, e.g., can't,
won't, didn't) (for contraction, see below).
The major characteristics of the pronunciation of the weak forms
are the following:
(1) The vowel reduces to one of the weak vowels, in most cases to /c/.
This is sometimes reflected in non-standard spelling, e.g., the <a> at the end
of wanna (=want to), gotta (=got to), gonna (=going to), kinda (=kind of),
cuppa (=cup of).
(2) Very often, the schwa is able to further reduce to zero, which
sometimes results in Syllabic Consonant Formation (SCF – discussed in
Chapter 5). Some of these vowelless pronunciations (n't, 's, 'd, 've, etc.) are
able to undergo contraction – that is, auxiliaries and the like attach to an
4
When it is interrogative (e.g., Who is it?). For the relative pronoun (e.g., the man who sold the world), there exists
an occasional weak form /ho/.
102
Connected speech
adjacent word. Contraction can also affect certain other words, cf. wanna,
gotta, etc. in (1) above. Rather exceptionally, it is possible to contract the
object pronoun us in imperative let's. Recall from Chapter 6 the rules of the
voice assimilation of the -s suffix, and note that contracted 's, irrespective of
what function word it is a contraction of, conforms to them as well.
(3) The consonants surrounding the vowel also become weak, and
delete easily, so we find a number of examples of optional consonant
deletion, discussed above, among weak-form words. Especially word-initial
/h/ is targeted by such deletions, as it was already mentioned, sometimes
traceable in non-standard spellings like should of been for should have been.
However, when /h/-initial weak-form words occur at the beginning of a
sentence, the pronunciation is always with /h/.
(4) Certain weak-form words are pronounced differently before
consonant- and vowel-initial words, including a(n), the, do, to, you. This is
because in English schwa cannot normally occur before another vowel, so
some other pronunciation (an extra /n/ in an, or final /i/ or /o/ in the other
cases) is chosen to avoid that situation. Also, remember that in the non-rhotic
accents like RP a word-final /r/ is only pronounced when followed by a
vowel-initial morpheme – such potential Linking-R's are given in brackets in
the table. Keep in mind, though, that all those /r/'s are obligatory in all
positions in the rhotic accents like GA.
(5) Weak forms, particularly those of prepositions and pronouns,
typically lose their independent word status in connected speech, which is
evident from phonological processes such as word-initial tapping and aitch-
dropping. The initial /t/ of the unstressed preposition to is weak and
frequently tapped in the relevant dialects in a phrase like lie to me /'Iairc'mi:/
(analogously to a single word like lighter /'Iairc(r)/), and the initial /h/ of the
103
Chapter 7
unstressed personal pronoun him can be deleted in beat him (i.e., beat 'im),
similarly to vehicle or shepherd discussed in footnote 1 above. We are led to
the conclusion that all of lie to, lighter, beat 'im, vehicle and shepherd
undergo phonological rules in the same way because, as far as pronunciation
is concerned, they all constitute single words.
The table in the Appendix summarizes the most common weak forms,
contrasting them with the corresponding strong forms. Most examples
equally apply to RP and GA, although sometimes GA supplies additional
possibilities. All such differences are indicated in the table. A closer
examination of the weak forms leads to the observation that some of them are
ambiguous, so their meaning only becomes clear from the context (e.g., /cv/
can correspond to either of or have). All further comments, which do not fit
into a chart, including weak forms only used in certain meanings, are given
as footnotes.
To sum up the discussion of this chapter, let us emphasize that a well-
definable group of phonological processes (L-darkening, tapping and other
/t/-phenomena, R-dropping) apply within and across words in a uniform
fashion. Moreover, Linking-R and Intrusive-R can be proven to be two forms
of virtually the same object, a hiatus-filler inserted after non-high vowels,
and as such, they properly fit into the general picture of hiatus resolution. We
have also seen how other processes like assimilation and consonant deletion
are present in connected speech, and how the reduction of unstressed function
words contributes to the isochronous stress pattern and rhythm of English
utterances. The following two chapters take the stress pattern of individual
words under scrutiny.
104
Connected speech
Appendix
Word Strong
form
Examples Weak
form(s)
Examples
1. the ði:
2. a, an ci, æn
It's not "a" cat, it's
"the" cat!
ðc, ði the /ðc/ dog, the /ði/ end
c, (c)n a dog, an end
3. some
5
s\m I'll get you some. s(c)m I'll get you some apples.
4. his
6
hiz It's his car, not mine. (h)iz what's-his-name
5. your =
you're
j5:(r), joc(r) Is this YOUR CV? jc(r) Mind your head!
6. (s)he,
we,
you
hi:, ji:, wi:,
ju:
All I want is YOU. (h)i, ji, wi
jo (GA
also jc)
I'll get you some apples.
I gotcha!
7. him him
8. her ha:(r)
Whom do you love:
him or her?
(h)im I love him.
(h)c(r),
a:(r)
I love her.
9. their
them
ðcc(r)
ðcm
10. us \s
It wasn't US, it was
THEM.
--
7
ð(c)m Do you hate them?
cs one of us is crying
11. there
8
ðcc(r) There you are! ðc(r)
(GA also
ðr)
There's a book on the
table.
12. at æt What's he getting
at?
ct Look at me.
13. for I5:(r) It's just what I long
for.
Ic(r), Ir, I
9
Stay for a week.
14. from Irom
(GA Ir\m)
Where are you from? Ircm He's from Barcelona.
15. of ov (GA \v) It's love I've a lot of. cv
10
one of us
5
This word can reduce when it is a neutral quantifier (e.g., There's some milk in the fridge),
but not in other senses, e.g., when it is contrasted (e.g., Some students know this but others
don't).
6
This only applies to the possessive determiner (e.g., This is his car). When his is a pronoun
(e.g., This car is his), it always has the strong form.
7
In GA, there is a weak form /ðcr/, which is used in RP only occasionally.
8
When this word is a demonstrative element (opposite of here), it is a (stressed) adverb and
therefore it occurs in its strong form only. Also, cf. their above.
9
In both RP and GA, the occasional weak form /Ir/ is only used before weak vowels, e.g.,
stay for a week /'stci Irc 'wi:k/. The weak form /I/ is rare and only appears in very casual or
rapid speech.
10
There is also an informal rapid-speech or non-standard pronunciation, used before
consonants only, /c/, sometimes spelt o' (as in standardized o'clock). Also, compare of and
105
Chapter 7
16. to
11
tu: Who did you give it
to?
tc, to to /tc/ me, to /to/ Ann
17. than
12
ðæn "Than" is spelt with
an "a" not an "e".
ð(c)n even better than
the real thing
18. and æno "And" is a
conjunction.
(c)n(o)
13
Twist and shout!
19. but b\t Don't say "but"! bct sad but true
20. that
14
ðæt What's that? ðct the book that we bought
21. or 5:(r) To be or not to be? c(r)
15
sooner or later
22. as æz as and when cz as good as it gets
23. have
has
had
hæv
hæz
hæo
Have you seen her?
Had I known him
earlier...!
(h)cv, v
(h)cz, z, s
(h)co, o
You've got to know.
She's got it. It's been a
year. You'd better stop!
24. can
could
kæn
koo
Can you dance?
Yes, you could.
k(c)n
kco
I can see.
You could be mine.
25. will
would
wiI
woo
Will Susan be there?
Would you like it?
(w)(c)I
(w)(c)o
Susan will be at home.
I'd rather sail away.
26. shall
should
jæI
joo
Shall I open the
window?
j(c)I
jco
I think you should work
harder.
27. must
16
m\st You MUST hold on! mcs(t) I must go now.
28. do
does
ou:
o\z
How do you do?
Yes, she does!
oo, o(c)
o(c)z
How do you do?
What does he do?
29. am,
are
was,
were
17
æm, o:(r)
woz
(GA w\z),
wa:(r)
I AM hungry!
He said he wasn't
sleepy but he was!
(c)m, c(r)
wcz,
wc(r)
(GA also
wr)
I'm hungry.
They were all drinking
in the pub.
off: the latter has no weak form, and is pronounced /oI/.
11
The preposition and the infinitival particle exhibit the same behaviour.
12
This word is either used as a preposition (e.g., He's older than me) or a conjunction (e.g.,
He's older than I thought), but it is not to be confused with the adverb then, which only has a
strong form /ðcn/.
13
The weak form /@nd/ is slightly more formal than /@n/.
14
This word only has a weak form when used as a conjunction (e.g., I know that you know it;
the book that we bought); when it is a demonstrative determiner (e.g., Who's that girl?) or
pronoun (e.g., What's that?), or a degree word (e.g., Not that bad) it is always pronounced in
its strong form.
15
This is an occasional weak form in RP, only used between numbers and in fixed phrases. In
GA, however, this reduction is quite common.
16
When expressing probability (e.g., [doorbell rings] This must be the milkman), this word is
less likely to appear in its weak form than when it is used in the sense of obligation (e.g., You
must try harder).
17
The verb be always behaves like an auxiliary verb, even when it is the only verb in the
sentence. However, its forms are always strong in three-word wh-questions containing a
personal pronoun, e.g., Who is it?, How are you?, Where were they?
106
Connected speech
30. been bi:n
(GA bin)
Where have you
been?
bin I've been busy all day.
+1. Saint scint He's a saint. s(c)n(t) Saint Paul's Cathedral
107
8. Word stress – Part 1: The degrees of stress
Before you study this chapter, check whether you are familiar with the
following terms: ambisyllabic, aspirated, CiV, closed syllable, derivation,
diacritic, foot, free variation, full vowel, function word, IPA, major stress,
monomorphemic, morpheme, morphology, productive suffix, pulmonic,
reduced vowel, syllabic consonant, syllable peak, tapping/flapping,
Trisyllabic Laxness, utterance, vocal cords
As it was already mentioned in Chapter 7, stress is one of the
suprasegmental (or prosodic) features of speech, which extend over more
than one sound segment. They include variations in pitch, loudness, tempo
and rhythm, out of which pitch and loudness play the most significant role in
the stress system of English.
Pitch roughly corresponds to the acoustic feature of frequency, the
rate of vibration of the vocal cords, which is produced by their stretching and
tensing: the tenser they are, the higher the rate of vibration, and the higher the
pitch. The distinctive use of patterns of pitch is called intonation, whose
most important function is to signal grammatical structure (e.g., clause
boundaries within sentences, and the different sentence types, especially
questions vs. statements), similarly to punctuation in writing. In Hungarian,
for example, intonation plays a pivotal role in the distinction between
segmentally identical statements and yes-no questions like Jani elment
'Johnny has left' vs. Jani elment? 'Has Johnny left?'. Chapter 10 is devoted to
intonation in English.
Certain languages, but neither English nor Hungarian, use pitch to
contrast not sentences but words, thus pitch becomes an essential feature of
Word stress – Part 1
the meaning of morphemes. This phenomenon is called tone, and such
languages are called tone languages. Many of the languages of South-East
Asia and sub-Saharan Africa, e.g., Beijing Mandarin Chinese and Thai in
Asia or Hausa in Africa, belong here.
Besides pitch, loudness is the other major ingredient of stress
prominence in English. The loudness of (strings of) speech sounds depends
on the size of the vibrations of the vocal cords caused by the varying degrees
of pulmonic air pressure. Together with pitch level and vowel quality,
loudness produces the relative prominence of syllables called stress. It is of
crucial importance to understand that stress is not an absolute feature of
syllables but rather it is relative, only relevant in comparison of several
syllables. It is possible to say that a syllable is stressed, but this always means
that it is more stressed (=stronger, more prominent) than the adjacent
syllable(s). Due to the fact that stress is an extremely complex phenomenon
(governed by a number of different factors) and the fact that it is relative,
there exist several degrees of stress, out of which four are linguistically
relevant in English. In fact, these various degrees come into being owing to
the unequal role played by pitch, rhythmic prominence (already mentioned in
Chapter 7), and the full or reduced quality of the syllable peak.
Recall from the previous chapter that all non-function words in
English contain at least one syllable that constitutes a rhythmic beat (called
major stress) – at the same time, function words are normally unstressed.
The primary source of this rhythmic prominence of major stress is the
loudness of the syllable, but the difference in pitch level causes a difference
between two types of major stresses. In suprasegmental or syllabification, for
example, there are two rhythmic beats (underlined), but one of them, namely
the second one, is more prominent owing to its highest pitch in the word. In
addition, only this syllable can carry the main stress of an utterance, e.g., Are
109
Chapter 8
these features suprasegmental? or This is the correct syllabification (cf. the
discussion of phrasal stress in Chapter 7). It is traditionally called primary
stress or main stress, for obvious reasons, while the other type of major
stress is usually referred to as secondary stress. Secondary stress is optional,
basically it only appears in longer English words under very specific
circumstances (see below in more detail). For example, the first syllable of
the word suprasegmental and the second syllable of syllabification are
secondary stressed. Another basic difference between primary and secondary
stress is that while the former can only appear once in a word (this is logical,
since it is, by definition, the most prominent syllable), there may be several
occurrences of secondary stress, depending on the length of the word. For
instance, the word contamination contains one such syllable (underlined),
whereas decontamination already contains two.
1
Syllables without rhythmic prominence also fall into two subtypes. In
most such cases, the whole syllable becomes weak and reduced, which means
that, on the one hand, the vowel is not full but one of /c i o/ – most
frequently, schwa. It is in these cases that Syllabic Consonant Formation
(discussed in Chapter 5) is possible. On the other hand, the consonants
surrounding this weak peak also become unstable, especially the consonant
preceding it. So much so that /h/, for example, systematically disappears
altogether (recall the examples vehicle /'vi:ckI/ and shepherd /'jcpco/ of
Chapter 7, but vehement /'vi:cmcnt/, annihilate /c'naicIcit/, Buddha /'booc/,
Birmingham /'ba:mincm/, etc. are analogous), and even if a consonant
remains pronounced in such a position, its syllabic status is vague, that is, the
consonant is ambisyllabic (cf. Chapter 2), with all the consequences of this.
1
Based on the observation that out of two (or more) successive secondary-stressed syllables
the first one is always slightly stronger than the other(s), some authors apply the term
"secondary stress" to that one only and refer to the others as "tertiary-stressed".
110
Word stress – Part 1
Such syllables are zero-stressed or completely unstressed. However, some
otherwise weak syllables contain an unreduced vowel, that is, under certain
(not exactly straightforward) circumstances the expected vowel reduction
fails to take place, as in the first syllable of activity /æk'tivcti/. This full-
vowelled, rhythmically or pitch-wise non-prominent stress is called tertiary
stress in this book. An alternative name is minor stress (as opposed to major
stress). Although such syllables are not prominent as far as suprasegmental
features go, still they are stronger than completely unstressed syllables in the
sense that they are characterized by neither vowel reduction nor consonant
weakening, the two elementary features of zero stress mentioned above.
Compare the final syllable of Abraham /'cibrchæm/ and Graham /'grcicm/ –
in the former the vowel is full and the /h/ is pronounced (this is what we call
tertiary stress), whereas in the latter the vowel is a schwa and the /h/ is
dropped (this is what we call zero stress). Compare the underlined /t/ in
hesitate, which is strong and therefore aspirated, with that of activity or
better, which is not – rather, it is tapped in the tapping dialects of English (as
an indication of its ambisyllabicity), yielding [æ
1
k'tºivcri] and ['b crc(r)].
The four degrees of word stress are summarized in the following
chart. As the shaded areas show, the basic difference between unstressed and
stressed syllables lies in the presence vs. absence of vowel reduction,
respectively, while the major stress – minor stress distinction is based on
loudness (rhythmic prominence).
111
Chapter 8
Stress
category
MAJOR MINOR UNSTRESSED
Stress degree primary secondary tertiary zero
Prominence full vowel full vowel full vowel --
loudness loudness
highest pitch
Examples suprasegmental
syllabification
annihilate
hesitate
Japan
suprasegmental
syllabification
hesitation
grammaticality
Japanese
suprasegmental
syllabify
annihilate
hesitate
activity
suprasegmental
syllabification
annihilate
grammaticality
Japan
There are three equivalent stress-marking conventions in phonology: the
use of numbers, diacritics, and IPA stress marks. In this book, we only use
their most widely accepted forms, which are shown in the table below. In the
IPA, the upper mark / ' / is used for primary stress, and the lower mark / ¡ / for
secondary stress. Sometimes the segments are not transcribed but rather the
spelt form of the word is supplemented by diacritics on top of the stressed
vowel letters: the acute accent (e.g., ó) signals primary stress, and the grave
accent (e.g., ò) secondary stress. Finally, the stress degrees of the syllables in
a word can be referred to with numbers, 1 standing for primary, 2 for
secondary, 3 for tertiary, and 0 for zero.
Stress
category
MAJOR MINOR UNSTRESSED
Stress degree primary secondary tertiary zero
Numbers 1 2 3 0
Diacritics acute accent grave accent - -
IPA stress
marks
upper mark lower mark - -
112
Word stress – Part 1
Accordingly, the stress pattern of suprasegmental can be indicated with
numbers as 20310, with accents as sùprasegméntal, or, accompanying an IPA
transcription, as /¡su:prcscg'mcntI/.
On the basis of the examples above, the careful reader must have
already noticed some of the general properties of English word stress. First,
no major stress occurs after the primary stressed syllable (i.e., secondary
stress always precedes primary stress). It follows that primary stress is always
the rightmost major stress, i.e., the last rhythmic beat is the strongest. This
prominence of the right edge is usually explained by the directionality of
primary stress placement: it is supposed to proceed from right to left, docking
onto the first potential site available (see the next chapter).
Second, there are no English words starting with two successive zero-
or tertiary stressed syllables – one of the first two syllables of a word must be
rhythmically prominent (i.e., primary or secondary stressed). This is the
prominence of the left edge, or, as sometimes it is referred to, the Early
Stress Requirement. Notice that the prominence of the right edge and the
prominence of the left edge are in potential conflict in longer words: in a
five-syllable word, for instance, where primary stress falls on the second-last
syllable, there remain three more syllables to the left, which cannot all be
unstressed. Consequently, either the first one (as in sùprasegméntal) or the
second one (as in contàminátion) will necessarily receive secondary stress. In
fact, this is the reason why secondary stresses are created: to produce a more
or less regular alternation of stressed and unstressed syllables, e.g., 20310 in
suprasegmental or 02010 in contamination. This tendency in English for a
regular iambic rhythm (that is, speech rhythm with metrical feet consisting
of one unstressed syllable followed by one stressed syllable) also manifests
itself in the dispreference of adjacent major stresses. Such situations, called
stress clashes, tend to be avoided: in most cases (as in the examples above),
113
Chapter 8
there is at least one zero or tertiary stressed syllable between any two primary
or secondary stresses.
The basic principles of the English stress system, discussed above,
determine the regularities of stress placement. Primary stress is dealt with in
the next chapter – here we turn to secondary stress assignment. It has been
mentioned that the "war" of the two word edges is the primary motivation for
the creation of secondary stresses: recall the Early Stress Requirement, as a
result of which in longer words, if primary stress falls on the third (or a later)
syllable, the first or the second syllable must be assigned secondary stress.
Monomorphemic English words tend to be rather short, so there are just a
few examples (including a number of place names) for underived secondary
stress; in most such words (e.g., Àbergavénny, àbracadábra, àgoraphóbia,
àlumínium, Àpalàchicóla, Kàlamazóo, màcaróni, pàraphernália,
sànatórium, Wìnnepesáukee) the first syllable receives secondary stress
irrespective of the number of syllables before the primary stress – if there are
more than three, as in Apalachicola, additional secondary stresses are created.
The problem, however, mainly arises in derived words. What usually
happens in such cases is that since suffixation has made the word longer,
primary stress shifts to the right, and the original primary stress reduces to
secondary. Bear in mind that the rightmost rhythmic beat is the strongest! As
such derived words preserve the rhythmic prominence of the original stress
pattern, this secondary stress is frequently referred to as Derivational
Secondary Stress. In fíction, for example, primary stress falls on the first
syllable, which reduces to secondary stress when fìctionéer is derived. The
following examples illustrate the same mechanism: adáptable – adàptabílity,
éducate – èducátion, impréssion – imprèssionístic, irrégular – irrègulárity,
jústify – jùstificátion, órchid – òrchidáceous, perípheral – perìpherálity. Here
again the number of syllables before the primary stress does not matter. If a
114
Word stress – Part 1
suffix is attached to a long word which already contains a secondary stress,
further secondary stresses are brought about, cf. ìndivídual and ìndivìduálity,
còmprehénsible and còmprehènsibílity. Whole chains of derivation illuminate
how former primary stresses turn into secondary: dífferent – dìfferéntiate –
dìfferèntiátion, ínstitute – ìnstitútion – ìnstitùtionalizátion.
There is one situation, however, in which Derivational Secondary
Stress is blocked: when it would result in stress clash. The pressure to avoid
adjacent major stresses and therefore maintain a (near-)iambic rhythm leads
to one of two possible solutions: the original primary stress reduces to tertiary
or zero, either with a secondary stress appearing to the left (this is called
Iambic Secondary Stress), or with no (new) secondary stress at all, the
original major stress being deleted and lost (Major Stress Deletion). All in
all, the output of both strategies is a stress pattern with the stresses evenly
distributed. Here are a couple of examples – the previously primary stressed
syllables are underlined, and their vowels are indicated with IPA symbols:
Iambic Secondary Stress Major Stress Deletion
adápt – àdaptátion /æ/
doméstic – dòmestícity /c/
enígma – ènigmátic /i/
fragmént (verb) – fràgmentátion /c/
horízon – hòrizóntal /i/
Japán – Jàpanése /c/
refórm – rèformátion /c/
transpórt – trànsportátion /5:/
áctive – actívity /æ/
ànthropólogy – ànthropológical /c/
átom – atómic /c/
clímate – climátic /ai/
Gérman – Germánic /a:/
frágile – fragílity /c/
víctory – victórious /i/
vírgin – Virgínia /c/
115
Chapter 8
Since the primary stress is placed in different ways in the case of different
suffixes (see below and in the next chapter), the same word can undergo
Derivational Secondary Stress formation in one case and Iambic Secondary
Stress formation or Major Stress Deletion in another, e.g., for córrelate, the
original major stress is preserved in còrrelátion but deleted in corrélative.
Another manifestation of the tendency to maintain iambic rhythm and
avoid stress clash characterizes connected speech. When a word with a
secondary and a primary stress (e.g., thìrtéen) forms a phrase with another
one (e.g., mén), based on the discussion of phrasal stress (Chapter 7) we
expect the final stress to be the strongest, while all the others are expected to
reduce their stress degree by one, that is, something like thirtèen mén, with a
321 stress pattern. Instead, what normally happens in English is that the stress
levels "switch round" in the first element of the phrase, the result being
thìrteen mén, i.e., 231, where the intervening tertiary stress (formerly the
primary stress – underlined) separates the major stresses. This phenomenon
has been widely studied and therefore has a whole range of names, e.g.,
stress shift, iambic reversal, or the rhythm rule, all of which highlight one
or another feature of the process: stress degrees are shifted to move rhythmic
beats away from each other and thus facilitate the iambic rhythm of the
phrase. Some linguists dub it the thirteen men rule, after this very frequent
example. It is important to keep in mind, though, that it does not only take
place in thirteen men, but occurs automatically in all phrases where the first
element has at least one secondary stress, e.g., àchromatic léns, àcademic
wríting, àfternoon téa, Chìnese chéckers, fùndamental fréquency,
ìnternational láw, Jàpanese lánguage, nèolithic víllage, Tènnessee Válley, or,
for some speakers, the Bèrlin Wáll or ìdeal pártners.
From the discussion of secondary stress, it should be clear that after
certain suffixes have been attached to a word, the original stress pattern may
116
Word stress – Part 1
change, as in átom – atómic; moreover, this is the primary source for the
creation of secondary stress, as in décorate – dècorátion. Obviously,
morphology plays a crucial part in the English stress system. Nevertheless, it
is necessary to distinguish between two types of morphological operation.
Consider the following examples:
éducate – éducating – èducátion
adápt – adápted – àdaptátion
díagnose – díagnoses – dìagnóstic
jóurnal – jóurnalist – jòurnalése
áutumn – áutumn-like – autúmnal
As you can see, when a new word is formed out of a base word, the original
stress pattern may or may not be preserved. In educate, the first syllable is
primary stressed, and so is it in the -ing form, whereas in the -ion form it
reduces to secondary and a different syllable receives the primary stress.
Therefore, we are forced to break down the family of suffixes into two
classes. Certain suffixes, e.g., -ing, -ed, -s, -ist, and -like above, are unable to
affect the stress pattern of the word they are part of – they are stress-neutral.
Most of them are of Germanic origin. Curiously enough, the list of these
suffixes coincides with the type referred to in Chapter 3 as productive.
Others, like -ion, -ic, -ese, and -al, systematically change the place and/or the
degree of the stresses because they require primary stress to fall on a specific
syllable – they are non-neutral or stress-fixing. Most of them are of Latin
origin (they are Latinate). Curiously enough, the list of these suffixes
coincides with the type referred to in Chapter 3 as non-productive. Notice
that at this point we are able to make a generalization: regular, productive
suffixes, which do not count in, e.g., Trisyllabic Laxness (recall examples
117
Chapter 8
like lazy-laziness), are stress-neutral, i.e., do not count in stress placement,
either. Non-productive suffixes, on the other hand, do count in both
Trisyllabic Laxness (recall vain-vanity) and stress assignment – they are
stress-fixing. This is a curious interplay between word structure and sound
pattern: suffixes seem to exhibit consistent behaviour in phonology.
The most common productive (stress-neutral) and non-productive
(stress-fixing) suffixes are illustrated below.
(1) Stress-neutral suffixes
Suffix Examples
-able consíder – consíderable, avóid – unavóidable
-dom mártyr – mártyrdom, tòpsy-túrvy – tòpsy-túrvydom
-ed adápt – adápted, édit – édited
-er
2
cómmon – cómmoner, advénture – advénturer
-ful bárrow – bárrowful, béauty – béautiful
-hood bróther – brótherhood, ádult – ádulthood
-ing éducate – éducating, ínterest – ínteresting
-ish (adj) ánimal – ánimalish, fórty – fórtyish
-ism álcohol – álcoholism, fanátic – fanáticism
-ist jóurnal – jóurnalist, phýsics – phýsicist
-less bóttom – bóttomless, defénce – defénceless
-like áutumn – áutumn-like, búsiness – búsinesslike
-ly cáreless – cárelessly, appárent – appárently
-ment devélop – devélopment, accómpany – accómpaniment
-ness cáreless – cárelessness, lùkewárm – lùkewármness
-s díagnose – díagnoses, ímage – ímages
-ship cénsor – cénsorship, dictátor - dictátorship
-some advénture – advénturesome, quárrel – quárrelsome
-wise óther – ótherwise, córner – córnerwise
(2) Stress-fixing suffixes and endings
Some of these are not clearly isolatable suffixes (perhaps not even
morphemes) but rather simple word endings which are present in recurrent
stress patterns. They fall into various subclasses, two of which are introduced
2
This suffix either forms a comparative adjective (like commoner) or an agentive noun (like
adventurer) – in both cases it behaves in the same fashion.
118
Word stress – Part 1
presently (and a third one in the next chapter). First, auto-stressed (or self-
stressed) suffixes and endings are primary stressed themselves.
Suffix/ending Examples
-ade lémon – lèmonáde, bàrricáde, cánnon – cànnonáde
-aire míllion – mìllionáire, quéstion – quèstionnáire
-ee réfuge – rèfugée, tráin – trainée
-een vélvet – vèlvetéen, séven – sèventéen
-eer/ier éngine – ènginéer, bombárd – bòmbardíer
-elle mozélle, nacélle
-enne comédian – comèdiénne, Cayénne
-esce àcquiésce, èffervésce
-ese jóurnal – jòurnalése, Japán – Jàpanése
-esque Róman – Ròmanésque, pícture – pìcturésque
-esse largésse, noblésse
-ette cigár – cìgarétte, cassétte
-eur/euse èntreprenéur, masséuse
-ine cuisíne, ravíne
-ique antíque, critíque, techníque, uníque
-itis lárynx – làryngítis, appéndix - appèndicítis
-oo/oon kàngaróo, cartóon
Second, the so-called pre-stressed suffixes and endings require primary
stress to fall on the syllable which immediately precedes them in the word.
For example, -ic is a typical (and very frequent) pre-stressed suffix: while in
díagnose the first syllable is primary stressed, in dìagnóstic it is the third
one – right before -ic itself. The same happens in acádemy – àcadémic and
átom – atómic. Some of these suffixes and endings are monosyllabic (e.g.,
-ic, -ics, -id, -ish (v/n)
3
), some are disyllabic (e.g., -ify/efy, -itude, -ity/ety,
-itive, -ible, -ular, -ulous), some contain the CiV configuration introduced in
Chapter 3, or the similar CuV sequence (e.g., -ion, -ial/ual, -ious/uous, -ian,
-uant).
3
Note that adjectival -ish is a stress-neutral suffix, and as such, is listed in the first chart
above.
119
Chapter 8
Suffix/ending Examples
-ial/ual tútor – tutórial, cóntext – contéxtual
-ian/ean Húngary – Hungárian, líbrary – librárian,
Cáesar – Caesárean, crustácean
-ible deléte – indélible, incrédible
-ic dynámic, ecónomy – èconómic
-icide ínsect – insécticide
-ics ecónomy – èconómics, ácrobat – àcrobátics
-id intrépid, insípid, pellúcid
-ify/efy ácid – acídify, exémplify
-ion opínion, sólve – solútion, éducate – èducátion,
adápt – àdaptátion
-ious/
-eous/uous
céremony – cèremónious,
órchid – òrchidáceous, innócuous
-ish (v/n) abólish, demólish, dimínish, estáblish
-itive compétitive, infínitive, ìntuítion – intúitive
-itude exáctitude, símilar – simílitude
-ity/ety compléxity, socíety, perípheral – perìpherálity,
ánxious – anxíety
-meter spéed – speedómeter, thermómeter
-uant contínuant
-ular mólecule – molécular, mándible – mandíbular
-ulous míracle – miráculous, metículous, rídicule – ridículous
After the story of secondary stress and the effect morphological structure has
on stress placement, let us mention tertiary stress briefly. Recall that tertiary
stress is in fact the prominence caused by the absence of vowel reduction.
Why certain otherwise unstressed vowels fail to reduce to /c/, /i/ or /o/ is
difficult – if not impossible – to explain: it appears to be quite irregular and
mostly unpredictable, although a number of tendencies are observable. For
example, the syllable whose vowel refuses to reduce is very often a closed
syllable (cf. Chapter 5) (e.g., actívity) or the vowel is long, either a long
monophthong (e.g., Germánic) or a diphthong (e.g., climátic). Unfortunately,
120
Word stress – Part 1
this does not mean that all such vowels are protected from reduction (cf.
infórm – ìnformátion /c/; fragmént – fràgmentátion /c/ versus condémn –
còndemnátion /c/). Word frequency may also influence this: the more
frequently a speaker uses a word, the more likely vowel reduction is. For
instance, the musical instrument trombone is usually pronounced
/trom'bcon/, the musicians who play the trombone, however, tend to have a
schwa in the first syllable (/trcm'bcon/).
There are only a few cases where tertiary stress appears
systematically. One is the so-called Alternating Stress Rule, which is dealt
with in Chapter 9, and which accounts for the 103 stress pattern of verbs like
dédicate and certain adjectives and nouns like ábsolute or húrricane. Another
situation when tertiary stress is expected is compound stress. Chapter 7
explains that primary stress in a compound word normally falls on the first
term. Logically, this is accompanied by reduction in the other term(s),
namely, they lose their original rhythmic prominence but retain their full
vowel. For example, when bláck and bóard, two separate words with their
obligatory primary stress (neither of them is a function word!), are combined,
board ceases to be major stressed but its long vowel /5:/ survives in
bláckboard /'bIækb5:o/. Therefore, its stress pattern is 13. The same applies
to ráinbow /'rcinbco/, lífestyle /'IaiIstaiI/, and súperman /'su:pcmæn/. This is
in sharp contrast with what we usually observe in underived words like
bládder /'bIæoc/ or blánket /'bIænkit/, or in words containing suffixes (other
than auto-stressed ones, of course) like blácking /'bIækin/ or blábber /'bIæbc/
– all exhibiting 10. Interestingly, a number of historical compounds have by
now given up their complex morphological structure and are pronounced
according to the regularities of simple words. The word cupboard, for
instance, only means the piece of furniture if pronounced with considerable
121
Chapter 8
vowel (and consonant) reduction /'k\bco/ (its stress pattern is 10, similarly to
cumber, cupper, cupping or cupful) - a /'k\pb5:o/ (with a 13 stress pattern) is
simply a board with cups. Original sheep /ji:p/ plus herd /ha:o/ has become
shepherd /'jcpco/, post /pcost/ plus man /mæn/ is postman /'pcosmcn/, black
/bIæk/ plus berry /'bcri/ is blackberry /'bIækb(c)ri/. Forehead has two
alternative pronunciations: one which follows the rules for compounds
/'I5:hco/, and another with a reduced second term /'Iorio/. In sum, the
morphological structure of a word is clearly reflected in its pronunciation:
only constructs with a primary stress and a tertiary stress are real compounds.
Let us conclude this chapter with a remark concerning the fact that,
unfortunately, most of the stress rules introduced above have exceptions.
Stress clash does occur, although only in a handful of words like sàrdíne,
thìrtéen or Chìnése. Derivational Secondary Stress can override the desired
iambic rhythm, too, as in eléctric – elèctrícity. Exceptions also exist to the
stress-fixing mechanism of suffixes, e.g., Árabic, ársenic, cátholic, chóleric,
lúnatic, pólitics, impóverish. In addition, the picture is further complicated by
the free variation of zero and tertiary stress in words like direct /oi'rckt/~
/oai'rckt/ as well as occasional dialectal differences between RP and GA, e.g.,
address (n) RP /c'orcs/ vs. GA /'æorcs/, advertisement RP /co'va:tismcnt/ vs.
GA /¡æovcr'taizmcnt/ (or /'æovcrtaizmcnt/), or words ending in -ary and
-ory like January RP /'o¸ænjocri/ vs. GA /'o¸ænjocri/ or laboratory
RP /Ic'borctri/ vs. GA /'Iæbrct5:ri/. The next chapter, on primary stress, will
face even more exceptions and subregularities.
122
9. Word stress – Part 2: Primary stress
Before you study this chapter, check whether you are familiar with the
following terms: allophonic rules, auto-stressed ending, clear-L, compound,
dark-L, Early Stress Requirement, function word, Iambic Secondary Stress,
neutral suffix, non-neutral suffix, peak, prefix, pre-stressed, primary stress,
prominence of the right edge, rhyme, secondary stress, self-stressed, stem,
suffix, tapping, tertiary stress.
This chapter deals with primary stress assignment in underived verbs, nouns,
and adjectives. Adverbs usually follow the rules for adjectives, and most of
them are formed with a derivational suffix (e.g., extréme – extrémely), so
they are not treated separately. The remaining word classes belong to the
category of function words, which was discussed in Chapter 7.
Let us start with the differences between the English and the
Hungarian stress system, some of which are also mentioned in Chapters 7
and 8 above. As opposed to Hungarian, where the first, i.e., leftmost, syllable
is stressed in all words, primary stress can fall on virtually any of the
syllables in English. What is more, according to the prominence of the right
edge, English primary stress prefers the right edge of the word in the sense
that in unsuffixed forms the strongest stress is not normally placed more than
three syllables away from the end, irrespective of the length of the word.
Another basic difference between the two languages lies in the
information types determining stress placement. On the one hand, Chapter
8 elaborated on the role of morphology in English – something unheard of in
Hungarian with respect to stress placement. On the other hand, the syntactic
class of the word also plays a role in English: function words behave
Chapter 9
differently from non-function words, what is more, nouns and verbs are
shown below to have established two distinct patterns. In addition, as has
been suggested before, each regularity has a considerable number of
exceptions – the stress pattern of these words has to be simply memorized
since it is unpredictable: it is an idiosyncratic feature of the lexical item.
Even phonology makes a much more complex contribution to stress
rules. While in Hungarian the only piece of phonological information
required for stress placement is the position of the syllable (the first syllable
in the word gets stressed automatically), in English not only is its position
relevant but so is its structure. More specifically, the English stress system is
quantity-sensitive: it is strongly influenced by the amount of material found
in syllables. In this respect, there are two basic syllable types: light and
heavy syllables. A syllable is light if it contains a short vowel and is not
closed by a consonant; all the other possibilities (with a long vowel – a long
monophthong or a diphthong – and/or with (a) final consonant(s)) make the
syllable heavy. Crucially, the consonant(s) preceding the peak do(es) not
count: a syllable like /ti/ is light in the same way as /tri/ or /stri/ or /i/, while
/it/ is heavy even though altogether it consists of fewer elements than /stri/.
Similarly, a syllable like /ci/ or /cit/ or /tci/ is characterized by one type of
behaviour (it is heavy) as opposed to, say, /brc/. Therefore, we can state that
although this phenomenon is traditionally referred to as syllable weight, in
fact, it is only governed by the number of elements in and following the peak
– the portion of the syllable that we call the rhyme (cf. Chapter 5) – that is,
the name rhyme weight is more appropriate. A piece of warning is in order
here: do not let spelling mislead you! The second syllable in variety or
horizon is /rai/ and thus heavy, whereas the second syllable in various or
horizontal is /ri/ and consequently light.
124
Word stress – Part 2
Rhyme weight is relevant to stress placement in that in English, heavy
syllables attract stress. E.g., the readers are invited to check for themselves
that the words a.ro.ma, e.nig.ma, al.ge.bra, in.du.stry all consist of three
syllables (divided by dots), out of which only one is heavy (underlined), and
that one is primary-stressed. However, word-final consonants normally do
not count: words like a.ban.don, GA ten.der, pa.ren.tal, can.cel, de.ter.mine
contain two heavy syllables, but stress always falls on the non-final one – the
word-final consonant is unable to make its syllable heavy. Such "invisible"
segments are usually referred to as extrametrical, being outside the scope of
meter, i.e., rhythm; in the rest of the chapter, we indicate extrametrical
material by putting it in parentheses, e.g., abando(n), tende(r), determi(ne).
Notice that the -e at the end of determine is not a sound but a silent letter
only: it does not really matter whether it is or is not included in the
parentheses.
The status of final consonants is of great significance since, as it was
mentioned in Chapter 8, the directionality of primary stress placement is right
to left. That is, stress rules start scanning the syllables with the last one and
proceed towards the beginning of the word, in such a way that primary stress
is assigned to the first heavy syllable available, but not later than the second
syllable from the end. If word-final consonants were visible to these stress
rules, all words ending in at least one consonant sound would end in a heavy
syllable and be primary stressed there – but this is not what we find: abandon
and the like are stressed on the second-last syllable.
Now the time has come to formulate the Main Stress Rule (MSR) for
verbs, nouns, and adjectives.
1
Let us start with verbs: in this case, the final
consonant (if there is one) is extrametrical. If the remaining syllable is heavy,
1
Bear in mind that primary stress assignment takes place within monomorphemic words –
more complex word forms, including compounds, are stressed according to the effect of the
two affix classes and the compound stress rule familiar from previous chapters.
125
Chapter 9
it is stressed; if it is light, the preceding syllable is stressed. In a verb like
torment, for example, there are two syllables: tor.ment. The final /t/ is
extrametrical: tor.men(t). Thus we are left with men as the final syllable,
which is heavy (it contains a short vowel plus a consonant), so it receives the
primary stress: tormént. That is why final stress is usual in verbs ending with
at least two consonants: eléct, seléct, arrést, adópt, lamént, etc. When a verb
ends in a single consonant, the length of the preceding vowel decides: if it is
long, as in unite /-nai(t)/, erase /-rci(z)/, achieve /-tji:(v)/, the final syllable is
still heavy, and therefore stressed; otherwise the second-last syllable is
stressed, e.g., trável, fínish, ínjure, astónish, vómit, consíder. Logically, if the
verb ends in a vowel, there is nothing to be extrametrical, and the length of
that vowel automatically determines the place of stress: in cárry, the final
vowel is short and the second-last syllable carries the stress, but the last one
does so in applý, the vowel in question being long /ai/. Notice that the
second-last syllable is never checked for rhyme weight: it does not need to be
because primary stress cannot move further to the left anyway (except for the
Alternating Stress Rule, to be discussed below) – recall the prominence of the
right edge. Therefore the same syllable is stressed in a.stó.ni(sh) and
con.sí.de(r) although there is a heavy syllable (underlined) in the latter.
2
There is one case when, predictably, the regularity described above
does not apply, due to a morphological effect not yet mentioned: we need to
take a detour here and discuss the role of verbal prefixes in stress
assignment. Similarly to suffixes, these prefixes can be divided into two basic
subtypes: neutral and non-neutral prefixes. Neutral prefixes never influence
the place of stress in the stem – instead, they are (secondary or tertiary)
stressed themselves. In fact, they attach to the stem so loosely that even the
2
It goes without saying that, as the example words above are not longer than three syllables,
primary stress placement satisfies the prominence of the left edge, too.
126
Word stress – Part 2
frequently occurring stress clash that they cause does not disturb their status,
cf. dèbúg, ùnplúg, upsét, òut-Hérod, rèwínd, etc. In contrast, certain
monosyllabic verbal prefixes are non-neutral: they resist primary stress and
consequently, they systematically overwrite the results of the above
mechanism. For instance, consider two verbs, vomit and omit. Segmentally,
they minimally differ: the former contains an extra consonant at the
beginning. In fact, since that consonant is a syllable-initial one it is not even
expected to affect stress placement – recall that only syllable-final consonants
are able to contribute to weight. Nevertheless, vómit has initial primary stress,
which conforms to the MSR as introduced above, as opposed to omít, which
has, rather surprisingly, final stress. The source of the difference in stressing
cannot be phonological in nature since there is no relevant pronunciation
difference between the two words on the segmental level. It follows, then,
that morphology is to blame: while vomit consists of a single morpheme,
omit can be analysed into a prefix, o- (also appearing in oppose, oppress,
occur) and a stem, -mit (also appearing in submit, remit, admit, permit,
commit), and in all such examples consistently it is the stem which carries
primary stress rather than the prefix. Therefore we conclude that stems enjoy
a priviledge over peripheral elements like prefixes: even if a monosyllabic
stem is light (e.g., mi(t)), it is assigned primary stress since the monosyllabic
prefix "throws back" the stress to it. Interestingly, very often this stress-
resistance of a prefix does not have the chance to manifest itself because the
stem is heavy, e.g., províde, retúrn, but the effect is visible in a host of other
examples such as omít, expél, commít, attách, etc.
Sometimes a disyllabic prefix or two monosyllabic prefixes are
attached to a monosyllabic stem. It should be self-evident that in such words
primary stress falls on the stem, and the first syllable must be secondary
stressed in accordance with the Early Stress Requirement, cf. ìnter-véne,
127
Chapter 9
còntra-díct, rè-pre-sént, cò-rre-spónd, etc. It is crucial that in these examples
the stem is monosyllabic: when the final syllable does not constitute the stem
alone, as in éx-ecute, ré-cognize, intér-rogate, the prefix fuses with the stem
totally and the so-called Alternating Stress Rule, to be introduced below,
comes into effect.
Verbs, then, tend to be end-stressed when the last syllable is heavy
(even without the final consonant), otherwise primary stress normally falls on
the second-last syllable, except when it is a verbal prefix. As usual, the
algorithm provided above ("the word-final consonant is extrametrical, the
remaining last syllable is stressed if heavy, else the preceding syllable is
stressed") suffers from its exceptions: sometimes a light final syllable is
stressed, as in caréss, posséss, GA haráss (mind you, the final double <ss>
stands for a short /s/, which is extrametrical!); at other times a long-vowelled
final syllable fails to be assigned primary stress, e.g., fóllow, hárrow,
swállow, hállow.
Now let us pay some attention to nouns. The nominal subclause of the MSR
differs from the verbal one in the portion of the word which is usually
extrametrical, that is, non-stressable. Namely, in nouns it is the whole final
syllable that does not take part in the stress placement procedure, at least
when it contains a short vowel. That is why disyllabic nouns are normally
stressed on the second-last (=first) syllable (e.g., táble, páttern, chílli,
Lóndon, trúmpet, férry, GA míssile /'misI/, etc.), except for just a handful of
words like evént, hotél, Japán, succéss, Berlín, etc. This is easily accounted
for with reference to the extrametricality of the last syllable. In all other
respects the MSR for nouns is the same as the MSR for verbs: do not
consider the final syllable – if the remaining rightmost syllable is heavy, it is
stressed; if it is light, the preceding syllable is stressed. That is why in longer
128
Word stress – Part 2
nouns the weight of the second-last syllable decides the fate of primary
stress: in a.ré.(na) /c'ri:nc/, a.ró.(ma) /c'rcomc/, con.sén.(sus), hi.á.(tus)
/hai'citcs/, ho.rí.(zon) /hc'raizn/, sy.nóp.(sis), u.tén.(sil), ve.rán.(da) it
(underlined) is heavy and therefore stressed; in A.mé.ri.(ca), a.ná.ly.(sis)
/c'næIisis/, á.ste.(risk), cí.ne.(ma), cu.rrí.cu.(lum), hy.pó.the.(sis)
/hai'po0csis/, já.ve.(lin), me.tró.po.(lis) it is light and consequently the
syllable to the left carries primary stress.
If the last syllable of a noun contains a long vowel, it is very often an
auto-stressed ending and receives primary stress accordingly (see Chapter 8),
e.g., brocáde, millionáire, questionnáire, nominée, enginéer, voluntéer,
kangaróo, machíne, Tennessée. Otherwise such nouns fall into one of two
categories: they follow either the verbal subclause of the MSR (e.g., GA
ballét, baróque, cigár, GA detáil, helló, Julý, políce, regíme, trombóne – cf.
uníte, applý), or the compound stress rule (RP bállet, RP détail, féllow, ménu,
RP míssile /'misaiI/, vénue – cf. bláckboard, ráinbow).
However, nouns with long-vowelled final syllables do not present the
only complication – perhaps all the possible exceptional configurations exist.
First, a light second-last syllable is primary stressed in va.ní.lla, spa.ghé.tti,
um.bré.lla, Vi.é.nna, pro.fé.ssor; Di.á.na, pi.á.no, and a handful of other
words. (Keep in mind that consonant doubling in spelling does not indicate
length in pronunciation!) Second, a heavy second-last syllable is skipped by
the MSR in chá.rac.ter, cá.len.dar, á.djec.tive, pá.ssen.ger, etc. Finally, in
some nouns primary stress exceptionally moves further away from the right
edge: the fourth-last syllable is stressed in, e.g., cémetery, cátegory, RP
labóratory, céremony, ágriculture, télevision, hélicopter, and the fifth-last in
véterinary.
129
Chapter 9
The difference between nouns and verbs becomes clearly noticeable in
segmentally (nearly) identical noun-verb pairs. The usual state of affairs in
such cases is the following: since in verbs only the final consonant is
extrametrical, while in nouns it is the whole final syllable, it logically follows
that primary stress will fall one syllable closer to the left edge in nouns. Thus
digest the verb has primary stress on the (heavy) final syllable: di.gés(t),
while digest the noun must have initial stress: dí.(gest). The same applies to
récord (n) – recórd (v), ímport (n) – impórt (v), úpdate (n) – updáte (v),
áccent (n) – accént (v), ségment (n) – segmént (v), súrvey (n) – survéy (v),
tránsport (n) – transpórt (v), cóntrast (n) – contrást (v), désert (n) –
desért (v), óbject (n) – objéct (v), etc. Naturally, the stress resistance of verbal
prefixes is only applicable to verbs but not to nouns, that is how word pairs
like rébel (n) – rebél (v) or pérmit (n) – permít (v) emerge.
Nevertheless, in a few cases the noun and the verb in such pairs have
the same stress pattern, either the noun copying the verbal stress pattern (as in
attáck, debáte, reséarch, surpríse, GA detáil, etc.) or the verb copying that of
the corresponding noun (as in áccess, cómfort, cómment, cóntact, RP détail,
ínterest, ínterview (a compound noun exemplifying 103) etc.).
The third major word category, adjectives have not developed a third form of
extrametricality but are divided between the nominal and the verbal patterns.
On the one hand, derived adjectives (ending in, e.g., -al, -ar, -ous, -ant, -ent)
behave like nouns and have an extrametrical final syllable: fa.mí.li.(ar),
fá.(mous), gé.ne.(rous), íg.no.(rant), pa.rén.(tal), pí.vo.(tal), vá.ri.(ous)).
These suffixes make the adjectives behave as nouns as far as stress rules go,
inasmuch as it is nouns whose last syllable is not normally stressable. On the
other hand, underived adjectives and adjectives ending in -ic, -id, -it usually
behave like verbs: a.frái(d), a.rách.ni(d), cér.tai(n), có.mmo(n), ex.plí.ci(t),
130
Word stress – Part 2
ex.tré(me), Pla.tó.ni(c), púr.p(le), sin.cé(re) /sin'sic(r)/, su.pré(me). You may
have noticed that the three endings belonging here have the same effect as if
they were pre-stressed – in fact, some are listed as such in Chapter 8. After
all, they are monosyllabic with a short vowel and an (extrametrical) single
consonant: they inevitably make a light final syllable, so it naturally follows
that the preceding syllable is stressed. Therefore, we are unable to distinguish
between the two possible analyses: they are either taken as pre-stressed
suffixes, or the adjectives which they produce are considered to behave as
verbs as far as stress rules go, inasmuch as it is verbs whose final consonant
is not normally visible in stress assignment. It is also noteworthy that a
number of underived adjectives are so undeniably verbal in nature that they
exhibit exactly the same stress pattern as the corresponding segmentally
identical verbs, e.g., corréct, compléte, GA abstráct.
In sum, adjective-forming suffixes and endings fall into two
categories: some of them, such as -al, -ar, -ous, etc., trigger the nominal
subclause of the MSR, whereas others, like -ic, -id, -it, trigger the verbal
subclause. What the two groups have in common is that they are non-neutral
suffixes in the sense the term is introduced in the previous chapter: they do
influence the way primary stress is placed in the word. Consequently, we can
state that, in addition to auto-stressed and pre-stressed, we have identified a
third class of non-neutral suffixes and endings, whose members fix the
position of the main stress by simply launching the application of the MSR
on the adjective. Some authors call them integrated suffixes.
At this point we are able to introduce the Alternating Stress Rule, which
goes as follows: if the last syllable of a verb is stressable (i.e., heavy even
without the final consonant), and the verb has more than two syllables,
primary stress moves to the third-last syllable, and the stress of the final
131
Chapter 9
syllable is reduced to tertiary. Examples include the verbs mánifest, éxercise,
hármonize, décorate, súpplement, cómplement, cómpliment, súbstitute,
éxecute, récognize, invéstigate, elíminate, anníhilate, exággerate, etc., all
with a (0)103 stress pattern. The Alternating Stress Rule is an almost
inviolable constraint on verbs (it only has a handful of exceptions like
contínue, contríbute, distríbute), and it overwrites the result of the MSR,
without respect for the prominence of the right edge. However, as mentioned
above, in words containing a verbal prefix the Alternating Stress Rule can
only apply if the final syllable of the verb is not a monosyllabic stem;
otherwise the MSR produces the expected output with the primary stress at
the end and a secondary stress at the beginning (according to the Early Stress
Requirement). For example, there is no stress alternation in èxtrapóse since it
is composed of a verbal prefix (underlined) plus a monosyllabic stem, as
opposed to, say, díagnose, where there is because of the absence of the verbal
prefix. In intérrogate, for instance, we experience the effect of the Alternating
Stress Rule as, although it contains a verbal prefix (underlined), the stem is
disyllabic; whereas in ìntervéne the same prefix attaches to a monosyllabic
stem and consequently the cooperation of the MSR and the Iambic Secondary
Stress Rule takes place. The same features characterize extrápolate vs.
èxtrapóse.
Sometimes the Alternating Stress Rule is extended to word classes
other than verbs, so certain three-syllable adjectives, such as ábsolute,
grándiose, RP óbsolete, also undergo it. In addition, a number of adjectives
and nouns simply copy the pronunciation of the corresponding verbs, thus
they also appear to be subject to stress alternation, e.g., súbstitute (v/n),
éxercise (v/n), mánifest (v/adj/n). Nouns with a long-vowelled final syllable
belong here, too: they are claimed above to often behave like verbs – this is
also true in the case of the Alternating Stress Rule. Nouns like ávenue,
132
Word stress – Part 2
Fáhrenheit, ánecdote, sácrifice, mínuscule, pédigree, Válentine, RP
stálactite/stálagmite illustrate this.
To conclude the discussion of the three major word classes with respect to
stressing, let us highlight a few additional pairs of remarkable segmentally
nearly identical nouns, adjectives, and verbs. When the adjective in such a
word pair is stressed as a verb, for example because it is underived, its stress
pattern is the mirror image of the noun, as in Áugust (n) – augúst (adj),
cóntent (n) – contént (adj), mínute (n) – minúte /mai'nju:t/ (adj). In a few,
exceptional examples the adjective receives nominal stress – then it is the
mirror image of the verb, e.g., presént (v) – présent (adj/n), perféct (v) –
pérfect (adj/n), suspéct (v) – súspect (adj/n). Certain endings characterize
both nouns/adjectives and verbs, but somewhat differ in the two cases (e.g.,
-ment, -ate). Complement, for instance, is always primary stressed on the first
syllable, however, the third vowel is a schwa in the noun (yielding 100) but
unreduced /e/ in the verb (103). This is because the last syllable in nouns is
not normally stressable (recall that it is extrametrical!); in contrast, in the
verb that syllable is heavy (even without the final /t/) and such three-syllable
verbs are expected to undergo the Alternating Stress Rule. Further examples:
cómpliment, dócument, súpplement. Exactly the same happens in words
ending in -ate: this suffix-like morpheme contains a full diphthong (/-cit/)
when final in a verb but just a schwa (/-ct/) when final in an adjective, e.g.,
delíberate (v-adj), in a noun, e.g., délegate (v-n), éstimate (v-n), or both, e.g.,
assóciate (v-n/adj), gráduate (v-n/adj), séparate (v-n/adj).
With respect to the above discussion of the English MSR, it cannot be left
unnoticed how intimately primary stress placement is connected to
syllabification. When a consonant is situated between two vowels in a
133
Chapter 9
morpheme, it is not at all indifferent whether it belongs to the syllable headed
by the first one, making it a heavy syllable, or to the one headed by the
second vowel, and being a syllable-initial consonant, it is incapable of
influencing stress assignment. All the regular cases treated above suggest that
it is the latter solution which is chosen, that is, in the intervocalic position
syllable-initial consonants are created. Take the word skeletal for example. It
is a derived adjective following the nominal pattern (see above), therefore the
final syllable is expected to be extrametrical, the last "visible" syllable is
checked for weight but only receives primary stress if it is heavy. If the
syllable divisions were located as the dots indicate in skel.et.al, the
underlined syllable would be classified as heavy and assigned primary stress:
*skelétal. Nevertheless, this adjective is stressed at the very beginning:
skéletal, which can only be accounted for if we suppose that the
syllabification is the following: ske.le.tal. The underlined syllable is light and
consequently the preceding, first syllable is primary stressed. We conclude
that single consonants are initial in the syllable whenever possible.
Moreover, two- or three-member consonant clusters get syllabified
into the following syllable, too, on condition that they constitute a well-
formed initial cluster. Compare two nouns, algebra and agenda, and
concentrate on the consonants between the second and third vowels. The /br/
in the former is a possible initial cluster (cf. the Sonority Principle in Chapter
5) while the /nd/ in the latter is not – the two cases are predicted to be
syllabified and therefore stressed differently: the final non-extrametrical
syllable is light in ál.ge.(bra) but heavy in a.gén.(da). Examples like
sý.mme.(try), RP quá.dru.(ple), á.de.(quate), illustrate that indeed as many
consonants are syllabified as initial as possible. The fact that the underlined
syllables are not stressed can only be due to their lightness; the fact that they
are light can only be due to the absence of a closing consonant.
134
Word stress – Part 2
There is, however, a problematic case: word-medial sC sequences do
not always appear to constitute syllable-initial clusters. While they do in
mí.ni.(ster), Mán.che.(ster), ín.du.(stry), ór.che.(stra) quoted above, their
members belong to separate syllables in se.més.(ter), A.lás.(ka),
a.spi.dís.(tra). Recall from Chapter 5 that /s/ takes part in the construction of
syllables in a special way in various respects, exhibiting far more
combinatorial possibilities than any other consonant, one consequence of
which is the curious fact that certain /s/+consonant clusters are found both at
the beginning and the end of words (i.e., syllables). For instance, while /br/ is
only possible initially and not finally (examples like brim exist, but *mibr
would be ill-formed), and /nd/ is only possible finally and not initially (lend
vs. *ndel), we see /st/ in both stab and bast, /sp/ in both spill and lisp, /sk/ in
both scut and tusk. We can conclude therefore that /br/ is unambiguously
syllable-initial, but the same does not hold true for, at least, /s/ plus voiceless
plosive sequences, which is likely to have contributed to the ambivalent
behaviour they exhibit word-medially with respect to stress placement.
It is crucial to see that when stress rules apply, syllabification seems
to be always exhaustive and straightforward – significantly, there is no
ambisyllabicity for stress rules. The /t/ in skéletal clearly belongs to the final
(extrametrical) syllable, and so is the one in vánity or héretic, and the second
/t/ (but perhaps the first one as well) in compétitor. The fact that allophonic
rules like the ones introduced in Chapter 2 treat these consonants as
ambisyllabic can only be decided after stress assignment has taken place
simply because it hinges on the stressedness of the vowels: consonants
followed by a stressed vowel are never ambisyllabic; consonants followed by
an unstressed vowel normally are. The derivation of the pronunciation of
words, then, happens in steps:
135
Chapter 9
syllabification: ske.le.tal va.ni.ty com.pe.ti.tor
stress assignment: ské.le.(tal) vá.ni.(ty) com.pé.ti.(tor)
ambisyllabicity: ské.l.e.t.al vá.n.i.t.y com.pé.t.i.t.or
allophony
3
: 'skcIcrcI 'væncri kºcm'pºcrcrc(r)
It is not only allophony rules that follow stress assignment in English but
certain morphological operations, too. For example, there are a couple of
stress-sensitive affixes, whose attachment to a base is determined by its
stress pattern. Nominal -al, forming abstract nouns out of verbs, strongly
prefers to be suffixed to an end-stressed word, e.g., trý – tríal, dený – deníal,
refúse – refúsal, rehéarse – rehéarsal, arríve – arríval, the only exception
being búry – búrial.
As far as the stress rules introduced above are concerned, a note is in
order here. English spelling is not always capable of reflecting the
pronunciation of vowels, although it can be crucial whether a vowel is long,
automatically producing a heavy syllable, or short, in which case rhyme
weight depends on what element follows it. This fact can cause problems to
students of English, who are frequently first faced with an unknown word in
its spelt form. For example, nothing indicates that the second vowel in
canary and museum is long – therefore its syllable is heavy and as a result,
primary stressed: ca.ná.(ry) /kc'nccri/, mu.sé.(um) /mjo'zi:cm/. Compare
apparatus and asparagus, two words showing spooky resemblance. Still,
since the third vowel is long in the former but short in the latter, they are
stressed differently: à.ppa.rá.(tus) /¡æpc'rcitcs/ vs. a.spá.ra.(gus)
/c'spærcgcs/. Unfortunately, vowel length is not consistently encoded in the
spelling of English.
3
The examples illustrate the pronunciations in a tapping dialect of English which
distinguishes clear and dark /l/.
136
Word stress – Part 2
This chapter has looked into primary stress assignment in underived
verbs and nouns as well as the different subtypes of adjectives. We hope that
the discussion faithfully reflects the complexity of this issue, being
influenced by syntactic, morphological and lexical factors beside the
phonological ones: verbs and nouns follow two distinct patterns; neutral and
non-neutral affixation exert various effects; and finally, all the regularities
have exceptions. In spite of this, we are able to identify the stress rules of
English as the generalizations which hold for the majority of the vocabulary
and which characterize newly borrowed or coined items. Clearly, only the
minority of the examples constitutes the cases we refer to as "irregular" even
if some of them happen to be highly frequent words in English and therefore
our impression of the proportions may be somewhat distorted. Perhaps this is
a situation where the exception proves the rule.
137
10. Sentence stress and intonation
Before you study this chapter, check whether you are familiar with the
following terms: adjective, adverb, conjunction, content word,
demonstrative pronoun, function word, interrogative pronoun, noun, pitch,
stress (primary/secondary), strong syllable, suprasegmental, verb, weak
syllable, Wh-question, Yes/No question
This chapter is concerned with some suprasegmental processes of English
which are often grouped under the umbrella term of intonation. In the first
part of the chapter we will be discussing the position of sentence stress while
the second part of the chapter will discuss intonation, the melody of
sentences. We have already seen the principles of assigning different degrees
of word stress in Chapters 8 and 9. Let us now continue with a description of
how the same thing works at sentence level.
If we want to show stress (and later intonation) at sentence level, we
will have to do things a little bit differently from what we have been doing
when transcribing the stressed syllables of isolated words. There are several
principles to be kept in mind:
 In isolation, every word has a primary stressed syllable (although the
stress of monosyllabic words is not indicated). In a sentence, many
words will have no stressed syllable at all, i.e., they occur in their
weak forms, e.g., can is realized as /kn/ instead of /kæn/. (Cf. Chapter
7.)
 In isolation, the stress of monosyllabic words is not shown. However,
if they receive sentence stress it is always indicated, e.g., This is the
car that I bought /'ðis iz ðc 'ko: ðct ai 'b5:t/. The words this, car,
Sentence stress and intonation
bought are stressed in the sentence as indicated by the stress marks
although they are all monosyllabic words.
To be able to cope with the assignment of sentence stress and, on the
basis of that, intonation, we have to clarify a few important basic notions.
One of the distinctions we have to make is between lexical/content words
and grammatical/function words. The former include the four basic
categories, nouns, verbs, adjectives and adverbs – including adverbial
particles like up – while the latter contain the rest of the categories,
prepositions, pronouns, auxiliaries, conjunctions. Also, some minor
categories can be compared to one of these two groups: demonstrative and
interrogative pronouns, e.g., this, that and what, where respectively, are
stressed like content words.
In the case of word stress we identified different degrees of stress in
Chapters 8-9, we may do so in sentences, too, and just like in words, in
sentences it is also the last stress that is the strongest. That is, the strongest
stress of a sentence falls on the last stressed syllable, which is called the tonic
– sometimes also called accent, nucleus or sentence stress, indicated by
underlining in this chapter. The tonic will have a special role in describing
intonation as intonation is nothing else but a falling or rising melody starting
on the tonic.
With the help of the tonic we may define some further concepts
relevant for our discussion: speech is divided into so-called tone-units – or
tone groups or intonation phrases –, which are parts of connected speech
ending in a tonic. That is, a tone-unit starts after a tonic and ends in a tonic.
Tone-units are normally realized by clauses as in the first three examples or
by longer phrases as in the second three examples below. The boundaries of
tone-units are usually indicated by vertical lines.
139
Chapter 10
|She 'wanted to 'face the 'problems on 'Tuesday.|
|He 'felt un'easy| but the 'others were en'joying them'selves.|
|I 'didn't really 'want to 'come| but 'here I 'am.|
|'No 'way!|
|At 'five o''clock.|
|'No|, 'only at the 'meeting.|
Besides the tonic, the tone-unit has the following parts: tail – the unstressed
syllables following the tonic, e.g., -day in the first example above –, the pre-
head – the unstressed syllables before the first stress, e.g., she in the first
example –, and the head – starting with the first stressed syllable and ending
with the last unstressed syllable before the tonic, e.g., wanted to face the
problems on in the first example above:
|She 'wanted to 'face the 'problems on 'Tuesday.|
Pre-head Head Tonic Tail
The tone unit may also be divided into other kinds of constituents which play
a very important role in determining the rhythm of the sentence. These units
of rhythm are called feet (cf. Chapter 7), the same name that is used for
rhythmic units in literature to determine the rhythm of poems. As it was
mentioned in Chapter 7, a foot is the sequence of a stressed syllable and all
the unstressed syllables following it up to the next stress. The sentence above
may be divided into feet the following way:
140
Sentence stress and intonation
She 'wanted to 'face the 'problems on 'Tuesday.
foot
0
foot
1
foot
2
foot
3
foot
4
It is clear that the first foot, foot
0
is an incomplete one as it only contains
unstressed syllables – if there is a pre-head, it is always an incomplete foot.
Recall from Chapter 7 that the special characteristic property of
English rhythm is that it is stress-timed. It means that the stressed syllables
follow each other at intervals of about the same length, which sounds like a
pulsating rhythm. This means that in the sentence above the time elapsing
between the stressed syllables 'wan- ... 'face ...'prob ... 'Tues is approximately
equal although there might be different numbers of unstressed syllables
between them. Since this rhythmic sequence of pulses is very different from
Hungarian, it is something to be practised a lot to get used to pronouncing
(sometimes many) weak syllables between the stressed ones.
When connecting words into a sentence it often happens that there
will be a sequence of three stressed syllables with zero or just one unstressed
syllable between them. In such cases the rhythm becomes jerky, staccato-like.
To avoid such stress clusters the middle one of the three stresses is deleted
and the syllable is pronounced as unstressed, a process called Rhythmic
Stress Deletion – this deletion of stress is indicated with a superscript zero in
the examples:
'good 'old 'days → 'good
0
old 'days
'very 'brave 'soldier → 'very
0
brave 'soldier
'cover the 'big 'news → 'cover the
0
big 'news
141
Chapter 10
In cases when a word with two stresses is followed by a word stressed on the
initial syllable it would result in three stressed syllables in a row. As a result,
rhythmic stress deletion will delete the stress in the middle. This way the
stress pattern of the first word has been changed as the primary stress has
shifted to the left from the last syllable, a process called Rhythmic Stress-
Shift:
the ¡best 'man 'asked /¡bcst'mæn/ → the 'best
0
man 'asked /'bcstmcn/
a ¡stone 'deaf 'guy /¡stcon'ocI/ → a 'stone
0
deaf 'guy /'stconoiI/
a ¡dark 'brown 'hat /¡oo:k'braon/ → a 'dark
0
brown 'hat /'oo:kbraon/
As is clear from the above, this may affect finally-stressed compounds and
longer words ending in two stressed syllables in a way that they will have
two slightly different stress patterns depending on whether the next word
starts with a stressed or unstressed syllable. Before words starting with an
unstressed syllable nothing happens, but before words starting with a stressed
one the final primary stress of the first word shifts one to the left.
In the following we take a look at the two major types of tonic
placement. The first type of tonic placement is the neutral, unmarked or
default type: it does not express emphasis or contrast. This is called neutral
tonic placement or neutral tonicity. The neutral tonic is normally placed on
the last content word but in some exceptional cases it may fall on an earlier
content word or on a function word.
Tonic on last content word
|'Everyone was 'there|
|We 'didn't 'want to 'talk about the 'details.|
142
Sentence stress and intonation
|He was 'finally ad'mitted to uni'versity.|
Tonic on an earlier content word (the skipped last content word italicized)
|He 'bought a 'new 'mountain bike.| No tonic on 2
nd
part of initially
stressed compound.
|It was 'nice, I think.| No tonic on afterthoughts, appended
remarks.
|We'll just 'stay here.| No tonic on common adverbs.
|'That's what the 'book says.| No tonic on "obvious predicates".
Tonic on a function-word
|'No, you 'can't.| Tonic on an auxiliary if no other
stressable word.
|'Where are you 'from?| Tonic on Prep in short sentences
without main verb.
|'This is 'mine| Tonic on possessive pronoun.
In the second major type of tonic placement the speaker wishes to emphasize
some part of the utterance, contrast a part of it with something or focus on
some new information, which may be achieved by placing the tonic at a
different place from where it would normally appear. The following two
sentences demonstrate that while the first sentence with neutral tonic
placement on the last content word does not emphasize or contrast any part of
the sentence, the second sentence with so-called dislocated tonic does.
|'Jim was 'there.| |'Jim was 'there.|
143
Chapter 10
As the underlining indicates, there has the tonic in the first sentence while
Jim has the tonic in the second. Accordingly, the first sentence has a neutral
interpretation, while the second sentence emphasizes that it was Jim who was
there, not somebody else. Let us now take a look at what might be the
reasons for having a dislocated tonic in a sentence.
Contrast
In many cases the tonic is placed on an earlier content word to express
contrast between what has been said and the word/expression bearing the
tonic. The two most common cases are when either it is a particular lexical
item, a certain word that we want to contrast with another one, e.g., a name
with another name, or negative polarity with positive polarity, i.e., negation
with assertion. If a word is contrasted, it is indicated in capital letters.
|I 'gave 'JACK that 'book on 'history 'yesterday.| ... and not somebody else.
|We 'visited a lot of MU' SE UMS in 'London.| ... and not night-clubs.
|I 'HAVE 'seen the 'film 'earlier.| ... contrary to what you
think/claim.
|I 'DID 'pass the 'test.| ... although that's not what
you think.
New information
If the end of the sentence contains information the speaker thinks is known
by the listener as old information, then the words describing this old
information will be de-stressed and stress (and the tonic) will be shifted
leftwards to some earlier word considered to carry new information. It most
typically happens in answering questions repeating some words from the
question – the old information skipped by tonic assignment is italicized.
144
Sentence stress and intonation
Is the book interesting? |It is 'VE RY interesting.|
Do you want to have lunch? |But I 'already 'HAD lunch.|
I'm from Hungary. |Oh, my FI'AN CÉE is Hun'garian, 'too.|
So far we have seen which part of a sentence carries the strongest stress – the
tonic –, how it relates to the rest of the clause – the tail, the head and the pre-
head –, how the head plays a role in determining the rhythm of the clause by
being divided into feet, and, finally, how tonic assignment may be performed
in neutral cases and in dislocated cases when the speaker wishes to
emphasize or contrast some part of the sentence. Now we turn to how these
units relate to intonation, the melody and the melody change of a sentence.
The intonation or melody of a sentence is the voice-height, or pitch.
On the one hand, pitch depends on what kind of intonation is used in the
pronunciation of a particular sentence. On the other hand, there are also non-
linguistic factors that influence pitch: age – children have a higher pitch than
adults –, sex – men normally have a lower pitch than women –, and the
emotional state of the speaker – excited speakers tend to have a higher pitch
than someone in a neutral mood. Every speaker has a limit to how high or
how low a pitch they may produce; these two are the upper and lower limits
of one's pitch range. This pitch range is different for each speaker but it does
not influence the understandability of their speech: it is not the absolute but
the relative pitch height that matters.
Pitch differences do not only occur between speakers but also
between languages. Hungarian, for instance, is said to have a much narrower
pitch range in general than that of English; that is, the highest pitch of an
average native English speaker is higher than that of an average native
Hungarian speaker, while the lowest pitch of an English speaker is generally
145
Chapter 10
lower than that of a Hungarian speaker. This is even noticed by the untrained
ear. For instance, Hungarian speakers often report that they find English
speech too emotional, excited and affected; English speakers, on the other
hand, find the speech of the average Hungarian – whether this average
Hungarian speaks English or Hungarian – too flat, monotonous or boring.
Intonation is the way the pitch changes in the tone-unit. Recall that
the last stressed syllable of the tone-unit, the tonic always has pitch change,
that is, the speakers' voice will either rise or fall on the tonic syllable. This
change associated with the tonic syllable is referred to as the tone
1
. The
melody of the tone is always continued in the tail of the tone-unit; the tail
will never contain another change in pitch. If we do not only consider the
pitch change realized on the tonic syllable, but rather the pitch changes
occurring throughout the whole tone-unit, we may talk about the tune or
intonation pattern of the sentence. In the following we briefly describe the
characteristic properties of the parts of the tone-unit followed by a discussion
of the tones and the typical meanings or functions associated with them. The
reader, however, has to be aware that this relationship is not a one-to-one
relationship, so the same communicative function is not always expressed by
the same tone and the same tone does not always express the same meaning.
The melody of the pre-head of the tone-unit – if there is one –
normally starts at a relatively low pitch which normally jumps high up on the
first stressed syllable, i.e., on the beginning of the head. The pitch usually
gradually falls throughout the head, which is called downdrift. Since it is not
the pitch change realized on the tonic, it does not count as falling intonation,
it is just a natural consequence of the fact that speakers are normally running
out of air and this way the velocity of the outflowing air is dropping, which
results in a lower pitch. The part of the tone-unit after the tonic, the tail – if
1
Note that this usage of tone is slightly different from tone in tone languages (cf. the
beginning of Chapter 8).
146
Sentence stress and intonation
there is one – is normally a simple continuation of the pitch change of the
tonic: if the pitch rises on the tonic, it will slowly, gradually rise on the tail,
too. If it falls on the tonic, then it will fall on the tail, too. This is
demonstrated by the graphic representations below. The arrow before the
tonic syllable indicates the pitch change on the tonic.
He be'lieved that they had 'seen the 'movie earlier.
'Have you 'seen this movie?
Let us now take a look at the four types of tone and the different functions
associated with them:
Type of tone Name of tone Tone contour Function, meaning
Falling Fall neutral, definiteness, finality
Rising Low rise indifference, encouragement
High rise Yes/No questions, inquiry
Fall-rise old information, implication
The falling tone
The falling tone is the most common, neutral tone used in English. It suggests
that speakers are simply conveying information. As a result, it is most often
used in plain statements, real, serious commands – as it expresses finality and
definiteness –, in exclamations – expressing that the speakers are sure of
147
Chapter 10
what they are saying –, and in Wh-questions, i.e., questions starting with a
question word (who, what, where, why, how, etc.).
I 'think we 'haven't met. 'This is the 'house where I grew up.
'Leave all the 'books on the desk! 'Put the 'gun on the ground!
'What a 'truly 'beautiful day! 'How 'absolutely 'fabulous an idea!
'How did you 'know he was my brother? 'What are you working on?
The low rising tone
The low rising tone is the most difficult for Hungarian learners of English as
in Hungarian it is only used in certain types of questions while in English it is
never used in this sentence type. Instead, it is always used to respond to
something that somebody said. Often it expresses indifference – an "I-don't-
care attitude" –, it is used in apologies, greetings and when saying thanks,
and also in cases of expressing encouragement.
(How was the film?) It was all

right. (Not too bad, not too good.)
(Where shall we go tonight?) 'We may 'go to the

cinema. (I don't care.)
148
Sentence stress and intonation

Thank you Ex

cuse me. Bye-

bye. You're

welcome.
'Don't you

worry! It 'doesn't

matter. It will be all

right.
The high rising tone
The high rising tone is either a high rise on the tonic, if there is no tail, or it is
accomplished on the tonic and the tail, if there is one. If there is no tail, i.e.,
the whole high rise is realized on the tonic syllable, then Hungarian speakers
tend to have serious problems with the height: the high rise produced by a
Hungarian speaker is simply not high enough for an English speaker; instead,
it will sound as a low rise, and will consequently express indifference or
boredom.
In English, if there is a tail, then the high rise continues from the tonic
throughout the tail and is evenly distributed over the syllables of the tonic
and the tail. This is also a serious difficulty, if not greater, for Hungarians as
such tonic+tail combinations may not be pronounced with a steady high rise
in Hungarian as the language does not permit a rise on consecutive syllables
within the same tone unit.
In Hungarian there is only a real rise if there is no tail. If there is a
tail, then depending on the number of its syllables, one of the following
things will happen: if the tail consists of just one syllable, it will be a rise-fall;
if the tail has two syllables, the first will rise, the second will fall; finally, if
the tail has more than two syllables, the second last will rise, the last one will
fall. All in all, whenever there is a tail in such Hungarian sentences, it will
have a rise-fall and not a steady rise as it does in English.
149
Chapter 10
The high rise is most often used in Yes/No questions, which do not
start with a question word. If something is said with a high rising intonation,
it is always a real question.
Have you 'ever been to England? Did you 'do your homework?
'Was it your grandmother? 'Is 'Johnny 'coming to the party?
The high rise is also found in echo-questions, which repeat what a speaker
has previously said.
' Where ? You 'met him at which station?
The fall-rise tone
The fall-rise is a combination of a fall from high or mid tone to low followed
by a low rise, i.e., a rise from low to mid. It is one of the tones that makes
English speech sound too theatrical or affected for the Hungarian ear but it is
not as difficult to learn to do correctly as the high rise or the low rise.
A fall-rise may be used for several purposes: on the one hand, it may
indicate that the speaker is not telling everything, but a part of the message is
only implied, the listener has to find it out from the context.
I've 'never 'seen your flat. (Invite me to see it.)
150
Sentence stress and intonation
It may also occur in a sentence made up of two tone-units, the first
expressing old information serving as background for the new information in
the second part of the sentence. The new information is pronounced with a
falling tone.
In our 'old car | there was 'enough 'room for 'six people.
In this chapter we have taken a look at two very important suprasegmental
aspects of English pronunciation: the stress patterns of sentences, especially
tone placement on the one hand, and the basic types of intonation and their
differences from Hungarian intonation on the other.
151
11. Letter-to-sound rules – Part 1: Consonants
Before you study this chapter, check whether you are familiar with the
following terms: allophone, allomorph, aspiration, clear/dark-L, coronal,
devoicing, digraph, glottalization, homorganic, loanword, morpheme (free
and bound), orthography, palatalization, palato-alveolar, place assimilation,
productive/non-productive, R-dropping, rhotic/non-rhotic, root/stem, suffix,
tapping/flapping, weak/strong forms of function words, Yod-dropping
This chapter mainly focuses on the regular correspondences between
consonant letters and sounds, and the rules regulating this relationship. This
is made necessary by the fact that the principles of English spelling (or,
orthography) are quite different from those of Hungarian. On the one hand,
the correspondences between Hungarian letters and sounds are much more
straightforward as spelling observes the phonemic principle more than in
English, i.e., it aims at setting up a one-to-one relationship between letters
and phonemes as much as possible, but at least much more successfully than
English spelling does. On the other hand, Hungarian mostly represents the
different pronunciation variants, allomorphs of a morpheme differently in
spelling, e.g., ház-hoz, kert-hez, föld-höz where the vowels of the three
different variants of this suffix are different in pronunciation and it is clearly
indicated in spelling, too. This way, the spelling will always tell us how to
pronounce the particular morpheme in question. English observes another
principle instead, that of morpheme identity: it prefers to keep the spelling
of a morpheme unchanged regardless of whether the particular morpheme is
pronounced with one allomorph or another, e.g., want-ed /'wontio/, kiss-ed
/kist/, play-ed /pIcio/ (cf. Chapter 6). This sometimes also happens in
Letter-to-sound rules – Part 1
Hungarian but not as often as in English. Thus, the two languages observe the
principles of spelling in very different ways – although they are clearly not
the two extremes on the scale.
In this chapter we are going to take a look at the regular pronunciation
of single consonant letters and consonant digraphs one by one, and also at the
letter-to-sound rules that regulate the connection between sounds and letters
as well as the exceptions that fail to obey these rules. The next chapter is
going to discuss the same for vowel letters and vowel digraphs.
1
Single consonant letters
Let us take a look at single consonant letters first. For each consonant letter
we are going to define what sound(s) it normally represents in what
environments, list exceptional cases and positions in which the letter is
typically silent. We have to note again that English lacks long or so-called
geminate consonants. Although doubled consonant letters do occur in
English, they are pronounced as short sounds as in letter /'Ictc(r)/, attack
/c'tæk/, ballet RP /'bæIci/ (GA /bæ'Ici/), recommend /¡rckc'mcno/, Higgins
/'higinz/. Long consonants are only pronounced if two identical consonant
sounds are put in adjacent positions at morpheme or word boundaries, i.e., if
a word or morpheme ends in a certain consonant and the next one starts with
the same as in disservice /ois'sa:vis/, unnatural /\n'nætjrcI/, greenness
/'gri:nnis/.
1
Throughout these two chapters transcriptions show RP pronunciations. Keep in mind that
GA is a rhotic accent (Chapter 2) with extensive Yod-dropping (Chapter 5) and frequent
tapping (Chapters 2 and 7). These and other systematic differences between RP and GA,
mentioned in previous chapters, are not indicated separately. However, full transcriptions are
given whenever the two accents differ more significantly.
153
Chapter 11
p It regularly represents the phoneme /p/ and all of its possible variants
– weakly or strongly aspirated, unaspirated, glottalized – as in plenty
/'pIcnti/, prayer 'words used in praying' /'prcc(r)/, pen /pcn/, pirate
/'pai(c)rct/, lap /Iæp/ [Iæp] or [Iæ1p], step /stcp/ [stcp] or [stc1p],
leopard /'Icpco/, super /'s(j)u:pc(r)/, supper /'s\pc(r)/.
It is regularly silent in word-initial position in pn- and ps- as in
pneumonia /nju:'mconjc/, pneumatic /nju:'mætik/, psychology
/sai'koIco¸i/, psychiatrist /sai'kaictrist/, psychopath /'saikcpæ0/.
It is irregularly silent in corps /k5:/, coup /ku:/, cupboard /'k\bco/,
raspberry /'ro:zbri/, receipt /ri'si:t/.
b It regularly represents the phoneme /b/ and its – devoiced or voiced –
allophones as in banana RP /bc'no:nc/ (GA /-næ-/), below /bi'Ico/,
label /'IcibI/, sober /'scobc(r)/, rob /rob/, stab /stæb/, rubber
/'r\bc(r)/, pebble /'pcbI/.
It is regularly silent in morpheme-final position after a nasal as in
numb /n\m/, bomb /bom/, climb /kIaim/, numbest /'n\mist/, bomber
/'bomc(r)/, bombed /bomo/, climbing /'kIaimin/. (Cf. Chapter 5.)
It is irregularly silent in certain -bt clusters as in debt /oct/, debtor
/'octc(r)/, doubt /oaot/, subtle /'s\tI/.
t It regularly represents the phoneme /t/ and its allophonic – weakly or
strongly aspirated, unaspirated, glottalized or flapped – variants as in
take /tcik/, tonight /tc'nait/, better /'bctc(r)/ ['bctc(r)] or ['bcrc(r)],
rotten /'rotn/, late /Icit/ [Icit], [Ici1t] or [Ici1], fantastic /Iæn'tæstik/.
It regularly represents the palatals /j/ and /tj/ in cases of lexical
palatalization (see rule at the end of Chapter 11) in words like action
154
Letter-to-sound rules – Part 1
/'ækjn/, literature /'Iitritjc(r)/, motion /'mcojn/, nature /'ncitjc(r)/,
picture /'piktjc(r)/, question /'kwcstjn/.
It is irregularly silent in words of French origin ending in -et as in
ballet RP /'bæIci/, beret RP /'bcrci/ (GA /bc'rci/), bouquet /bu:'kci/ or
/bco'kci/, buffet RP /'boIci/ (GA /bc'Ici/), cabaret /'kæbcrci/,
Chevrolet RP /'jcvrcIci/ (GA /¡jcvrc'Ici/).
It is irregularly silent in consonant clusters in words like boatswain
/'bcosn/ (also spelled bosun), Christmas /'krismcs/, forecastle
/'IcoksI/, listen /'Iisn/, often /'oIn/ (this word is pronounced by some
speakers as /'oItcn/), wrestle /'rcsI/, tsar /zo:(r)/.
d It regularly represent the phoneme /d/ and its allophonic – devoiced,
flapped – variants as in damage /'oæmio¸/, delete /oi'Ii:t/, rider
/'raioc(r)/, ['raioc(r)] or ['rairc(r)], sender /'scnoc(r)/, madder
/'mæoc(r)/, ['mæoc(r)] or ['mærc(r)], bend /bcno/, recommend
/¡rckc'mcno/.
It regularly represents the phoneme /t/ in the past tense suffix after
stem final voiceless consonants other than /t/ as in backed /bækt/,
kissed /kist/, laughed RP /Io:It/ (GA /IæIt/), squashed /skwojt/,
stepped /stcpt/ (for the pronunciation rule of the past tense suffix, see
Chapter 6).
It regularly represents the palatal /o¸/ in cases of Palatalization (see
below) in words like educate /'co¸okcit/, gradual /'græo¸ocI/,
grandeur /'græno¸c(r)/, soldier /'scoIo¸c(r)/.
It is irregularly silent in words like grandmother /'grænm\ðc(r)/,
155
Chapter 11
grandpa /'grænpo:/, sandwich /'sænwitj/ or /'sænwio¸/.
k It regularly represents the phoneme /k/ and its allophonic – weakly or
strongly aspirated, unaspirated and glottalized – variants as in kettle
/'kctI/, king /kin/, baker /'bcikc(r)/, poker /'pcokc(r)/, banking
/'bænkin/, thank /0ænk/.
It is regularly silent in word-initial kn- cluters as in knave /nciv/, knife
/naiI/, knitting /'nitin/, knock /nok/, knowledge /'noIio¸/, knuckle
/n\kI/.
c It regularly represents the phoneme /k/ and its – aspirated, unaspirated
and glottalized – variants as in cat /kæt/ cover /'k\vc(r)/, account
/c'kaont/, vicar /'vikc(r)/, acne /'ækni/.
It regularly represents the phoneme /s/ as in city /'siti/, lucid
/'I(j)u:sio/, face /Icis/, racing /'rcisin/, dice /oais/ (see the discussion
of Velar Softening below).
It regularly represents the phoneme /j/ in cases of Palatalization (see
below) as in vicious /'vijcs/, musician /mju:'zijn/, facial /'IcijI/, social
/'scojI/, ocean /'cojn/.
It irregularly represents the phoneme /tj/ in words of Italian origin
like cello /'tjcIco/, concerto /kcn'tjcctco/.
It is irregularly silent in Connecticut /kc'nctikct/, endictment
/in'oaitmcnt/, muscle /'m\sI/, czar /zo:(r)/.
g It regularly represents the phoneme /g/ and its devoiced variant as in
gallop /'gæIcp/, get /gct/, goulash RP /'gu:Iæj/ (GA /'gu:Io:j/), linguist
/'Iingwist/, longer /'Iongc(r)/, beggar /'bcgc(r)/, bigger /'bigc(r)/, hug
156
Letter-to-sound rules – Part 1
/h\g/.
It regularly represents the phoneme /o¸/ (see the discussion on Velar
Softening below) and its – devoiced – variants as in engineer
/¡cno¸i'nic(r)/, gym /o¸im/, ginger /'o¸ino¸c(r)/, harbinger
/'ho:bino¸c(r)/, huge /hju:o¸/.
It is irregularly pronounced as /¸/ in French loanwords as in beige
/bci¸/, garage RP /'gæro:¸/ (GA /gc'ro:¸/), collage /kc'Io:¸/, regime
RP /rci'¸i:m/ (GA /rc'¸i:m/).
It is regularly silent in morpheme-final position after a nasal as in
sing /sin/, singing /'sinin/, singer /'sinc(r)/, belong /bi'Ion/, belonged
/bi'Iono/. But it is irregularly pronounced in morpheme-final position
after a nasal in the comparative and superlative forms of the following
three adjectives: long /Ion/, longer /'Iongc(r)/, longest /'Iongist/,
young /j\n/, younger /'j\ngc(r)/, youngest /'j\ngist/, strong /stron/,
stronger /'strongc(r)/, strongest /'strongist/. (Cf. Chapter 5.)
It is regularly silent in word-initial and word-final gn clusters as in
gnome /ncom/, gnu /nu:/; sign /sain/, resign /ri'zain/.
j It regularly represents the phoneme /o¸/ and its devoiced variant as in
jet /o¸ct/, jockey /'o¸oki/, cajole /kc'o¸coI/, Don Juan /'oon 'o¸u:cn/.
It irregularly represents the phoneme /h/ in some Spanish
geographical names like Baja /'bo:ho:/.
Note that this consonant letter is never pronounced as /j/!
f It is regularly pronounced as /I/ as in final /'IainI/, forget /Ic'gct/, café
RP /'kæIci/ (GA /kæ'Ici/), reference /'rcIrcns/, coffee /'koIi/, strife
157
Chapter 11
/straiI/, stuff /st\I/, staff RP /sto:I/ (GA /stæI/).
It is irregularly pronounced as /v/ in of RP /ov/ (GA /\v/) (in its
strong form) or /cv/ (in its weak form). (Cf. Chapter 7.)
v It is regularly pronounced as /v/ and its devoiced variant as in veal
/vi:I/, vanity /'væniti/, lover /'I\vc(r)/, never /'ncvc(r)/, Denver
/'ocnvc(r)/, elves /cIvz/, wives /waivz/, grave /grciv/, jive /o¸aiv/.
It never represents the phoneme /w/!
s It regularly represents the phonemes /s/ and /z/ depending on the
environment:
Word-initially it regularly represents the phoneme /s/ as in singer
/'sinc(r)/, silence /'saiIcns/, Sudan RP /su:'oo:n/ (GA /-'oæn/), senior
/'si:nic(r)/.
Word-finally it regularly represents /s/ as in hazardous /'hæzcocs/,
cactus /'kæktcs/, crisis /'kraisis/, minus /'maincs/, bus /b\s/; but it
irregularly represents /z/ in word-final position in proper names and
function words, i.e., in words like is /iz/, was RP /woz/ (GA /w\z/) or
/wcz/, has /hæz/ or /hcz/, his /hiz/, Jones /o¸conz/, James /o¸cimz/,
Charles /tjo:Iz/.
Between vowel letters it regularly represents /z/ as in music
/'mju:zik/, desert (n) /'oczct/, cousin /'k\zin/, phase /Iciz/, close (v)
/kIcoz/, bosom /'bozcm/, busy /'bizi/; but it irregularly represents /s/
between vowel letters, for instance in base /bcis/, basic /'bcisik/, case
/kcis/, bison /'baisn/, promise /'promis/, goose /gu:s/, house /haos/,
close (adj) /kIcos/.
158
Letter-to-sound rules – Part 1
Between a root vowel and an affix vowel it normally represents /s/ as
in dis-integrate /ois'intcgrcit/, dis-agree /¡oisc'gri:/, mis-understand
/¡mis\noc'stæno/, bi-sect /bai'sckt/, be-side /bi'saio/; but it irregularly
represents /z/ in words like divis-ible /oi'vizibI/, pre-sume
/pri'z(j)u:m/, dis-ease /oi'zi:z/, de-sign /oi'zain/ (the hyphens indicate
morpheme boundaries).
It regularly represents /s/ when doubled, ss, as in kiss /kis/, bass
/bcis/, message /'mcsio¸/, passing RP /'po:sin/ (GA /'pæsin/),
assassin /c'sæsin/, but it irregularly represents /z/ in words like
scissors /'sizcz/, dissolve /oi'zoIv/, dessert /oi'za:t/, possess /pc'zcs/.
It regularly represents /s/ after n, l, and r (silent in the non-rhotic
accents) as in course /k5:s/, horse /h5:s/, universe /'ju:niva:s/, insist
/in'sist/, tense /tcns/, false /I5:Is/, pulse /p\Is/.
It regularly represents /z/ in final -es when not a regular suffix as in
species /'spi:ji:z/, Hercules /'ha:kjoIi:z/, analyses /c'næIcsi:z/, crises
/'kraisi:z/, Mercedes /ma:'scioi:z/.
It regularly represents /s/ or /z/ in the regular suffix -(e)s. For the rules
of its pronunciation, see Chapter 6.
It regularly represents the palatalized variants of the above sounds, /j/
and /¸/, in all the possible environments (for Palatalization see below)
as in mission /'mijn/, sure /joc(r)/, mansion /'mænjn/, version
RP /'va:jn/ (GA /'var¸n/), vision /'vi¸n/, measure /'mc¸c(r)/, fusion
/'Iju:¸n/.
z It regularly represents the phoneme /z/ and its devoiced variant as in
159
Chapter 11
zoo /zu:/, zeal /zi:I/, razor /'rcizc(r)/, Gonzo /'gonzco/, buzz /b\z/.
m It regularly represents the phoneme /m/ as in matter /'mætc(r)/,
meringue /mc'ræn/, hammer /'hæmc(r)/, summer /'s\mc(r)/, plumb
/pI\m/, bottom /'botm/.
It is irregularly silent in the word-initial mn- cluster in mnemonic
/ni'monik/.
n It regularly represents the phoneme /n/ as in number /'n\mbc(r)/,
notion /'ncojn/, penny /'pcni/, fence /Icns/, pin /pin/.
It regularly represents the phoneme /n/ when followed by k or g (at
least in spelling) as in ink /ink/, sing /sin/, singing /'sinin/, language
/'Iængwio¸/, pink /pink/, banquet /'bænkwit/.
It is irregularly silent in final -mn clusters as in autumn /'5:tcm/,
solemn /'soIcm/, condemn /kcn'ocm/.
l It regularly represents the phoneme /I/ and its – clear and dark –
allophones (see Chapter 2) as in light /Iait/, level /'IcvI/, building
/'biIoin/, follow /'IoIco/, fell /IcI/, people /'pi:pI/, final /'IainI/.
It is irregularly silent before consonants in words like folk /Icok/, talk
/t5:k/, walk /w5:k/, yolk /jcok/, salmon /'sæmcn/, almonds /'o:mcnoz/.
r It regularly represents the phoneme /r/ as in rifle /'raiII/, raccoon
/rc'ku:n/, redial /ri:'oaicI/, burial /'bcricI/, borrow /'borco/, caring
/'kccrin/.
It is regularly made silent before consonants and a pause by the R-
Dropping Rule (see Chapter 2) as in cart /ko:t/, flair /IIcc(r)/, barn
/bo:n/, steer /stic(r)/.
160
Letter-to-sound rules – Part 1
Note that it is silent in iron /'aicn/ (cf. footnote 1 in Chapter 4).
y It regularly represents the phoneme /j/ as in yet /jct/, yoghurt /'jogct/,
mayonnaise /¡mcic'nciz/, junkyard /'o¸\nkjo:o/.
It often functions as a single vowel letter, almost like a variant of <i>,
as in cry /krai/, analysis /c'næIisis/, bicycle /'baisikI/ or, after a vowel
letter, as a member of vowel digraphs like <ay>, <ey>, <oy> as in
bay /bci/, key /ki:/, coyote /k5i'coti/ (see Chapter 12).
w It regularly represents the phoneme /w/ and its – devoiced – variants
as in want /wont/, reward /ri'w5:o/, away /c'wci/, watt /wot/, witch
/witj/.
It is regularly silent in initial wr- clusters as in writer /'raitc(r)/, wrong
/ron/, wretched /'rctjio/, wrist /rist/.
It is irregularly silent in words like who /hu:/, whom /hu:m/, whose
/hu:z/, whole /hcoI/, answer RP /'o:nsc(r)/ (GA /'ænscr/), sword /s5:o/,
two /tu:/.
Note that when following a vowel letter, it often forms part of a vowel
digraph as in row /rco/ or /rao/, coward /'kaoco/. For details see the
next chapter.
For the pronunciation of the digraph wh, see below.
h It regularly represents the phoneme /h/ as in head /hco/, hollow
/'hoIco/, history /'histri/, ahead /c'hco/, cohesion /kco'hi:¸n/.
It is regularly silent in words like Shah /jo:/, blah-blah /'bIo:bIo:/,
yacht /jot/, vehicle /'vi:ikI/, annihilate /c'naicIcit/.
It is irregularly silent in words like honest /'onist/, hour /aoc(r)/.
161
Chapter 11
For the rule on the deletion of /h/, see below.
x It regularly represents the sequence /ks/ and its palatalized variant
(see the rule of Palatalization below) as in axe /æks/, expand
/ik'spæno/, exit /'cksit/, boxing /'boksin/, tax /tæks/, anxious
/'ænkjcs/, luxury /'I\kjcri/.
It regularly represents the sequence /gz/ and its palatalized version
/g¸/ when followed by a stressed vowel as in executive /ig'zckjotiv/,
example RP /ig'zo:mpI/ (GA /-'zæm-/), exist /ig'zist/, exempt
/ig'zcmpt/, exult /ig'z\It/, luxurious /I\g'¸ocrics/.
It regularly represents the phoneme /z/ when word-initial as in xerox
/'zicroks/, xylophone /'zaiIcIcon/, Xavier /'zævic/, xenophobia
/¡zcnc'Icobic/, Xena /'zi:nc/.
q It regularly represents the phoneme /k/ and its – weakly or strongly
aspirated, unaspirated or glottalized – variants as in quotation
/kwco'tcijn/, quickly /'kwikIi/, quart /kw5:t/, clique /kIi:k/, antique
/æn'ti:k/, liqueur RP /Ii'kjoc(r)/ (GA /Ii'kar/), liquid /'Iikwio/, lacquer
/'Iækc(r)/.
Finally, we must consider two vowel letters that may often represent the
consonant /w/ in certain environments.
u It may regularly represent the phoneme /w/ in the combinations qu,
ngu, su in words like language /'Iængwio¸/, acquaint /c'kwcint/,
aquarium /c'kwccricm/, banquet /'bænkwit/, persuade /pc'swcio/,
dissuade /oi'swcio/, suite /swi:t/, quest /kwcst/, question /'kwcstjn/.
162
Letter-to-sound rules – Part 1
o It may irregularly represent the phoneme /w/ or the phoneme
combination /w\/ in words like choir /kwaic/, one /w\n/, once
/w\ns/, and in some words of French origin containing -oir, -ois as in
reservoir /'rczcvwo:(r)/, bourgeois /'boc¸wo:/, memoirs /'mcmwo:z/.
Let us now turn to those cases when two or three consonant letters represent a
phoneme regularly, i.e., to digraphs and trigraphs.
Consonant digraphs and trigraphs
Before we start discussing consonant digraphs, we must emphasize once
more that although there are a great many English words containing two
identical consonant letters next to one another, these are normally
pronounced as a single short consonant unless they belong to two different
morphemes (see above). In the following, we only discuss cases in which the
two consonant letters are different.
ch It regularly represents the phoneme /tj/ and its glottalized variant as
in chocolate /'tjokIit/, bachelor /'bætjcIc(r)/, beach /bi:tj/, chunk
/tj\nk/, munch /m\ntj/, cheque/check /tjck/.
It irregularly represents the phoneme /j/ in words of French origin
like machine /mc'ji:n/, moustache RP /mc'sto:j/ (GA /'m\stæj/),
champagne /¡jæm'pcin/, chauffeur RP /'jcoIc(r)/ (GA /joo'Iar/),
chauvinism /'jcovinizcm/, chic /ji:k/, and also in Chicago /ji'ko:gco/,
Chevrolet /'jcvrcIci/, Michigan /'mijigcn/.
It regularly represents the phoneme /k/ and its allophones, mostly in
words of Latin and Greek origin as in chaos /'kcios/, chameleon
163
Chapter 11
/kc'mi:Iicn/, character /'kæriktc(r)/, charisma /kc'rizmc/, chemical
/'kcmikI/, choir /kwaic/, Christian /'kristjcn/, Munich /'mju:nik/,
echo /'ckco/, Czech /tjck/.
tch It regularly represents the phoneme /tj/ as in catching /'kætjin/, fetch
/Ictj/, latch /Iætj/, wretched /'rctjio/.
rh It regularly represents the phoneme /r/, i.e., we may say that the letter
<h> is regularly silent in this combination in words like rhyme /raim/,
rhythm /'riðm/, rheumatism /'ru:mctizm/, rhino /'rainco/, myrrh
/ma:(r)/.
sh It regularly represents the phoneme /j/ as in shooting /'ju:tin/, fashion
/'Iæjn/, cushion /'kojn/, bushes /'bojiz/, crush /kr\j/, hush /h\j/,
Bolshevik RP /'boIjcvik/ (GA /'booI-/).
ph It regularly represents the phoneme /I/ as in phoneme /'Iconi:m/,
allophone /'æIcIcon/, Humphrey /'h\mIri/, pamphlet /'pæmIIit/,
photograph RP /'Icotcgro:I/ (GA /-græI/).
th This digraph regularly represents the dental fricatives /0/ and /ð/.
Unfortunately there is no rule predicting when it stands for which.
However, we can say that in the majority of the cases, especially in
"international" words of Greek origin, it is normally /0/ except for
rhythm /'riðm/, and that in function words it is pronounced as /ð/,
e.g., they /ðci/, that /ðæt/, those /ðcoz/.
/0/: thinking /'0inkin/, bath RP /bo:0/ (GA /bæ0/), cathedral
/kc'0i:orcI/, healthy /'hcI0i/, Thursday /'0a:zoi/, fifth /IiI0/, length
/Icn0/, method /'mc0co/.
/ð/: bathe /bcið/, feather /'Icðc(r)/, this /ðis/, these /ði:z/, the /ðc/,
164
Letter-to-sound rules – Part 1
brother /'br\ðc(r)/, soothe /su:ð/.
It irregularly represents the phoneme /t/ in a few words, typically in
proper names: Thomas /'tomcs/, Thames /tcmz/, Anthony /'æntcni/,
thyme /taim/.
kh It regularly represents the phoneme /k/ as in khaki RP /'ko:ki/ (GA
/'kæki/).
gh It irregularly represents two phonemes, /g/ and /I/, the former before
vowels as in ghoul /gu:I/, ghost /gcost/, ghetto /'gctco/, gherkin
/'ga:kin/, the latter in a few words as in enough /i'n\I/, rough /r\I/,
toughness /'t\Inis/, laughing RP /'Io:Iin/ (GA /'IæIin/), cough /koI/.
It is irregularly silent in many words and indicates the length of the
preceding vowel as in sight /sait/, nightingale /'naitingciI/, fought
/I5:t/, weight /wcit/, although /5:I'ðco/, daughter /'o5:tc(r)/, height
/hait/.
wh It regularly represents the phoneme /w/ as in where /wcc(r)/, why
/wai/, what /wot/, whale /wciI/, wheel /wi:I/, whether /'wcðc(r)/,
whine /wain/.
Note that in some dialects of English (especially in some American
dialects and in conservative British, e.g., Scottish pronunciations) it
represents a voiceless labiovelar, /\/.
2
For these speakers there is a
difference between which /\itj/ and witch /witj/, where /\cc(r)/ and
wear /wcc(r)/.
qu It regularly represents the phoneme /k/ word-finally as in cheque
/tjck/, antique /æn'ti:k/, clique /kIi:k/.
2
This sound is similar to the sequence of a /h/ and a /w/.
165
Chapter 11
It regularly represents the phoneme combination /kw/ in other
positions as in queen /kwi:n/, question /'kwcstjn/, request /ri'kwcst/,
banquet /'bænkwit/.
It irregularly represents the phoneme /k/ in queue /kju:/, quay /ki:/,
liquor /'Iikc(r)/, liqueur RP /Ii'kjoc(r)/.
gu It regularly represents the phoneme /g/ (for the pronunciation rules of
g see below) as in guerrilla /gc'riIc/, guest /gcst/, guardian /'go:ojcn/,
colleague /'koIi:g/, guy /gai/.
It regularly represents /gw/ in the combination ngu as in language
/'Iængwio¸/, distinguish /oi'stingwij/.
In some words gu is actually a sequence of g + u and is pronounced
as /gju:/ or /gjo/ as in argument /'o:gjomcnt/, Jaguar /'o¸ægjoc(r)/.
ck It regularly represents the phoneme /k/ as in back /bæk/, hacker
/'hækc(r)/, reckon /'rckcn/, docking /'ookin/, sucker /'s\kc(r)/.
cz It regularly represents the phoneme /tj/ as in Czech /tjck/,
Czechoslovakia /¡tjckcsIc'vækic/, czardas RP /'tjo:oæj/ (GA
/'tjoroo:j/).
dg It regularly represents the phoneme /o¸/ in environments where g
would represent /o¸/ as in edge /co¸/, hedge /hco¸/, badger
/'bæo¸c(r)/, gadget /'gæo¸it/, budget /'b\o¸it/, bridge /brio¸/.
It irregularly represents the phoneme sequence /dg/ in some words as
in Edgar /'cogc(r)/.
xc It regularly represents the phoneme sequence /ks/ before the vowel
letters e, i, y as in excited /ik'saitio/, excellent /'ckscIcnt/, exception
166
Letter-to-sound rules – Part 1
/ik'scpjn/.
sc It regularly represents the phoneme /s/ before the vowel letters e, i, y
as in science /'saicns/, scenery /'si:ncri/, sci-fi /'saiIai/, scissors
/'sizcz/.
In the last part of this chapter we take a look at the rules that regulate some of
the letter-to-sound correspondences mentioned above.
Consonant rules
Lexical palatalization
Lexical palatalization is a rule that operates inside a word, i.e., a lexical item,
and regulates the pronunciation of the consonant letters <t>, <d>, <s>, <c>,
<x> representing the alveolar obstruents /t/, /o/, /s/, /z/ before an underlying
/j/ phoneme represented by the vowel letters <i> or <u> in certain
environments. It is an obligatory process independent of style, speech
situation or tempo (in contrast to cross-word palatalization, discussed in
Chapter 7).
1. Palatalization by <i>
An alveolar obstruent will be palatalized before <i> if the vowel letter if the
vowel letter does not represent a stressed vowel and it is followed by another
vowel letter. It is also important that palatalization does not apply in word-
initial position (for exceptions see Palatalization by <u>). This environment
of palatalization is often referred to as CiV as the alveolar consonant, i.e., C,
is followed by the vowel letter <i> and another vowel letter, i.e., V, hence the
name CiV. (We have seen a different effect of the same environment in CiV
Laxness and CiV Tenseness in Chapter 3.) The vowel letter <i> is usually not
167
Chapter 11
pronounced at all, e.g., soCIAl /'scojI/ (the relevant letters of the words will
be capitalized).
alveolar C

unstressed <i>

V letter

so c i a l /'scojI/
an c i e nt /'cinjnt/
mi ss i o n /'mijn/
vi s i o n /'vi¸n/
an x i o us /'ænkjcs/
men t i o n /'mcnjn/
ques t i o n /'kwcstjn/
sol d i e r /'scoIo¸c(r)/
Note that because of the above requirements there is no lexical palatalization
if the vowel letter <i> represents a stressed vowel, e.g., soCIEty /sc'saicti/, or
if it is not followed by another vowel letter, e.g., construcTIVe /kcn'str\ktiv/.
2. Palatalization by <u>
An alveolar obstruent will also be palatalized before <u> if the vowel letter
represents an unstressed vowel and it is followed by another vowel letter or a
consonant+vowel letter combination. Palatalization by <u> does not apply in
word-initial position, except in the words sugar /'jogc(r)/ and sure /joc(r)/. It
logically follows from the above that there is no palatalization if <u>
represents a stressed vowel, e.g., asSUME /c'sju:m/, or if <u> is not followed
by another vowel letter or consonant+vowel letter combination but two
consonant letters or one consonant letter in word-final position, e.g.,
constrUCT /kcn'str\kt/, cacTUS /'kæktcs/. The word maTURE /mc'tjoc(r)/ is
168
Letter-to-sound rules – Part 1
exceptional as Lexical Palatalization does apply although <u> is stressed (but
it is usually /mc'tocr/ in GA).
alveolar C

unstressed <u>

V letter

u s u a l /'ju:¸ocI/
ca s u a l /'kæ¸ocI/
vi s u a l /'vi¸ocI/
ac t u a l /'æktjocI/
sen s u a l /'scnjocI/
alveolar C

unstressed <u>

C letter

V letter

na t u r e /'ncitjc(r)/
litera t u r e /'Iitritjc(r)/
mea s u r e /'mc¸c(r)/
cen s u r e /'scnjc(r)/
The dropping of <b> and <g>
We have already noted in the discussion above (as well as in Chapter 5) that
the consonants b and g are often dropped in certain positions. As it will be
clear from what follows, the two consonants are affected by the very same
letter-to-sound rule. These voiced non-coronal stops are dropped if they are
preceded by a homorganic nasal and are in morpheme-final position. It
follows, then, that the two stops are not dropped in morpheme-initial and
internal positions.
b dropped b pronounced g dropped g pronounced
climber /'kIaimc(r)/
number (adj) /'n\mc(r)/
thumb /0\m/
timber /'timbc(r)/
number (n) /'n\mbc(r)/
sombrero /som'brccrco/
singer /'sinc(r)/
hanging /'hænin/
belonged /bi'Iono/
fungus /'I\ngcs/
bingo /'bingco/
Bangor /'bængc(r)/
169
Chapter 11
The dropping of <h>
The consonant h has a very restricted distribution in both English and
Hungarian. In both languages the h is silent in word-final position and before
consonants.
h silent in Hungarian
céh /tsc:/, juh /ju/, csehnek
/'tjrnrk/, méhtől /'mc:to:I/
h silent in English
Shah /jo:/, Sarah /'sccrc/, John
/o¸on/, yacht /jot/
In a great many words in Hungarian the letter h is pronounced before a vowel
or in final position. Note, however, that in final position it is not a glottal
fricative, /h/, that occurs in pronunciation but a voiceless velar fricative, /x/
(the same sound as the so-called Ach-Laut in German), as in doh /oox/,
potroh /'potrox/, jacht /joxt/, Bachtól /'boxto:I/.
Another difference between the two languages lies in the behaviour of
h before vowels: in Hungarian h is always pronounced before vowels while
in English, as mentioned in Chapter 7, h is only pronounced before stressed
vowels. Before unstressed vowels it is always deleted in English (recall
examples like véhicle vs. vehícular), except in word-initial position, where it
is pronounced even before unstressed vowels, e.g., in both hállow and helló.
h pronounced in Hungarian h pronounced in English
ház /ha:z/, juhéj /'juhc:j/, csehek
/'tjrhrk/, méhek /'mc:hrk/
historical /hi'storikI/, ahead /c'hco/,
height /hait/, Soho /'scohco/
In some words, of typically French origin, the h is irregularly silent in initial
position as in honest /'onist/, honour /'onc(r)/, heir /cc(r)/, hour /aoc(r)/, and
170
Letter-to-sound rules – Part 1
before a stressed vowel in exháust, exhíbit, exhílarate, exhórt and all their
derivatives.
Velar Softening
Velar Softening regulates the pronunciation of the consonant letters c and g,
which have two regular pronunciations, a "hard" one, a velar stop, and a
"soft" one, a coronal sibilant: c may be pronounced as /k/ or /s/ while g may
represent /g/ or /o¸/. According to the rule, c and g are pronounced soft, i.e.,
as a coronal sibilant, before the vowel letters <e>, <i>, <y> regardless of
whether the vowel letter is pronounced and how it is pronounced, i.e., it is a
purely graphic rule only based on spelling.
c regularly pronounced as /s/ g regularly pronounced as /o¸/
cellar /'scIc(r)/, facilitate /Ic'siIitcit/,
cyber /'saibc(r)/, dance RP /oo:ns/
(GA /oæns/)
fragile RP /'Iræo¸aiI/ (GA /'Iræo¸I/),
sergeant /'so:o¸cnt/, stingy /'stino¸i/,
gyroscope /'o¸aircskcop/
There are quite a few cases when c and (especially) g fail to be pronounced
soft in this environment, for instance:
c irregularly pronounced as /k/ g irregularly pronounced as /g/
soccer /'sokc(r)/, Celtic /'kcItik/,
sceptical /'skcptikI/
get /'gct/, give /'giv/, hunger
/'h\ngc(r)/, finger /'Iingc(r)/, begin
/bi'gin/, girl /ga:I/
In other positions, i.e., before other vowel letters, before consonant letters
and in word-final position c and g are normally pronounced hard, as a velar
stop, although exceptions exist, e.g., Caesar, gaol, margarine, veg, etc. Note
171
Chapter 11
that in morpheme-final position after a nasal, g is not pronounced (see
above).
c regularly pronounced as /k/ g regularly pronounced as /g/
catarrh /kc'to:(r)/, function /'I\nkjn/,
culinary /'k\Iincri/, cancer
/'kænsc(r)/
bogus /'bcogcs/, language /'Iængwio¸/,
distinguish /oi'stingwij/, jungle
/'o¸\ngI/
We should also remember that root-final g is not softened if a regular,
productive suffix starting with <e>, <i>, or <y> is added as in bigger
/'bigc(r)/ and not */'bio¸c(r)/, longest /'Iongist/ and not */'Iono¸ist/, bagged
/bægo/ and not */bæo¸o/.
There are cases, though, when a non-productive suffix is added to the
stem, a suffix which is normally placed after a bound and not a free stem. In
such cases, if the stem ends in c or g (which is, of course, pronounced hard in
final position if no suffix follows) and the non-productive suffix begins with
<e>, <i>, or <y>, then the stem-final consonant changes into a coronal
sibilant, i.e., into its soft pronunciation: Velar Softening as a process has
taken place. In just the other way round, if a stem ends in a c or g in their soft
pronounciation when followed by a suffix then in the unsuffixed form they
will be present with their hard pronunciation.
<c> electri[k] - electri[s]ity
indu[k]tion – indu[s]e
dedu[k]tion – dedu[s]e
<g> analo[g]ous – analo[o¸]y
ma[g]us – ma[o¸]ic
lo[g]o – lo[o¸]ic
172
Letter-to-sound rules – Part 1
Yod-Dropping
This rule was introduced in Chapter 5 as a phonotactic restriction on
homorganic consonant clusters, however, it may as well be conceived of as a
letter-to-sound rule. Although it refers to the deletion of a consonant sound
/j/, it is used to distinguish between two very similar vowel pairs of English,
the Plain-Tense /ju:/-/u:/ and their Broken-Tense variants /joc/-/oc/.
The assumption underlying this distinction is that /j/+/u:/ or /j/+/oc/
sequences are not really combinations of two separate sounds but form one
unit, one complex vowel, like a diphthong. The main reason for this is that
the combination /j/+/u:/ has interesting phonotactic characteristics as it shows
a very special behaviour in syllable structure. If we list all the possible two or
three-member consonant clusters that may start a syllable in English, then we
will find that whenever the last member of such a cluster is /j/, it is always
followed by the vowel /u:/ or /oc/. Of course, it cannot be a coincidence and
the most obvious explanation for this state of affairs is that /j/ and /u:/ or its
Broken-Tense variant /oc/ form one unit, /ju:/ and /joc/. This complex vowel
is regularly represented by <u>, <eu>, <ew>, <ue>, <ui> in spelling.
However, it often happens that although one of these possible spellings
occurs, in pronunciation we have /u:/ or /oc/, i.e., /j/ is missing. This is due to
the rule of Yod-Dropping, which deletes the /j/ of the complex vowel /ju:/ in
certain environments.
1. Obligatory Yod-Dropping
Yod-Dropping is obligatory in RP after palato-alveolars, /j, ¸, tj, o¸, r/ and
consonant+/I/ sequences as in the words parachute /'pærcju:t/, luxurious
173
Chapter 11
/I\g'¸ocrics/, mature RP /mc'tjoc(r)/, June /o¸u:n/, July /o¸u:'Iai/, rude /ru:o/,
rumour /'ru:mc(r)/.
Recall, however, that in GA Yod-Dropping is much more extensive as
it applies after all coronal consonants – dentals, alveolars, palato-alveolars.
As a result of this, many words are pronounced differently in (conservative –
see below) RP and in GA.
RP GA
enthusiasm /in'0ju:ziæzcm/ /in'0u:ziæzcm/
new /nju:/ /nu:/
tuna /'tju:nc/ /'tu:nc/
dubious /'oju:bics/ /'ou:bics/
super /'sju:pc(r)/ /'su:pcr/
exuberant /ig'zju:bcrcnt/ /ig'zu:bcrcnt/
illusion /i'Iju:¸n/ /i'Iu:¸n/
2. Optional Yod-Dropping
In RP, there is a tendency to also drop the /j/ in some environments,
especially in the speech of speakers belonging to the younger generations.
Elderly speakers still often retain the Yod in these words. This version of
Yod-Dropping is optional, it depends on style and speech tempo. It applies
after the consonants /s, z, I/ as in super /'s(j)u:pc(r)/, suit /s(j)u:t/, assume
/c's(j)u:m/, exuberant /ig'z(j)u:bcrcnt/, presume /pri'z(j)u:m/, illusion
/i'I(j)u:¸n/, lukewarm /'I(j)u:kw5:m/, lewd /I(j)u:o/.
3. The absence of Yod-Dropping
It has also been mentioned in Chapter 5 above that if the complex vowel /ju:/
occurs in a completely unstressed syllable, Yod-Dropping is prohibited not
just in RP but also in GA, where Yod-Dropping is otherwise obligatory in a
174
Letter-to-sound rules – Part 1
much wider range of environments than in RP. Thus, the rule cannot apply in
words like value /'væIju:/, consulate /'konsjoIct/, annual /'ænjocI/, menu
/'mcnju:/.
In this chapter we have seen the regular and irregular pronunciation values of
single consonant letters and consonant digraphs, as well as the positions in
which they are silent. Then we have also seen the most important letter-to-
sound rules that refer to the pronunciation of consonant letters.
175
12. Letter-to-sound rules – Part 2: Vowels
Before you study this chapter, check whether you are familiar with the
following terms: CiV, digraph, free U, lax (plain/broad), laxing rules,
orthography, Pre-R Breaking, Pre-R Broadening, stem, suffix, tense
(plain/broken), Yod-Dropping, Vowel Shift, Trisyllabic Laxness
This chapter deals with the area of the English language that has driven many
language learners crazy throughout the years: the relationship between the
spelling and pronunciation of English vowels. This is probably the area that is
the most difficult for language learners, especially Hungarian learners as its
principles are very different from those found in Hungarian. Hungarian letter-
to-sound rules for vowels are very simple, each vowel letter represents one
vowel sound and each vowel sound is represented by one vowel letter; there
are a few minor alternations both in length and quality but they are not
significant. In English, however, each vowel sound may be represented by
quite a few vowel letters or digraphs, and each vowel letter and digraph may
stand for a few vowel sounds.
The main reason for this many-to-many relationship between English
vowel letters and sounds originates in the fact that, as introduced in Chapter
3, there are two major types of full vowel in English, tense and lax, and each
vowel letter has tense and lax pronunciations as well. Moreover, tense vowels
are further divided into two subclasses, Plain-Tense and Broken-Tense
vowels, while lax vowels are classified into the Plain-Lax and Broad-Lax
subcategories. Logically, each vowel letter will have not just a tense and a lax
pronunciation but a Plain-Tense, a Broken-Tense, a Plain-Lax and a Broad-
Lax pronunciation. It is these four different pronunciations that we turn to
Letter-to-sound rules – Part 2
first, followed by a discussion of the pronunciation values of vowel digraphs,
the rules determining the pronunciation of vowel letters and finally the many
different kinds of regular and irregular exceptions to leave the best for last.
The regular sound values of single vowel letters are as follows:
<a> <e> <i> = <y> <o> <u>
Plain-Tense /ci/ /i:/ /ai/ /co/ /(j)u:/
Broken-Tense /cc/ /ic/ /ai(c)/ /5:/
3
/(j)oc/
Plain-Lax /æ/ /c/ /i/ /o/ /\/
Broad-Lax /o:/ /a:/ /a:/ /5:/
2
/a:/
A few important remarks are due here concerning the table above. First, we
have to note an interesting peculiarity of English, namely that one of its
vowel phonemes, /o/ does not have a regular representation in spelling – we
will only find it in the last section of the chapter, containing irregularities.
Second, the vowel letter <o> actually has only three different pronunciations
as its Broken-Tense and Broad-Lax pronunciations are phonetically identical.
Third, the vowel letter <u> has six different pronunciations instead of the
expected four since it is also affected by the rule of Yod-Dropping, i.e., in the
tense values there is a yodless and yodful pronunciation (see the discussion
below and in the previous chapter). Fourth, it is very easy to remember the
Plain-Tense value for each vowel letter as it is the pronunciation used to
name the letter in the alphabet or to spell a word letter by letter.
Let us now take a look at the pronunciation values of vowel
digraphs.
177
Chapter 12
<ai>=<ay> <ei>=<ey> <ea> <ee> <ie>
Plain-Tense /ci/ /ci/ /i:/ /i:/ /i:/
Brkn-Tense /cc/ /cc/ /ic/ /ic/ /ic/
<oa> <oo> <eu>=<ew> <ui> <ou>=<ow>
Plain-Tense /co/ /u:/ /(j)u:/ /(j)u:/ /ao/
Brkn-Tense /5:/
3
/oc/ /(j)oc/ /(j)oc/ /ao(c)/
<oi>=<oy> <au>=<aw>
Plain-Tense /5i/ Plain-Lax -
Brkn-Tense /5i(c)/ Broad-Lax /5:/
1
Again, some generalizations may be found in the tables above. First, vowel
digraphs regularly represent tense vowel sounds with one exception only,
<au>=<aw>. Second, the letters <i> - <y>, and <u> - <w> play the same role
in the digraphs. As it will be clear from the examples below there is even a
tendency (although not a rule) to predict where we find which.
Let us now take a look at some examples for the above sound values
of single vowel letters and vowel digraphs.
<a> <e> <i> = <y> <o> <u>
Plain-Tense mate /mcit/ scene /si:n/ bite /bait/ sole /scoI/ cute /kju:t/
rude /ru:o/
Broken-
Tense
care /kcc(r)/ here /hic(r)/ fire /Iaic(r)/ sore /s5:(r)/ cure /kjoc(r)/
sure /joc(r)/
Plain-Lax bat /bæt/ bet /bct/ bit /bit/ bond /bono/ but /b\t/
Broad-Lax car /ko:(r)/ her /ha:(r)/ firm /Ia:m/ born /b5:n/ burn /ba:n/
178
Letter-to-sound rules – Part 2
<ai>=<ay> <ei>=<ey> <ea> <ee> <ie>
Plain-Tense bay /bci/ obey /c'bci/ beat /bi:t/ bee /bi:/ believe /bi'Ii:v/
Brkn-Tense fair /Icc(r)/ heir /cc(r)/ fear /Iic(r)/ beer /bic(r)/ pier /pic/
<oa> <oo> <eu>=<ew> <ui> <ou>=<ow>
Plain-Tense boat /bcot/ boot /bu:t/ few /Iju:/
drew /oru:/
suit /sju:t/
fruit /Iru:t/
house /haos/
how /hao/
Broken-
Tense
boar /b5:(r)/ boor /boc(r)/ Europe
/'jocrcp/
Jewry
/'o¸ocri/
Muir /mjoc/ our /aoc(r)/
Bowra
/'bao(c)rc/
<oi>=<oy> <au>=<aw>
Plain-Tense boy /b5i/ Plain-Lax -
Broken-
Tense
Moira
/'m5i(c)rc/
Broad-Lax claw /kI5:/
As for the digraphs ending in <i> or <y> and <u> or <w> we can claim that
there is a tendency to use <i> and <u> inside a word, e.g., fair, heir, Europe,
our, Moira, sauce, and to use <y> and <w> in word-final position, e.g., bay,
obey, drew, how, boy, claw. However, one should be careful as this is not an
exceptionless rule, rather a tendency and there are quite many exceptions for
it, e.g., town, Bowra, Jewry, powder, drown, bowl.
In the following we enumerate and discuss the rules which are
responsible for the letter-to-sound correspondences in English vowels. The
first such rule concerns the distinction between the vowels /ju:/, /joc/ and /u:/,
/oc/, that is the yodful and yodless variants. The basic rule is that the five
graphic representations, <u>, <eu>, <ew>, <ue>, <ui> normally stand for the
variants starting with /j/. All other representations normally stand for the
variants without /j/, typically <oo> and in some irregular cases <o> and <ou>
(for irregular cases, see below). However, in many environments the vowel
179
Chapter 12
letters and digraphs normally representing /ju:/ or /joc/ are pronounced
without /j/, a case often referred to as Yod-Dropping. The details of this rule
may be found at the end of Chapter 11.
Before actually discussing the pronunciation rules of vowel letters, we
must introduce a pair of notions referring to the positions of vowel letters in
the orthographic word. The two types of graphic positions are called Free
Position and Covered Position. In the examples below the vowel letter in
question is in underlined boldface and the letters of the relevant environment
are in capitals. (V = vowel letter, C = consonant letter, S = stop, L = liquid;
the hyphen indicates the position of the stressed vowel letter.)
Free Position Covered Position
(1) -V (2) -CV (3) -SLV (4) -# (5) -CC (6) -C#
diAl
goIng
diE
doEr
baKE
muTE
caRE
hoCUs-
poCUs
aPROn
cyCLone
maPLE
oGRE
my#
so#
be#
fly#
baNK
thuNDer
yaCHT
fiRSt
spaM#
fiT#
heR#
spoT#
As it can be seen in the table above, vowel letters are said to be in free
graphic position if they are followed by another vowel letter, a
consonant+vowel letter, two consonant letters representing a stop and a liquid
sound plus a vowel letter and also when they are word-final. Single vowel
letters are in covered graphic position if they are followed by two consonant
letters (which are either not followed by a vowel letter, or if they are, then
they do not represent a stop and a liquid sound) and if they are followed by a
word-final consonant letter.
Note that there are some problematic cases. On the one hand, the
consonant letter <x> normally represents the sound sequence /ks/, so it must
180
Letter-to-sound rules – Part 2
be counted as a sequence of two consonant letters. As a result, words like
taXi will belong to class (5) in the table. On the other hand, if we recall that
consonant digraphs regularly represent a single consonant sound, they must
be counted as one letter. As a result again, words like goPHer will belong to
class (2) and buSH to (6). Bear in mind that these are graphic positions for
letters – whether these letters are pronounced or not does not matter. That is
why dial and die, for instance, belong to the same category. The word yacht
(/jot/) contains a silent digraph followed by another consonant letter, but is
totally identical to bank (in class 5) in this respect.
There are two rules that help us decide the pronunciation of vowel
letters on the basis of whether they are in a free or covered position. If we
take a look at the words in columns (1) to (4), we can see that they all contain
a stressed (plain or broken) tense vowel. Thus, the Free Position Basic Rule
(FPBR) can be stated as follows:
FPBR: Stressed single vowel letters in free position are normally pronounced
as tense. (See exceptions below.)
On the other hand, columns (5) and (6) both contain words in which
the stressed vowels are (plain or broad) lax. On the basis of this we can state
the Covered Position Rule (CPR) as follows:
CPR: Stressed single vowel letters in covered position are normally
pronounced as lax. (See exceptions below.)
Remember that these two rules only refer to stressed single vowel
letters – unstressed or weak vowels behave differently, while vowel digraphs
usually represent tense vowels, i.e., a rule is unnecessary in their case. Also,
181
Chapter 12
note that the expression "stressed single vowel letter" is just a short hand for
a single vowel letter representing a stressed vowel.
In the following part of the chapter we take a look at exceptions, i.e.,
those cases when a vowel letter is not pronounced according to the two rules
above. Exceptions may fall into two different types: those that simply do not
obey the two rules but the vowel is pronounced with one of its regular
pronunciations (discussed in the very first table above) – this type of
exception is often called a tenseness reversal – and those that involve a
vowel letter pronounced as an irregular sound, i.e., it has a sound value which
is not one of its four regular pronunciations – a kind of exception often
referred to as a quality deviation. We address the two types in this order
below.
Tenseness reversals
In some cases the stressed single vowel letter is not pronounced with the
tense/lax value predicted by the FPBR or the CPR, i.e., it is pronounced with
a tenseness/laxness which is just the opposite of what is expected on the basis
of the rules.
Tenseness reversal exceptions to the Covered Position Rule
bind /baino/, bold /bcoIo/, both /bco0/, chamber /'tjcimbc(r)/, comb /kcom/,
danger /'ocino¸c(r)/, fight /Iait/, find /Iaino/, fold /IcoIo/, gross /grcos/,
island /'aiIcno/, kind /kaino/, manger /'mcino¸c(r)/, mild /maiIo/, most
/mcost/, old /coIo/, range /rcino¸/, Ruth /ru:0/, sign /sain/, soldier
/'scoIo¸c(r)/, told /tcoIo/, truth /tru:0/, etc.
182
Letter-to-sound rules – Part 2
Tenseness reversal exceptions to the Free Position Basic Rule
In the case of the Free Position Basic Rule we may add one more clause to
the rule: stressed single vowel letters in free graphic position are pronounced
tense unless they are laxed by one of the laxing rules. These laxing rules are
those that we have already discussed in Chapter 3 in connection with the
Vowel Shift. Recall that in the Vowel Shift a tense vowel of a stem becomes
lax if a certain kind of suffix is added. The original stem vowel is tense,
which means that if it is a regular word then the vowel letter representing the
tense vowel must be in a free graphic position. If a suffix is added, the
stressed vowel may still be in free position but its pronunciation becomes lax
because of one of the laxing rules. Let us repeat the most important features
of the laxing rules for the convenience of the reader now focussing on the
spelling of the stem and the suffix. Before we start enumerating the laxing
rules, we must remember that the vowel /(j)u:/ is non-laxable in any position,
i.e., it is a regular exception to all the laxing rules below.
Trisyllabic Laxness
If the stressed vowel is in at least the third-last syllable of the word then it
must be lax. This is a result of the Trisyllabic Laxing rule, an active
phonological rule which applies if a suffix is added to the stem. In the present
context, it is simply applied to any word even without adding suffixes.
183
Chapter 12
<a> <e> <i> = <y> <o>
Canada
laminate
salivate
Capricorn
cabaret
radical
president
hesitate
regiment
federate
general
Hemingway
miracle
similar
cinema
pitiful
militant
typical
solitude
domino
mahogany
positive
solitary
dominant
In some words the stressed single vowel letter remains tense as required by
the FPBR in spite of the fact that it is in a trisyllabic position, e.g., isolate,
microphone, notify, nightingale, omega, etc. Also, recall from Chapter 3 that
regular, productive suffixes are not counted when determining whether a
word serves as an input to the rule. Free <u> is non-laxable; this is illustrated
by examples like cubicle, puritan, enumerate.
Laxing by ending
In some words the stressed single vowel letter in free graphic position is
pronounced lax despite the FPBR because one of the so-called laxing endings
follows.
<a> <e> <i> = <y> <o>
manic
static
habit
tablet
establish
parish
metric
intrepid
edit
level
Eric
perish
clinic
timid
optimistic
limit
critic
lyric
polish
tonic
solid
shovel
novel
comet
184
Letter-to-sound rules – Part 2
There are a few words in which the stressed vowel is followed by one of the
laxing endings but still it is pronounced as tense as in basic, strategic, label,
navel, secret. Recall also that the ending -ish is only laxing when producing a
noun or a verb but it is non-laxing if it makes an adjective, hence the
difference between Polish /'pcoIij/ vs. polish /'poIij/ and Swedish /'swi:oij/
vs. finish /'Iinij/.
1
Free <u> is non-laxable; this is illustrated by examples like
Punic, Cupid, rubric, unit.
Laxing by free <u>
The stressed single vowel letter occurring in free graphic position is regularly
pronounced lax if it is followed by a free vowel letter <u> in the next
syllable, i.e., a <u> followed by a vowel or a consonant+vowel combination.
<a> <e> <i> = <y> <o>
gradual
value
sensual
schedule
visual
ritual
soluble
module
A typical exception to this laxing rule is the ending -ure, which often attaches
to a stem whose vowel is pronounced tense in spite of the free <u> of the
suffix, as in closure, erasure, nature, etc. Free <u> itself is non-laxable; this
is illustrated by examples like usual.
CiV laxing
The rule of CiV laxing only applies if there is a stressed vowel letter <i> or
<y> which is followed by a consonant letter + another vowel letter <i> + one
more vowel letter. The rule does not apply to any of the other vowel letters
(cf. CiV tensing below). Its application may be witnessed in words like idiot,
1
The word Spanish is exceptional.
185
Chapter 12
idiom, decision, revision, dominion, Syria. The rule does not have any
exceptions.
Before continuing to a rule which works just in the opposite direction as
compared to the ones we have just discussed, let us note that there is a further
restriction on all laxing rules, namely, that vowels occurring before another
vowel, i.e., prevocalic vowels, are non-laxable, they have to remain tense
even before suffixes, even if one of the laxing rules could apply. That is, as
explained in Chapter 3, Prevocalic Tenseness is stronger than any of the
laxing rules. The following examples could as well be subject to either
Trisyllabic Laxness or Laxing by ending, still, their stressed vowel
(underlined) is tense.
<a> <e> <i> = <y> <o>
prosaic
archaic
laity
mosaic
nucleic
spontaneity
simultaneity
deity
variety
diet
sobriety
anxiety
heroic
stoic
echoic
poet
CiV tensing
Recall that the rule of CiV tensing is the mirror image of CiV laxing in two
senses: firstly, it applies to all vowel letters except <i> or <y> – remember
that CiV laxing only applies to these. Secondly, it requires that the stressed
vowel letter followed by CiV be tense, while CiV laxing enforces the
opposite. A very important feature of this rule is that it overrides the laxing
rules, i.e., it blocks their application and applies instead resulting in tense
vowels in an environment where some of the laxing rules, in most cases
Trisyllabic Laxness, should apply.
186
Letter-to-sound rules – Part 2
<a> <e> <o>
maniac
Australia
radiate
serious
serial
senior
notion
Gloria
phobia
There are no examples listed for the vowel letters <i/y> and <u>. For the
former we noted that it is made lax in this environment; for the latter we have
already mentioned that it is an exception to all the laxing rules, that is, in free
graphic position it will always be tense even if a laxing rule could apply (cf.
fusion, union, Muriel, etc.). Since CiV tensing makes a vowel tense, it is not
necessary to indicate it for a vowel letter which is always pronounced as
tense in free positions anyway.
The rule has a few exceptions in which the vowel letters <a>, <e>,
<o> are followed by CiV in spelling but they are pronounced as lax – but
remember, this cannot be the result of CiV laxing as that rule only applies to
the vowel letter <i/y>! Exceptions to CiV tensing include words like Daniel,
Slovakia, special, national, precious, RP patriot etc.
Irregular tenseness reversals
There are cases where the stressed single vowel letter occurring in free
graphic position is not made lax by any of the laxing rules but, for some
reason, it is still lax. This typically happens in the last or only syllable of a
word, which is pronounced as if the final silent <e> was not present at the
end, and also in the second-last syllable of the word.
187
Chapter 12
Irregular tenseness reversal in last or
only syllable
Irregular tenseness reversal in
second-last syllable
<a> are, have
<e> were, allege
<i> live (v), give
<o> gone, RP shone
2
<a> cabin, Latin, salad, atom
<e> lemon, tenant, very, devil
<i> city, pity, linen, consider
<o> body, copy, orange, forest
<u> punish, study, Dublin, public
Quality deviations
In a great number of words the single vowel letters or vowel digraphs are not
pronounced with one of their regular pronunciation values given in the tables
at the beginning of this chapter. Since these are not simple exceptions from
the FPBR or the CPR, i.e., they do not belong to any of the tenseness reversal
cases discussed above, they are often called quality deviations as the
graphemes deviate from their own regular qualities and take on the quality of
some other vowel grapheme. As a result, the spelling and pronunciation of
these words is not predictable by any of the rules we have seen, they must be
memorized as exceptions.
1. Isolated deviating words
There are a few sporadic quality deviations which are isolated in the sense
that there are very few examples for them, e.g., <a> = /c/ as in any, many, ate
RP /ct/ (cf. GA /cit/), Thames (exhaustive list!), <u> = /i/ as in busy, business,
<u> = /c/ as in bury, burial, or <oa> = /5:/ as in broad, abroad (cases of
broadness without r).
2
GA /joon/ is regular.
188
Letter-to-sound rules – Part 2
2. Groups of deviating words
Some quality deviations are unpredictable but much more common than the
ones above and can be classified according to some pronunciation or spelling
characteristics. (A few of these groups have been mentioned in Chapter 4.)
Deviations due to neighbouring sounds/letters
WANT-words <a> = /o/ want, was, wash, swan, quality, quantity, squash
WAR-words <a> = /5:/ war, thwart, dwarf, quarter, swarm, warmth
CALL-words <a> = /5:/ call, fall, bald, talk, alter, stalk, walk, Balkans
ASK-words <a> = /o:/ ask, dance, fast, past, class, path, last, example
CALM-words <a> = /o:/ calm, palm, calf, halve, balm, psalm
WORK-words <o> = /a:/ work, worth, world, word, worth, Wordsworth
Foreignisms (French or Italian loanwords with spelling imitating the original)
MACHINE-words <i> = /i:/ machine, clique, kilo, ski, pizza, visa
CREPE-words <e> =/ci/ crepe, fete, suede, régime, café, née, fiancé(e)
CHAUFFEUR-words <au> = /co/ chauffeur, mauve, chauvinism, sauté
SOUP-words <ou> = /u:/ soup, group, route, souvenir, rouge, douche (also
some original English words like you, youth, wound (n))
MEMOIR-words <oi> = /wo:/ memoir, bourgeois, reservoir
Traditional spellings with <e>, <ea>
CLERK-words <e> =/o:/ clerk, sergeant, Derby, Berkeley
3
BREAD-words <ea> = /c/ bread, health, meant, pleasant, jealous
EARTH-words <ea> = /a:/ earth, heard, pearl, earn
BEAR-words <ea> = /cc/ bear, swear, tear (v)
3
GA clerk /kIark/ and Derby /'oarbi/ are regular. Also, the place in California called Berkeley
is pronounced /'ba(r)kIi/.
189
Chapter 12
Traditional spellings with <o>, <oo>, <ou>, <u>
LOVE-words <o> = /\/ love, come, onion, mother, London, among
MOVE-words <o> = /u:/ move, do, prove, tomb
LOOK-words <oo> = /o/ look, book, crook, good, wool
TROUBLE-words <ou> = /\/ trouble, country, courage
4
, young
SOUL/BOWL-words <ou/ow> = /co/ soul, shoulder, bowl, know
THOUGHT-words <ough> = /5:/ thought, brought, nought
PUT-words <u> = /o/ put, full, butcher, cushion
However numerous these exceptions may seem, the majority of English
words, including new coinages, do conform to the basic letter-to-sound rules
introduced in the first half of the chapter.
4
GA /'kario¸/ – cf. the exceptions to the Carrot-Rule in Chapter 4.
190
Bibliography
Selected readings in English phonetics and phonology
Carney, Edward (1994) A survey of English spelling. London: Routledge.
Carr, Philip (1999) English phonetics and phonology. An introduction.
Oxford: Blackwell.
Cook, Vivian (2004) The English writing system. London: Hodder Arnold.
Giegerich, Heinz J. (1992) English phonology: An introduction. Cambridge:
CUP.
Gimson, A.C. (1980) An introduction to the pronunciation of English. 3rd
edition. London: Edward Arnold.
Harris, John (1994) English sound structure. Oxford: Blackwell.
Kovács, János and Siptár, Péter (1986/1989) Angolra hangolva. Budapest:
Tankönyvkiadó.
Kovács, János and Siptár, Péter (2000) Újra angolra hangolva. Budapest:
Helikon.
Kreidler, Charles W. (1989) The pronunciation of English: A coursebook in
phonology. Oxford: Blackwell.
Ladefoged, Peter (1993) A course in phonetics. 3rd edition. New York:
Harcourt Brace Jovanovich.
Ladefoged, Peter (2001) Vowels and consonants. An introduction to the
sounds of languages. Oxford: Blackwell.
Nádasdy, Ádám (2003) Practice book in English phonetics and phonology.
Budapest: Nemzeti Tankönyvkiadó.
Nádasdy, Ádám (2006) Background to English pronunciation. Budapest:
Nemzeti Tankönyvkiadó.
Odden, David (2005) Introducing phonology. Cambridge: CUP.
191
Bibliography
Roach, Peter (1983/1991/2000) English phonetics and phonology: A practical
course. Cambridge: CUP.
Shockey, Linda (2003) Sound patterns of spoken English. Oxford: Blackwell.
Wells, John C. (1982) Accents of English. Volume 1. Cambridge: CUP.
Yavaş, Mehmet (2006) Applied English phonology. Oxford: Blackwell.
192
Subject index
This index lists all the technical terms introduced and/or highlighted as
significant in the book. They appear in boldface in the text. Whenever
thought to be relevant and useful, the RP pronunciation of the term is given in
slashes and the Hungarian equivalent in italics. The first number refers to the
chapter, the number(s) following the colon indicate(s) the page(s) where the
term is (first) mentioned.
/5:/ 3:34
accent /'ækscnt/ akcentus 1:3,
10:139
accidental gap véletlen hiány 5:61
acoustics /c'ku:stiks/ akusztika 1:7
Adam's apple ádámcsutka 6:78
affricate /'æIrikct/ affrikáta,
zárréshang 2:15
Aitch-Dropping /h/-elhagyás 7:97,
11:170
allophone /'æIcIcon/ allofón 1:7
Alternating Stress Rule
Hangsúlyváltakozás 9:131
alternation alternáció, váltakozás
1:8
alveolar ridge /'æIvi¡coIc 'rio¸/
fogmeder 2:16
ambisyllabic /¡æmbisi'Iæbik/
ambiszillabikus 2:26
approximant /c'proksimcnt/
approximáns, közelítő hang
2:16
articulation /o:¡tikjo'Icijn/
artikuláció, hangképzés 1:6
ASK-words 4:59, 12:189
aspiration /¡æspi'rcijn/ aspiráció,
hehezet 2:22, 6:86
assimilation /c¡simc'Icijn/
asszimiláció, hasonulás 7:95,
see place assimilation, voice
assimilation
asterisk csillag 1:8
auditory phonetics /'5:oitri
Ic'nctiks/ percepciós fonetika
1:7
Subject index
auto-stressed /¡5:tco'strcst/
önhangsúlyos 8:119
back vowel hátulképzett
magánhangzó 3:35
backing diphthong 3:36
b-dropping b-elhagyás 11:169
BEAR-words 12:189
Black English fekete angol 1:3
brackets, types of zárójelek típusai
1:10
BREAD-words 12:189
Broadness without r Szélesedés r
nélkül 4:58
CALL-words 4:59, 12:189
CALM-words 4:59, 12:189
Carrot-Rule Sárgarépa-szabály
4:56
cartilage /'ko:tcIio¸/ porc 6:78
Celtic languages /'kcItik/ kelta
nyelvek 1:2
central vowel centrális
magánhangzó 3:35
centring diphthong 3:36
CHAUFFEUR-words 12:189
CiV Laxing /¡si:ai'vi:/ CiV Lazítás
3:42, 12:185
CiV Tensing CiV Feszesedés 3:42,
12:186
clear-L világos /l/ 2:20
CLERK-words 12:189
close vowel zárt magánhangzó
3:35
closed syllable zárt szótag 5:69
closing diphthong 3:36
cluster /'kI\stc/ mássalhangzó-
kapcsolat/torlódás 2:24
Cockney /'kokni/ 1:3
compensatory lengthening
/¡kompcn'scitri/ pótlónyúlás
4:49, 54
complementary distribution
/¡kompIi'mcntri/ kiegészítő
eloszlás 1:8
completely unstressed see zero
stress
compound stress összetett szavak
hangsúlyozása 7:99, 8:121
consonant chart mássalhangzós
tábla 2:16
consonant deletion 7:97
content words see lexical
continuant /kcn'tinjocnt/
kontinuáns, folyamatos 2:14
194
Subject index
contraction /kcn'trækjn/
összevonás 7:102
contrast in intonation 10:144
controversy of length marking
3:31
coronal /'korcncI/ koronális 2:13
covered position fedett helyzet
12:180
Covered Position Rule 12:181
CPR see Covered Position Rule
CREPE-words 12:189
cross-word palatalization 7:96
cultural relatedness kulturális
kapcsolat 1:2
dark-L sötét /l/ 2:20
degrees of stress hangsúly
fokozatai 8:109
Derivational Secondary Stress
Derivációs Másodlagos
Hangsúly 8:114
devoicing /oi:'v5isin/
zöngétlenedés 2:24, 6:81
digraph /'oaigro:I/ digráf,
kettősbetű 11:163
diphthong /'oiI0on/ diftongus,
kettőshangzó 3:31
diphthongs of RP RP diftongusai
3:35
dislocated tonic elmozdított tonik
10:143
downdrift lecsúszás 10:146
DRAMA-words 4:59
dropping of <b> b-elhagyás
11:169
dropping of <g> g-elhagyás
11:169
dropping of <h> h-elhagyás
11:170
duration /ojo'rcijn/ hosszúság,
időtartam 3:31
Early Stress Requirement Korai
hangsúly követelménye 8:113
EARTH-words 12:189
egressive airstream /i'grcsiv
'ccstri:m/ egresszív/kilégzett/
exspirációs levegőáram 1:6
extrametrical /¡ckstrc'mctrikI/
extrametrikus 9:125
falling tone ereszkedő/eső dallam
10:147
fall-rise tone ereszkedő-emelkedő
dallam 10:150
195
Subject index
Finno-Ugric /¡Iinco'ju:grik/
finnugor 1:1
flap see tap
flapping see tapping
foot láb 7:99, 10:140
fortis /'I5:tis/ fortisz képzésű 6:81
FPBR see Free Position Basic
Rule
free position szabad helyzet 12:180
Free Position Basic Rule 12:181
fricative /'Irikctiv/ frikatíva,
réshang, spiráns 2:15
Fricative Devoicing
Frikatívazöngétlenedés 7:96
front vowel /Ir\nt/ elölképzett
magánhangzó 3:35
fronting diphthong 3:36
full vowel teljes magánhangzó
3:33
function word see grammatical
function word see strong forms
function word see weak forms
GA see General American
g-dropping g-elhagyás 11:169
geminate /'o¸cminct/ gemináta,
kettőzött mássalhangzó 1:11,
11:153
General American 1:3
genetic relatedness genetikai
kapcsolat, rokonság 1:1
Gimsonian system /gim'sconicn/
Gimson-féle átírás 1:9
glide félmagánhangzó, siklóhang
2:15
glottal /'gIotI/ glottális,
hangszalaghang 2:13
glottal reinforcement glottális
megerősítés 6:87
glottal replacement glottális
behelyettesítés 6:87
glottalization glottalizáció 2:24,
6:87
glottis /'gIotis/ hang(szalag)rés
2:16, 6:79
grammatical differences between
British and American English
1:4
grammatical/function word
funkciószó, viszonyszó 10:139
graphic position grafikus helyzet
12:180
groups of deviating words 12:189
half-close vowel félig zárt
magánhangzó 3:35
196
Subject index
half-open vowel félig nyílt
magánhangzó 3:35
hard palate /'ho:o 'pæIct/ kemény
szájpadlás 2:16
head fej 10:140
heavy syllable see light
hiatus /hai'citcs/ hiátus,
hanghézag 3:43
hiatus-filler hiátustöltő 7:95
high rising tone magas emelkedő
dallam 10:149
high vowel magas nyelvállású
magánhangzó 3:35
homorganic /¡hom5:'gænik/
homorganikus 5:71
Hungarian terminology for vowels
3:36
Hungarian vowels 3:37
iambic reversal see stress shift
iambic rhythm /ai'æmbik/
jambikus ritmus 8:113
Iambic Secondary Stress Jambikus
Másodlagos Hangsúly 8:115
ill-formed rosszul formált 1:8
Indo-European /¡inoco¡jocrc'pi:cn/
indoeurópai 1:1
information types determining
stress placement 9:123
integrated suffix integrált
szuffixum 9:131
International Phonetic Alphabet
nemzetközi fonetikai ábécé 1:9
intonation intonáció, hanglejtés,
beszéddallam 1:6, 8:108,
10:138
intonation pattern see tune
intonation phrase intonációs frázis
10:139
Intrusive-R /in'tru:siv/
intrúzív/betoldott r 2:19, 7:93
IPA see International Phonetic
Alphabet
Jonesian notation /'o¸conzicn/
Jones-féle átírás 1:9
labial /'IcibicI/ labiális, ajakhang
2:13
labio-velar /¡Icibico'vi:Ic/ labio-
veláris 2:17
larynx gégefő 6:78
lateral laterális 2:14
Latinate see stress-fixing
lax laza 3:33
197
Subject index
Laxing by free U Lazítás szabad
helyzetű U előtt 3:42, 12:185
laxing ending lazító végződés 3:41,
12:184
laxing processes lazítási
folyamatok 3:39, 12:183
L-darkening l-sötétedés 2:21, 7:91
lenis /'Ii:nis/ lénisz képzésű 6:81
letter-to-sound rules betű-hang
szabályok 1:11, 3:41, 11, 12
lexical palatalization lexikális
palatalizáció 11:167
lexical/content word fogalomszó,
tartalmas szó 10:139
lexicon /'Icksikcn/ lexikon,
szókincs 1:12
liaison /Ii'cizn/ liaison, hangkötés
7:95
light/heavy syllable könnyű/nehéz
szótag 9:124
Linking-R kötő r 2:19, 7:93
liquid likvida, folyékony hang 2:15
LOOK-words 12:190
loudness hangerő, hangosság
8:109
LOVE-words 12:190
low rising tone alacsony emelkedő
dallam 10:148
low vowel alacsony nyelvállású
magánhangzó 3:35
low-starting diphthong 3:36
L-vocalization l-vokalizáció 2:22
MACHINE-words 12:189
magas (hangrendű) 3:36
main stress főhangsúly 8:110
Main Stress Rule
Főhangsúlyszabály 9:125
Main Stress Rule for adjectives
Főhangsúlyszabály
melléknevekre 9:130
Main Stress Rule for nouns
Főhangsúlyszabály főnevekre
9:128
Main Stress Rule for verbs
Főhangsúlyszabály igékre
9:126
major stress erős hangsúly 7:98,
8:109
Major Stress Deletion erős
hangsúly törlése 8:115
manner of articulation képzési mód
1:6
manners of articulation of
consonants mássalhangzós
képzési módok 2:14
198
Subject index
mély (hangrendű) 3:36
MEMOIR-words 12:189
mid vowel középső nyelvállású
magánhangzó 3:35
minimal pair minimális pár 1:7
minor stress gyenge hangsúly,
félhangsúly 8:111
monophthong /'moncI0on/
monoftongus, tiszta
magánhangzó 3:30
monophthongization
monoftongizáció 4:49
monophthongs of RP RP
monoftongusai 3:35
morpheme identity szóelemzés
elve11:152
morpheme structure condition
morfémaszerkezeti megkötés
3:40
morphology morfológia, alaktan
1:11
MOVE-words 12:190
MSR see Main Stress Rule
narrow diphthong szűk diftongus
3:36
nasal /'ncizI/ nazális 2:14
nasals nazális mássalhangzók 2:15
neutral prefixes semleges
prefixumok 9:126
neutral tonicity semleges tonicitás
10:142
new information in intonation
10:140
noncontinuant nem
kontinuáns/folyamatos 2:14
non-laxable nem lazítható 3:38
non-neutral prefixes nem semleges
prefixumok 9:127
non-neutral see stress-fixing
non-productive nem produktív
3:39, 8:117
non-rhotic /¡non'rcotik/ nem
rotikus, r-elhagyó 2:17
noun-verb pairs főnév-ige párok
9:130
nucleus /'nju:kIics/ (dallam)csúcs
10:139
obstruent obstruens, zörejhang
2:15
occasional cross-word
palatalization 7:96
open syllable nyílt szótag 5:69
open vowel nyílt magánhangzó
3:35
199
Subject index
optional consonant deletion 7:97
oral /'5:rcI/ orális 2:14
oral stop orális zárhang 2:15
orthography /5:'0ogrcIi/
ortográfia, helyesírás 1:11
palatalization palatalizáció 7:96,
11:167
palatalization by <i> 11:167
palatalization by <u> 11:168
palate see hard palate
phoneme /'Iconi:m/ fonéma 1:7
phonemic principle /Ic'ni:mik/
kiejtés/hangjelölés elve 11:152
phonetic classes of vowels
magánhangzók fonetikai
osztályai 3:30
phonetics /Ic'nctiks/ fonetika 1:7
phonological classification of
vowels /¡Iconc'Ioo¸ikI/
magánhangzók fonológiai
osztályozása 3:32
phonology /Ic'noIco¸i/ fonológia
1:7
phonotactic restrictions
fonotaktikai megkötések 5:69
phonotactics /¡Iconc'tæktiks/
fonotaktika, hangelrendezés
5:63
phrasal stress szócsoportok/
szószerkezetek hangsúlyozása
7:99
pitch /pitj/ hangmagasság 8:108,
10:145
pitch range hangmagasság-
terjedelem 10:145
place assimilation képzési hely
szerinti hasonulás 7:96
place of articulation képzési hely
1:6
places of articulation of
consonants mássalhangzós
képzési helyek 2:13
places of articulation of vowels
magánhangzók képzési helyei
3:34
plosive /'pIcosiv/ explozíva 2:15
post-stress syncope hangsúlyt
követő szinkópa 5:76
pre- and postvoicing elő- és
utózönge 6:82
Pre-cluster Laxness Lazítás
mássalhangzócsoport előtt
3:42
200
Subject index
Pre-fortis Clipping Rövidülés
fortisz előtt 6:88
pre-head elő-fej 10:140
Pre-R Breaking Törés r előtt 4:46
Pre-R Broadening Szélesedés r
előtt 4:52
pre-stress syncope hangsúlyt
megelőző szinkópa 5:76
pre-stressed elő-hangsúlyos 8:119
Prevocalic Tenseness/¡pri:vc'kæIik/
Prevokalikus Feszesedés 3:43,
12:186
primary stress elsődleges
hangsúly, főhangsúly 8:110
productive /prc'o\ktiv/ produktív
3:39, 8:117
progressive /prc'grcsiv/
progresszív, előreható 6:83
prominence of the left edge 8:113
prominence of the right edge 8:113
prosodic /prc'sooik/ see
suprasegmental
pulmonic airstream /p\I'monik/
pulmonikus levegőáram 1:6
PUT-words 12:190
quality deviation 12:182, 188
quantity-sensitive hosszúság-
érzékeny 9:124
R-dropping r-elhagyás 2:17, 7:93
Received Pronunciation elfogadott
kiejtés 1:3
reduced vowel /ri'oju:st/ redukált/
sorvadt magánhangzó 3:32
regressive /ri'grcsiv/ regresszív,
hátraható 6:83
rhotic accent /'rcotik/ rotikus/
rotacizáló/R-ejtő akcentus 2:17
rhyme /raim/ rím 5:74
rhyme weight rímsúly 9:124
rhythm rule see stress shift
rhythmic beat ritmikus ütem 7:98
Rhythmic Stress Deletion
Ritmikus Hangsúlytörlés
10:141
Rhythmic Stress Shift Ritmikus
Hangsúlyeltolódás 10:142
rounded vowel ajakkerekítéses
magánhangzó 3:35
RP see Received Pronunciation
rule szabály 1:8
SAUCE-words 4:59, 12:179
schwa /jwo:/ svá 1:9
201
Subject index
secondary stress másodlagos
hangsúly, mellékhangsúly
8:110
secondary stress assignment
másodlagos hangsúlykijelölés
8:114
segment szegmentum 1:6
segmental feature szegmentális
jegy 1:6
self-stressed see auto-stressed
sentence stress see nucleus
short-long vowel pairs rövid-
hosszú magánhangzó-párok
3:32
sibilant /'sibiIcnt/ szibiláns,
sziszegő-susogó hang 2:17
single consonant letters 11:153
single vowel letters 1:11, 12:177
smoothing /'smu:ðin/ simulás 4:49
soft palate lágy szájpadlás 2:16
sonorant /'soncrcnt/ szonoráns,
zengőhang 2:15
sonority /sc'noriti/ szonoritás,
hangzósság 2:15
sonority peak szonoritási csúcs
5:64
Sonority Principle Szonoritási Elv
5:65
sonority scale szonoritási skála
2:15
SOUL/BOWL-words 12:190
sound correspondence
hangmegfelelés 1:3
sound value hangérték 12:177
SOUP-words 12:189
spelling helyesírás 1:10
spelling differences between
British and American English
1:4
standard varieties standard/
köznyelvi változatok 1:2
stem szótő 1:12
stop zárhang 2:14
stress hangsúly 1:6, 7:98, 8:109
stress clash hangsúlyütközés 8:113
stress shift hangsúlyeltolódás
8:116
stress-fixing hangsúlyrögzítő
8:117, 118
stress-marking conventions 8:112
stress-neutral suffix hangsúlyra
nézve semleges szuffixum
8:117, 118
stress-sensitive affixes hangsúly-
érzékeny affixumok/toldalékok
9:136
202
Subject index
stress-timed hangsúlymértékes
7:100
strong forms of function words
funkciószavak erős alakjai
7:100, 105
suffix /'s\Iiks/ szuffixum, toldalék
1:12
suprasegmental feature szupra-
szegmentális/prozódiai jegy
1:6, 7:98, 8:108
syllabic consonant /si'Iæbik/
szótagalkotó mássalhangzó
2:22, 5:66
Syllabic Consonant Formation
5:75
syllabification /si¡IæbiIi'kcijn/
szillabifikáció, szótagolás
9:133
syllable boundary szótaghatár
2:25
syllable weight szótagsúly 9:124
syllable-timed szótagmértékes 7:99
syncope /'sinkcpi/ szinkópa 5:76
syntax /'sintæks/ szintaxis,
mondattan 1:11
tail farok 10:140
tap, flap egyperdületű/legyintett r,
érintőhang 2:23
tapping, flapping 2:24, 7:92
tense feszes 3:33
tenseness reversal 12:182
tertiary stress /'ta:jcri/ harmad-
lagos hangsúly, félhangsúly
8:111, 120
thirteen men rule see stress shift
THOUGHT-words 4:59, 12:190
tone /tcon/ tónus 8:109, 10:146
tone group dallamhordozó 10:139
tone-unit dallamhordozó 10:139
tonic placement 10:142
transcription transkripció,
fonetikus átírás 1:9
trigraph /'traigro:I/ trigráf,
hármasbetű 11:163
triphthong /'triI0on/ triftongus,
hármashangzó 3:31
Trisyllabic Laxness /¡traisi'Iæbik/
Harmadszótagi Rövidülés
3:39, 12:183
TROUBLE-words 12:190
tune dallam 10:146
typological relatedness nyelv-
tipológiai kapcsolat 1:2
203
Subject index
underived secondary stress
deriválatlan másodlagos
hangsúly 8:114
ungrammatical agrammatikus 1:8
unrounded vowel réses
magánhangzó 3:35
Uralic /jo'ræIik/ uráli 1:1
utterance megnyilatkozás 2:28
velar /'vi:Ic/ veláris 2:13
Velar Softening Veláris Puhulás
11:171
velarized velarizált 2:20
velum see soft palate
verbal prefixes igei prefixumok
9:126
vibration (of the vocal cords)
hangszalagok rezgése 6:79
vocabulary differences between
British and American English
1:4
vocal cords /¡vcokcI 'k5:oz/
hangszalagok 2:16, 6:78
vocal folds see vocal cords
voice assimilation zöngésségi
hasonulás 6:83, 7:95
voiced /v5ist/ zöngés 1:6, 6:80
voiceless zöngétlen 1:6, 6:80
voicing zöngésség, zönge 2:13
vowel digraphs magánhangzó-
digráf 1:11, 12:177
vowel reduction magánhangzó-
redukció/sorvadás 7:100
vowel shift magánhangzó-
eltolódás 3:37
vowels magánhangzók 3:30
WANT-words 12:189
WAR-words 12:189
weak forms of function words
funkciószavak gyenge alakjai
7:100, 102, 105
West Germanic /o¸a:'mænik/
nyugat-germán 1:1
wide diphthong széles diftongus
3:36
WORK-words 12:189
yod-dropping j-elhagyás 5:64, 71,
11:173
zero stress hangsúlytalan 8:111
204

Kiadta a Bölcsész Konzorcium
A Konzorcium tagjai: • Eötvös Loránd Tudományegyetem • Pécsi Tudományegyetem • Szegedi Tudományegyetem • Debreceni Egyetem • Pázmány Péter Katolikus Egyetem • Berzsenyi Dániel Főiskola • Eszterházy Károly Főiskola • Károli Gáspár Református Egyetem • Miskolci Egyetem • Nyíregyházi Főiskola • Pannon Egyetem • Kodolányi János Főiskola • Szent István Egyetem
Szerkesztette: Lektor: Balogné Bérces Katalin Dávid Gyula Siptár Péter

A kötet megjelenése az Európai Unió támogatásával, a Nemzeti Fejlesztési Terv keretében valósult meg: A felsőoktatás szerkezeti és tartalmi fejlesztése – HEFOP-3.3.1-P.-2004-09-0134/1.0

ISBN 963 9704 34 2 © Bölcsész Konzorcium. Minden jog fenntartva!

H-1088 Budapest, Múzeum krt. 4/A. tel.: (+36 1) 485-5200/5772 – dekanbtk@ludens.elte.hu

Bölcsész Konzorcium HEFOP Iroda

Contents
Preface…………………………………………………………………... 1. English pronunciation: phonetics and phonology……………………. 2. The phonology of English consonants: an introduction……………… 3. The phonology of English vowels: an introduction………………….. 4. R-influence on vowels……………………………………………….. 5. The English syllable………………………………………………….. 6. Laryngeal features……………………………………………………. 7. Connected speech…………………………………………………….. 8. Word stress – Part 1: The degrees of stress………………………….. 9. Word stress – Part 2: Primary stress…………………………………. 10. Sentence stress and intonation……………………………………… 11. Letter-to-sound rules – Part 1: Consonants…………………………. 12. Letter-to-sound rules – Part 2: Vowels……………………………... Bibliography……………………………………………………………. Subject index…………………………………………………………… vi 1 13 30 45 61 78 91 108 123 138 152 176 191 193

v

Preface
This book contains twelve chapters introducing the basic characteristics of the pronunciation of standard English, and it is designed for a one-term course with twelve weekly topics elaborated on in ten to twenty pages on average. It is the authors' intention to keep both the amount of material covered and the students' reading load to the absolute minimum: the book leaves, on purpose, considerable time and space for practice, revision and assessment as well as for the inclusion of the personal preferences of the instructor teaching the course. Primarily for pedagogical reasons, but also out of space limitations, the book describes the pronunciation of the standard dialects of English only, though reference is made, whenever relevant to the topic, to non-standard regional varieties, too. As usual in similar textbooks on English used in Hungarian higher education, the description focuses on standard British English pronunciation ("RP"), as this is the accent which most Hungarian students of English appear to be familiar with. However, an attempt is made to include the characterization of standard American English ("GA") as well, especially where the two reference accents significantly differ, both because neither of the standards should be considered inferior to the other and neglected, and because Hungarian students seem to be exposed to American English at least as much as, if not more than, to British English. In addition, the authors believe that all students holding a degree in English are required to be aware of what pronunciation differences are to be expected between native speakers of different linguistic backgrounds. Once the major varieties are introduced, it is generally accepted that the two together present a good starting point for the discussion of other dialects.

vi

segment. a certain amount of preliminary knowledge of the basic notions of linguistics is assumed since most BA programmes contain an introductory tier with at least one course in elementary linguistics. syllable. bound morpheme. diacritic. diphthong. free variation. stem. voicing.Throughout the book. As it has been mentioned several times. At the end of this section you will find a list of the terms which are considered to be elementary. Before you study this book. all the discussions aim to be as self-explanatory and selfcontained as possible. vowel. Wh-question. suffix. their static version can be handed out in class – this is totally at the instructor's discretion. phonetics. allomorph. monomorphemic. adverb. triphthong. Yes/No question vii . Alternatively. lexical content word. assimilation. stress. conjunction. the book grants considerable freedom to the instructor. The book is accompanied by a digital material. places of articulation. it covers so little phonological theory that it leaves the way open for advanced studies in both BA and MA programmes to elaborate on all the topics in more detail. function word. speech organs. verb. monosyllabic. articulation. loanword. manners of articulation. schwa. noun. distribution. morpheme. minimal pair. complementary distribution. idiomatic. which contains exercises to practice and revise the topics covered by the readings. Also. consonant. root. monophthong. homophone. Nevertheless. spelling. interrogative pronoun. compound. free morpheme. demonstrative pronoun. check whether you are familiar with the following terms: adjective. and ample opportunities for extension. allophone. These exercises are primarily meant to be used in their interactive form as home practice for the weekly readings.

for example: apart from the core of the word stock. and "pronunciation" on the other. Namely. It is the most popular language learnt and used as a second or foreign language. As far as linguistic structure is concerned. Finnish.1. among others. Flemish. English pronunciation: phonetics and phonology This is a book on the pronunciation of English. However. Estonian. This results in numerous linguistic differences between the two languages. they exhibit very few of the similarities one would expect from two languages that have evolved from a common ancestor. Compare English to German. we have to explain what we mean by "English" on the one hand. languages can not only be related genetically – English is not only related to the other Indo-European languages (and most closely. North America. As you know. English and Hungarian could not be farther from each other. of course. from a historical point of view. cognate to. is genetically related to a number of tongues spoken all over Europe and Asia: from the Indian subcontinent to Western and Southern Europe. In contrast. English shows more resemblances to Chinese (with its comparative lack of different word endings) than to any of . Hungarian is of Uralic (more precisely. No matter how obvious our topic might seem. Frisian. to West Germanic German. and Yiddish). Finno-Ugric) origin. Afrikaans. it needs considerable clarification. Australia and New Zealand as well as parts of Africa and Asia. and the Samoyed languages. Therefore. which is why the Hungarian student of English (as well as the English student of Hungarian) is faced with so many difficulties. and as such. Lapp. English has as many as 400 million native speakers in the British Isles. Dutch. it is a member of the Indo-European family of languages.

Asterix and company). English is so related to North American Indian languages: although they share neither early history nor (much) linguistic structure. English does not only have contacts with non-Germanic languages outside the British Isles.g. Even though the amount of English influence on Hungarian vocabulary is obviously larger. a discussion of which is beyond the present purposes. Languages. even in its homeland English lives side by side with a couple of Celtic languages. Besides its intricate pattern of connections to other languages and its dominant status on the linguistic map of the world. László Bíró). Due to a series of historical events. both of which are equally accepted by the societies of their respective countries. (Unfortunately. we are able to identify a handful of English words as words of Hungarian origin. It should be kept in mind that these are only distantly related to English. in both directions. Therefore. which belong to another branch of the IndoEuropean family: Welsh (spoken in Wales). cf. too. typologically. This is the way in which English can be considered as related to Hungarian. that is. e. and are languages in their own right. from the name of the inventor. English is very special in at least one more respect. English has developed two standard varieties. including the well-known loanwords coach (from kocsi 'carriage') and biro ('ball-point pen'. still spoken in parts of Ireland). Gaulish. and Scottish Gaelic (in the north-west of Scotland. they have borrowed a number of place names and terms from each other.Chapter 1 the other members of the Indo-European family. that is. Finally.. two forms. can also be related to each other according to what linguistic type they belong to. all the Celtic languages formerly spoken on the European continent are now extinct. Irish (or Irish Gaelic. all languages are naturally related to others culturally through the contacts they come into. then. One is Standard British English in 2 . especially the Hebrides Islands).

rather. e. pass. a GA speaker pronounces //. can't. ask. it is a general expression to refer to the pronounced form of any variety of any language. whereas that of the USA is often referred to as General American. most of which concern vowels. or the Szeged dialect of Hungarian). staff. Australian English. glass. 3 . etc.g. bath. Further examples are given in the chart below: address advertisement ate clerk figure inquiry laboratory 1 typical RP // // // // // // // typical GA // // // // // // // Notice that this sense of the word accent is much wider than in everyday use. As this book is exclusively concerned with pronunciation. there are differences which are not as general as that but only affect certain individual words.g. class. past. e. that is. Every speaker has an accent.. dance. On the other hand. abbreviated to RP. where it basically coincides with what linguists refer to as a foreign accent. chance. the African American vernacular). Here. Black English (that is. henceforth we will concentrate on the pronunciation varieties (called accents1) of the two standards. The standard accent of England is traditionally referred to as Received Pronunciation (where received means 'accepted'). a number of systematic sound correspondences can be identified. after. the standard accent (standard English pronunciation. The two accents differ in many ways. standard Hungarian pronunciation. RP // in lot and dog corresponds to a somewhat longer // in GA. have // in GA. whenever an RP speaker uses the vowel // as in know.. the other is Standard American English in the USA. go. last.English pronunciation: phonetics and phonology England. a couple of words pronounced with // in RP. or GA for short. For example.g. working class London English (called Cockney)..) is just another accent in the same way as geographically or otherwise localizable forms (e. path. half. On the one hand. in contrast. grass. boat.

words ending in -our. -ise and -re in British English end in -or. -ize and -er in American 4 . there exists a certain amount of vocabulary. AmE apartment baggage bill cab candy closet cookie corn diaper elevator eraser fall BrE flat luggage (bank) note taxi sweets wardrobe biscuit maize nappy lift rubber autumn AmE faucet french fries garbage gasoline hood (of a car) line (potato) chips sidewalk the first floor truck trunk (of a car) vacation BrE tap (potato) chips rubbish petrol bonnet queue (potato) crisps pavement the ground floor lorry boot holiday In contrast to pronunciation and vocabulary differences. The following table lists some of the notions that American English (AmE) and British English (BrE) use different words for. the two systems of spelling and grammar do not deviate considerably. spelling and grammatical differences as well. there are two types again: some are systematic (e.. we will mention just a few. Since these are irrelevant to the discussion in the rest of this book.g. As to differences in spelling.Chapter 1 leisure lieutenant (n)either schedule shone tomato vase Z // // /()/ // // // // // // // /()/ // // // // // Although the two varieties of English mostly differ in pronunciation.

either. smell. colour/color. learn. leveling – BrE traveller. -ize is used henceforth. levelling). but its occasional differences from GA (and some other accents) are not left unmentioned. But what elements is pronunciation composed of? Let us first take a look at the basic mechanism that is used to produce speech sounds in English and in most Indo-European languages and also in Hungarian. plow. However. program. in AmE final -l is not usually doubled. AmE stay home – BrE stay at home. gaol. BrE cheque. there are small differences in the use of prepositions. quit and wet are regular in British English but irregular in American English (all three forms being the same). e. AmE Monday through Friday – BrE Monday to Friday. jail. and the past participle of get is gotten in American English. as this form is getting so widespread that even major British publishers recommend it to their authors. e. theatre/theater. typical BrE Have you got a problem? In addition.. spill and spoil are typically irregular while in American English they are regular. Grammatical differences do not abound. 5 . Finally. in British English the past tense and past participle of burn. realise/realize2.g. either.English pronunciation: phonetics and phonology English.. we follow the British conventions for spelling. AmE traveler.. dream. pyjamas. dive is regular in British English but irregular in American English (dive/dove/dived). in words with alternative -ise/ize. As to verb forms. spell. perhaps the most conspicuous concerns the usage of have.g. AmE He just went home. as in AmE Do you have a problem? vs. tyre correspond to AmE check. e. centre/center. tire. Turning back to pronunciation. programme.. got in British English. American English uses simple past tense in some cases where British English has present perfect. The first phase 2 In this book. e. pajamas. some characterize individual words only. in the rest of the book the main emphasis falls on RP since it is the pronunciation most students of English as a foreign language are familiar with all over the world.. AmE meet with sy – BrE meet sy.g.g. lean. e.g. fit. plough.

8. sentences. larger chunks of pronunciation also have characteristics of their own – these are the so-called suprasegmental features. The features mentioned above characterize individual segments. but the combination of consonants and vowels called 6 . of the articulation of English (more precisely. and 9) and intonation (Chapter 10). phrases. RP) vowels in Chapter 3. that is. where it is modified in various ways by the movements of the speech organs called articulators. ejectives and clicks. They are named so because in some sense of the word they are situated "above" segments. As the air leaves the lungs it continues upward in the windpipe (trachea). It is important to note as other languages might use other kinds of airstream mechanisms to produce certain speech sounds – e. meaning that the source of the air to be used in speech is the lungs ("pulmonic") and that the direction of the airflow is outward ("egressive"). All speech sounds can be classified according to where in the vocal tract this modification takes place (the so-called place of articulation) and how exactly this modification is carried out (the manner of articulation).Chapter 1 in the process of articulation (speech production) is called a pulmonic egressive airstream mechanism. we can specify how active the vocal cords are: whether they vibrate (in voiced sounds) or not (in voiceless sounds). place and voicing are the so-called segmental features of speech. Notice that it is never a single consonant or vowel which is stressed.. As a third term in the description of speech sounds. and enters the so-called vocal tract.g. which are not found in Indo-European languages or in Hungarian. they affect elements which are higher up in the hierarchy of linguistic units: syllables. and of voicing and related phenomena in Chapter 6. speech sounds – manner. implosives. However. A more detailed discussion of the articulation of English consonants is found in Chapter 2. The two most significant suprasegmental features are stress (discussed in Chapters 7.

too. in the rest of the book we will concentrate on articulation as it comprises a minimally necessary element of the physical properties of speech. but a whole phrase or sentence. The same is not true for these sounds in English: in skin it is plain but in kin it is aspirated. a plain [] sound and its aspirated version [] (with an extra puff of air following the consonant – see Chapter 6) are separate phonemes in. Although this is not clear from the above discussion.. a positional variant (an allophone) of a phoneme in another. it is never a single consonant or vowel which has a characteristic intonational contour. phonetics does not only deal with the process of articulation. but it only uses them as tools to achieve the ultimate objective: describe the functions the segments have in speech.. the same phonetic object. where a lot of word pairs (called minimal pairs) like // and // are distinguished by this very feature – and it does matter which word you mean as the former means 'ear' while the latter 'mime'. Two phonemes always enter into such a relation that they contrast and distinguish words. A different point of view is taken by the branch of linguistics called phonology. and the various systems and patterns they constitute. which is at the same time sufficient for the present purposes. For example. that is.g. 7 . nevertheless.English pronunciation: phonetics and phonology syllable. the relationships they contract with each other. The scientific study of the segmental and suprasegmental features of speech is called phonetics. may serve as an independent unit (a phoneme) in one language but only as a form. For example. allophones never do so but are predictable instead. e. It also deals with speech and sounds. All the three aspects of pronunciation are of equal importance. the same sound. i. Hindi. speech production.e. but it is concerned with acoustics (the way speech travels in the air in the form of sound waves) and speech perception (sometimes referred to as auditory phonetics). and it borrows the terms and notions of phonetics.

are in complementary distribution. Phonology also attempts to handle cases when a sound appears in different forms in different environments. that is. the two types of English k above can be argued to stand in such a correlation: a common underlyer k is realized as aspirated at the beginning of words. or the rules of etiquette. say. though. the rules of language are more like the rules (or "laws") of physics or football: it is the rules which constitute the system. and we also know how often drivers and pedestrians break these rules without traffic as such coming to an end). A system in which all k's are plain cannot be (native) English. *[].Chapter 1 however.. The word rule therefore denotes 8 . Languages do not exist without their rules – in fact. which cannot exist without these rules. as plain after an s. The rules of a language are not like that. In contrast. they mutually exclude each other. and a ball game in which the players are allowed to catch the ball cannot be football. there used to be a time when cars were already used but no traffic signs had been invented yet. traffic does exist without the Highway Code (in fact. in which case they are always plain.. There is no physical world without. and will henceforth be indicated with an asterisk (*). this is totally predictable as all word-initial k's are aspirated unless they are preceded by a s. i. at least in standard English. ungrammatical). when phonemes or allophones alternate. e. A note of warning is in order here. Because of this it is impossible for them to appear in identical phonological positions.g. the rules define the languages..g. and it is possible to show (some kind of) human behaviour without respect to the rules of politeness (and how many people do so at least in certain situations!). e. The word rule should not be taken here in the same sense as in the case of. the rules of the Highway Code. Instead. gravitation. pronunciations like that of kin with a plain k are ill-formed (or. For example. Such alternations in linguistics are commonly referred to as rules.e.

That is why no minimal pairs exist for these two sounds: they mutually exclude each other. with the tongue resting in its neutral position. Sometimes these two viewpoints arrive at totally different conclusions. whereas the latter can only occur in strong. both types exist in Englsh: the first solution characterizes the so-called Jonesian notation (found in. the one at the beginning of about (called schwa) and the vowel of bird. e. with //.. compare two vowels of English. A great number of further examples illustrate that the phonetic and the phonological classes of sounds do not necessarily coincide. Phonetically. the old bilingual Országh-dictionaries) whereas the second one is utilized in the socalled Gimsonian system (e.. For instance. e. For example.. while phonology is concerned with their function and patterning within a linguistic system.English pronunciation: phonetics and phonology the observation of some systematic regularity rather than a regulation which must be obeyed by all good citizens. someone primarily governed by phonetic criteria will choose the same symbol for about and bird. they are each other's opposites: the former can only occur in weak. phonologists vary as to which symbols to use in their description. Actually. Phonologically. Therefore. we can state that phonetics treats speech sounds from the viewpoint of their physical properties. No matter to what extent the transcription system of the International Phonetic Alphabet (IPA for short) is based on universal agreement. two phonetically nearly identical objects are evaluated by phonology as two distinct elements. The two different points of view may also influence the notational conventions analysts use. To summarize the discussion so far. stressed syllables. they are almost identical: in both cases the airstream entering the vocal tract is only slightly modified. unstressed syllables.g.g..g. e. the latest editions of Oxford Advanced 9 . while phonology-oriented researchers will transcribe only about with a schwa (//) and mark the birdsound differently. however. //.g.

but the same is impossible with the latter: /()/ is never *//. The differences of phonetic detail versus phonological analysis in transcription can be highlighted by enclosing the two types in different brackets: usually square brackets [ ] denote a phonetic transcription while slant brackets / / stand for a phonological one. subconsciously of course. the former is. native speakers are typically unaware of phonetic facts like the near physical identity of the two vowels. they learn this when they acquire their mother tongue. that the first vowel in police is not the same as the one in pearly: only the former is weak enough to be dropped. and anybody who is ignorant of aspiration must be a nonnative speaker. or the exact articulatory gestures involved in their production (not to mention their acoustics). They also know.) The two transcription systems vary in several other respects as well. for instance. As opposed to this. Chapter 5). English kin [] can be phonologically transcribed as // since the aspirated k is simply a phonetic variant (allophone) of the phoneme /k/. E. so that // frequently becomes // (cf. (You may have already guessed that both are named after the person who devised them. but the latter is not. part of linguistic knowledge.Chapter 1 Learner's Dictionary). Native speakers know when to aspirate their k's because they are native speakers.. Not only the phonetic aspect of a language is considered to lie outside the linguistic knowledge of native speakers – so is the history of the language (would you as a native speaker of Hungarian know that your mother tongue is a Finno-Ugric language had you not been taught this at school?) as well as the written form. 10 . A crucial difference between phonology and phonetics lies in their status concerning native speakers: namely. this book favours Gimson's version of the IPA.g. the spelling.

and commonly given between angle brackets. phonology. we are forced to make constant reference to other components as well. <u> (in bury). Emma). and many cultures have never employed an orthography). with single letters. That component of linguistic knowledge that we are concerned with here is. there are not too many one-to-one spelling-to-pronunciation correspondences. that human language is primarily spoken: children learn to speak first and are explicitly taught to read and write somewhat later (in fact. iron. as <e> (in bed). we are very often first confronted with an unknown word in its spelt form and consequently we tend to overestimate its role: pronounce long consonants (called geminates) – erroneously – for double consonant letters (e. or letter-tosound rules. most phonological processes have exceptions. the rules of sentence formation (syntax). since phonology seems to be heavily influenced by the rules of word formation (morphology) and. some native speakers never acquire the spelt form of their language and remain illiterate). which cannot be accounted for by 11 . Wednesday). therefore. On the one hand.g. The same sound. Nevertheless. can be spelt.g. and speech comes first in the history of language itself. for at least two reasons. see Chapters 11-12). too (writing systems have emerged to represent already existing spoken languages. or pronounce silent letters (e. to a more limited extent. or with digraphs (combinations of two graphic symbols) like <ea> (in head) (for more detail on letter-tosound rules.English pronunciation: phonetics and phonology The relationship between English pronunciation and spelling (also called orthography.. In addition. Notice however. and not the other way round. <a> (in many). Thus writing must be conceived of as a derived version of the spoken language. On the other hand.. < >) is worthy of interest in this book. say /e/. being non-native learners of English. the spelling system of English is a mixture of several different traditions and no major reforms have affected it for centuries – as a result.

Chapter 1 the rules but must be stored in the speakers' memory (called the lexicon). 12 . person. (On stress placement. they are derived from the same stem. the fact that stress falls on the first syllable in blackbird ('feketerigó') but on the second in black bird ('fekete madár') is due to the syntactic difference between a single (compound) noun and a phrase composed of an adjective and a noun. For example.. be able to recognize in native speech. together with all the morphological. Finally. which every student of English in higher education is expected to be aware of. the fact that stress falls on the first syllable in the noun present but on the second in event is an example of a mere idiosyncracy: the stress pattern of event must be memorized in the lexicon as irregular. and consciously use in order to improve their pronunciation. using suffixes of different types).) In the remaining eleven chapters follows a description of the main phonetic and phonological features of standard English pronunciation (RP and GA). see Chapters 8 and 9. the fact that stress falls on the first syllable in personal but on the second in personify is due to the morphological difference between these words (viz. syntactic and lexical conditions.

and it is dealt with in considerable detail in Chapter 6. soft palate (velum). velar (the back of the tongue moving towards the soft palate or velum). and the manner of articulation. larynx. deletion. vocal cords In this chapter we take a brief look at what phonological processes affect the consonants of English. oral cavity. especially those of RP and GA. According to their articulation. Voicing results from vocal cord vibration. GA. schwa. and glottal (involving some kind of manipulation of the opening between the vocal cords in the larynx called glottis). stress. compound. For the present purposes it is sufficient to say that when the vocal cords vibrate. the consonants of English can be classified along three terms: voicing. RP. The phonology of English consonants: an introduction Before you study this chapter. suffix. orthography. the articulatory classification of consonants will only be touched upon and not discussed in great detail. we get voiceless sounds. the place of articulation. and when they do not. syllable. . Since we assume a knowledge of the basics of phonetics. free variation. and which aspect of sound structure determines their behaviour. we get voiced sounds. nasal cavity. morpheme. which are illustrated in the following diagram showing the cross-section of the head. The place and manner of articulation refer to where and how the airflow is obstructed during the production of the consonant. check whether you are familiar with the following terms: allophone. coronal (involving the blade of the tongue called the corona).2. There are four major places of articulation: labial (involving a lip or both lips). homophone.

e. According to the manner of articulation. and therefore it is generally assumed to belong to the continuants.Chapter 2 VELAR CORONAL LABIAL GLOTTAL These places of articulation can be further divided into subcases. These are included in the consonant chart below. we get stops. i.e. which normally results in noncontinuant articulation... some consonants are oral (the air escapes through the oral cavity and the mouth). noncontinuant sounds. is special: although the tip of the tongue touches the alveolar ridge. If the articulation involves a total obstruction of the air in the larynx or the oral cavity. or.g. 14 . whether a labial makes use of both lips (bilabial) or just the lower one plus the upper teeth (labiodental). the nose). //. For example. Otherwise the sound is continuant. if the air is stopped for a short period. that is. in this case the air is able to escape along the sides of the tongue (the name of this manner is lateral). there are several possible divisions of consonants. in other words. One consonant. others are nasal (the air escapes through the nasal cavity.

however.The phonology of English consonants: an introduction The major subclasses of consonants. The following chart includes all places and manners of articulation relevant to the description of English consonants. all sonorants are voiced. When consonants appear in pairs. these consonants are called obstruents. The remaining classes (nasals. stem from a further aspect of the manner of articulation: the degree of openness of the vocal tract. liquids. The most sonorous consonants are the glides. 15 . and this order of sound classes constitutes a sonority scale. with the resulting relative loudness of sounds called sonority. From plosives up to (and including) fricatives the obstruction is considerable. degree of sonority ——————————————————————————————→ oral stops (plosives) and affricates – fricatives – nasal stops – liquids – glides (semivowels) (– vowels) As you can see. the one on the left is voiceless. the smaller the degree of obstruction). the following classes can be set up. but the most sonorous sounds are the vowels. Accordingly. oral stops and affricates are the least sonorous as their production involves complete obstruction to the airflow. That is. Unpaired obstruents are voiceless. glides) are the sonorant consonants because they are dominated by sonority. sonority increases from left to right. the more sonorous the sound is. The more open the vocal tract (that is. the one on the right is voiced.

Chapter 2 16 .

The consonant // is produced with two important articulatory gestures. and Irish English. what processes they undergo and trigger. are deleted in certain accents of English is traditionally called R-dropping. rho). therefore this word contains three sound segments only. In contrast. For this reason. and American English in New England and parts of Southern US are nonrhotic. it is sometimes termed labio-velar.) The result is // for talk and..g. and a velar gesture on the other. Australian English. which is not true for. Scottish English.The phonology of English consonants: an introduction The obstruents in the shaded cells (     ) are the hissing and hushing sounds called sibilants. These include GA. Southern British English (including Cockney and London English). that is. Welsh English. // for York. we turn to their phonology. with the rhyming words underlined: You can hear it in my accent when I talk I'm an Englishman in New York The two words. that is. the speakers of GA. After this brief introduction to the phonetics of consonants. The process whereby certain spelt <r>'s are not pronounced. Those accents of English whose speakers pronounce all orthographic <r>'s are referred to as rhotic accents (after the name of the corresponding letter in the Greek alphabet. talk and York. let us see one of the most salient differences between RP and GA: the pronunciation of orthographic <r>. First. Consider a pair of rhymes in the song An Englishman in New York by British pop musician Sting. more importantly. RP. rhyme because Sting (along with millions of English speakers) does not pronounce the <r> in York. that is. (Notice that the <l> in talk is not pronounced in any accent of English. and consequently appears in the chart twice: it involves considerable lip rounding on the one hand. Canadian English. e. New Zealand English. their speakers drop the <r> when it is followed by a consonant 17 . British English in North and South-West England. South African English.

is part of a suffix.. The most frequent silent letter is <e>. e. as in column (b) – in such cases the <r> really is before a consonant and behaves accordingly. the following vowel. the letter <h> is not pronounced. bears. In (g). is unpronounced.e.Chapter 2 or a pause (i. Notice 18 . In the word rhyme in (e).g. so the // is followed by a vowel. which enables the <r> to be pronounced. no // before a consonant (a) (b) York tired party bird allergy leopard particular bears iron aren't feared retirement fires rarely before a pause (c) (d) your tire car err refer teacher particular bear bore care lyre restore more centre pronounced // before a vowel (e) (f) (g) ring crow tiring routine rhyme refer restore retirement rarely pray tribute shrimp Africa poetry arrive boring error referee fiery furry rarest In non-rhotic accents like RP. which makes good use of those three rhyming words. as in the examples in column (a). Notice that Jamaican English is also non-rhotic. no // is pronounced before a consonant. (You may be able to recall a Bob Marley song entitled Iron Lion Zion. The following chart illustrates this with a few examples. Very often the letter <r> is immediately followed by a vowel letter which is not pronounced. then. When the <r> is followed by a vowel. e.. so it can be misleading when some other vowel letter. it is pronounced even in non-rhotic accents.g. //. nothing in speech). Sometimes the consonant is a suffix. however.. be it at the beginning (column (e)) or the middle (column (f)) of the word. and they only pronounce it when it is followed by a vowel.) The <r>'s in (c) and (d) are in final position – again. the <o> in iron. silent letters (in (d)) do not count.

more – maw. but they disappear when they are final or when they are followed by a consonantinitial morpheme like -(e)d (the <e> is mostly silent). which pronounce all <r>'s that are present in spelling. spar – spa. etc. Hence the difference between tire/ tired vs tiring. Intrusive-R is only found in non-rhotic accents. aren't sounds the same as aunt. remember that non-rhotic accents are named so not because they do not pronounce any // sounds. however. so sore sounds the same as saw. rumbaing //. and there is no <r> in spelling. Note that it never matters whether the // is spelt as a single letter <r> or double <rr>. Further examples include farther – father. blaher // 'more mediocre'. In fact. -er/-or. err vs error. refer vs referee.. source – sauce. subpoenaing // 'summoning sy to appear in a lawcourt'. and it only appears at (certain) morpheme boundaries. (those //'s are often called Linking-R). -ment. This is called Intrusive-R. gnawing //. sort – sought. baahing // (of sheep). Also. as in retirement and rarely in column (e) above. fort – fought. i. pour sounds the same as paw. We learn more about this in Chapter 7.The phonology of English consonants: an introduction that word-final <r>'s are like ghosts: sometimes you see them. 19 .g. etc. e. most non-rhotic speakers (perhaps except only for conservative RP. after certain vowels. there are some cases when rhotic speakers do not pronounce an //. You "see" them when they are followed by a vowel-initial morpheme like -ing. RP spoken by older generations) pronounce one. An interesting consequence of R-dropping in non-rhotic accents is that new homophones emerge. court – caught. but because they do not pronounce certain orthographic <r>'s. -ee. in sawing //. bore vs boring. career – Korea.e. tuner – tuna. as opposed to rhotic accents. -y... and fur vs furry. guffawing // 'giving a noisy laugh'. -ly. sometimes you do not.

It's softer or Call Maria.Chapter 2 When a word ends in an <r>. In certain positions. Roughly.. this is usually called dark-L. that is. in RP the /l/ is only pronounced clear when followed by a vowel. and it becomes dark (the IPA 20 . Now we turn to another consonant. e. schwa insertion. I do or There's a spider. i. He doesn't care. care about. (cats) claw at (the furniture).. that is. even a following vowel-initial word. I'm scared. it can not only "escape" being dropped when a vowel-initial suffix is attached to the word as in the examples in column (g) in the chart above. when the <l> is pronounced. I need her. In several dialects of English. Amanda Avon – Amanda Raven. it becomes velarized. however. your eyes.. but in fact any vowel-initial morpheme following the word is able to produce the same effect. the letter <l> is also sometimes not pronounced. law and (order). the traditional name of this sound is clear-L (or light-L).g.. As exemplified by the word talk mentioned above. This can even happen between two sentences. (the giant) panda is (an endangered species). Intrusive-R appears between the two words in visa application. (the) Shah of (Persia). etc. In other positions. (to) err is (human). two allophones. the spa is broken – the spar is broken. under the same conditions as between a word and a suffix. its articulation involves the movement of the tongue towards the soft palate (velum). (the) idea is. Gloria Estefan. but this time we are only concerned with the sound //.e. put the tuna in the box – put the tuner in the box. //. tire us. it has two possible pronunciations. Similarly. the /l/ is the same alveolar lateral as in Hungarian. Therefore. Further homophones arise. Such homophones are only possible in the non-rhotic accents of English exhibiting Intrusive-R. vanilla ice – vanilla rice. Linking-R is heard in phrases like more exciting. etc. centre of. e. and between the two sentences in Try that sofa.g.

the /l/ is clear because it is prevocalic. so <l> and <ll> behave identically since they are both pronounced in the same way: short /l/. bef. If we compare the examples killed – kill – killing and tallness – tall – taller. or may result from the reduction of an unstressed // (e. million //). which may be part of the complex vowel // or its reduced counterparts /. or it precedes a /j/ sound (column (e)). /. it is clear in feel at home. Spelling does not count. the pronunciation of word-final /l/ is determined by the following segment in the same way. The following chart shows the details. killed. Do not let silent letters (e. In the three columns on the right. dark-L bef. stale) mislead you. the /l/ is at the end of the word school. In schoolyard. a cons. dark in spell and spell this word but 21 . While it is dark in feel and feel me. Within and across sentences... we find that word-final /l/ is only dark when followed by a consonant or a pause. the /l/ is dark when it is followed by a consonant sound (column (a)) or when final in the word (column (b)). a pause (a) (b) spilt pill belch Albert else killed tallness bell rebel stale kill tall clear-L before a vowel (c) (d) lip slip look Linda lateral libido lullaby splendid Ashley colon killing taller before // (e) value cellular million evaluate volume schoolyard As you can see above.The phonology of English consonants: an introduction symbol of dark-L is []) when a consonant or pause follows. but it is followed by a /j/ in the compound.g. either word-initially (column (c)) or medially (column (d)).g. and therefore it is clear. This is the rule of L-darkening. but it becomes clear when a vowel-initial suffix follows.

Finally. where /l/ is usually dark in all positions). which is always dark and which no following vowel can "rescue". This phenomenon is referred to as aspiration. Word-initially and before a stressed vowel. and is transcribed [] (syllabicity is indicated in the IPA by a short vertical line under the main symbol). since they exhibit a wide range of allophones. feel [].. In words like cycle []. with word-initial prestress plosives having the strongest possible aspiration. shelf []. so it is also dark in cycling []. the distinction between clear and dark-L may not be present at all (as in GA.g. For some reason. In other accents of English. milk []. the /l/ is found in a syllable that lacks a vowel: notice that in the second syllables of these examples. in physical reality. Let us emphasize at this point that the discussion of L-darkening refers to RP only. word-medial pre- 22 . e. There is one type of word-final /l/.e. and a detailed description of its articulatory basis is given in Chapter 6. syllabic /l/ is always dark. Since in such cases the /l/ is considered to take up the role of the vowel and constitute the syllable. smile and smile back with smile at me. In several non-standard varieties of English. as the consonant /l/ is replaced by a vowel.Chapter 2 clear in spell it. Phonetically. it is usually referred to as a syllabic /l/. however. or channel []. This is called L-vocalization.. or Channel Islands [ ]. or may have slightly different conditions. martial []. especially voiceless plosives. we take a look at plosives in English. even if it is followed by a vowel sound in the next syllable. Compare kill and kill Bill with kill you. dark-L is often articulated as an [o]-like vowel. martial arts [ ]. i. the voiceless plosives (= /p t k/) are followed by a short [h]like sound. the /l/ constitutes the syllable along with another consonant. there are various degrees of aspiration.

In (a)-(c). and the 23 . The chart also shows that on the basis of the behaviour. and other word-internal ones and those in word-final position having even less.e. aspirated unaspirated 1 unaspirated 2 ——————————————————————————————→ (a) (b) (c) (d) (e) (f) pát repéat potáto léopard ráp spíll póker tén tíger kíll cút suppórt retúrn detér índicàte políce tomáto todáy cajóle clípper tomáto vánity quáker póker gállop cút suppórt póke láck wásp stóp stándard scúll skín raccóon collápse The columns (a)-(f) give examples of the degrees of aspiration from strong in (a) to zero in (f). i.. the possible pronunciations. however.The phonology of English consonants: an introduction stress plosives and unstressed initial plosives having somewhat less.. they are definitely unaspirated: after an /s/. The optionality of aspiration in (d)-(e) means that in the positions in question other allophones can also appear. differ in that aspiration is either optional ("unaspirated 1" in (d)-(e)) or impossible ("unaspirated 2" in (f)). Besides a plain unaspirated plosive. The accents on top of vowel letters in the example words denote stress (the acute accent <´> means stronger stress than the grave accent <`> – see Chapter 8 for the degrees of stress in English). only aspirated plosives are pronounced by speakers of English. This is illustrated in the chart below. we can identify a phonologically relevant binary distinction between aspirated and unaspirated. of the plosives. In one case. however.e. before an unstressed vowel). the IPA symbol of which is []. The other columns. in GA or informal-colloquial British English /t/ is frequently realized as a so-called tap (or flap) in the cases in (d) (i. and it is just the degree of the aspiration which distinguishes the subcases.

) Examples include tomato []. etc. the /t/ is glottalized and the /l/ is fully voiced. word-finally. simplicity. which is. One is the short []-like sound following the plosive. In contrast. so it does not undergo aspiration. sleep []. little []. In words like attractive. In addition. (See Chapter 6. not a voiceless plosive but a voiced one. attract. (In fact. trombone. that is. In both cases the voiceless plosive is followed by a sonorant consonant. The other manifestation of aspiration is the devoicing of a following sonorant consonant. In pray [].) Turning back to aspiration. If we take consonant sequences (called clusters) under closer scrutiny. in examples such as Atlantic. the voiceless plosives /  / (and also //) are usually unreleased and/or are accompanied by a short closure of the vocal cords called the glottal stop (symbolized by [] in the IPA). matter [()]. not []. Notice how minimal the difference is between train and chain. In // sequences. liqueur. nobody []. This cannot be an accident. butterfly [].. mentioned above. twist. queen []. of course. match []. the underlined sonorant consonants are voiceless. most native speakers of English would insist that the /t/ in attractive belongs to the 24 . and the phenomenon is accordingly dubbed glottalization. crucially. but there are no such examples with /tl/. plug []. betray. thousands of English words start with a /tr/ cluster (tray. Where can the difference come from? Notice that.Chapter 2 process is referred to as tapping (or flapping). we make an interesting observation. and the resulting [] sounds very much like a //.g. this is also true for /d/. cube [].). trip. Scotland. In (e). the /t/ is aspirated and thus the /r/ is devoiced. it actually has two realizations. the // is aspirated and therefore the // is devoiced. as if it was an affricate. E. vanity [].

ppórt re. at the same time.ccóon collápse 25 .lan. syllable-) final or medial but certainly not initial.càte políce tomáto todáy cajóle clí. This yields a.The phonology of English consonants: an introduction second syllable of the word along with the /r/. while in /tl/ the /t/ is at the end of one syllable and the /l/ is at the beginning of the next one.t.péat potáto léo.er pó.pp.tive (the double <t> in spelling stands for a single /t/ sound) but At.k. while in (b) there is a unanimous agreement among speakers as to the syllable boundary. and why /tl/ never occurs in that position (the /t/ is at the end of a syllable and not at the beginning). customarily indicated by a dot) between the /t/ and the /l/. In (e) and (f). Now. In conclusion. If we (or rather: native speakers of English) syllabify the example words given above and repeated here for convenience (only the relevant syllable divisions are indicated).ttrac.p. This also explains why /tr/ is possible at the beginning of words (after all. we identify a syllable boundary to the left of all the "aspirated" cases: in (a) and (c).o váni.k.er gállop cút suppórt póke láck wásp stóp stándard scúll skín ra.túrn de.(that is. the beginning of the word is. whereas in Atlantic there is a syllable division (or syllable boundary.tic.t.y quá.er tomá.tér índi. it appears that the pronunciation of the /t/ (and the other voiceless plosives as well) depends on its position within the syllable: it is aspirated when syllableinitial but glottalized when syllable-final. /tr/ is a frequent syllable-initial cluster.ard ráp spíll póker tén tíger kíll cút su. this clearly does not hold: the plosive is either word. the beginning of the (first) syllable). the plosive is at the beginning of the word. aspirated unaspirated 1 unaspirated 2 ——————————————————————————————→ (a) (b) (c) (d) (e) (f) pát re.

ambisyllabic /t/ and /d/ may even be tapped.rr.e. The chart illustrating R-dropping is repeated presently.rr.y fu.ly pronounced // before a vowel (e) (f) (g) ring crow ti.ing routine rhyme refer restore retirement rarely pray tribute shrimp A. (In the chart. let us revisit the allophonies discussed earlier.frica poe.est before a pause (c) (d) your tire car err refer teacher particular bear bore care lyre restore more centre 26 . and to such consonants as ambisyllabic. we have indicated this hesitation by dots assigned to both possible locations.pard.r. Syllabic position initial ambisyllabic final Pronunciation strongly aspirated weakly aspirated or tapped unaspirated glottalized For the sake of experiment. others for leo. We can conclude.or refe.r.ticular bears iron aren't feared retire.Chapter 2 Column (d) is.ty bird aller. problematic.ment fires rare.) For this reason.. then. native intuition fails to make unambiguous judgments: some speakers would opt for leop.gy leopard par.r.try a. yet others "feel" as if the /p/ belonged to both syllables – i. the syllabification of such consonants is ambivalent.y ra. When a single consonant is followed by an unstressed vowel. and investigate into syllable boundaries in the examples.ing e. Our findings are summarized in the following table.ard.r. no // before a consonant (a) (b) York tired par.ree fie.rrive bo. however. phonologists often refer to such a situation as ambisyllabicity. that ambisyllabic voiceless plosives can be plain or weakly aspirated.

and the examples in column (b)) or part of a final cluster (as in the rest of the words in (a)). the /l/ is always at the end of a syllable: it is either absolute final (as in.shley co. e.l.l.ness bell cancel stale kill ill clear-L before a vowel (c) (d) lip slip look Linda lateral libido lullaby splendid A.. 27 .uate vo. Albert.. leopard).ular mi.l.ll. whether absolute initial (in (e) and in arrive and referee) or part of an initial cluster (in (f)). Syllabic position initial ambisyllabic final Pronunciation pronounced pronounced dropped Notice that initial and ambisyllabic consonants have something in common: they (can) occupy the beginning of the syllable. dark-L bef.ion eva. a cons. illness. Syllable-initial /r/.ll.bert else killed ill. party in (a).ue ce.ll. and so does ambisyllabic /r/ (in (g)).g. a pause (a) (b) spilt pill belch Al. bef.yard In the first two columns.ume schoo.l. belch or else: the <e> in the -ed suffix is silent.l.g. Notice that killed goes under exactly the same rubric as spilt.. escapes being dropped.er before // (e) va. Now here is the chart for L-darkening.on ki. which can be absolute final (as in.ll. so the /l/ is immediately followed by the final consonant /d/. e. and all the example words in (c)-(d)) or part of a final cluster (as in.ing i.The phonology of English consonants: an introduction The position of the relevant syllable divisions reveal that R-dropping affects syllable-final /r/.g. e.

The only exception is tapping. the syllabification of such consonants is not unambiguous – 28 .. spell it and get up syllabified? As we have just stated. they are. or the /t/ in get) in phrases and sentences. the /l/ in feel. When we look at the "fate" of word-final consonants (e. feel at home. the /l/ is syllable-initial (including cases when it is part of an initial cluster) or ambisyllabic.e. Consequently. it is always ambisyllabic since it may end the syllable to the left. syllable-final. but notice that they can also start the following syllable – after all. but it may as well form an initial cluster with the following // – words like lucid //. But how are strings like more exciting. Apparently. whereas for a plosive to be (considerably) aspirated it must exclusively occupy the beginning of the syllable. the /r/ in more.. Therefore. In column (e). /l/ is clear and /r/ is pronounced in at least partial initial position. /l/ and /t/. just like in the allophony of plosives and in R-dropping. things become particularly exciting. ludicrous // exemplify //. the ambisyllabic situation patterns with the initial rather than the final position. at least in one possible pronunciation (the other alternative does not contain the // – see Yod-dropping in Chapter 5). there are a whole lot of English words beginning with /r/.g.Chapter 2 In the remaining three columns. obviously. lucrative //. the underlined consonants can be syllable-final. In isolation or at the end of an utterance (i. a stretch of speech uttered without a pause). your eyes. which is possible when the /t/ or /d/ is ambisyllabic but impossible when it is clearly initial. Syllabic position initial ambisyllabic final Pronunciation clear clear dark Careful readers must have noticed that in L-darkening. we can conclude the following. however.

and discover some of its further aspects.l. however.re. as in Channel Islands. but in Chapter 5 we take up the discussion of syllable structure again.g. Zelefánt. e. or the Hungarian translation of Heffalump. fee. The next two chapters introduce the phonology of English vowels. In sum. in the cartoon Micimackó és a Zelefánt. where ambisyllabicity is ruled out. The same is not true. Hungarian. it.) This straightforwardly explains why such /r/'s are pronounced and such /l/'s are clear. they are necessarily always syllable-final. It follows. of phrases like more beautiful and feel me. and how these processes are driven by the position the consonants occupy within the syllable. and as a result. it must be tapped in the so-called tapping dialects like GA or informal-colloquial British English.g. Zenga zének 'the song resounds'. not a joke. since no English syllable can start with one (that is why no English word starts with one).The phonology of English consonants: an introduction they are ambisyllabic. is always dark. then. the /z/ in az 'the' is readily illustrated by the title of a Miklós Vámos book. e. (This is also reflected by the fact that phrases like a nice cream and an ice-cream are indistinguishable. eyes. you. such /r/'s are dropped and such /l/'s are dark. what is more. as well: the native intuition concerning the position of. Therefore. This prediction is supported by the facts: in GA. that syllabic /l/. As far as syllabic consonants are concerned.. In light of the above discussion. In fact.r. at all is usually pronounced as a tap..ll. this phenomenon is not only observable in English but in other languages. get up. it can only be moderately aspirated. the underlined /t/ in right away. this chapter has shown what major phonological processes affect the consonants of English. exciting. Pooh's Heffalump Movie). spe. you could have guessed by now what happens to a final /t/ in a phrase like get up: since it is ambisyllabic. the relevant syllable boundaries can be located as mo. 29 . at home. for example. (Remember to ignore the silent <e> at the end of more.

check whether you are familiar with the following terms: back. vowels differ from consonants in their behaviour. half-close. too: while consonants typically occur in syllable marginal positions – they appear at the peripheries of the syllable –. monophthong. tense. Let us first examine what phonetic classes may be defined in the English vowel system. vowel sounds may be classified according to two types of factors: phonetic and phonological. suffix.e. lax. As suggested in Chapter 1. the articulators do not form a complete or partial closure or a narrowed passage in the way of the exhaled air. close. rounded. In some vowels the position of the tongue is relatively stable during articulation. On the other hand. the position that the tongue occupies at the beginning of the vowel differs significantly from what it occupies at the end of the vowel. place of articulation. such vowels are called monophthongs. unrounded. central. stress. low.e. vowel shift. i. triphthong.3. manner of articulation. Vowels differ from consonants in two very important ways: they are articulated without any kind of obstruction in the oral cavity – i. In the first case. mid. full vowel. though. consonant cluster. open... weak vowel In this chapter we take a look at English vowel sounds and their possible classifications. diphthong. . classification is based on some articulatory characteristics while in the second it is some aspect of vowel behaviour that serves as the basis for classification. compare them with the Hungarian vowel system and see what typical vowel alternations occur in English. front. half-open. schwa. syllable. The phonology of English vowels: an introduction Before you study this chapter. vowels form the very core of the syllable and occur in syllable central position. In other vowels. high.

Chapter 2 – typically lack triphthongs – and even some of the diphthongs as we will show in Chapter 4.g. . . Whenever we refer to long vowels. We may also think of this difference as a difference in how many vowels are found within one syllable: in monophthongs there is one – e. e.. However. . i. length is not a stable property. Note that length in English varies depending on the environment – i. . . . one of the so-called short monophthongs...The phonology of English vowels: an introduction some tongue movement is involved. Such vowels are referred to as diphthongs (and triphthongs). /. The vowels of RP are the following: Short vowels . / –. Note though that triphthongs are not found in all dialects of English: those dialects that pronounce all underlying /r/'s – the so-called rhotic dialects.  Diphthongs ..g.e.. /. . . /. we may refer to the controversy of length marking: the vowel length of monophthongal – or pure – vowels is indicated with a colon. Note that diphthongs and triphthongs are just as long as long monophthongs.g. /. . vowels may be short – e. / – or long – e. . . / – while in triphthongs there are three – e.  Triphthongs To further demonstrate that length is not a purely phonetic property of English vowels. /. On the other hand. . e. . we always mean long monophthongs. the vowel // is just as long in actual pronunciation as any of the long monophthongs or diphthongs and it even undergoes the very same shortening process as long vowels do 31 . u. in diphthongs there are two – e. . cf. /. .g. .  Monophthongs Long vowels . For more on length alternations. diphthongs and triphthongs together.g.  . . see Chapter 6. / – depending on their duration: long vowels are approximately twice as long as short ones.

that justifies the use of the different symbols. However. //-// kör 'circle' vs. The vowel // is categorized as a short vowel because it behaves like other short vowels do. that is. Consequently. there is no English short-long vowel pair in which the qualities of the two vowels are the same.Chapter 3 (see Chapter 6). in this case it is the full vowel-weak vowel distinction.. kőr 'hearts (in cards)'.1 On the other hand.g. its length is not indicated in transcription with the colon. while there is a short // and a long /i/. /i/-/i/ Sirok (a placename) vs. there is no //. /y/-/y/ tüze 'his/her/its fire' vs. the two major phonological classes are based on the type of syllable the vowel appears in. This is also reflected in the phonetic symbols used to indicate them. to be discussed presently. English short-long vowel pairs always involve a quality difference. Another important note concerning vowel length is due here: while most Hungarian short-long vowel pairs consist of vowels of more or less the same quality with just a length difference (e. English behaves quite differently from Hungarian as far as stressed and unstressed syllables are concerned. etc. However. English unstressed syllables have reduced 1 For the degrees of stress. 32 . second. tűz-e 'does he/she/it stitch?'. sírok 'I cry'). where the quality of the vowels is the same. it may be the first. third. similarly. As far as phonological classifications of vowels are concerned. mentioned in Chapter 1. there is no //. The phonetic length of // may be due to the fact that during its production the lower jaw and the tongue are in their most open position. see Chapter 8. while there is a short // and a long /u/. The only exception to this rule is the vowel pair //-//. a gesture which might take long enough to cause a perceivable length difference. syllable of an English word that carries primary stress. while it is always the first syllable of the word that carries the main stress in Hungarian. On the one hand.

However. . Thus. // and // – are lax. which. Non-high long monophthongs – that is //. the vowel is lax (phonetically). tense and lax vowels (in a phonological sense) occur in different types of environment. also including // and //.  Lax . . . the vowel is tense (in a phonetic sense). in unstressed syllables only weak vowels – //. .The phonology of English vowels: an introduction vowels only.) As we will show below. Tense i. may also function as full vowels. all the other vowels of English. . to refer to a certain kind of vowel behaviour. Within the class of full vowels we may distinguish two subclasses: tense and lax vowels. . u.e. .. 33 . Whenever these muscles are tense. except in the case of //3. e. 1. we will use these terms in a purely phonological sense. 3 . i. . besides occurring in unstressed syllables. weaker in energy and closer to schwa // in place of articulation. In a phonetic sense. in the sense that these vowels are shorter. . against the back wall of the pharynx (throat).e. One has to be very careful when using these two terms as they are often used as phonetic labels. too. .. . (We might just as well call the two types of vowel Type1 and Type2 was it not for our wish to follow the tradition. . . . // and // – may be found while in stressed syllables we may only find so-called full vowels – i. when such muscle tenseness is not present. 2 Monophthongs Diphthongs and triphthongs There are a few generalizations to be drawn on the basis of the above table: all short vowels are lax and all diphthongs and triphthongs are tense while long monophthongs are divided between the two classes. these terms refer to the muscle bundles located at the backmost part of the tongue.

 long . . horn. clause. . in other cases it is lax. word-final position or followed gorgeous. where we take a look at their behaviour before <r>. door. . cord. horde. war. north. adore. .  . sport <or> in spelling followed by a before. . //1 Lax // 2 No <r> in spelling broad. fault <ar> or <or> in spelling in for. <aur> in soar. . historian. pour. quarter. boar. . shore // 3 Tense letter <oar>. roar. aura. u. Before we actually discuss these we have to point out that places of articulation are not as clearcut for vowels as for consonants for the very simple reason that while in consonants the place of articulation refers to the articulators producing some 34 . .Chapter 3 The behaviour of // is twofold: sometimes it is tense. . saw. <our>. <oor>. .  . . morning. called. bored. Shaw. Laura The behaviour and alternations of tense and lax vowels are discussed below and also in Chapter 4. spelling four. (pronounced or silent) vowel more. There are two general types of spellings that indicate a lax /:/ while a further set of spellings represents the tense variant. bought. floor.  Weak vowel .  - Let us now turn to the places of articulation of vowels.  tense long i. ball. e. Full vowel lax Monophthong Diphthong Triphthong short . store. The following table sums up what we have discussed about the manner of articulation and the behaviour of vowels so far. by a consonant letter dwarf.

while the ones inbetween as mid. in vowels it is simply inapplicable as they do not involve any kind of obstruction. The places of articulation of the diphthongs of RP are the following: Front Central Back unrounded unrounded unrounded rounded Close Half-close Half-open Open  a // bay // bye // boy // bound // boat e a  35 . close vowels are often referred to as high. except for //. open vowels as low. As it can be seen from the above table the following generalizations may be drawn: front and central vowels are unrounded while back vowels are rounded. we will use three criteria to classify vowels according to horizontal tongue position. vertical tongue position and lip rounding. The places of articulation of the monophthongs of RP are the following: Front unrounded // beat // bit /e/ bet // bat Central unrounded // ago // burn // but Back unrounded rounded // boot // put // bought // bar // Bob Close Half-close Half-open Open We have to note that besides the terms used in the table above.The phonology of English vowels: an introduction degree of obstruction. Instead of referring to obstruction sites.

those that end in // are fronting (and closing) while those ending in // are backing (and closing). while the rest. In all other diphthongs the second component is more close than the first. magas and 36 . Hungarian also has front rounded vowels. On the one hand. // are narrow (and closing). unlike English.. Attention must also be paid to the fact that the traditional Hungarian terminology might be misleading: the so-called "magas (hangrendű)" vowels are actually front – and not high –. we may distinguish them according to their second component: if it is a schwa //. closing diphthongs may be classified according to the articulatory distance between the two components: the diphthongs //. . // are the so-called low-starting or wide diphthongs.  As shown in the following table. then we talk about centring diphthongs. while "mély (hangrendű)" vowels are central or back – and not low –.Chapter 3 Front Central Back unrounded unrounded unrounded rounded Close Half-close Half-open Open  e  // beer // bear // boorish Diphthongs may be classified according to several factors. //. On the other hand.  Closing Fronting Backing    . i. Hungarian back vowels are all rounded – note that // is a central vowel.e. In addition. and these are thus called closing diphthongs. //. This is summarized in the table below: Centring Narrow Wide .

The phenomenon is called vowel shift. The places of articulation of Hungarian vowels are the following: Front unrounded rounded /i/ hív /y/ tűz /i/ ki /y/ üt /e/ kér // nő // kör // kert Central unrounded /a/ ház Back rounded /u/ út /u/ kulcs /o/ tó /o/ hoz // kar Close Half-close Half-open Open Let us now turn back to phonology and the discussion of tense and lax vowels. but are metaphors for the acoustic effect made by the vowel. when a word is suffixed.The phonology of English vowels: an introduction mély do not refer to tongue height. The vowel shift is thus a case of tense-lax alternations. In English. The tense-lax vowel pairs are as follows: 37 . unlike in Hungarian. One of the differences between them is in what positions they may appear in a word. a historical version of which – the Great Vowel Shift – applied to English long vowels around the 15th century. One such alternation involves the change of an original tense vowel into a lax one. Tense vowels of word stems become lax in certain environments. often it is the pronunciation of the word stem that changes and not that of the suffix.

and // have no lax counterparts and as a result do not participate in the alternation (they are non-laxable). //-// join-juncture. etc. There are a few untypical pairings that may occur: /a/-// pronounce-pronunciation. Note that in the examples relevant to the present discussion. //-// 4. namely /(j)/. //-// vane-vanity meter-metric secret-secretary 3. //-// 3. Also. /u/// do-does. tense vowels differ according to what follows them) but Pre-R Broadening does not (that is. //-// There are two types of word pair: one in which the stressed vowel is followed by the letter <r> and one in which it is not. the same lax vowels appear in both the first and the last columns of the table). //-// 2. It is also clear from the table that some of the tense vowels. there are a few cases that involve some alternation but 38 . //. Pre-R Breaking does apply for tense vowels (that is.Chapter 3 Regular type 1. As vowels may be influenced by a following <r> – Pre-R Breaking for tense vowels and Pre-R Broadening for lax vowels. //-// Bible-biblical holy-holiday know-knowledge sole-solitude Vowel letter compare-comparison A prepare-preparatory barbarian-barbaric imperial-imperative E severe-severity hero-heroine satire-satirical I or Y tyrant-tyranny lyre-lyrical historian-historical O explore-exploratory flora-florist Before R 1. for details see Chapter 4 – we have to consider pre-R cases separately. //-// keep-kept final-finish decide-decision 4. //-// grade-gradual sane-sanity 2.

etc.. purify. grade-gradual. As we have noted above. in which a stressed vowel in (at least) the third-last syllable must be lax – e. or a non-productive suffix which is only added to certain stems of a class and therefore has fewer examples. 39 . i. What makes this rule problematic is that there is a great number of exceptions. nightingale. //-// empire-imperial. there are irregular exceptions. verb. e. e. //-// break-breakfast – or they involve lax-lax or tense-tense alternations – //-// example-exemplify. Also. the rule is sensitive to the morphological structure of the word: it applies if certain suffixes are attached to the stem but not if others are added. notify.g.g. too. That is. which can be added to almost all members of a category (noun. Besides these. Probably the most influential such laxing process is Trisyllabic Laxness. this rule has regular exceptions: the tense vowels /u/. sane-sanity. stupefy.. Abraham. adjective. productive suffix. isolate. unity. Some of the typical examples of the two suffix classes are shown in the following table. //-// clearclarity. both regular and irregular. In a few cases other tense vowels may also occur in trisyllabic environments. /j/ are regular exceptions.g.. compare-comparison. Let us now turn to the environments in which vowel shift may occur.The phonology of English vowels: an introduction it is either not one of the regular vowel pairs above – e.) to produce a large number of words.g. it seems that the syllables of certain suffixes are counted when we count the three syllables from the end of the word while others are not.. they freely occur in trisyllabic environments.e. Whether to count the syllables of the suffix or not depends on whether the suffix is a regular.. /ju/ and their variants //.

several. succulent. cannibal. autism // irritate. intelligent // company. positive. tidy-tidiness -ly total-totally. decorate. frivolous // opera. sane-sanity -al crime-criminal -ative provoke-provocative -ible divide-divisible. stamina. gullible. fantasy. janitor.g. stimulate. 40 . lazy-lazily -ary/-ery/-ory advise-advisory -ing pilot-piloting Non-productive suffixes counted in Trisyllabic Laxness -ity grave-gravity. audible. e..Chapter 3 Productive suffixes not counted in Trisyllabic Laxness -ness lazy-laziness. harmony. pertinent. parsimony // orthodox. oracle Short // is missing from the charts above simply because it is so rare in present-day standard English that it is almost impossible to find relevant examples. bulletin. similar. permanent. courtesy // participle. homonym. auditor. // animal. character // penetrate. dominate. carnival. demonstrate. eat-edible We must mention here that Trisyllabic Laxness is not just an active phonological rule that applies to certain roots if they are followed by certain suffixes but also a so-called morpheme structure condition. miracle. a passive constraint that requires that a stressed vowel which is in at least the third syllable from the end of the word must be lax – even if no suffix is added to it. limerick. Gulliver // terminal. sonorant. separate.

g. -ish (n/v).The phonology of English vowels: an introduction Another laxing process applies if a so-called laxing ending is added to the word stem: a stressed syllable followed by one of the laxing endings must be lax. -ic. too. some roots resist laxing. the ending -ic causes the laxness of the stressed vowels in the names Eric. For instance. patron []. stressed /u/ and /ju/ vowels are not affected by this laxing process: cube-cubic.g. stupe-stupid. -id. Swedish (adj). satire-satirical.. vanish (v) and greenish (adj). and thus a stressed vowel followed by at least two 41 . Cupid. As indicated in brackets. as in metre-metric. The first two examples are nouns and verbs and thus the stressed vowel must be lax as opposed to the other two examples which are both adjectives and. if it is an adjective. Examples include monosyllabic suffixes typically spelled with <i> or <e>: e. On the one hand. Patrick although the very same stressed vowels would be tense where they are followed by some other kind of ending.. as a result. -et. e. It is important that this rule does not only apply if the endings are separate morphemes attached to a root but also if they are just part of the root. Compare the sample words finish (n/v). there are exceptions to this laxing rule. Just as in the case of trisyllabic laxness. the vowels /u/. final-finish. etc. It is clear then that this rule is not just an active phonological rule but also a letter-to-sound rule that determines how letters must be pronounced depending on the environment. /ju/ and their variants //. era []. The third relevant laxing rule is triggered by the presence of a consonant cluster – a sequence of at least two consonants – immediately after the stressed vowel. base-basic. the suffix does not influence the pronunciation of the stressed vowel.g. -el. e. -it. There are irregular exceptions as well. the ending is non-laxing and the preceding stressed vowel may remain tense. -ish (n/v) is a laxing ending only if the word ending in -ish is a noun or a verb. /j/ are regular exceptions.. However.

The reason for this is that CiV tensing is more powerful and robust than the laxing rules.intervention.g.. that is they tell us how to pronounce vowel letters in certain environments in spelling.Chapter 3 consonants must be lax: e. manic-mania. rite-ritual. Celia.g. grade-gradual. e. senior. Gloria. receive-reception. they stay tense before a free U.g.g. other vowel letters must be pronounced with a tense vowel. For this reason we just mention them here very briefly and they will be discussed in detail in Chapters 11-12.. There are two more rules that may cause the laxness of a vowel but they are clearly not phonological rules but letter-to-sound rules.g. This regularity is sometimes dubbed Pre-cluster Laxness.. e. etc. hence the name of the rule. The other letter-to-sound rule causing laxness is the so-called CiV Laxing rule.. etc. In all the sample words above the stressed vowel is the third-last vowel from the end of the word. Interestingly. and thus it can override their effect. It is important to note that all the other vowel letters undergo CiV Tensing in the same environment.e. where we discuss letter-to-sound rules exhaustively. the stressed vowel letter <i/y> is followed by the CiV configuration in spelling. e. i. Of course. this rule is able to block the application of the laxing rules. intervene. etc. familiar. decide-decision. the vowels /(j)/ and /(j)/ are regular exceptions. idiot. The first such rule is Laxing by free U. e. revise-revision. a letter U followed by a vowel letter (as in venue. Trisyllabic Laxness does not apply and make them lax. Syria. etc. Albania.. use-usual. which requires that if the stressed syllable is followed by a free U – roughly. That is. still. there are exceptions to the 42 . which forces a stressed vowel letter – spelled with <i> or <y> – to be pronounced lax // when followed by a consonant letter.. another letter <i> and one more vowel letter. Just like in all other laxing rules. radio. statue) – then the stressed vowel must be pronounced lax.

It is called Prevocalic Tenseness.g. for instance. namely /  / in the examples above. e. There are numerous examples where the originally tense stressed vowel becomes lax although none of the above environments may be blamed for the change. if stressed. the surviving effects of older rules which are no longer active in the language. the underlined portions of Noam.The phonology of English vowels: an introduction CiV tensing rule as well. That is. e. it is only relevant to pronunciation: compare Leo to mean or people – the underlined vowel letter is pronounced as a separate sound in Leo only. national. Besides CiV Tenseness. mead-meadow. in which the stressed vowel is lax even though it is followed by CiV. In such cases we may only say that these are unexplained..g.g. The situation when two separate vowels (the centres of their respective syllables) are adjacent is generally referred to as hiatus. read (present)-read (past). shade-shadow. either Trisyllabic Laxness or Prevocalic Tenseness could in principle take effect. Italian. is always tense. hiatus. In this chapter we saw that.g. but it is the latter that "wins". special. Daniel. Leo.. Similarly to CiV Tenseness. e. although sometimes English chooses a tense vowel systematically. Prevocalic Tenseness is also stronger than the laxness rules: in variety. and consequently Prevocalic Tenseness is not applicable. 43 . in mean and people there is no hiatus.. in the others it combines with the following vowel letter to represent a single sound. react. as its effect is to ensure that all stressed vowels preceding other vowels are tense. in many situations tense vowels are replaced by their lax counterparts. there is another regularity in English which requires a vowel in a certain position to be tense. which are of course reduced. e. Prevocalic Tenseness does nothing but describe the observation that in English the first member of a hiatus. idiosyncratic cases of vowel shift. etc. Note that it does not apply to unstressed vowels.. life-live (v). In addition.

Some of these also affect the classification of vowels.g. 44 . tone – tonic RP // – //. the vowel behaves as lax in the same way in the two accents: cf. There are only a few minor differences in the vowel inventories.. most of it is valid in the case of GA. still. GA // – //. together with the tenseness and laxness rules. e. The tense-lax distinction applies to GA in the same way. e. e. apricot. with just a handful of examples where the two dialects diverge. the rest of the dialectal deviations are caused by the differing distribution of /r/.Chapter 3 Although the above discussion of the phonetic and phonological classification of vowels concentrated on RP. As we will see in the next chapter. odd. recall from Chapter 1 that all RP //'s correspond to // in GA.. pronounced (irregularly) with // in RP but very often (conforming to Trisyllabic Laxness) with // in GA.g.. RP // in lot.g. wash is usually long and unrounded // in GA. etc. too.

non-productive suffix. RP. a rhotic accent. . . In General American. foreignism. Tense i. morpheme. triphthong This chapter mainly focuses on the behaviour of full vowels before an /r/. are most easily distinguished by whether an /r/ is pronounced in all positions or not. . To remind the reader of the most important vowel classes that will be referred to we repeat one of the tables from Chapter 3 for convenience. . . homophone. all /r/'s are pronounced while in Received Pronunciation. Linking-R. hiatus. the phonological and letter-to-sound rules related to this behaviour and some further phenomena concerning vowels. . . non-rhotic accent. . Intrusive-R. . phoneme. rhotic accent. full vowel. . centring diphthong. As it is demonstrated in Chapter 2 the two main accent types of English. nasal. letter-to-sound rule. R-dropping. minimal pair. . Besides this. 2 Monophthongs Diphthongs and triphthongs . briefly mentioned in the previous chapter. e. rhotic and non-rhotic accents. 1. fricative. diphthong. tense. . a non-rhotic variant. . R-influence on vowels Before you study this chapter. these – and other – dialects may also be distinguished by the behaviour of stressed vowels before an /r/. lax. complementary distribution. labial. GA. . low-starting diphthong. only prevocalic ones are.4. productive suffix. check whether you are familiar with the following terms: allophone. 3 .  Lax . u. distribution. monophthong.

all diphthongs and triphthongs are tense. i. see Chapter 3.. namely that all short vowels are lax. which behaves in an ambiguous way: sometimes it is tense. 46 . in other cases it is lax. Let us first consider the behaviour of tense vowels and the rule called Pre-R Breaking. Non-low-starting    (j)   (j) 3 Low-starting       Plain-Tense Broken-Tense The rule of Pre-R Breaking seems to be a very simple allophonic rule at first sight: the members of the Plain-Tense – Broken-Tense vowel pairs appear to occur in complementary distribution: Broken-Tense vowels only appear before r within the same word while Plain-Tense vowels occur everywhere else but never before r within the word. except for //. the environments in which they may occur. nonhigh long monophthongs are lax. Tense vowels may be further classified into two subgroups on the basis of their distribution.Chapter 4 Recall that we have come up with a few generalizations in Chapter 3.e. For the details of this controversy.

// // // // // // // [] [] [] [] [] [] [] [] [] [] [] [] [] [] 1 Be careful with the word iron since its second vowel letter. allophones of the same phoneme. 47 .R-influence on vowels Plain-Tense bead [] tea [] cohesion [] cute [] futile [] unity [] baby [] staple [] Rumanian [] stone [] cloakroom [] broken [] Broken-Tense beard [] tear (n) [] adherence [] curious [] furious [] Europe [] bare [] staring [] Hungarian [] story []3 roaring []3 glorious []3 Plain-Tense fight [] tonight [] pine [] pint [] town [] cloud [] Downing [] moist [] Broken-Tense fire [] admire [] pirate [] or [] iron1 [] hour [] flour [] dowry [] or [] Moira [] or [] Since the members of the pairs are in complementary distribution and are phonetically quite similar to each other. <o> is silent. we may just as well assume that they are variants. and the pronunciation of the <r> is determined accordingly: dropped in RP //. but not in GA /()/.

The differences between these two groups are twofold: on the one hand. //. that is. appearing in lighter shaded cells in the table. This tense //3 variant historically derives from the centring diphthong *// now always pronounced as //. Also. biro // or //. //. whenever a tense vowel from this group is followed by an r in the same word it is always replaced by its Broken-Tense counterpart. in words like fire /()/ the r is wordfinal. in lowstarting diphthongs it is not always so: in low-starting diphthongs Breaking is only obligatory if the r is at the end of a word or followed by a productive suffix (cf. it is always a triphthong. While in non-low-starting tense vowels it is always obligatory. the so-called low-starting diphthongs . if we think of a long monophthong as consisting of two identical short components (as opposed to diphthongs whose two components are different). //. in non-low-starting tense vowels the broken tense variant is typically a centring diphthong except for //. On the other hand. in these four vowel phonemes the second half of the vowel is changed into //. there is a difference between the nature of Breaking in the two vowel groups.g. pirate /t/ or //. //. e. in the darker shaded cell in the table. [].Chapter 4 The vowel phonemes in the upper row of the chart on p.//. then Breaking is optional. in fired // and firing // it is followed by a productive suffix (-ed and -ing) and as a result the stressed vowel always has to be realized by a Broken-Tense vowel. and the vowel may be Plain or Broken-Tense. 48 . In low-starting diphthongs the broken tense variant contains an extra element. and the rest of the vowels . //. that is. For instance. //.46 are divided into two major classes: the last three. Chapter 3).. if the low-starting diphthong is followed by an r which is morpheme-internal or followed by a nonproductive suffix.//. Also.

Triphthong fire [()] tired [] hours [] dowry [] Middle component dropped [()] [] [] [] Last component dropped + first component lengthened [()] [d] [] [] It is interesting to note that in many dialects of English. schwa [] may be dropped: this process is known as monophthongization. Smoothing. of the triphthong is dropped in casual speech.. see below). in faster speech even the last component. Smoothing influences the triphthongs resulting from the abovementioned mechanism of obligatory or optional Breaking of low-starting diphthongs. the first part is lengthened. [] or []. and its extreme form. To make up for the loss of the second and third components of the triphthong. a process often referred to as compensatory lengthening (for more detail. Typically the middle component. the low-starting diphthongs // and // may also be simplified. the complete monophthongization of diphthongs or triphthongs.R-influence on vowels A process that is closely related to Pre-R Breaking is the simplification of Broken-Tense vowels in fast casual speech. replaced by a long monophthong. in a nonpre-R environment: 49 . i. for instance in Southern dialects of American English.e.

the tendency does not only influence the Broken-Tense but also the Plain-Tense variant of the vowel. 50 .Chapter 4 Diphthong why [] I'm [] wow [] about [t] Second component dropped [] [] [] [t] Another similar process by which certain diphthongs become simplified concerns the pronunciation of the diphthongs /(j)/ and //. that is. // is also being replaced by []. /()/ [] poor [()] purify [] Europe [] rural [] tourist [] bureau [] // [] stairs [] parent [] hairy [] repair [()] Hungarian [] fairness [] Note. is left uninfluenced. as in fuel or ritual. the same sequence arising from hiatus. like Australian English for instance. that this monophthongization only affects those /()/'s which are the result of Breaking. In some dialects. The tendency especially in the speech of younger speakers of RP is to pronounce /(j)/ as /()/ and // as a long half-open unrounded front //. however.

therefore we will simply follow the traditional practice of indicating Plain-Tense and Broken-Tense vowels separately in phonological transcriptions. it is possible to analyse all Broken-Tense vowels as the outputs of R-influence. that the plain and broken vowels are independent phonemes. in such a way that the trigger itself (the /r/) is subsequently deleted if the conditions of R-dropping are met. On the other hand. Pre-R Breaking also qualifies as a letter-to-sound rule: whenever a tense vowel is followed by the letter <r> within the word. On the one hand. descriptions of GA do not normally consider Pre-R Breaking as either a phonological rule or a letter-to-sound regularity – the occasional appearance of the schwa is usually taken to be the result of an optional schwa-insertion rule taking place before syllable-final /r/. the status of Pre-R Breaking in English phonology is not straightforward. beer //. respectively. It is also shown in the transcriptions that consequently. Some might argue. on the basis of minimal pairs like bee // vs. notice that the spelling of Broken-Tense vowels always involves an <r> (cf. beard). GA // and /()/. beard //. stone – story GA // – //). beer. and it practically never occurs before a syllable-initial /r/ (e. the lack of the rule of R-dropping results in the absence of apparent minimal pairs like bee and beer. which behaves in the same way as in RP. All in all. is one of the most salient allophonic rules affecting RP vowels.. As regards GA.R-influence on vowels Pre-R Breaking. which means two things.g. then. It also follows that smoothing and 51 . As a further result. not even in the case of non-low-starting tense vowels (except for the //-// pair. that is. beard //. hairy //). etc. cf. it is broken. A consequence of this is that the GA inventory of diphthongs is much smaller than that of RP (no centring diphthongs) and triphthongs are missing altogether. However. beer //. at least in non-rhotic accents like RP. Pre-R Breaking is never obligatory in GA. bead // vs.

On the other hand. . i. Short (lax) vowels all belong to the former group while the three long lax vowels all fall into the latter as indicated in the following table: Plain-Lax Broad-Lax    2      The rule of Pre-R Broadening seems to be very similar to Pre-R Breaking as Broad-Lax vowels will replace their Plain Lax counterparts before r. cannot be considered an allophonic rule as the distribution of Plain-Lax and BroadLax vowels is not complementary. Pre-R Broadening. Let us now turn our attention to the other major group of full vowels. the two types of vowel do appear in the same environment – with certain limitations. that this is not so as it will be clear from the discussion below. and in stairs and hairy the monophthong is automatically created if the schwa is not inserted (cf.. Note. and their behaviour before r. namely /.Chapter 4 monophthongization are not as extensive in GA: fire is always /()/.e. //. . the 52 . sure and poor are usually /()/ and /()/. it will turn out that Pre-R Broadening is a practical rule concerning the relationship between the spelling and pronunciation of vowel letters before r. from the table above it is obvious that four of the Plain Lax vowels. however. //). Also. / share a Broad Lax counterpart. which also makes Pre-R Broadening different from Pre-R Breaking as in the latter all Plain-Tense vowels had a Broken-Tense counterpart of their own. /()/. Lax vowels may also be divided into two major groups: Plain-Lax vowels and Broad-Lax vowels. respectively. unlike Pre-R Breaking. Since these Plain-Lax – Broad-Lax vowel pairs do not occur in complementary distribution. Instead.

its now empty position becomes available for the vowel before it.g. car. i. burp – or it is word-final and is followed by a pause – e.R-influence on vowels sample word pairs have been set up on the basis of spelling: the Plain-Lax – Broad-Lax vowel pairs are represented by the very same vowel letter in the pairs. it is not pronounced in non-rhotic accents like RP. As a result of this one might easily find a very attractive explanation for the lengthening component of broadening: since the r is dropped in these environments... for. stir.. That is. Thus we can conclude that Pre-R Broadening does apply if the r after the Lax vowel is silent. either because it is followed by a consonant – e. bird. abort. Plain-Lax cat [] fan [] bad [] fond [] bond [] clock [] stem [] send [] head [] Broad-Lax car [] far [] bar [] for []2 abort []2 lord []2 stern [] serve [] heard [] Plain-Lax fit [] bingo [] stick [] hut [] cutlery [] spun [] put [] bush [] buffet [] Broad-Lax firm [] bird [] stir [] hurt [] curl [] spur [] purr [] burst [] burp [] Having taken a look at the examples containing a Broad-Lax vowel. it is dropped because of the R-Dropping rule (Chapter 2).g. stern. we may notice that although there is always an r in spelling in these words. the vowel lengthens to make up for the loss of 53 . purr.e.

can be dropped before a consonant. can be accounted for with reference to compensatory lengthening.. its "place" is preserved. Notice that the effect of compensatory lengthening is very similar to the so-called law of mass preservation in the physical world: we have the same amount of material – that is. i. it takes twice the time to pronounce. Both star and hard originally contain four short segments. the same number of timing units – on both sides of the equation. l. and the preceding lax vowel lengthens by becoming linked to this empty timing unit. and as a result the preceding vowel is lengthened. note that however attractive this 54 . Pre-R Broadening. where the other liquid. The process is very similar to what frequently happens in certain non-standard varieties of Hungarian. then.Chapter 4 the r in the word – the kind of process referred to above as compensatory lengthening. If a sound segment is linked to one timing unit. XXXX → XXXX         →    XXXX → XXXX          →   The diagrams above demonstrate compensatory lengthening: the X's stand for timing units. it is long. XXXX → XXXX           →   bolt 'shop' XXXX → XXXX          →   zöld 'green' In such Hungarian examples the l is deleted but its timing unit is retained. Nevertheless. as the broken lines indicate. When the r in the words is dropped.e. it is short while if it is linked to two.

Let us take a look at some sample words containing such a sequence: 55 . only they are short: //. However. in many cases it can be shown to be a component of this rule. stern. While there is evidence that there is an /r/ phoneme in words like star as the word final /r/ is often realized as a Linking-R (see Chapter 2). it cannot be true in all cases. consequently. but not compensatory lengthening. words like hard pose a problem for babies aquiring a non-rhotic dialect like RP. no r's are dropped. has taken place. if there is no /r/. compensatory lengthening is impossible. then it is actually not dropped and thus the vowel is not lengthened because the r has been dropped. // – Broadening. lord. They will always hear such words pronounced without r as there is no environment in which the r of hard would be present in actual pronunciation. //. and as a result they will have to assume that these words do not contain an /r/ phoneme. //. hurt as in RP.R-influence on vowels explanation may be. While it seems that we have to give up our idea of compensatory lengthening as a motivation for Pre-R Broadening. and it still is a useful kind of explanation when teaching pronunciation. The next question that we turn to is what happens if the r following the lax vowel is realized in pronunciation. Therefore in GA we find the same vowels in car. All the more so as a major difference between RP and GA can only be accounted for if we separate Broadening proper (influencing the quality of the target vowel) and compensatory lengthening (responsible for vowel quantity). firm. //. As GA is a rhotic accent.

the r is syllable-final (cf. a few irregular words are exempt from it – that is. curry. Jerry. We have seen so far that normally it is Broad-Lax vowels that occur before an r.e. Broadening does take place even though the following /r/ is not 56 . on the other hand.e. for instance. courage. The Carrot-Rule itself is not without exceptions. either. marriage. i.. Thus we may claim that Pre-R Broadening only applies if the r is dropped. Chapter 2).. bury. although in most cases it applies in the same way as in RP. In GA.. where the r is followed by a pronounced vowel. mirror. i. If the /r/ is not silent.. the lax vowel before it will be a Plain-Lax vowel as the Carrot-Rule will block the application of Broadening. The Carrot-Rule is often indicated in spelling by the doubling of the r. e.g. it results in a group of regular exceptions. miracle.g.Chapter 4 [] barrier carrot chariot Harry marriage narrow wheelbarrow [] borrow corridor Morris sorrow sorry tomorrow torrent [e] berry bury Jerry merit serendipity terrible terror [] irritate lyrical miracle mirror pirouette pyramid spirit [] burrow courage current curry furrier (n) furrow hurry There are two possible conclusions that can be drawn on the basis of the data in the table: on the one hand. borrow. this regular absence of Broadening is usually referred to as the Carrot-Rule. however. although it does not always happen. As the absence of Broadening is typical in words like carrot. it is also clear that Pre-R Broadening does not apply in any of the above words. e. it does not contain sample words containing an [] before a pronounced [r] as this vowel does not regularly appear in such a position with a few exceptions like courier /()/.

current. i. currency. on the other hand. say. the Carrot-Rule should block Broadening. which exhibits the same pattern as.. E. discussed above. concerning words like occurrence and current. courage. Non-productive suffixes. occurrence. Broadening affects both occur (with a syllable-final /r/) and occurring (with productive -ing). the discussion of Pre-R Breaking above) or a vowel-initial word. This is the case when the r is followed by a vowel-initial productive suffix (again! – cf. behave as if they were not separate morphemes and the word was morphologically simple.R-influence on vowels syllable-final. worry.2 syllable-final /()/ /b()/ /()/ // followed by non-productive suffix occurr+ence // barr+en err+or cler+ical // /()/ // occur bar err clergy 2 Note. the differences between RP and GA. current. hurry.. and squirrel RP // vs. GA //. in contrast to occurrence (with non-productive -ence). For example. 57 .g. curry. In certain cases when the r is pronounced. it does not do so and as a result Broadening will apply resulting in a Broad-Lax vowel before a pronounced r. turret.e. all with // in RP but // in GA. however. Murray.

. Let us now take a look at the other half of the story and see some examples in which Broad-Lax vowels appear in environments other than before r. -ing. then the Carrot-Rule will block the application of Broadening as expected. 58 . they will remain Plain-Lax. -ence.3 3 Most of these examples are repeated in Chapter 12 as groups of deviating words. -y –.g.. -ical –.e. However. Thus it seems that one half of the original suggestion concerning Broadening has already been borne out: before an r lax vowels are not always Broad-Lax as in some cases. i. as a result of the Carrot-Rule. then the Carrot-Rule will not be able to block Broadening.g. which will hence normally apply to the vowel making it Broad-Lax. if the r precedes a productive suffix – e.. cases of Broadness without r.Chapter 4 syllable-final /()/ /bl()/ /()/ /()/ /()/ /()/ followed by productive suffix occurring // blurring referring furry barring starring // // // // // occur blur refer fur bar star It is clear from the tables above that if a non-productive suffix follows – e.

bra //. psalm //. lawn //. Some of the examples. halve have // in GA. thought //. e. fallen //. broad //. stalk //. can’t //. talk //. 4 5 Recall from Chapter 1 that all these words are pronounced with // in GA. 6 In GA. 4.4 3. 2. pass //. Note irregular salmon / /. French or Italian pronunciation (so called DRAMA-words): bourgeois //.5 4. brass //. before a voiceless consonant or a nasal (so-called SAUCE-words): author /()/. ball //. dawn //. fought //. claw //. The vowel letter <a> followed by a voiceless fricative or a nasal+consonant cluster (so called ASK-words): ask //. sought //. The vowel letter <a> followed by a silent <l>+labial consonant cluster (so called CALM-words): almond //. Irregular cases: abroad //. drama //. sauce //. sonata //. raw //. aunt //. Irregular cases: father /()/. too. spa //. bath //. caught //. path //. Foreignisms imitating the original Greek. calm //. rather /()/. lather and rather contain //.6 // 1. The vowel letter a followed by a pronounced l+consonant or nothing or by silent l+k (so called CALL-words): bald //. palm //. calf.R-influence on vowels // 1. Shah //. tall //. last //. task //. balm //.g. half. walk //. 3. saw //. halve //. wall //. laugh //. class //. lather /()/. -au or -aw word-finally. gratis //. water /()/. calling //. -ough or -augh (so-called THOUGHT-words): bought //. 59 . law //. half //. dance //. 2. calf //..

kernel. This means that this word actually falls under the same rubric as. 7 Although this word is generally considered to be an exception. Broadening is not always predictable (sometimes it takes place without a potential trigger being present in the word). on the one hand. //. words like roar and raw. the /r/ is in syllable-final position. spar and spa. baa 'make the bleat of a sheep' and bar are totally indistinguishable for a non-rhotic speaker – a fact which contributes to the emergence of the so-called Intrusive-R. one with a Broad vowel without an /r/. Consequently. with an r in one member but with no r in the other. that is. Only in one word: colonel //. For example. and on the other hand. pore and paw.7 These examples of Broadness without r illustrate that. although in non-rhotic English only. 60 .Chapter 4 // 1. and the second <o> is silent. say. numerous pairs of homophones exist. mentioned in Chapter 2 but treated in detail in Chapter 7. the corresponding rhotic pronunciation. shows that in fact the <l> after the stressed vowel represents an /r/. and dropped in non-rhotic accents like RP.

obstruent. plosive. glide. as it is shown below. there are always thousands of combinations whose absence cannot be accounted for: they are potential words but have no meaning. rhotic accent. velar In this chapter. nasal. this is not the only way it affects the patterning of speech sounds. monosyllabic. suffix. let us cite the first stanza of Lewis Carroll's Jabberwocky. missing items) in the vocabulary by accident only and may gain some meaning later on.5. liquid. The English syllable Before you study this chapter. check whether you are familiar with the following terms: coronal. Such "nonsense words" are sometimes referred to as accidental gaps since they are gaps (that is. You may have already noticed that in languages in general only a very small portion of theoretically possible sound sequences is used as actual words. fricative. However. On the one hand. homophone. distribution. together with the Hungarian translation (by István Tótfalusi). morpheme. a nonsense poem in his book entitled Through the Looking Glass. we take a closer look at the structure of English syllables. the syllable is one of the major factors determining the restrictions on sound sequences. . In Chapter 2 it was demonstrated that the syllable plays a significant role in defining what positions host the targets of phonological processes like aspiration or R-dropping. As an example.

bröftyent a mamsi plény. nyalkás brigyók turboltak. certain accidental gaps do become part of the language with time. respectively. now it is generally accepted as a term to refer to unsolicited. however. therefore it would not be impossible to imagine them as. usually commercial e-mail sent to a large number of addresses. while brillig and plény are acceptable as words of English and Hungarian. No English or Hungarian words start with // or //. that word-finally you observe just the opposite: // or // is possible (cf. nyamlongott mind a pirityók. do not panic – however well-formed English words brillig and mimsy and gimble look. and the slithy toves Did gyre and gimble in the wabe: All mimsy were the borogoves. A GRUFFACSÓR Nézsonra járt. they are nonexistent. they really sound like English and Hungarian words. it appeared in a skit on the British television series Monty Python's Flying Circus. virtually the whole text). kerb // in the rhotic accents of English. although they contain exactly the same segments. a sound sequence is not a potential word as it contains some combination which is systematically rejected by the language. And the mome raths outgrabe. a nonsense trademark can start a life of its own. Not all gaps are accidental. as it happened in the case of spam (once a trademark for a canned meat product.g. though. For example. and it is completely unlikely that some ever will. and it is even used as a verb). just like the quasi-Hungarian words nézson and brigyó and plény. e. dialectal forms of existing words. If you do not understand the italicized words (that is. What is crucial is the fact that there is no principled reason for their nonexistence. In a great many cases. Notice. say. or Hungarian talp 62 .. not even as trademarks or internet terminology. purrtak a zepén. In fact.Chapter 5 JABBERWOCKY 'Twas brillig. neither rbillig nor lpény would be.

// and //. // for //. both languages seem to impose strict restrictions on what sounds can appear in what order in what position. are phonetically so close to the glides // and //. N = nasal. cf. Thus you can find inscriptions like GAIVS IVLIVS CAESAR for Gaius Julius Caesar. As we will see below. No wonder glides are also called semivowels! The intuition that there is no clear dividing line between vowels and glides is also reflected in the choice of the ancient Roman alphabet to represent both with the same symbol.. heart. <kn->. who or which. G = glide. 63 . or <wh-> in spelling never stand for clusters because one or the other letter remains silent. lie). P = plosive). or // for //. psychology (cf. not letters. L = liquid.The English syllable 'sole') but // or // is not. and English spelling can sometimes be misleading. gnome (cf. For example. O+O stop Spain ski sphere O+N snake snore shmuck schnook O+L slay shrimp plead trap O+G %suit %tune queen swing V+G1 eye tow G+L %fire %hour file owl /r/+/l/ %earl %girl L+N %earn %harm elm kiln N+F ounce nymph F+P east raft clasp ask 1 Diphthongs can be analysed as vowel-glide sequences. Hungarian gnóm). we discover the major phonotactic restrictions in English. viz.g. Bear in mind that we are talking about sounds here. that some transcription systems denote them with the symbols of the glides. all single consonants except for // can start such a morpheme (e. <gn->. Hungarian knédli 'steamed dumpling'). These restrictions are called phonotactics in phonology. Notice that the second members of closing diphthongs.. knife (cf. F = fricative. <ps->. The chart below illustrates some of the most frequent two-member combinations of sounds on either edge of English monosyllabic morphemes (O = obstruent. pit. In sum. e.g. In what follows. respectively. Hungarian pszichológia). V = vowel. Such sequences of letters are not taken into consideration either. so there are almost as many consonant+vowel sequences as the number of consonants multiplied by the number of vowels – therefore they are not included in the chart.

Chapter 5 The percentage mark (%) customarily indicates that the given example only applies to certain speakers – in most cases it shows dialectal variation. on both sides of syllables. and it is elaborated on below and in Chapter 11. while in words like suit the yod is only pronounced by conservative (that is. e. and then sonority decreases..// . which is only pronounced in rhotic accents. older) speakers of RP (//) but not by younger speakers or speakers of GA in general (//). sonority increases towards the vowel. On the basis of the examples. we arrive at the following order in which sound segments are usually organized in the syllable: obstruents .plosives The careful reader may have noticed that this list is more or less symmetrically organized. and sonorants inbetween.g. Moreover. having similar groups of consonants on either end (namely.nasals/liquids/glides . it bears a spooky resemblance to the sonority scale discussed in Chapter 2 and repeated here for convenience. vowels in the centre. RP //). earl GA // (vs.nasals .// . which forms a sonority peak. or. This phenomenon.glides . In words like tune the yod (/j/) is not pronounced in GA (//) so the example is only relevant to RP (//).fricatives . obstruents). In the chart above it mainly refers to /r/. is called yod-dropping.vowels . degree of sonority ——————————————————————————————→ oral stops (plosives) and affricates – fricatives – nasal stops – liquids – glides (semivowels) (– vowels) Therefore we can make the following generalization: within syllables. sonority increases towards 64 . the absence of a yod in certain positions.

and another plosive at the end. This can be schematically represented as follows. then the vowel. V G L N F P      tramp Further examples: V G L N F P        blind  GA quirk  GA swarm        Notice what happens in words like tender or button: since there are two vowels. for example. Henceforth we will call this the Sonority Principle. that is. a nasal. Let us illustrate with a few examples how the Sonority Principle describes the structure of well-formed syllables. The difference between the schwa-ful and the schwaless pronunciations is that in the latter case the second sonority peak is not a vowel but a consonant (the /n/). 65 . two syllables! Even if the /n/ is syllabic in button. The word tramp //. the number of syllables is unchanged. then comes a liquid. starts with a plosive. there are two sonority peaks. the number of sonority peaks. that is.The English syllable the vowel.

66 . can only be pronounced with two syllables as it contains two such peaks (a vowel and a (syllabic) /m/). What we have seen above. The conditions on syllabic consonant formation in English are discussed towards the end of this chapter. It also follows that not only vowels can occupy the sonority peak. thus the Sonority Principle needs reformulating: on both sides of syllables. e. for instance. resulting in // or //.g. then. which is a vowel or a syllabic consonant. there are as many syllables as sonority peaks.. The English word rhythm. directly follows: in a word. tramp // (mentioned above). It is very interesting that the above definition of the Sonority Principle can be turned inside out and translated as the definition of the syllable: it is a phonological unit which contains a sonority peak. sonority increases towards the peak. If we shuffle the segments in a well-formed syllable.Chapter 5 V G L N F P     tender   button       button   A simple definition of syllabic consonants ensues: they are consonants functioning as the sonority peak in a syllable. we arrive at the same conclusion: these must be disyllabic words.

however. there exist a number of examples where the Sonority Principle fails. V G L N F  P       stop   ski  apse    fox  In addition. i. still. 67 . The Sonority Principle is.. in a few cases segments of equal sonority follow each other within the syllable. Consider the following words: they all contain two sonority peaks. and consequently sonority neither rises nor falls. is much graver: while // is a possible (disyllabic) word of English.e. an accidental gap. therefore. the same is not true for the other: // starts with a syllabic /r/. all speakers of English insist that they are monosyllabic. The difference between (hypothetical) // and (hypothetical) //.The English syllable V G L  N F P      rhythm  //  //       The difference between rhythm and (hypothetical) // is that the latter is simply non-existent. one of the major factors determining and explaining what qualifies as a well-formed English syllable. However. and for independent reasons English words never start with a syllabic consonant.

We will see below that /s/ takes part in the construction of syllables in a special way in several further respects. the word-final examples fall into two categories: they are either monomorphemic (i. we can still conceive of it as a generalization describing the majority of the data. that even these exceptions are constrained: the problems are caused by obstruents.Chapter 5 V G L N F  P     sphinx    act  Notice that the word sphinx. it starts with flat sonority rather than the expected rise. however.e. and treat stop and the like as exceptions.g. hit-s or fif-th.. is doubly problematic: on the one hand.. it ends in a sonority rise rather than the expected fall. more specifically /s/. for example. The following chart summarizes the possible exceptions to the Sonority Principle and gives a couple of examples.. initial fall stop ski Spain final rise simple complex fox hits apse axe lads eighth flat sonority initially sphinx sphere svelte flat sonority finally simple complex act fifth adopt corrupt ached robbed As you can see. or the problematic segments straddle a morpheme boundary (i. e.g. the word is morphologically complex). Although words like these contradict the Sonority Principle. on the other. morphologically simple).e. 68 . in most cases fricatives. It is intriguing. fox or act. e..

this is four. the Sonority Principle serves us with a considerably reliable definition of the syllable. which is possible in all the languages of the world: one starting with a single consonant and ending with a vowel. In languages like English. some languages permit no consonants and therefore all syllables end in a vowel (such syllables are called open). for example. Compare rhythm and fox. and numerous violations of the Sonority Principle are attested. it is possible to find consonants at the end of any syllable (making it a so-called closed syllable). in others there cannot be more than a single consonant at the beginning. the Slavic languages like Russian) where (almost) any combination of their consonants is possible. there are strict conditions on sound sequences. usually with a strict adherence to the Sonority Principle. And finally. this very simple configuration is also the first to emerge during the 69 . the role of the Sonority Principle is far from uniform. discussed above: both contain two sonority peaks but only one of them (rhythm) is judged by speakers of English to be disyllabic. As to the syllable-final position. but they also contain violations of the Sonority Principle. that is. there is one syllable type which is universal. In certain languages each syllable must start with a consonant. (You may have noticed that the clusters at the end of texts or sixths are not only too long. which means that some syllables contain such "monster clusters" as // or // in texts or sixths.e. there are a few languages (e. Besides being universal. From a cross-linguistic perspective.. others (like Italian or Japanese) differentiate between wordinternal and final syllables. but there is always a limit on the maximal number of consonants. i.The English syllable In sum. In most languages.. and only have syllable-final consonants in one of the two types. Yet others (like English or Hungarian) allow for clusters syllable-initially but only certain types. with a sonority rise. In English.g. although sometimes it is overridden by native intuition.) Nevertheless.

but there are no syllables without a peak (in Hungarian.). etc. we provide a brief description of what language-specific phonotactic restrictions accompany the Sonority Principle. pray. As we have seen. are ruled out. cube.. and only a single consonant is pronounced. In what follows. however. the letter <x> at the beginning of words like Xerox.. without a vowel). English I/eye // or Hungarian ő '(s)he').g. the centre of the syllable is the sonority peak. fry. (Such spelling-to-pronunciation 70 . *ps. Two-member clusters usually consist of an obstruent and an approximant. skirt. slip. Just list the words Hungarian kids learn first. /s/. tube.Chapter 5 process of language acquisition. etc. Let us now turn back to the discussion of the English syllable. The warning is still in effect that you should not let words like pneumonia.g. quick. trip. there are syllables with a single vowel and no consonants (e.g.g.. but in pronunciation they are simplified. which is usually a vowel. stink. clean. fling. that is. xylophone and Xanadu does not denote a /ks/ sequence but a single /z/.. Similarly. Gwen. sphere. Certain consonants like // and // are relatively infrequent in this position. and in fact in English (and Hungarian) this peak is the only obligatory constituent – that is. and this very often leads to the sonority sequencing violations mentioned above. play. sport. One consonant. snip. *gn and *kn. The spelling of the words pterodactyl and mnemonics suggest initial clusters of flat sonority. psycho. dry. can be combined with any of the others except for voiced obstruents and /r/ (e. The English peak can be preceded by zero to three consonants and followed by zero to four. *pn. it can be any consonant except //. cry. when babies learn their mother tongue. swim.). Recall that it is usually /s/ that is to blame! A few rising-sonority clusters. puke. e. twin. and you will see. If there is a single consonant before the peak. since these obey the Sonority Principle (e. gnu and knight mislead you – they only start with a consonant cluster in spelling.

The English syllable regularities are discussed in detail in Chapter 11.) The nonexistence of these clusters of rising or flat sonority is curious because apparently they are completely acceptable in Hungarian (cf. the Hungarian equivalents of the above words, or the examples given earlier), although their foreign origin is evident. There is another set of rising-sonority clusters which is unattested in English, but this time the same holds for Hungarian, and in fact, we will be able to find a principled explanation for why they are so unpopular. These include, e.g., *tl, *tn, *pw, *fw – no English (or Hungarian) examples are available for them. What these clusters have in common is that they are homorganic, i.e., their members share the place of articulation. Both /t/ and /l/ are coronal, and so is /n/; both /p/ and /w/ are labial, and so is /f/. Although a whole lot of other homorganic clusters exist, e.g., /, , , , / plus the /s/+coronal clusters (recall that /s/ can combine with almost all other consonants), there is a clear dispreference for clustering consonants to share the place, one manifestation of which is a phenomenon referred to above, Yod-dropping. There is an absolute ban on /j/, the coronal glide, to appear after coronal //. There are no English syllables beginning with /, , , , /. After coronal /l/, it is again impossible to find a yod if the /l/ is preceded by another consonant, that is, when it is part of a cluster: e.g., */-/. Following a single /l/, the yod can "survive" dropping but only in conservative RP, cf. lucid //, lucrative //, ludicrous //. Even in RP, however, the pronunciations without the yod, i.e., //, //, //, are more frequent, and the same applies to // in words like suit /~/, super /~/, Zeus /~/, presume /~/. In GA, this tendency to drop the yod has

71

Chapter 5 become generalized to take place after all coronals – not only /, , / but /, , , / too. That is why new is // in RP but // in GA, tuna is // in RP but // in GA, dude is // in RP but // in GA. It is only in GA that the title Looney Tunes can refer to lunatic toons (cartoon characters) since both are pronounced //, as opposed to RP, where tune is //. In contrast, the yod is rather stable in both dialects in unstressed syllables, e.g., after a lone /l/ in value //, after an /s/ in capsule //, although after an /n/ as in avenue both options are available in GA /~/. All in all, /j/ is gradually disappearing after the other coronals, which can be considered as another illustration of the dispreference of homorganic syllable-initial clusters.2 As it has been mentioned above, the maximal number of syllableinitial consonants in English is three. The three-member sequences are, however, heavily constrained: they always begin with /s/ (again, it is /s/!), which is followed by a legitimate two-member cluster (strength, spring, square, splash, %stew RP //, etc.). Since all such syllables contain the /s/+(voiceless) plosive+approximant sequence, they always violate the Sonority Principle. Turning to the syllable-final position, we can state that any single consonant except for /h/ can occupy it. In addition, in non-rhotic accents like RP, /r/ is also banned at the end of syllables, as it was discovered in Chapter 2 – therefore the rule of R-dropping can be treated as a phonotactic restriction characterizing non-rhotic accents only. In two-member clusters after the peak, we usually find nasal/liquid+consonant sequences, which exhibit falling sonority, e.g., lamp, month, land, mince, help, bulb, elf, %carp, %herb,
2

You find further examples of Yod-dropping in Chapter 11, where it is discussed again from a slightly different point of view: as a letter-to-sound rule. In addition, it is argued there that the yod is in fact part of a complex vowel //.

72

The English syllable %search, film, %harm, %curl, etc. Notice that within the class of liquids /r/ systematically "pretends" to be more sonorous than /l/: -rl is possible (at least in rhotic accents) but -lr is not. When two obstruents compose a syllable-final cluster, one of them is usually /s/ (again!): /s/+obstruent in grasp, last, risk, etc., obstruent+/s/ in lapse, axe, etc. Flat sonority contours are also attested (apt, act, etc.) but the second consonant is always a coronal. In three-member strings (prompt, against, next, etc.) the third member is always a coronal obstruent, and in morphologically complex words additional combinations yielding the "monster clusters" with four consonants in a row can also be formed (ending in -ed, -s, -th – all coronals). The examples of final clusters we have seen up to this point also appear word-medially, e.g., /mp/ is found in both lamp and campaign, /lm/ in film and helmet, /st/ in last and asterisk, and /pt/ in apt and chapter. There are, however, certain word-internal consonant clusters which are impossible word-finally. In such cases, the consonant cluster suggested by the spelling undergoes simplification, and remains simple even if a suffix is attached to the word. For instance, // is well-formed within words like cognate, dignity, magnet, signature, resignation, but the // is deleted in sign and resign as well as in signing and resigning. The same goes for /mn/: it is acceptable in alumnus, amnesty, chimney, insomnia, damnation, hymnal, autumnal, but simplified (with the /n/ lost) in damn and damning, hymn, autumn. Homorganic nasal+voiced plosive sequences are also highly restricted unless the consonants are coronal: // is unmodified irrespective of its position (cf. lend, bind, wound, and candle, tender, boundary) whereas // and // only survive word-internally (amber, ambulance, bombard; finger, anger) but not finally (bomb, bomber, bombing; long, strong, sing,

73

Chapter 5 bang, singer, singing, banger). The distribution of the velar nasal is particularly intriguing: it does not normally appear between vowels in morphologically simplex forms like finger or anger (*fi[]er, *a[]er) – with just a few exceptions such as hangar. Next to a morphological boundary, however, it is rather frequent in such position, as we have seen above (singer, singing, banger, etc.). In this respect, what happens in the comparative and superlative forms of adjectives is surprising: the simplified cluster of the positive forms long, strong, young is "regained" in longer, stronger, younger; longest, strongest, youngest (all with //).3 Besides the restrictions on syllable-initial and -final consonant sequences, there is an additional type of phonotactic constraint, namely, one which applies to the vowel and the following consonant(s) together. Since in poetry this part of the syllable determines whether two words rhyme, phonologists conventionally refer to it as the syllable rhyme. There are several restrictions on the English rhyme, e.g., // can only be followed by coronal consonants (shout, crowd, south, town, etc.); // can only be followed by alveolars (exploit, void, voice, noise, coin, coil, moist, point); a long vowel is only possible before a consonant cluster if the cluster is made up of coronals (mind, boost, faint, etc.); and in word-final open syllables (i.e., without a closing consonant) the vowel has to be either long (monophthong or diphthong, e.g., taboo, array, RP near) or unstressed (happy, comma, etc.). Before discussing the restrictions concerning the peak, let us take another look at syllable and word edges, and the asymmetry between them. On the one hand, in word-final clusters more consonants are possible than in word-medial ones; what is more, they frequently violate the Sonority Principle (cf. sixth, text), which also holds for word-initial clusters (cf. stop, Spain, screw, strip). On the other hand, it is a well-known fact that not all
3

Examples like these are repeated in Chapter 11, in the discussion of silent letters.

74

the consonant following the schwa must be more sonorous than the one 75 . however. button [-]. where an alternative pronunciation (mainly used in slow. in most cases. it is /n/ or /l/. The process. finally [-]. As far as diphthongs go. national [-].The English syllable combinations of well-formed syllables yield a well-formed word. syllabic consonant formation (SCF) is only possible in unstressed syllables. careful speech) contains a schwa followed by a non-syllabic version of the consonant. Besides vowels. Recall from Chapter 2 that syllabic consonants are indicated with a subscript [ ]. what happens is that the schwa drops out but the number of syllables is preserved since the following consonant steps up to act as a peak instead. Second. For instance. their second members can only be one of three vowels. Finally. First. in which case the final consonant is syllabic []. either monophthong or diphthong. let us see some of the phonotactic constraints on the syllable peak. we find that they are heavily restricted: not all the possible combinations of the vocalic segments of English exist. In RP. Basically. In most cases it is occupied by a vowel. Moreover. the word table has two possible pronunciations. has a number of conditions. table [-]. e. so the attempt at joining the apparently well-formed right edge /kst/ of a syllable like text with the apparently well-formed left edge /str/ of a syllable like strip will result in the string /kststr/ unattested word-internally. certain consonants can also constitute the peak of the syllable. in which case they are syllabic consonants. one with a schwa [] and one without..g. faculty [-]. the consonant following the schwa must be a sonorant. /  / – this number is radically smaller than the number of English monophthongs. It seems impossible to talk about phonotactic restrictions without making reference to the position of the syllable within the word.

and it must be more sonorous than the one preceding it. yielding disyllabic //.. e.g. SCF is possible... This sonority condition does not hold if the first consonant is /r/: barrel [] is well-formed although both consonants are liquids. but not in vanity (the /t/ is not a sonorant). barren [] is possible alongside examples like banner GA [].g. the consonant following the schwa must be a sonorant. In addition. consequently a peak is lost. Intriguingly. is not as restricted: although it always occurs in initial syllables. Schwa can also drop out in such a way that the number of syllables is NOT preserved – a vowel is lost. In non-rhotic English (including RP) /r/ can only become syllabic word-internally. For post-stress syncope to take place. //.Chapter 5 preceding it. weak (cf. The underlined vowel can not only 76 . different. Such straightforward examples of vowel loss are traditionally referred to as syncope. That is how the underlined vowels in camera.. better [] or []) or even in stressed syllables (e. family. bird []). //. natural []. consequently a syllable is lost. compare this with column //. however. In camel. e. //. The same applies to kennel ([] or []) versus melon [] and not *[]. because /l/ is more sonorous than /m/. yielding []. separate (adj) can be deleted. where it is not. the following vowel must be unstressed. Pre-stress syncope. SCF is not the only form of schwa deletion. the conditions on syncope are more strict after a stressed vowel than before it. though. the consonants surrounding the target schwa do not necessarily obey the sonority constraint.g. felony (nasals are less sonorous than liquids) or separate (v) (the third syllable contains a full vowel). that is. but in rhotic English (especially GA) /r/ can also become syllabic word-finally (e.g. Chapter 3).

similarly to the overall structure of the syllable. however. support as sport. parade as prayed. and police may only differ in the final consonant from please. suffice. Both are. Interesting new homophones emerge: terrain may sound the same as train. potato. 77 .The English syllable be omitted from words like terrain or parade but also in suppose. The difference between syncope and SCF. etc. in most cases (except for pre-stress syncope) governed by some kind of sonority condition. is that the number of syllables in the word is affected in the former but not in the latter. then. Sapir as spear.

root. the vocal cords/folds and all the different phenomena that are related to the operation of the larynx. (a comparison of English and Hungarian) voice assimilation. aspiration and glottalization. which can close the windpipe. syllabic consonant. allophone. the air continues upward in the windpipe up to the larynx – the front. glottalization. organs of speech. Pre-fortis Clipping. As it can be seen in the diagram. Having left the lungs. pulmonic egressive airstream. This includes voicing and voicelessness. sibilant. two bundles of muscle. phoneme. utterance. glottal stop. Before discussing these. voiced. larynx. the . voicing In this chapter we take a look at the articulatory role of the glottis. devoicing. glottis. frequency. and a circular one serving as the frame for the larynx –. the larynx consists of some cartilages – one fixed in the front and two movable ones in the back. and the effect of voicelessness on preceding vowels. check whether you are familiar with the following terms: allomorph. voice assimilation. devoicing. the oral and nasal cavities. shield-like part of which is called Adam's apple – then into the pharynx and the supraglottal cavities. which may produce numerous different effects. vocal cords/folds. suffix. Recall from Chapter 1 that the basic mechanism that is used to produce speech sounds in English and Hungarian is a pulmonic egressive airstream mechanism.6. the epiglottis. plus one more on the top not shown in the diagram. Laryngeal features Before you study this chapter. In the larynx it has to pass between the vocal cords/folds. hiatus. voiceless. let us take a look at the structure and parts of the larynx and their different configurations. aspiration.

the glottis. In 79 . voiceless consonants are articulated. This is repeated periodically at a rate of about 120 times in average in males and about 220 times in average in females (that is. and the opening between them. that is. Depending on how tense the muscles of the vocal cords are. This mechanism is very similar to when children put a blade of grass or a piece of a leaf between their two thumbs and then blow it producing a high pitch trumpet-like noise. they return into their original position. When the vocal cords are wide apart then silent breathing is produced. this way producing vibration. This way the vocal cords will close or open the glottis to different degrees. the basic frequency in males is about 120 Hz and about 220 Hz in females). a voiceless glottal [h] sound is produced. voiced sounds. When the vocal cords are slightly pulled together but still do not close the glottis and no vibration is produced. because of their elastic nature. If the vocal cords are pulled a bit closer together than in the case of voiceless consonants.Laryngeal features vocal cords connecting the cartilages. In another configuration the elastic vocal cords are pulled together and the outflowing air pushes them apart and then. the two cartilages in the back will move closer together or further apart.

then it is easy to see that obstruents occur in voiced-voiceless pairs or to put it very simply: in the obstruent part of the table there are always voicedvoiceless pairs of stops. Let us now take a look at how English makes use of the qualities voiced and voiceless in the different classes of speech sounds. we can say that voicing is not a distinctive quality in vowels. it does not distinguish vowels from each other. fricatives and affricates at each place of articulation. and in Hungarian as an extralinguistic device to express surprise in [] or to optionally break up a hiatus – the sequence of two vowels – in words like kiiktat [kt]. If one takes a look at the table of manners and places of articulation in Chapter 2.e. Voiceless vowels are only used when whispering.. Finally. fricatives and affricates. consider vowels first: as all vowels are always produced as voiced. it is possible to produce a total closure. The glottal fricative does not have a voiced counterpart in English – as opposed to Hungarian. the very beginning. This leaves us with obstruents: stops. the first few milliseconds of the vowel being voiceless – occur after aspirated stops. 80 . a complete obstruction to stop the air in the larynx. periodic way. i. This way a glottal stop is produced. To start with the easy part. partially devoiced vowels – vowels which have lost part of their voicing. Sonorant consonants behave in a very similar way to vowels: they are always voiced by default and they only become partially devoiced after aspirated stops. the sound often heard in the pronunciation of words like bottle [] or [] in British dialects (mentioned in Chapter 2 and discussed in more detail below). The only exceptions seem to be the glottal stop and /h/. a topic already discussed in Chapter 2 and to be discussed later in this chapter.Chapter 6 both cases it is a flexible. elastic string – the vocal cords or the blade of grass – that is forced to move in a fast.

. Lenis obstruents are weak and often lose their underlying voiced quality.. konyha []. e.e. are strong. they are phonologically voiced and may be realized by voiced or voiceless speech sounds in actual pronunciation depending on the environment. and are thus always realized by voiceless speech sounds. The most interesting aspect of the voicing of obstruents is the stability of voicelessness and the relative instability of voicedness in English. they only behave as if they were voiced) in many cases. As English voiced obstruents seem to be voiced only phonologically (i.g. on the other hand. utterance-final positions and next to fortis obstruents. [z].g. but not a separate voiced glottal // phoneme.. The devoicing process affecting lenis consonants typically applies in utterance-initial. Fortis obstruents. two other terms have been introduced instead of voiced and voiceless: lenis (Latin for 'weak') and fortis (Latin for 'strong'). e. The socalled voiced obstruents of English are very often realized by a partially or fully devoiced allophone – this devoicing is represented in transcription by a small circle below the symbol of the sound. 81 .Laryngeal features where the phoneme /h/ does have a voiced allophone [] occurring between sonorants.

The voicing of obstruents normally starts before the closure is produced and voicing only ends after the closure ceases – Hungarian is said to have pre. i. before or after a pause.e. in utterance-initial and -final position. partially or fully voiceless. This is probably the consequence of the fact that there is a timing difference between the beginning or end of vocal cord vibration (voicing) and the beginning or end of the closure. while in final position voicing stops first and only after that is the stop released. Note that Hungarian is different in this respect as there is no such devoicing in initial or final position. lenis obstruents often devoice at least partially: in initial position it is typically the beginning of the obstruent that is voiceless while in final position it is the end. the preceding obstruents become devoiced. In column (c) all the sample words contain a lenis obstruent followed by a fortis one. In initial position closure is produced first and voicing starts only a few milliseconds later. It is similar to what happens in identical Hungarian clusters. 82 .Chapter 6 Utterance-initial (a) Bravo! [b] Good! [] Zany! [] Damn! [] Very much! [] Utterance-final (b) Mad! [] Go ahead! [] Think big! [] Bob! [] Leave! [] Next to a fortis sound (c) (d) obtain [t] cheesecake [] bigfoot [] egghead [] roadster [()] matchbox [] baseball [] cookbook [] life gear [()] Shoot back! [t ] In (a) and (b).and postvoicing in obstruents.. As a result of the influence of the fortis (voiceless) sounds.

whenever a Hungarian voiced and voiceless obstruent occur adjacently. It is called regressive if the sound that changes precedes the sound that influences it. We may call this a case of voice assimilation by which the voicing of one sound becomes identical to that of a neigbouring one. The difference between the lenis+fortis case in the English and Hungarian examples – columns (c) and (e) – is manyfold: devoicing is not necessarily complete in English. it seems that in English it is the result of assimilation that is fixed – it is always voicelessness –. devoicing is not obligatory in English but it is in Hungarian – but assimilation is regressive in both languages. but it is in Hungarian as indicated by the difference in the symbols.Laryngeal features English (c) obtain [t] cheesecake [] bigfoot [] egghead [] roadster [()] Hungarian (d) (e) rabtól [] rézkarc [] hangfal [] éghez [] roadster [] (f) matchbox [] baseball [] tökből [] afgán [] kertből [] matchbox [] baseball [] cookbook [] life gear [()] Shoot back! [t ] As it can be seen. one of them has to change its voice value. in Hungarian it is the fortis obstruent that assimilates to the lenis one – voicing applies. As for the difference between the fortis+lenis cases – columns (d) and (f) – the difference is even bigger. Similarly. whenever a fortis and a lenis obstruent of English occur adjacently. That is. It is not just a matter of degree – partial or complete – and nature – optional or obligatory – but also a matter of value: in English the lenis obstruent assimilates to the fortis one – devoicing applies –. and it is called progressive in the opposite case. in English we see progressive devoicing while in Hungarian we see regressive voicing. Consequently. while in 83 . one of them changes its voice quality.

as these are the ones that appear after vowel-final roots. '3rdSg' or 'possessive' and -ed 'past tense' or 'past participle'. These forms then assimilate to the root-final consonants. Normally it is assumed that the basic forms of these suffixes are /z/ and /d/. respectively.Chapter 6 Hungarian it is the direction – it is always regressive. // legs // tabs // heads // means // girls // ways // shows // // kicks // blokes // taps // turnips // hats // laughs // baths // // churches // judges // bushes // garages // kisses // buzzes // stretches // 84 . English ► ► ► ► ► optional partial or complete its result is always devoicing may be regressive or progressive initial or final devoicing may apply ► ► Hungarian ► ► ► obligatory complete its result may be devoicing or voicing always regressive no initial or final devoicing There is one more special area of English voice assimilation that we have to mention here: the assimilation of the suffixes -s 'plural'. The following table sums up the differences between voice assimilation and devoicing in English and Hungarian.

// begged // robbed // advised // depraved // damaged // contained // filled // followed // // clicked // ripped // laughed // passed // kissed // hushed // stretched // attached // // wanted // mended // protected // beheaded // located // paraded // navigated // vaccinated // In the last column. //. We have to remember. after sibilant consonants. the // allomorph is used. that this voice assimilation is different from the devoicing cases in that it is always complete. // is used after the sounds //. After all other voiced root-final phonemes the basic variant // is used while after all other 85 . after root-final alveolar stops // and //. progressive and obligatory. The first column of the table shows that the basic form // is used after all non-sibilant voiced sounds – both consonants and vowels – while the second column demonstrates that // occurs after voiceless non-sibilants. //. as an // is inserted between the two alveolar stops. that is.Laryngeal features As can be seen in the last column of the table. though. as it would be difficult to pronounce the sibilant // of the suffix after the root-final sibilants. //. we can say that the suffix consonant progressively assimilates to the rootfinal consonant. //. The suffix -ed behaves in a very similar way to -s presented above. All in all. and //.

in spay and spray. there is. i. the two forms of aspiration are not unrelated. when they zigzag.Chapter 6 voiceless root-final phonemes a // allomorph occurs. quick []. The two parallel lines symbolize the vocal cords. This assimilation process is also always complete. and its voicelessness spreads onto the vowel. In bay. as a result of which the vocal cord vibration characteristic of all vowels starts much later than the release of the plosive and the onset of the vowel. As you know. We know that it has two forms. Therefore.. its vocal cord vibration lags well behind the release of the plosive. we have a voiced plosive. In Chapter 2. their articulation involves vocal cord vibration. but it is left unexplained what exactly aspiration is phonetically. In words like pay []. on the contrary. Now we turn to laryngeal processes other than voicing. which is in fact not a separate sound segment but merely the acoustic impression that we get due to the first half of the following vowel being devoiced. when they are straight. which is. This is schematically represented in the diagram below. identical with /h/ (= open vocal cords plus no considerable obstruction to the airflow in the oral cavity). aspiration is mentioned as one of the processes affecting the voiceless plosives /  /. what is heard right after the burst of the plosive is a short period with a voiceless vowel (= open vocal cords plus no considerable obstruction to the airflow in the oral cavity). Whether it is a vowel or a consonant. they are the same: the voicelessness of the plosive spreads onto the following segment. One is the short []-like sound following the plosive. all vowels are normally voiced. The other manifestation of aspiration is the devoicing of a following sonorant consonant in words like play []. there is no vibration. obligatory and progressive. true []. a voiceless unaspirated one owing to the 86 . phonetically. the /p/ is voiceless. Of course.e.

button []. and no gesture above the larynx. especially London English. This is called glottal reinforcement. the voiceless plosives /  / (and also //) are in certain positions accompanied by a short closure of the vocal cords.. which devoices the following vowel and sonorant.1 Another process caused by laryngeal activity is glottalization. a voiceless aspirated /p/. bat []. e.g. teacher [()]. actor [()]. As it is explained in Chapter 2. i. this can even happen in words like let []. butter [] (or []) or city [].g. 1 We ignore the slight devoicing of /b/ at the beginning of bay.. especially before a syllabic /n/. in several non-standard varieties of English. glottal replacement can take place. 87 . respectively. by the so-called glottal stop ([]).. Sometimes. e.e. that is. in pay and pray. What connects these examples to the previous discussion is the fact that the production of the glottal stop involves the movement of the vocal cords only. a /t/ can be completely replaced by it.Laryngeal features preceding /s/.

it patterns exactly like phonologically long vowels do. i. diphthong or triphthong – is immediately followed by a fortis consonant or a nasal and a fortis consonant within the word.Chapter 6 A final rule that belongs to laryngeal processes – although not a strictly laryngeal one – is the way fortis consonants influence the phonetic length of the preceding vowel. Remember (from Chapter 3) that the phonologically short but phonetically long vowel // behaves identically in this respect. the phonetic length of the vowel is not indicated with the colon originally. The case of diphthongs is similar: their length is encoded in the combination of 88 .e. hence the name of the process. The change in vowel length is indicated in narrow. phonologically short vowels but it is important to remember that there is no change in the quality of the vowels at all. Pre-fortis Clipping. If a long vowel – monophthong. Long vowels become approximately as short as real. long monophthong balloon [] believe [] store [()] star [()] stir [()] span [] carnival [] spawn [] shortened long monophthong boot [] speak [] sports [] start [] shirt [] attacked [] dance [] daunting [] long diphthong obey [] decide [] allowing [] towed [] cleared [] destroy [] staring [] curious [] shortened diphthong mistake [] tonight [] bounce [] boat [] fierce [] catering [] spouse [] biting [] Note that in the case of the vowel // we cannot indicate shortening as the vowel is classified as a phonologically short vowel and.. phonetic transcription with the symbol [] instead of [] after the vowel. then the vowel will be shortened or clipped. as a result.

where the consonant following the vowel is not only voiceless but also fortis. the distinction between /t/ and /d/ may be lost in certain positions... and the // is much shorter in atom [] than in Adam []. aspiration and glottalization. To sum up the discussion of laryngeal processes.Laryngeal features two symbols rather than a length mark – since neither of the two elements is lost through clipping. which constitute some of the most significant allophonic rules that English consonants undergo.. but also in producing individual segments like /h/ or the glottal stop. e.g. devoicing. In the so-called tapping dialects of English.g.g. utterance-final position. In addition. we can state that the vocal cords play a crucial role in the articulatory process not only in determining the voicedness of speech sounds. said []. the // is much shorter in write [] than in ride []. Similarly. e. there is a 89 . e. Recall that lenis obstruents systematically become devoiced/voiceless in. short in writing [] but long in riding []. both being replaced by a tap. for instance. The duration of the vowel ultimately becomes the indirect indicator of the nature of the following consonant. they are responsible for certain phenomena. we are again unable to show this phonetic shortening in our transcriptions. We have also seen that although both Hungarian and English obstruents take part in voicing assimilation. short in seater [()] but long in seeder [()]. seed []. What is intriguing about Pre-fortis Clipping is that it is clearly conditioned by the fortis character of the following consonant. as opposed to words like seat []. rather than its voicelessness. mentioned in Chapter 2. a preceding long vowel remains long. Still. but the length of the vowel is still there to show the fortis/lenis character of the original consonant: short in atom [] but long in Adam [].

90 . Finally. it has been demonstrated how complex an interaction exists between the vocal cord activity characterising a consonant and the phonetic length of the preceding vowels. it always proceeds from right to left). the direction of the assimilation is fixed (namely. it always results in voicelessness). whereas in English the output of the process is fixed (namely.Chapter 6 huge difference between the two languages: in Hungarian.

compound. dark-L. when the following morpheme – suffix or word – starts with a vowel. We have already mentioned a number of such processes. First. Miller or The mill is closed). for example. it is clear (e. Connected speech Before you study this chapter. morpheme. clear-L. full vowel. very often several sentences one after the other. However. dental. high vowel. yod This chapter deals with the phenomena that characterize connected speech. R-dropping. assimilation. combinations of words rather than individual words uttered in isolation. reduced vowel. that word-final consonants are always ambisyllabic when followed by a vowel in the next word. suprasegmental features. Recall. rhotic. and choose their pronunciation variant accordingly. linking-R. Elsewhere includes the ambisyllabic position. velar. function word.g. nasal. alveolar.g.g. plosive. glottalized. the /l/ at the end of mill.. syncope. too. we found that syllablefinal /l/ is dark. allophone. that is. diacritic. palatal. but it is clear elsewhere. non-rhotic. is dark when the word is final in the utterance (e. devoicing. suffix. aspirated. Bear in . tapping/flapping. in the discussion of L-darkening in RP. which are recapitulated below. Therefore. These cross-word processes are of crucial importance since rarely do we pronounce a single word only – normally we use phrases and sentences. segment. idiomatic. hiatus..7. utterance. stressed/unstressed. Where's the mill?) or when the next segment is a consonant (e. especially in Chapter 2. glottal stop. ambisyllabic. intrusive-R. It's Mr. check whether you are familiar with the following terms: accent.. There are two mills here or The mill was closed). also from Chapter 2.

as in at ease. cf. Second. as in a tease. just like retúrn. This is a logical consequence of the fact that peaks are never ambisyllabic – after all. Word-initial /t/ is always aspirated and never tapped. aspirated [] (syllable-initially)..g. tapped/flapped. and replaced by the tap/flap [] (in ambisyllabic position).. aspirated. recall that /t/ has several allophones in the dialects of English. but in get úp. after an /s/). It is tapping in GA or informalcolloquial British English that we are concerned with here most. Similarly. Chapter 6) (syllable-finally). since this is the process which clearly affects ambisyllabic consonants only. they define syllables (cf. e. rather exceptionally. the aspiration-killing effect of a preceding /s/ can 92 . where the next vowel is unstressed. whereas in RP the string is. The same contrast is found in the pair might I (tap) vs. plump eye (unaspirated). Let us emphasize that this is an isolated. just as dark in cancel a meeting as in cancel the meeting. not only do we find tapping in get alóng. The expression at all is pronounced differently in the two standard varieties of English: the expected unaspirated pronunciation of the /t/ is only found in GA (of course. and all word-final consonants followed by (any!) vowel undergo the process.Chapter 7 mind that syllabic /l/ is dark. plum pie (aspirated) vs. whatever it may be followed by. too. In sum. glottalized [] or replaced by a glottal stop [] (cf. my tie (aspiration). and consequently aspiration appears: []. word-final /t/ is never aspirated but may be tapped. glottalized. Chapter 5). /t/ has the following major allophones: plain [] (e. with tapping: []). irregular example. treated as if it was a single word. While within words a consonant must be followed by an unstressed vowel to be ambisyllabic (the /t/ is tapped in átom but not in atómic). across words this stress-sensitivity ceases to exist. and word-final plosives in general do not normally become aspirated.g. therefore their affiliation cannot be ambiguous.

Chapter 3). an /r/ which is absent from spelling and from the rhotic accents of English. or. (v) Both Linking-R and Intrusive-R are always sandwiched between two vowels: they are preceded by a (non-high) vowel and followed by another vowel in the next morpheme. a broad lax vowel. a non-high vowel. a schwa (as the discussion on the R-influence affecting preceding vowels in Chapter 4 shows).Connected speech only be exerted if the /s/ is in the same syllable as the following plosive: the /t/ is plain in both stake and mistake but aspirated in miss Tom. destroy. that is. both always pop up between vowels in a hiatus (cf. R-dropping. it follows that Linking-R always follows one of /   /. We observe a few interesting facts when we compare Linking-R and Intrusive-R: (i) They are phonetically identical. in unstressed final syllables. That is. a wordfinal <r> which does not undergo R-dropping because the next morpheme starts with a vowel. (iv) It is a general feature of Intrusive-R in all the non-rhotic accents exhibiting it that it does not appear in a random fashion. namely /   /. Besides L-darkening and tapping. (iii) Since a word-final <r> can only be preceded by a broken tense vowel. but after certain vowels only. (ii) Both of them characterize the non-rhotic accents of English only – linking and intrusion go hand in hand with R-dropping. i. which "saves" it. 93 . This is called Intrusive-R. that is. the hiatus. they break up. We have also seen that sometimes a "historically unmotivated" /r/ shows up between two morphemes.. after a non-high vowel. in fact. You may be able to recall from Chapter 2 the phenomenon called Linking-R. there is a third rule which applies across words in the same fashion as word-internally.e.

/. manna and manner. spa and spar. …spar is… //. spar. spa. (You may only face the fact that they are spelt differently when you start learning to read and write at school. respectively. and have the same function (to fill a hiatus). although both are pronounced with the same sound. manner. For you. and the next morpheme begins with a vowel. manna analogously to pore. an /r/ suddenly appears between them: …pore is… //. after a non-high vowel). it should also be clear that the existence of both crucially depends on the presence of the R-dropping rule. words like paw and pore. Doesn't this remind you of the sufferings you underwent in primary school while trying to memorize that gólya 'stork' is written with <ly> but bója 'buoy' with a <j>.1 pore // // /r/ paw // // /r/ spar // // /r/ spa // // /r/ manner // // /r/ manna // // /r/ before a pause before a cons. spar. Suppose you are a speaker of non-rhotic English. it should be clear that Linking-R and Intrusive-R are virtually the same: they appear in the same position (intervocalically. Let us illustrate how Intrusive-R must have come into being. On the other hand. …manner is… //. before a vowel 1 Recall the discussion of Broadness without r in Chapter 4. /j/?) You also notice that when words like pore. are homophones: /. You conclude that whenever a word ends in /   /. . an /r/ is inserted inbetween. manner are followed by a vowel-initial element.Chapter 7 How can all these five observations be accounted for in the simplest way? On the one hand. 94 . You start treating paw.

although it is not obligatory. First.. it is the yod.Connected speech Therefore. irrespective of the relative order of the two. If the first vowel is high and front. e. it is /w/. although in such cases it is not a /r/ which is inserted but a semivowel. //.g. that is. e. it is not always complete. as it is described in Chapter 6. that is.g. There are two basic phenomena which need mentioning: assimilation and deletion. being velar (just like //). Can they be similarly broken up by a hiatus-filler consonant? The answer is yes. Notice that the choice of the glide is not random. e. Linking-R and Intrusive-R are both the manifestations of the same process of hiatus-filling after word-final non-high vowels. produced by the front surface of the tongue (just like //). link'. whereas the former characterize all the dialects of English. facilitating the smooth transition between the vowels. Such phenomena are frequently referred to as liaison. The output of voice assimilation in English is always devoicing. The major difference between the hiatus-filling glides and /r/ is that the latter only has this function in non-rhotic accents. If the first vowel is high and back. either: /j/ is coronal. As a result. Assimilation processes are of several different types. you and me //. and it is more limited than what we find in Hungarian.g. a form of voice assimilation is observable in English.. //. After the discussion of what processes affect vowels meeting across morpheme boundaries in connected speech. and now we are faced with a further example illustrating it..g. a voiceless consonant affects a voiced one. me and you //. The question arises what happens in hiatuses when the first member is not a non-high vowel. Footnote 1 in Chapter 5 mentioned the close connection between the high vowels and the glides. let us turn our attention to what happens to consonants in such situations. a French noun meaning 'connection. the direction of the 95 . e. while /w/ is formed in the back of the oral cavity.

This is brought about by a /j/ that follows either an alveolar fricative (/s/ or /z/) turning it into its postalveolar equivalent (// or //. missed Jane is [ ] rather than *[ ].. catch Bill is [ ] rather than *[ ]. blood pudding [b p]). including the dentalization of alveolar /t d n l/ when they are followed by dental // or // (this is indicated by the diacritic [  ].g..g. there is a phenomenon which is unattested in Hungarian: (occasional) cross-word palatalization.Chapter 7 assimilation is not fixed – it can be either regressive or progressive. respectively). yielding []. too. the velar assimilation of /t d/ (e. However. or an alveolar plosive (/t/ or /d/) turning it into a postalveolar 96 . These cases are not problematic to Hungarian speakers as such processes automatically take place in Hungarian.. This does not only apply in words within a compound but also across words within the utterance. what’s this is [ ] rather than *[ ]. e. in question [ ]). Progressive devoicing is what poses particular difficulties for Hungarian speakers of English since it subsumes cases where the opposite happens in Hungarian. whose directionality coincides with Hungarian. is most likely for fricatives and affricates (therefore it is also referred to by certain authors as "Fricative Devoicing"). yielding *[] or *[] (this latter is in fact the standard Hungarian pronunciation of the word).g. it could [k k].g. Besides voice assimilations. Thus in the oftcited example matchbox. rather than the other way round. have to [ ] (or even [ ]). as in his tie [ ].. bad company [g k]). Regressive voice assimilation. and nasal place assimilation (Green Peace [m p]. live show [ ]. Matt thanked [ ]). English exhibits a variety of regressive place assimilations. the /b/ assimilates to the voiceless affricate. the labial assimilation of /t d/ (e. eight pence [p p]. e.

etc. ease your pain [  ] or [ ]. I don't know.) Assimilations. granny) for instance. first knight. especially when more than two consonants "pile up". or herd with an /h/ vs. a /h/ is very often silent in the function words he-him-his. are not the only processes affecting consonants across word boundaries. or a /d/ between an /n/ and another consonant. véhicle without one. Guns and Roses.. In addition. We told her. the diminutive form hankie). where the underlined consonant letters are normally unpronounced. could you see [ ]. The expression "occasional" refers to the fact that this assimilation is optional (characterizing faster. colloquial speech rather than slow and careful pronunciation) and only applies on certain restricted occasions. 'er. next day.. Optional consonant deletion is just as frequent. which is sometimes reflected by spellings like 'im. Namely. Christmas.g. e. too. you. exactly and grandmother (cf.. find me. as in Wednesday. e.g. respectively). why don't you love me [   ].Connected speech affricate (// or //.2 2 In fact. etc. gran. however. You may have noticed that such "congestions" get simplified word-internally. send Jim. see Chapter 11.. handkerchief (cf. A few examples: this year [ ] or []. her. vehícular with a pronounced /h/ vs.g. (For word-internal palatalization. 'em. That'll teach them may remain unpronounced. aitches are not only dropped at the beginning of unstressed function words but also word-internally before an unstressed vowel. your. Who is he?. for obvious reasons. Thus the underlined consonants in I met him. In addition. it produces palato-alveolar /   / only if a word ending in one of the alveolar obstruents /   / and a function word beginning with /j/ (e. Saint Paul. shépherd without one. see below). have-has-had and so is the voiced dental in them (for the so-called weak forms of function words. mind your head [ ]. plus a few other common words including year and usual) are combined. yet. This is called Aitch-Dropping. rock and roll. it is a wide-spread 97 . cf. The same happens across words to a /t/.

Of course. leave me alone. nouns. One of the most important of such features. Curiously enough. some of the processes mentioned above only apply when the second of the two words juxtaposed is a function word: palatalization is possible in miss you but not in. In sum. i. The way stress is placed in English words is dealt with in Chapters 8-9 – for the time being. perhaps. cf.Chapter 7 Finally. than the neighbouring syllables. intonation. yielding 'ouse instead of standard house. this generalization can feature of non-standard pronunciation in all dialects (but. When words are combined into phrases. the second syllable of ideas is stronger than green. it is worthy of mention that the final /v/ of words like give or leave can also be deleted if they are followed by an unstressed function word (e. 98 ..g. those that characterize larger strings like syllables or utterances. so here we can concentrate on the other one: stress. more prominent. will be devoted a whole chapter to later (Chapter 10). verbs.. in the sentence Colourless green ideas sleep furiously. segmental phenomena – let us scrutinize the suprasegmental features of connected speech. green is stronger than the first syllable of colourless. which makes it stronger. /v/deletion is possible in leave me alone but not in leave Maureen alone. the strongest phrasal stress normally falls on the final element. suffice it to say that all non-function words (that is.e. say. more extensively in England than the US) to "drop one's aitches" at the beginning of content words as well. usually the (final) major stress of the final element is even stronger than the others. contracted gimme from give me). and when phrases are combined into sentences. and the major stress of furiously is stronger than that of either sleep or ideas. for example. For example. miss Yolanda. the strongest major stress of the final phrase in the string receives the greatest emphasis. adjectives and adverbs) contain at least one syllable that constitutes a rhythmic beat (called major stress). After the discussion of the cross-word phenomena affecting individual speech sounds (segments) – that is.

Similarly.Connected speech be overridden if the speaker wishes to put extra emphasis on another word in the phrase or sentence for some reason. the shorter the stressed vowel and the more compressed the 99 . thus producing a stress pattern which is the mirror image of the usual phrasal stress pattern. but not in Hungarian) rhythmic beats occur at more or less equal intervals: the greater the number of following unstressed syllables is. and consequently the time elapsing between two major stresses is approximately the same. This is traditionally referred to as stress-timed rhythm..) The rule applies in longer compounds as well. Further details on compound stress are adduced in the next chapter. The alternation of rhythmic beats and weak (unstressed) syllables produces the rhythm of speech. (Notice how much this resembles Hungarian examples like eladó lány 'marriageable girl'.e. What follows from this is the fact that in English (and similar languages. so colourless may as well become the most prominent in. versus eladólány 'salesgirl'. when the various degrees of stress are discussed. and therefore the basic unit of speech rhythm is the syllable (this is called syllable-timed rhythm). a big wig is simply a wig which is big (it is a phrase consisting of an adjective and a noun). compound stress. but a redskin is a North American Indian.g.. while a bigwig is an important person (a compound with its idiomatic meaning). is normally assigned to the first term. which is a compound. a red skin is a skin which is red. Colourless green ideas sleep furiously and not coloured ones! As opposed to phrasal stress. A major difference between English and Hungarian lies in the type of rhythm they exhibit. i. e. so pet is the strongest term in both pet shop and pet shop boys. which is a phrase. While in Hungarian each syllable is pronounced in about the same time. For example. in English it is the sound string stretching from one major stressed syllable up to the next one (the so-called foot). maximal prominence within a compound word.

Perhaps the most effective strategy whereby syllables can be "squeezed" is vowel reduction. Of course. as we have already seen above. . /). There are roughly forty words in English that have two basic forms: one which is the usual. (r). e.g. The list of the most common such words is given in tabular form in the Appendix at the end of 100 . a. which is only used in certain specific situations (see below). . this reduction process characteristically manifests itself in the reduction of unstressed function words. to. . the same word exhibits several different weak forms). In the example sentence above none of the function words (he. . unstressed pronunciation (called the weak form – very often. syncope in Chapter 5) so that camera to fits into more or less the same time span as. whereas camera tends to get compressed to disyllabic cam'ra (cf. his . and under such circumstances these function words may contain unreduced vowels (/. Due to the stress-timed rhythm of English. Let us see the details./.g. For illustration. and another. for. consider the following sentence: He | gave a | digital | camera to | George | Clooney for his | birthday The rhythmic beats are indicated by underscores. that is. the strings between any two such boundaries are much the same in duration.. digital. from which a number of consequences ensue.Chapter 7 unstressed syllables become.. the replacement of full vowels with the weak (reduced) vowels /  / (see Chapter 3). stressed pronunciation (called the strong form or full form). George is pronounced considerably long. . ()/) contains anything other than those three vowels. and the vertical lines denote foot boundaries. e. any word can in principle be stressed in an utterance for special emphasis. In connected speech.

when it is cited or quoted (e.g. although articles are generally weak even there.g. It's time to // go on). in which strong forms are often used in normally unstressed positions. sentence-initially as well (e. prepositions [12-17]. It has already been repeated several times that even otherwise unstressed words can become prominent for purposes of emphasis or contrast.. and. answer this question or It's at // the corner. not on // the corner). which may even lead to misunderstanding. as many as 95% of the occurrences of a function word in native speech are weak). when a preposition is followed by a 3 One exception is singing. In addition.. and some foreigners (including the typical Hungarian speaker of English) do this. moreover. the noun saint when part of compound proper names – [+1]) also show this kind of dual behaviour. I can do it if you want me to //). for example – accordingly.Connected speech the chapter. conjunctions [17-22] and auxiliaries [23-30]). although certain highly frequent major category words (e.g. or it is simply emphasized (You must // hold on! or He does // do the homework regularly!).. To // err is human). Don't say "but"! //). it is crucial for learners of English to be familiar with the use of weak forms or else they are likely to have difficulty comprehending (native) speakers who do use them (statistically. the strong form is chosen when the word is contrasted or co-ordinated with another one (e. whereas the strong forms occur at the end of the sentence (e. In addition..g.3 The unstressed.g. 101 . It is possible to use only strong forms in speaking... weak forms are normally used sentence-medially (e. but only Jack will //.g. Most of them belong to the closed class of function words (determiners and pronouns [1-11 in the table in the Appendix]. with the exception of auxiliaries [23-30]. the unnecessary lack of reduction creates the impression of emphasis. but native speakers find such "all-strong-form" pronunciations unnatural and foreign-sounding. Both of them can //.

those. Have you seen them? //). For the relative pronoun (e. 'd. e. however. need). the schwa is able to further reduce to zero. there exists an occasional weak form /h/. of course. Finally. might. may.g. I'm looking at you / /. 's. there are a few function words that have a strong form only. have a number of exceptions. they have the strong pronunciation at the beginning of the sentence (Can // you dance? as opposed to John can // dance the tango). in most cases to //. First. the <a> at the end of wanna (=want to). Some of these vowelless pronunciations (n't. 102 .g. usually. didn't) (for contraction. as it has been mentioned above. see below)..g. these. also of the pronoun) is used (e. pronouns (that. e. up)..Chapter 7 pronoun at the end of a sentence. I can't // (or cannot //) dance) and. gotta (=got to). The major characteristics of the pronunciation of the weak forms are the following: (1) The vowel reduces to one of the weak vowels.g. cuppa (=cup of).. combined with not) – the very nature of negation involves emphasis (e. usually the strong form of the preposition (and. kinda (=kind of). which sometimes results in Syllabic Consonant Formation (SCF – discussed in Chapter 5).. e. (2) Very often.) are able to undergo contraction – that is. on. These general rules. the man who sold the world). This is sometimes reflected in non-standard spelling. and the negative particle not (but it shortens to n't when contracted with certain auxiliaries..g..g. Who is it?). object pronouns [7-10] are not normally full even sentence-finally (e. Second. prepositions (in. who4). though not always. auxiliaries and the like attach to an 4 When it is interrogative (e. can't. It's at // the corner). when). auxiliaries (did. etc. conjunctions (though... won't. gonna (=going to).g.e. auxiliary verbs never have the weak pronunciation in their negative form (i. off. 've.g.. cf.

so we find a number of examples of optional consonant deletion. (3) The consonants surrounding the vowel also become weak. cf. though. among weak-form words. the. sometimes traceable in non-standard spellings like should of been for should have been. gotta.and vowel-initial words. which is evident from phonological processes such as word-initial tapping and aitchdropping. Also. to. Keep in mind. remember that in the non-rhotic accents like RP a word-final /r/ is only pronounced when followed by a vowel-initial morpheme – such potential Linking-R's are given in brackets in the table. in (1) above. that all those /r/'s are obligatory in all positions in the rhotic accents like GA. and note that contracted 's. (4) Certain weak-form words are pronounced differently before consonant. The initial /t/ of the unstressed preposition to is weak and frequently tapped in the relevant dialects in a phrase like lie to me // (analogously to a single word like lighter /()/). conforms to them as well. and the initial /h/ of the 103 . and delete easily. etc. wanna.Connected speech adjacent word. Contraction can also affect certain other words. typically lose their independent word status in connected speech. particularly those of prepositions and pronouns. (5) Weak forms. discussed above. it is possible to contract the object pronoun us in imperative let's. This is because in English schwa cannot normally occur before another vowel. irrespective of what function word it is a contraction of. do. Especially word-initial /h/ is targeted by such deletions. However. as it was already mentioned. including a(n). you. Rather exceptionally. the pronunciation is always with //. Recall from Chapter 6 the rules of the voice assimilation of the -s suffix. when /h/-initial weak-form words occur at the beginning of a sentence. so some other pronunciation (an extra /n/ in an. or final // or // in the other cases) is chosen to avoid that situation.

Moreover.e. We have also seen how other processes like assimilation and consonant deletion are present in connected speech. are given as footnotes. they all constitute single words. Linking-R and Intrusive-R can be proven to be two forms of virtually the same object. although sometimes GA supplies additional possibilities. // can correspond to either of or have). Most examples equally apply to RP and GA. lighter. contrasting them with the corresponding strong forms. as far as pronunciation is concerned. All such differences are indicated in the table. The table in the Appendix summarizes the most common weak forms. beat 'im).g. a hiatus-filler inserted after non-high vowels. tapping and other /t/-phenomena. A closer examination of the weak forms leads to the observation that some of them are ambiguous. beat 'im. including weak forms only used in certain meanings. All further comments. and as such.Chapter 7 unstressed personal pronoun him can be deleted in beat him (i. vehicle and shepherd undergo phonological rules in the same way because. and how the reduction of unstressed function words contributes to the isochronous stress pattern and rhythm of English utterances. We are led to the conclusion that all of lie to. they properly fit into the general picture of hiatus resolution. The following two chapters take the stress pattern of individual words under scrutiny.. so their meaning only becomes clear from the context (e.. which do not fit into a chart. 104 . R-dropping) apply within and across words in a uniform fashion. let us emphasize that a welldefinable group of phonological processes (L-darkening. To sum up the discussion of this chapter. similarly to vehicle or shepherd discussed in footnote 1 above.

This car is his). 5 at for from of  ()  (GA )  (GA ) What's he getting at? It's just what I long for. it is a (stressed) adverb and therefore it occurs in its strong form only. () --7 ()  () (GA also )  ().g. 7 In GA. There you are! Do you hate them? one of us is crying There's a book on the table. Is this YOUR CV? All I want is YOU. it always has the strong form... Where are you from? It's love I've a lot of.   () ()   () Examples It's not "a" cat.. The weak form // is rare and only appears in very casual or rapid speech.g. When his is a pronoun (e. 9  10 Examples the // dog. This is his car). sometimes spelt o' (as in standardized o'clock).Connected speech Appendix Word 1.    (). () () () () (). the occasional weak form // is only used before weak vowels. the // end a dog. when it is contrasted (e. Some students know this but others don't).. It's his car. 6 This only applies to the possessive determiner (e. //.. one of us 12. There's some milk in the fridge).g. stay for a week /  /. 9 In both RP and GA. an end I'll get you some apples. 10. . Also. e. 9.. 3. 11. their above. 6. an some5 his6 your = you're (s)he. .g. compare of and 105 . it's "the" cat! I'll get you some. e. 8. used before consonants only. () .g. 8 When this word is a demonstrative element (opposite of here). the a. we. 7. Also. it was THEM. 14. . He's from Barcelona. . Look at me. there is a weak form //. but not in other senses. 2.   (GA also ) () ()(). I gotcha! I love him. 15. Weak form(s) . 10 There is also an informal rapid-speech or non-standard pronunciation. you him her their them us there 8 Strong form  .g. Stay for a week. Whom do you love: him or her? It wasn't US. which is used in RP only occasionally. 5.  . not mine. 13. what's-his-name Mind your head! I'll get you some apples. This word can reduce when it is a neutral quantifier (e. cf. I love her. 4.

In GA. They were all drinking in the pub. 28.. 15 This is an occasional weak form in RP. Who is it?. however. 13 The weak form /@nd/ is slightly more formal than /@n/. 29. I'd rather sail away. 16 When expressing probability (e. Susan will be at home. You must try harder).Chapter 7 16. I know that you know it. You could be mine.. I must go now. 22. to // Ann even better than the real thing Twist and shout! sad but true the book that we bought sooner or later as good as it gets You've got to know.g. 17. when it is a demonstrative determiner (e. to11 than12 and but that14 or as have has had can could will would shall should must16 do does am.g. I think you should work harder. He's older than I thought). 25. [doorbell rings] This must be the milkman). e.g. 27. 26.. 21. How do you do? What does he do? I'm hungry. .g. 11 The preposition and the infinitival particle exhibit the same behaviour.  () ()()13   ()15  (). 24. 19. () .g. Will Susan be there? Would you like it? Shall I open the window? You MUST hold on! How do you do? Yes.. you could. or a degree word (e. 23. only used between numbers and in fixed phrases. You'd better stop! I can see. which only has a strong form //.g. She's got it. How are you?.  ().. 14 This word only has a weak form when used as a conjunction (e. were17      ()              . the book that we bought). () Who did you give it to? "Than" is spelt with an "a" not an "e". Don't say "but"! What's that? To be or not to be? as and when Have you seen her? Had I known him earlier. Not that bad) it is always pronounced in its strong form. What's that?). this reduction is quite common. and is pronounced //. "And" is a conjunction.! Can you dance? Yes. () () (). However. this word is less likely to appear in its weak form than when it is used in the sense of obligation (e. () (GA also ) to // me. she does! I AM hungry! He said he wasn't sleepy but he was! . 12 This word is either used as a preposition (e...g. 18.  (). off: the latter has no weak form. Who's that girl?) or pronoun (e.  ()  ()() ()() ()  () . 20.g.g. It's been a year. He's older than me) or a conjunction (e.. even when it is the only verb in the sentence.. its forms are always strong in three-word wh-questions containing a personal pronoun. 17 The verb be always behaves like an auxiliary verb.. Where were they? 106 . but it is not to be confused with the adverb then. are was. ()  (GA )..

been Saint  (GA )  Where have you been? He's a saint. +1.Connected speech 30. Saint Paul's Cathedral 107 .  ()() I've been busy all day.

for example. check whether you are familiar with the following terms: ambisyllabic.8.. tapping/flapping. use pitch to contrast not sentences but words. In Hungarian. derivation. vocal cords As it was already mentioned in Chapter 7. out of which pitch and loudness play the most significant role in the stress system of English. major stress. The distinctive use of patterns of pitch is called intonation. free variation. monomorphemic. but neither English nor Hungarian. Word stress – Part 1: The degrees of stress Before you study this chapter. similarly to punctuation in writing. statements). which is produced by their stretching and tensing: the tenser they are. stress is one of the suprasegmental (or prosodic) features of speech. which extend over more than one sound segment. function word. pulmonic. IPA. morphology. especially questions vs. Pitch roughly corresponds to the acoustic feature of frequency. thus pitch becomes an essential feature of . Jani elment? 'Has Johnny left?'. diacritic.g. reduced vowel. the rate of vibration of the vocal cords. CiV. productive suffix. the higher the rate of vibration. tempo and rhythm. syllable peak. Certain languages. and the different sentence types. clause boundaries within sentences. closed syllable. intonation plays a pivotal role in the distinction between segmentally identical statements and yes-no questions like Jani elment 'Johnny has left' vs. full vowel. loudness. whose most important function is to signal grammatical structure (e. Trisyllabic Laxness. They include variations in pitch. utterance. Chapter 10 is devoted to intonation in English. and the higher the pitch. foot. aspirated. morpheme. syllabic consonant.

The loudness of (strings of) speech sounds depends on the size of the vibrations of the vocal cords caused by the varying degrees of pulmonic air pressure. namely the second one. The primary source of this rhythmic prominence of major stress is the loudness of the syllable. Many of the languages of South-East Asia and sub-Saharan Africa. In suprasegmental or syllabification. Together with pitch level and vowel quality. loudness is the other major ingredient of stress prominence in English. for example. e. belong here. these various degrees come into being owing to the unequal role played by pitch. This phenomenon is called tone.g. In fact. Due to the fact that stress is an extremely complex phenomenon (governed by a number of different factors) and the fact that it is relative. is more prominent owing to its highest pitch in the word. It is possible to say that a syllable is stressed.g. but the difference in pitch level causes a difference between two types of major stresses. there exist several degrees of stress. Besides pitch. In addition. rhythmic prominence (already mentioned in Chapter 7). Beijing Mandarin Chinese and Thai in Asia or Hausa in Africa.Word stress – Part 1 the meaning of morphemes. loudness produces the relative prominence of syllables called stress. but one of them. there are two rhythmic beats (underlined). Are 109 . Recall from the previous chapter that all non-function words in English contain at least one syllable that constitutes a rhythmic beat (called major stress) – at the same time. only this syllable can carry the main stress of an utterance. but this always means that it is more stressed (=stronger. and the full or reduced quality of the syllable peak. more prominent) than the adjacent syllable(s). and such languages are called tone languages. only relevant in comparison of several syllables. It is of crucial importance to understand that stress is not an absolute feature of syllables but rather it is relative. function words are normally unstressed.. e. out of which four are linguistically relevant in English..

whereas decontamination already contains two. with all the consequences of this. 1 Based on the observation that out of two (or more) successive secondary-stressed syllables the first one is always slightly stronger than the other(s). basically it only appears in longer English words under very specific circumstances (see below in more detail). there may be several occurrences of secondary stress. on the one hand. some authors apply the term "secondary stress" to that one only and refer to the others as "tertiary-stressed". since it is. and even if a consonant remains pronounced in such a position. its syllabic status is vague. while the other type of major stress is usually referred to as secondary stress. which means that. by definition. the discussion of phrasal stress in Chapter 7). the consonants surrounding this weak peak also become unstable. Buddha //. are analogous). the most prominent syllable). For example. annihilate //. for obvious reasons. Secondary stress is optional. For instance. the whole syllable becomes weak and reduced. depending on the length of the word. for example. the vowel is not full but one of /  / – most frequently. especially the consonant preceding it. Birmingham //. the first syllable of the word suprasegmental and the second syllable of syllabification are secondary stressed. the word contamination contains one such syllable (underlined). So much so that /h/. Chapter 2). 110 . In most such cases. schwa. On the other hand. It is traditionally called primary stress or main stress.1 Syllables without rhythmic prominence also fall into two subtypes. the consonant is ambisyllabic (cf.Chapter 8 these features suprasegmental? or This is the correct syllabification (cf. but vehement //. Another basic difference between primary and secondary stress is that while the former can only appear once in a word (this is logical. It is in these cases that Syllabic Consonant Formation (discussed in Chapter 5) is possible. etc. systematically disappears altogether (recall the examples vehicle // and shepherd // of Chapter 7. that is.

which is not – rather. under certain (not exactly straightforward) circumstances the expected vowel reduction fails to take place. rhythmically or pitch-wise non-prominent stress is called tertiary stress in this book. The four degrees of word stress are summarized in the following chart. which is strong and therefore aspirated. still they are stronger than completely unstressed syllables in the sense that they are characterized by neither vowel reduction nor consonant weakening. 111 . absence of vowel reduction. Although such syllables are not prominent as far as suprasegmental features go. with that of activity or better. respectively. while the major stress – minor stress distinction is based on loudness (rhythmic prominence). Compare the final syllable of Abraham // and Graham // – in the former the vowel is full and the /h/ is pronounced (this is what we call tertiary stress). Compare the underlined /t/ in hesitate. yielding [] and [()]. the basic difference between unstressed and stressed syllables lies in the presence vs. An alternative name is minor stress (as opposed to major stress). that is. as in the first syllable of activity //. whereas in the latter the vowel is a schwa and the /h/ is dropped (this is what we call zero stress). As the shaded areas show. the two elementary features of zero stress mentioned above. This fullvowelled. it is tapped in the tapping dialects of English (as an indication of its ambisyllabicity). some otherwise weak syllables contain an unreduced vowel.Word stress – Part 1 Such syllables are zero-stressed or completely unstressed. However.

. and IPA stress marks.. diacritics.Chapter 8 Stress category Stress degree Prominence Examples MAJOR primary full vowel loudness highest pitch suprasegmental syllabification annihilate hesitate Japan secondary full vowel loudness suprasegmental syllabification hesitation grammaticality Japanese MINOR tertiary full vowel suprasegmental syllabify annihilate hesitate activity UNSTRESSED zero -suprasegmental syllabification annihilate grammaticality Japan There are three equivalent stress-marking conventions in phonology: the use of numbers. 2 for secondary. and 0 for zero. Stress MAJOR secondary 2 grave accent lower mark MINOR tertiary 3 UNSTRESSED zero 0 - category Stress degree primary Numbers 1 Diacritics acute accent IPA stress upper mark marks 112 .g. In the IPA. ò) secondary stress.g. 3 for tertiary. 1 standing for primary. and the lower mark /  / for secondary stress. Finally. ó) signals primary stress. In this book. Sometimes the segments are not transcribed but rather the spelt form of the word is supplemented by diacritics on top of the stressed vowel letters: the acute accent (e. and the grave accent (e. the stress degrees of the syllables in a word can be referred to with numbers. the upper mark /  / is used for primary stress. we only use their most widely accepted forms. which are shown in the table below.

tend to be avoided: in most cases (as in the examples above). there remain three more syllables to the left.e. It follows that primary stress is always the rightmost major stress. Second. or. as //.e. called stress clashes. Such situations. speech rhythm with metrical feet consisting of one unstressed syllable followed by one stressed syllable) also manifests itself in the dispreference of adjacent major stresses. as sometimes it is referred to.Word stress – Part 1 Accordingly. First. Notice that the prominence of the right edge and the prominence of the left edge are in potential conflict in longer words: in a five-syllable word. or. the Early Stress Requirement. for instance. This prominence of the right edge is usually explained by the directionality of primary stress placement: it is supposed to proceed from right to left. there are no English words starting with two successive zeroor tertiary stressed syllables – one of the first two syllables of a word must be rhythmically prominent (i.. the last rhythmic beat is the strongest. the careful reader must have already noticed some of the general properties of English word stress.. docking onto the first potential site available (see the next chapter). with accents as sùprasegméntal. accompanying an IPA transcription. Consequently. e.. primary or secondary stressed). either the first one (as in sùprasegméntal) or the second one (as in contàminátion) will necessarily receive secondary stress.g. This is the prominence of the left edge. the stress pattern of suprasegmental can be indicated with numbers as 20310. which cannot all be unstressed.. 113 . no major stress occurs after the primary stressed syllable (i. secondary stress always precedes primary stress). i.e. This tendency in English for a regular iambic rhythm (that is. 20310 in suprasegmental or 02010 in contamination. this is the reason why secondary stresses are created: to produce a more or less regular alternation of stressed and unstressed syllables. where primary stress falls on the second-last syllable. On the basis of the examples above. In fact.

perípheral – perìpherálity. discussed above. The problem. mainly arises in derived words. sànatórium. órchid – òrchidáceous. àlumínium. The following examples illustrate the same mechanism: adáptable – adàptabílity. as a result of which in longer words. pàraphernália. this secondary stress is frequently referred to as Derivational Secondary Stress. jústify – jùstificátion. primary stress falls on the first syllable. if primary stress falls on the third (or a later) syllable. màcaróni. however. Here again the number of syllables before the primary stress does not matter.Chapter 8 there is at least one zero or tertiary stressed syllable between any two primary or secondary stresses. What usually happens in such cases is that since suffixation has made the word longer.. àgoraphóbia. Àbergavénny. primary stress shifts to the right. Bear in mind that the rightmost rhythmic beat is the strongest! As such derived words preserve the rhythmic prominence of the original stress pattern. for example. and the original primary stress reduces to secondary. additional secondary stresses are created. in most such words (e. àbracadábra. Àpalàchicóla. It has been mentioned that the "war" of the two word edges is the primary motivation for the creation of secondary stresses: recall the Early Stress Requirement. Wìnnepesáukee) the first syllable receives secondary stress irrespective of the number of syllables before the primary stress – if there are more than three.g. so there are just a few examples (including a number of place names) for underived secondary stress. the first or the second syllable must be assigned secondary stress. Monomorphemic English words tend to be rather short. which reduces to secondary stress when fìctionéer is derived. impréssion – imprèssionístic. irrégular – irrègulárity. In fíction. The basic principles of the English stress system. Primary stress is dealt with in the next chapter – here we turn to secondary stress assignment. Kàlamazóo. éducate – èducátion. If a 114 . as in Apalachicola. determine the regularities of stress placement.

There is one situation.Word stress – Part 1 suffix is attached to a long word which already contains a secondary stress. in which Derivational Secondary Stress is blocked: when it would result in stress clash. the output of both strategies is a stress pattern with the stresses evenly distributed. however. còmprehénsible and còmprehènsibílity. Whole chains of derivation illuminate how former primary stresses turn into secondary: dífferent – dìfferéntiate – dìfferèntiátion. cf. and their vowels are indicated with IPA symbols: Iambic Secondary Stress adápt – àdaptátion // doméstic – dòmestícity // enígma – ènigmátic // fragmént (verb) – fràgmentátion // horízon – hòrizóntal // Japán – Jàpanése // refórm – rèformátion // transpórt – trànsportátion // Major Stress Deletion áctive – actívity // ànthropólogy – ànthropológical // átom – atómic // clímate – climátic // Gérman – Germánic // frágile – fragílity // víctory – victórious // vírgin – Virgínia // 115 . Here are a couple of examples – the previously primary stressed syllables are underlined. All in all. the original major stress being deleted and lost (Major Stress Deletion). or with no (new) secondary stress at all. either with a secondary stress appearing to the left (this is called Iambic Secondary Stress). ínstitute – ìnstitútion – ìnstitùtionalizátion. ìndivídual and ìndivìduálity. The pressure to avoid adjacent major stresses and therefore maintain a (near-)iambic rhythm leads to one of two possible solutions: the original primary stress reduces to tertiary or zero. further secondary stresses are brought about.

Chìnese chéckers. e. that it does not only take place in thirteen men. ìnternational láw. àcademic wríting.e. thìrtéen) forms a phrase with another one (e. àfternoon téa. it should be clear that after certain suffixes have been attached to a word.. all of which highlight one or another feature of the process: stress degrees are shifted to move rhythmic beats away from each other and thus facilitate the iambic rhythm of the phrase. àchromatic léns.g. i.. the result being thìrteen mén. with a 321 stress pattern. Another manifestation of the tendency to maintain iambic rhythm and avoid stress clash characterizes connected speech. When a word with a secondary and a primary stress (e. This phenomenon has been widely studied and therefore has a whole range of names. while all the others are expected to reduce their stress degree by one. Some linguists dub it the thirteen men rule. after this very frequent example..Chapter 8 Since the primary stress is placed in different ways in the case of different suffixes (see below and in the next chapter). the same word can undergo Derivational Secondary Stress formation in one case and Iambic Secondary Stress formation or Major Stress Deletion in another. the original major stress is preserved in còrrelátion but deleted in corrélative. the original stress pattern may 116 ... It is important to keep in mind. where the intervening tertiary stress (formerly the primary stress – underlined) separates the major stresses. or the rhythm rule. mén). From the discussion of secondary stress.g.g. something like thirtèen mén. for some speakers. stress shift. e. Instead. Jàpanese lánguage. what normally happens in English is that the stress levels "switch round" in the first element of the phrase. based on the discussion of phrasal stress (Chapter 7) we expect the final stress to be the strongest. iambic reversal. but occurs automatically in all phrases where the first element has at least one secondary stress. that is.. nèolithic víllage. though. Tènnessee Válley.g. 231. the Bèrlin Wáll or ìdeal pártners. e. or.g. for córrelate. fùndamental fréquency.

g. and -al. the list of these suffixes coincides with the type referred to in Chapter 3 as non-productive. which do not count in. and -like above. systematically change the place and/or the degree of the stresses because they require primary stress to fall on a specific syllable – they are non-neutral or stress-fixing. the first syllable is primary stressed. In educate. Certain suffixes.g. Curiously enough. the original stress pattern may or may not be preserved. e.Word stress – Part 1 change. -ing. Consider the following examples: éducate – éducating – èducátion adápt – adápted – àdaptátion díagnose – díagnoses – dìagnóstic jóurnal – jóurnalist – jòurnalése áutumn – áutumn-like – autúmnal As you can see.. -ic. -ese. this is the primary source for the creation of secondary stress. productive suffixes. whereas in the -ion form it reduces to secondary and a different syllable receives the primary stress. Most of them are of Germanic origin. e. Others. as in átom – atómic. Therefore. Most of them are of Latin origin (they are Latinate). when a new word is formed out of a base word. as in décorate – dècorátion. Trisyllabic Laxness (recall examples 117 . Notice that at this point we are able to make a generalization: regular. are unable to affect the stress pattern of the word they are part of – they are stress-neutral. the list of these suffixes coincides with the type referred to in Chapter 3 as productive.. -ist. it is necessary to distinguish between two types of morphological operation. Nevertheless. we are forced to break down the family of suffixes into two classes. -ed. Obviously. moreover. Curiously enough. -s. morphology plays a crucial part in the English stress system. like -ion. and so is it in the -ing form.

búsiness – búsinesslike cáreless – cárelessly. accómpany – accómpaniment cáreless – cárelessness. They fall into various subclasses. are stress-neutral.dictátorship advénture – advénturesome.e. avóid – unavóidable mártyr – mártyrdom. ímage – ímages cénsor – cénsorship.. lùkewárm – lùkewármness díagnose – díagnoses. either. fanátic – fanáticism jóurnal – jóurnalist. defénce – defénceless áutumn – áutumn-like. phýsics – phýsicist bóttom – bóttomless. do not count in stress placement. tòpsy-túrvy – tòpsy-túrvydom adápt – adápted. ínterest – ínteresting ánimal – ánimalish. i. This is a curious interplay between word structure and sound pattern: suffixes seem to exhibit consistent behaviour in phonology. dictátor . ádult – ádulthood éducate – éducating. Non-productive suffixes.Chapter 8 like lazy-laziness). édit – édited cómmon – cómmoner. (1) Stress-neutral suffixes Suffix -able -dom -ed -er2 -ful -hood -ing -ish (adj) -ism -ist -less -like -ly -ment -ness -s -ship -some -wise Examples consíder – consíderable. quárrel – quárrelsome óther – ótherwise. 118 . córner – córnerwise (2) Stress-fixing suffixes and endings Some of these are not clearly isolatable suffixes (perhaps not even morphemes) but rather simple word endings which are present in recurrent stress patterns. on the other hand. do count in both Trisyllabic Laxness (recall vain-vanity) and stress assignment – they are stress-fixing. béauty – béautiful bróther – brótherhood. The most common productive (stress-neutral) and non-productive (stress-fixing) suffixes are illustrated below. advénture – advénturer bárrow – bárrowful. fórty – fórtyish álcohol – álcoholism. two of which are introduced 2 This suffix either forms a comparative adjective (like commoner) or an agentive noun (like adventurer) – in both cases it behaves in the same fashion. appárent – appárently devélop – devélopment.

some are disyllabic (e. cartóon Second. critíque.g. in dìagnóstic it is the third one – right before -ic itself. -ic. First. Cayénne àcquiésce. techníque. and as such. -itude. auto-stressed (or selfstressed) suffixes and endings are primary stressed themselves. ravíne antíque. -uant). bàrricáde. appéndix . pícture – pìcturésque largésse. noblésse cigár – cìgarétte. -ular. -ible. -ify/efy. tráin – trainée vélvet – vèlvetéen. Japán – Jàpanése Róman – Ròmanésque.. is listed in the first chart above. -ious/uous. Some of these suffixes and endings are monosyllabic (e. -ity/ety..g. Suffix/ending -ade -aire -ee -een -eer/ier -elle -enne -esce -ese -esque -esse -ette -eur/euse -ine -ique -itis -oo/oon Examples lémon – lèmonáde. -ish (v/n)3). cassétte èntreprenéur. èffervésce jóurnal – jòurnalése. -itive. bombárd – bòmbardíer mozélle. uníque lárynx – làryngítis.appèndicítis kàngaróo.g. -id. -ulous). 3 Note that adjectival -ish is a stress-neutral suffix. or the similar CuV sequence (e. 119 . -ics. -ic is a typical (and very frequent) pre-stressed suffix: while in díagnose the first syllable is primary stressed. cánnon – cànnonáde míllion – mìllionáire. some contain the CiV configuration introduced in Chapter 3. séven – sèventéen éngine – ènginéer.Word stress – Part 1 presently (and a third one in the next chapter). -ial/ual. masséuse cuisíne.. The same happens in acádemy – àcadémic and átom – atómic. For example. quéstion – quèstionnáire réfuge – rèfugée. nacélle comédian – comèdiénne. the so-called pre-stressed suffixes and endings require primary stress to fall on the syllable which immediately precedes them in the word. -ion. -ian.

either a long monophthong (e. although a number of tendencies are observable. socíety. exémplify opínion. thermómeter contínuant mólecule – molécular.. let us mention tertiary stress briefly. demólish. líbrary – librárian. actívity) or the vowel is long. ácrobat – àcrobátics intrépid.. ìntuítion – intúitive exáctitude. insípid. estáblish compétitive. sólve – solútion. incrédible dynámic. climátic). infínitive.. rídicule – ridículous After the story of secondary stress and the effect morphological structure has on stress placement. Why certain otherwise unstressed vowels fail to reduce to //. dimínish. Cáesar – Caesárean. perípheral – perìpherálity. éducate – èducátion. crustácean deléte – indélible. símilar – simílitude compléxity. the syllable whose vowel refuses to reduce is very often a closed syllable (cf. cóntext – contéxtual Húngary – Hungárian. Recall that tertiary stress is in fact the prominence caused by the absence of vowel reduction. órchid – òrchidáceous. ecónomy – èconómic ínsect – insécticide ecónomy – èconómics. metículous. innócuous abólish.Chapter 8 Suffix/ending -ial/ual -ian/ean -ible -ic -icide -ics -id -ify/efy -ion -ious/ -eous/uous -ish (v/n) -itive -itude -ity/ety -meter -uant -ular -ulous Examples tútor – tutórial. ánxious – anxíety spéed – speedómeter.g.g. mándible – mandíbular míracle – miráculous. For example. adápt – àdaptátion céremony – cèremónious. 120 . Germánic) or a diphthong (e. // or // is difficult – if not impossible – to explain: it appears to be quite irregular and mostly unpredictable. Unfortunately. Chapter 5) (e. pellúcid ácid – acídify.g.

fragmént – fràgmentátion // versus condémn – còndemnátion //). two separate words with their obligatory primary stress (neither of them is a function word!). For example. which is dealt with in Chapter 9. the musicians who play the trombone. this is accompanied by reduction in the other term(s). One is the so-called Alternating Stress Rule. For instance. namely. however. There are only a few cases where tertiary stress appears systematically. board ceases to be major stressed but its long vowel // survives in bláckboard //. and which accounts for the 103 stress pattern of verbs like dédicate and certain adjectives and nouns like ábsolute or húrricane. Therefore. infórm – ìnformátion //. Logically. only means the piece of furniture if pronounced with considerable 121 . The word cupboard. its stress pattern is 13. the more likely vowel reduction is. for instance. Another situation when tertiary stress is expected is compound stress.Word stress – Part 1 this does not mean that all such vowels are protected from reduction (cf. Word frequency may also influence this: the more frequently a speaker uses a word. they lose their original rhythmic prominence but retain their full vowel. tend to have a schwa in the first syllable (//). This is in sharp contrast with what we usually observe in underived words like bládder // or blánket //. and súperman //. Interestingly. a number of historical compounds have by now given up their complex morphological structure and are pronounced according to the regularities of simple words. the musical instrument trombone is usually pronounced //. of course) like blácking // or blábber // – all exhibiting 10. Chapter 7 explains that primary stress in a compound word normally falls on the first term. or in words containing suffixes (other than auto-stressed ones. when bláck and bóard. are combined. The same applies to ráinbow //. lífestyle //.

.g. chóleric. GA // or laboratory RP // vs. In addition. on primary stress. Árabic. or words ending in -ary and -ory like January RP // vs. cátholic. e. 122 . Forehead has two alternative pronunciations: one which follows the rules for compounds //. similarly to cumber. most of the stress rules introduced above have exceptions. will face even more exceptions and subregularities.. Let us conclude this chapter with a remark concerning the fact that. too. ársenic. Stress clash does occur. e. and another with a reduced second term //.a // (with a 13 stress pattern) is simply a board with cups. black // plus berry // is blackberry /()/. impóverish. the picture is further complicated by the free variation of zero and tertiary stress in words like direct //~ // as well as occasional dialectal differences between RP and GA. Exceptions also exist to the stress-fixing mechanism of suffixes. Original sheep // plus herd // has become shepherd //. advertisement RP // vs. cupper.g.Chapter 8 vowel (and consonant) reduction // (its stress pattern is 10. In sum. The next chapter. post // plus man // is postman //. thìrtéen or Chìnése. GA //. lúnatic. as in eléctric – elèctrícity. although only in a handful of words like sàrdíne. GA //. GA // (or //). unfortunately. cupping or cupful) . address (n) RP // vs. pólitics. the morphological structure of a word is clearly reflected in its pronunciation: only constructs with a primary stress and a tertiary stress are real compounds. Derivational Secondary Stress can override the desired iambic rhythm.

primary stress can fall on virtually any of the syllables in English. Adverbs usually follow the rules for adjectives. nouns. English primary stress prefers the right edge of the word in the sense that in unsuffixed forms the strongest stress is not normally placed more than three syllables away from the end. rhyme. extréme – extrémely). tertiary stress. and adjectives. the syntactic class of the word also plays a role in English: function words behave .. where the first. Chapter 8 elaborated on the role of morphology in English – something unheard of in Hungarian with respect to stress placement. and most of them are formed with a derivational suffix (e. self-stressed. clear-L. Another basic difference between the two languages lies in the information types determining stress placement. suffix. pre-stressed. so they are not treated separately. i. compound. On the other hand. dark-L.. On the one hand. Let us start with the differences between the English and the Hungarian stress system. This chapter deals with primary stress assignment in underived verbs. check whether you are familiar with the following terms: allophonic rules.g. leftmost. some of which are also mentioned in Chapters 7 and 8 above. Word stress – Part 2: Primary stress Before you study this chapter. non-neutral suffix.9. peak. primary stress. which was discussed in Chapter 7. syllable is stressed in all words. tapping. Iambic Secondary Stress. according to the prominence of the right edge. The remaining word classes belong to the category of function words. What is more.e. Early Stress Requirement. prominence of the right edge. prefix. function word. As opposed to Hungarian. neutral suffix. auto-stressed ending. secondary stress. stem. irrespective of the length of the word.

we can state that although this phenomenon is traditionally referred to as syllable weight.Chapter 9 differently from non-function words. whereas the second syllable in various or horizontal is // and consequently light. In addition. Similarly. Therefore. the consonant(s) preceding the peak do(es) not count: a syllable like // is light in the same way as // or // or //. A piece of warning is in order here: do not let spelling mislead you! The second syllable in variety or horizon is // and thus heavy. each regularity has a considerable number of exceptions – the stress pattern of these words has to be simply memorized since it is unpredictable: it is an idiosyncratic feature of the lexical item. 124 . More specifically. as has been suggested before. the name rhyme weight is more appropriate. //. nouns and verbs are shown below to have established two distinct patterns. in English not only is its position relevant but so is its structure. while // is heavy even though altogether it consists of fewer elements than //. A syllable is light if it contains a short vowel and is not closed by a consonant. a syllable like // or // or // is characterized by one type of behaviour (it is heavy) as opposed to. all the other possibilities (with a long vowel – a long monophthong or a diphthong – and/or with (a) final consonant(s)) make the syllable heavy. it is only governed by the number of elements in and following the peak – the portion of the syllable that we call the rhyme (cf. Crucially. While in Hungarian the only piece of phonological information required for stress placement is the position of the syllable (the first syllable in the word gets stressed automatically). say. Even phonology makes a much more complex contribution to stress rules. what is more. the English stress system is quantity-sensitive: it is strongly influenced by the amount of material found in syllables. there are two basic syllable types: light and heavy syllables. Chapter 5) – that is. in fact. In this respect.

Word stress – Part 2 Rhyme weight is relevant to stress placement in that in English, heavy syllables attract stress. E.g., the readers are invited to check for themselves that the words a.ro.ma, e.nig.ma, al.ge.bra, in.du.stry all consist of three syllables (divided by dots), out of which only one is heavy (underlined), and that one is primary-stressed. However, word-final consonants normally do not count: words like a.ban.don, GA ten.der, pa.ren.tal, can.cel, de.ter.mine contain two heavy syllables, but stress always falls on the non-final one – the word-final consonant is unable to make its syllable heavy. Such "invisible" segments are usually referred to as extrametrical, being outside the scope of meter, i.e., rhythm; in the rest of the chapter, we indicate extrametrical material by putting it in parentheses, e.g., abando(n), tende(r), determi(ne). Notice that the -e at the end of determine is not a sound but a silent letter only: it does not really matter whether it is or is not included in the parentheses. The status of final consonants is of great significance since, as it was mentioned in Chapter 8, the directionality of primary stress placement is right to left. That is, stress rules start scanning the syllables with the last one and proceed towards the beginning of the word, in such a way that primary stress is assigned to the first heavy syllable available, but not later than the second syllable from the end. If word-final consonants were visible to these stress rules, all words ending in at least one consonant sound would end in a heavy syllable and be primary stressed there – but this is not what we find: abandon and the like are stressed on the second-last syllable. Now the time has come to formulate the Main Stress Rule (MSR) for verbs, nouns, and adjectives.1 Let us start with verbs: in this case, the final consonant (if there is one) is extrametrical. If the remaining syllable is heavy,
1

Bear in mind that primary stress assignment takes place within monomorphemic words – more complex word forms, including compounds, are stressed according to the effect of the two affix classes and the compound stress rule familiar from previous chapters.

125

Chapter 9 it is stressed; if it is light, the preceding syllable is stressed. In a verb like torment, for example, there are two syllables: tor.ment. The final /t/ is extrametrical: tor.men(t). Thus we are left with men as the final syllable, which is heavy (it contains a short vowel plus a consonant), so it receives the primary stress: tormént. That is why final stress is usual in verbs ending with at least two consonants: eléct, seléct, arrést, adópt, lamént, etc. When a verb ends in a single consonant, the length of the preceding vowel decides: if it is long, as in unite /-()/, erase /-()/, achieve /-()/, the final syllable is still heavy, and therefore stressed; otherwise the second-last syllable is stressed, e.g., trável, fínish, ínjure, astónish, vómit, consíder. Logically, if the verb ends in a vowel, there is nothing to be extrametrical, and the length of that vowel automatically determines the place of stress: in cárry, the final vowel is short and the second-last syllable carries the stress, but the last one does so in applý, the vowel in question being long //. Notice that the second-last syllable is never checked for rhyme weight: it does not need to be because primary stress cannot move further to the left anyway (except for the Alternating Stress Rule, to be discussed below) – recall the prominence of the right edge. Therefore the same syllable is stressed in a.stó.ni(sh) and con.sí.de(r) although there is a heavy syllable (underlined) in the latter.2 There is one case when, predictably, the regularity described above does not apply, due to a morphological effect not yet mentioned: we need to take a detour here and discuss the role of verbal prefixes in stress assignment. Similarly to suffixes, these prefixes can be divided into two basic subtypes: neutral and non-neutral prefixes. Neutral prefixes never influence the place of stress in the stem – instead, they are (secondary or tertiary) stressed themselves. In fact, they attach to the stem so loosely that even the
2

It goes without saying that, as the example words above are not longer than three syllables, primary stress placement satisfies the prominence of the left edge, too.

126

Word stress – Part 2 frequently occurring stress clash that they cause does not disturb their status, cf. dèbúg, ùnplúg, upsét, òut-Hérod, rèwínd, etc. In contrast, certain monosyllabic verbal prefixes are non-neutral: they resist primary stress and consequently, they systematically overwrite the results of the above mechanism. For instance, consider two verbs, vomit and omit. Segmentally, they minimally differ: the former contains an extra consonant at the beginning. In fact, since that consonant is a syllable-initial one it is not even expected to affect stress placement – recall that only syllable-final consonants are able to contribute to weight. Nevertheless, vómit has initial primary stress, which conforms to the MSR as introduced above, as opposed to omít, which has, rather surprisingly, final stress. The source of the difference in stressing cannot be phonological in nature since there is no relevant pronunciation difference between the two words on the segmental level. It follows, then, that morphology is to blame: while vomit consists of a single morpheme, omit can be analysed into a prefix, o- (also appearing in oppose, oppress, occur) and a stem, -mit (also appearing in submit, remit, admit, permit, commit), and in all such examples consistently it is the stem which carries primary stress rather than the prefix. Therefore we conclude that stems enjoy a priviledge over peripheral elements like prefixes: even if a monosyllabic stem is light (e.g., mi(t)), it is assigned primary stress since the monosyllabic prefix "throws back" the stress to it. Interestingly, very often this stressresistance of a prefix does not have the chance to manifest itself because the stem is heavy, e.g., províde, retúrn, but the effect is visible in a host of other examples such as omít, expél, commít, attách, etc. Sometimes a disyllabic prefix or two monosyllabic prefixes are attached to a monosyllabic stem. It should be self-evident that in such words primary stress falls on the stem, and the first syllable must be secondary stressed in accordance with the Early Stress Requirement, cf. ìnter-véne,

127

Chapter 9 còntra-díct, rè-pre-sént, cò-rre-spónd, etc. It is crucial that in these examples the stem is monosyllabic: when the final syllable does not constitute the stem alone, as in éx-ecute, ré-cognize, intér-rogate, the prefix fuses with the stem totally and the so-called Alternating Stress Rule, to be introduced below, comes into effect. Verbs, then, tend to be end-stressed when the last syllable is heavy (even without the final consonant), otherwise primary stress normally falls on the second-last syllable, except when it is a verbal prefix. As usual, the algorithm provided above ("the word-final consonant is extrametrical, the remaining last syllable is stressed if heavy, else the preceding syllable is stressed") suffers from its exceptions: sometimes a light final syllable is stressed, as in caréss, posséss, GA haráss (mind you, the final double <ss> stands for a short /s/, which is extrametrical!); at other times a long-vowelled final syllable fails to be assigned primary stress, e.g., fóllow, hárrow, swállow, hállow. Now let us pay some attention to nouns. The nominal subclause of the MSR differs from the verbal one in the portion of the word which is usually extrametrical, that is, non-stressable. Namely, in nouns it is the whole final syllable that does not take part in the stress placement procedure, at least when it contains a short vowel. That is why disyllabic nouns are normally stressed on the second-last (=first) syllable (e.g., táble, páttern, chílli, Lóndon, trúmpet, férry, GA míssile //, etc.), except for just a handful of words like evént, hotél, Japán, succéss, Berlín, etc. This is easily accounted for with reference to the extrametricality of the last syllable. In all other respects the MSR for nouns is the same as the MSR for verbs: do not consider the final syllable – if the remaining rightmost syllable is heavy, it is stressed; if it is light, the preceding syllable is stressed. That is why in longer

128

129 .(ma). etc. cu.(sis).(lum).(da) it (underlined) is heavy and therefore stressed. á. GA detáil.len. a heavy second-last syllable is skipped by the MSR in chá. Tennessée. enginéer.bré. cá.(sus).rán.ssor.(ma) //. and a handful of other words. voluntéer. hi.tti. trombóne – cf. millionáire. (Keep in mind that consonant doubling in spelling does not indicate length in pronunciation!) Second.rí.ri.rac.tén.na. RP míssile //. spa. baróque. helló.nóp. u. However. If the last syllable of a noun contains a long vowel. nominée. cí. á.(risk). cigár.(sis) //.lla.nna.Word stress – Part 2 nouns the weight of the second-last syllable decides the fate of primary stress: in a. machíne.(lin). ménu.é. pi. in some nouns primary stress exceptionally moves further away from the right edge: the fourth-last syllable is stressed in. nouns with long-vowelled final syllables do not present the only complication – perhaps all the possible exceptional configurations exist. First. or the compound stress rule (RP bállet. cémetery. céremony..ghé.ssen. uníte.(na) //. it is very often an auto-stressed ending and receives primary stress accordingly (see Chapter 8). vénue – cf.g. RP détail.tive.(ca).ró. RP labóratory. con.sén.ter. hy.ré. pá. regíme.tró. me.ger. in A. e. kangaróo.á.lla. hélicopter.fé. télevision. ho.rrí.ná.(tus) //. pro.the.(lis) it is light and consequently the syllable to the left carries primary stress. Julý.no.ly. féllow. ágriculture. a light second-last syllable is primary stressed in va. sy. cátegory. políce.á.ste. a. brocáde.ve.g.(sil). já. GA ballét. questionnáire. Di.. Finally.(sis) //. applý). ve. bláckboard.ní.djec.á.. a. Otherwise such nouns fall into one of two categories: they follow either the verbal subclause of the MSR (e.mé.dar. um.po.ne.g.(zon) //. and the fifth-last in véterinary.pó. e.cu. ráinbow). Vi.

có. cómment.(ous)). derived adjectives (ending in. -id. either the noun copying the verbal stress pattern (as in attáck. while digest the noun must have initial stress: dí. a. ex.(tal). etc.mí. -ent) behave like nouns and have an extrametrical final syllable: fa. -al.rén. fá. etc. These suffixes make the adjectives behave as nouns as far as stress rules go. pa. it logically follows that primary stress will fall one syllable closer to the left edge in nouns. cóntrast (n) – contrást (v).no. Nevertheless. e. pí.(mous). cér.frái(d). -ar. áccent (n) – accént (v).ni(d). cómfort. On the other hand.rách.(rant). -ant.ri. cóntact. the stress resistance of verbal prefixes is only applicable to verbs but not to nouns. -it usually behave like verbs: a.tai(n).). vá. On the one hand. surpríse. gé. GA detáil. that is how word pairs like rébel (n) – rebél (v) or pérmit (n) – permít (v) emerge. óbject (n) – objéct (v).g. ségment (n) – segmént (v). íg.gés(t).(ar). tránsport (n) – transpórt (v). -ous.Chapter 9 The difference between nouns and verbs becomes clearly noticeable in segmentally (nearly) identical noun-verb pairs. The third major word category. Naturally. ímport (n) – impórt (v). Thus digest the verb has primary stress on the (heavy) final syllable: di.(tal). désert (n) – desért (v).) or the verb copying that of the corresponding noun (as in áccess. The usual state of affairs in such cases is the following: since in verbs only the final consonant is extrametrical.ne. adjectives have not developed a third form of extrametricality but are divided between the nominal and the verbal patterns.(gest). súrvey (n) – survéy (v). while in nouns it is the whole final syllable.(rous). debáte. RP détail.mmo(n). underived adjectives and adjectives ending in -ic. 130 . inasmuch as it is nouns whose last syllable is not normally stressable.ci(t).vo.li. reséarch.. ínterest. ínterview (a compound noun exemplifying 103) etc. The same applies to récord (n) – recórd (v). úpdate (n) – updáte (v). in a few cases the noun and the verb in such pairs have the same stress pattern.plí.

Consequently. we can state that.. we have identified a third class of non-neutral suffixes and endings. they are monosyllabic with a short vowel and an (extrametrical) single consonant: they inevitably make a light final syllable. trigger the verbal subclause. and the verb has more than two syllables. whose members fix the position of the main stress by simply launching the application of the MSR on the adjective. Some authors call them integrated suffixes. After all. inasmuch as it is verbs whose final consonant is not normally visible in stress assignment.tó. You may have noticed that the three endings belonging here have the same effect as if they were pre-stressed – in fact. su. In sum.tré(me). and the stress of the final 131 . Therefore. in addition to auto-stressed and pre-stressed. we are unable to distinguish between the two possible analyses: they are either taken as pre-stressed suffixes.e. primary stress moves to the third-last syllable. some are listed as such in Chapter 8.. trigger the nominal subclause of the MSR. -it.Word stress – Part 2 ex..ni(c). whereas others. heavy even without the final consonant). What the two groups have in common is that they are non-neutral suffixes in the sense the term is introduced in the previous chapter: they do influence the way primary stress is placed in the word. adjective-forming suffixes and endings fall into two categories: some of them. It is also noteworthy that a number of underived adjectives are so undeniably verbal in nature that they exhibit exactly the same stress pattern as the corresponding segmentally identical verbs.cé(re) /()/. -id. compléte. which goes as follows: if the last syllable of a verb is stressable (i. sin. púr.p(le). -ous. e.g. so it naturally follows that the preceding syllable is stressed. -ar. etc. GA abstráct. like -ic. or the adjectives which they produce are considered to behave as verbs as far as stress rules go. corréct.pré(me). At this point we are able to introduce the Alternating Stress Rule. such as -al. Pla.

as mentioned above. thus they also appear to be subject to stress alternation. éxercise. mánifest (v/adj/n). all with a (0)103 stress pattern. The Alternating Stress Rule is an almost inviolable constraint on verbs (it only has a handful of exceptions like contínue. there is no stress alternation in èxtrapóse since it is composed of a verbal prefix (underlined) plus a monosyllabic stem. Examples include the verbs mánifest. éxercise (v/n). without respect for the prominence of the right edge. décorate. otherwise the MSR produces the expected output with the primary stress at the end and a secondary stress at the beginning (according to the Early Stress Requirement). in words containing a verbal prefix the Alternating Stress Rule can only apply if the final syllable of the verb is not a monosyllabic stem. etc. e. cómpliment. In addition. súbstitute. elíminate. èxtrapóse. éxecute. and it overwrites the result of the MSR. so certain three-syllable adjectives.Chapter 9 syllable is reduced to tertiary. distríbute). a number of adjectives and nouns simply copy the pronunciation of the corresponding verbs. 132 . say. súbstitute (v/n). invéstigate. In intérrogate. Nouns like ávenue. However.. we experience the effect of the Alternating Stress Rule as. anníhilate. grándiose. cómplement. For example. hármonize. exággerate. also undergo it. the stem is disyllabic. such as ábsolute. Sometimes the Alternating Stress Rule is extended to word classes other than verbs.g. too: they are claimed above to often behave like verbs – this is also true in the case of the Alternating Stress Rule. whereas in ìntervéne the same prefix attaches to a monosyllabic stem and consequently the cooperation of the MSR and the Iambic Secondary Stress Rule takes place. Nouns with a long-vowelled final syllable belong here.. súpplement. díagnose. The same features characterize extrápolate vs. where there is because of the absence of the verbal prefix. although it contains a verbal prefix (underlined). contríbute. for instance. récognize. RP óbsolete. as opposed to.

Certain endings characterize both nouns/adjectives and verbs. e. delíberate (v-adj). To conclude the discussion of the three major word classes with respect to stressing. perféct (v) – pérfect (adj/n). séparate (v-n/adj). In a few.g.. the third vowel is a schwa in the noun (yielding 100) but unreduced /e/ in the verb (103). but somewhat differ in the two cases (e. suspéct (v) – súspect (adj/n). -ment.g. is always primary stressed on the first syllable. cóntent (n) – contént (adj). This is because the last syllable in nouns is not normally stressable (recall that it is extrametrical!). pédigree. Further examples: cómpliment. Válentine. dócument. for example because it is underived.. exceptional examples the adjective receives nominal stress – then it is the mirror image of the verb. as in Áugust (n) – augúst (adj). for instance.g. mínuscule.g. -ate). in the verb that syllable is heavy (even without the final /t/) and such three-syllable verbs are expected to undergo the Alternating Stress Rule. Exactly the same happens in words ending in -ate: this suffix-like morpheme contains a full diphthong (/-/) when final in a verb but just a schwa (/-/) when final in an adjective. however. délegate (v-n). assóciate (v-n/adj). it cannot be left unnoticed how intimately primary stress placement is connected to syllabification. éstimate (v-n). e. adjectives. súpplement. gráduate (v-n/adj). in a noun. e.. in contrast. ánecdote. Complement. sácrifice. With respect to the above discussion of the English MSR. or both. e. RP stálactite/stálagmite illustrate this. and verbs. let us highlight a few additional pairs of remarkable segmentally nearly identical nouns...Word stress – Part 2 Fáhrenheit.g. presént (v) – présent (adj/n). When a consonant is situated between two vowels in a 133 . mínute (n) – minúte // (adj). When the adjective in such a word pair is stressed as a verb. its stress pattern is the mirror image of the noun.

al. illustrate that indeed as many consonants are syllabified as initial as possible. it is incapable of influencing stress assignment.(bra) but heavy in a. á. too.mme.gén. and concentrate on the consonants between the second and third vowels. Compare two nouns. All the regular cases treated above suggest that it is the latter solution which is chosen. this adjective is stressed at the very beginning: skéletal.dru. the Sonority Principle in Chapter 5) while the /nd/ in the latter is not – the two cases are predicted to be syllabified and therefore stressed differently: the final non-extrametrical syllable is light in ál.ge. first syllable is primary stressed. on condition that they constitute a wellformed initial cluster. algebra and agenda. and being a syllable-initial consonant. Take the word skeletal for example. 134 . Moreover.(quate). it is not at all indifferent whether it belongs to the syllable headed by the first one. therefore the final syllable is expected to be extrametrical. the fact that they are light can only be due to the absence of a closing consonant. the underlined syllable would be classified as heavy and assigned primary stress: *skelétal.Chapter 9 morpheme. or to the one headed by the second vowel. Nevertheless. Examples like sý.le.or three-member consonant clusters get syllabified into the following syllable.(try). RP quá. two. making it a heavy syllable. The fact that the underlined syllables are not stressed can only be due to their lightness. We conclude that single consonants are initial in the syllable whenever possible. The /br/ in the former is a possible initial cluster (cf. that is.(da). in the intervocalic position syllable-initial consonants are created.de. It is a derived adjective following the nominal pattern (see above).tal.et. which can only be accounted for if we suppose that the syllabification is the following: ske.(ple). If the syllable divisions were located as the dots indicate in skel. The underlined syllable is light and consequently the preceding. the last "visible" syllable is checked for weight but only receives primary stress if it is heavy.

a.(stry). syllables). /sp/ in both spill and lisp. exhibiting far more combinatorial possibilities than any other consonant. a problematic case: word-medial sC sequences do not always appear to constitute syllable-initial clusters.Word stress – Part 2 There is.che.(ster).spi.(ster). however. It is crucial to see that when stress rules apply. The /t/ in skéletal clearly belongs to the final (extrametrical) syllable. The fact that allophonic rules like the ones introduced in Chapter 2 treat these consonants as ambisyllabic can only be decided after stress assignment has taken place simply because it hinges on the stressedness of the vowels: consonants followed by a stressed vowel are never ambisyllabic. which is likely to have contributed to the ambivalent behaviour they exhibit word-medially with respect to stress placement.ni. happens in steps: 135 . For instance. A. there is no ambisyllabicity for stress rules.lás.dís. while /br/ is only possible initially and not finally (examples like brim exist. /s/ plus voiceless plosive sequences.més.du. While they do in mí. then. Mán. ór. and the second /t/ (but perhaps the first one as well) in compétitor. We can conclude therefore that /br/ is unambiguously syllable-initial. ín.(stra) quoted above.che. at least. and /nd/ is only possible finally and not initially (lend vs. but the same does not hold true for. consonants followed by an unstressed vowel normally are. syllabification seems to be always exhaustive and straightforward – significantly. but *mibr would be ill-formed). *ndel). and so is the one in vánity or héretic. one consequence of which is the curious fact that certain /s/+consonant clusters are found both at the beginning and the end of words (i.(tra). their members belong to separate syllables in se. we see /st/ in both stab and bast.. /sk/ in both scut and tusk.e.(ka). The derivation of the pronunciation of words.(ter). Recall from Chapter 5 that /s/ takes part in the construction of syllables in a special way in various respects.

Compare apparatus and asparagus. in which case rhyme weight depends on what element follows it. 136 .ni.ra. or short.ti.ni.(um) //. e. This fact can cause problems to students of English.(ry) //. primary stressed: ca.pé. refúse – refúsal. Still. dený – deníal. two words showing spooky resemblance. although it can be crucial whether a vowel is long.(ty) vá. vowel length is not consistently encoded in the spelling of English. As far as the stress rules introduced above are concerned. trý – tríal.tal ské. who are frequently first faced with an unknown word in its spelt form.l. since the third vowel is long in the former but short in the latter.g.rá. arríve – arríval. 3 The examples illustrate the pronunciations in a tapping dialect of English which distinguishes clear and dark /l/.ná. the only exception being búry – búrial.al  va.pe.t.t. strongly prefers to be suffixed to an end-stressed word.t.ti.spá. they are stressed differently: à.t. For example.le. a note is in order here.e. Nominal -al. rehéarse – rehéarsal.ty vá.Chapter 9 syllabification: stress assignment: ambisyllabicity: allophony3: ske.y  com.(tus) // vs. there are a couple of stress-sensitive affixes. For example.i.or () It is not only allophony rules that follow stress assignment in English but certain morphological operations.ppa. too. automatically producing a heavy syllable.tor com. whose attachment to a base is determined by its stress pattern.pé.le.(tor) com. English spelling is not always capable of reflecting the pronunciation of vowels.sé. a. forming abstract nouns out of verbs. Unfortunately..n.(tal) ské.(gus) //. mu. nothing indicates that the second vowel in canary and museum is long – therefore its syllable is heavy and as a result.i.

Perhaps this is a situation where the exception proves the rule. neutral and non-neutral affixation exert various effects. and finally. we are able to identify the stress rules of English as the generalizations which hold for the majority of the vocabulary and which characterize newly borrowed or coined items. 137 . We hope that the discussion faithfully reflects the complexity of this issue. being influenced by syntactic. In spite of this.Word stress – Part 2 This chapter has looked into primary stress assignment in underived verbs and nouns as well as the different subtypes of adjectives. only the minority of the examples constitutes the cases we refer to as "irregular" even if some of them happen to be highly frequent words in English and therefore our impression of the proportions may be somewhat distorted. morphological and lexical factors beside the phonological ones: verbs and nouns follow two distinct patterns. Clearly. all the regularities have exceptions.

g. verb. interrogative pronoun. suprasegmental. adverb. if they receive sentence stress it is always indicated. We have already seen the principles of assigning different degrees of word stress in Chapters 8 and 9. e. car. This is the car that I bought /      /.e. There are several principles to be kept in mind:  In isolation. many words will have no stressed syllable at all. we will have to do things a little bit differently from what we have been doing when transcribing the stressed syllables of isolated words.)  In isolation.10. . weak syllable. Wh-question. the melody of sentences. the stress of monosyllabic words is not shown. noun. The words this. conjunction. strong syllable. However. Let us now continue with a description of how the same thing works at sentence level. can is realized as // instead of //. function word. content word. demonstrative pronoun..g. e. (Cf. stress (primary/secondary). In the first part of the chapter we will be discussing the position of sentence stress while the second part of the chapter will discuss intonation. Sentence stress and intonation Before you study this chapter.. Chapter 7.. pitch. If we want to show stress (and later intonation) at sentence level. In a sentence. Yes/No question This chapter is concerned with some suprasegmental processes of English which are often grouped under the umbrella term of intonation. every word has a primary stressed syllable (although the stress of monosyllabic words is not indicated). check whether you are familiar with the following terms: adjective. i. they occur in their weak forms.

that and what. That is. we may do so in sentences. indicated by underlining in this chapter. and just like in words. a tone-unit starts after a tonic and ends in a tonic. too. With the help of the tonic we may define some further concepts relevant for our discussion: speech is divided into so-called tone-units – or tone groups or intonation phrases –. adjectives and adverbs – including adverbial particles like up – while the latter contain the rest of the categories. auxiliaries. conjunctions. in sentences it is also the last stress that is the strongest. pronouns. e. nouns. the strongest stress of a sentence falls on the last stressed syllable. prepositions. this. To be able to cope with the assignment of sentence stress and. Tone-units are normally realized by clauses as in the first three examples or by longer phrases as in the second three examples below. intonation. some minor categories can be compared to one of these two groups: demonstrative and interrogative pronouns. The boundaries of tone-units are usually indicated by vertical lines. on the basis of that. where respectively.Sentence stress and intonation bought are stressed in the sentence as indicated by the stress marks although they are all monosyllabic words. we have to clarify a few important basic notions. The former include the four basic categories. In the case of word stress we identified different degrees of stress in Chapters 8-9. which is called the tonic – sometimes also called accent.g. One of the distinctions we have to make is between lexical/content words and grammatical/function words. Also. verbs. nucleus or sentence stress. The tonic will have a special role in describing intonation as intonation is nothing else but a falling or rising melody starting on the tonic. That is.. which are parts of connected speech ending in a tonic. 139 . are stressed like content words.

The sentence above may be divided into feet the following way: 140 . and the head – starting with the first stressed syllable and ending with the last unstressed syllable before the tonic. Besides the tonic. e. As it was mentioned in Chapter 7.Chapter 10 She wanted to face the problems on Tuesday. No way! At five o'clock. only at the meeting. These units of rhythm are called feet (cf. -day in the first example above –. a foot is the sequence of a stressed syllable and all the unstressed syllables following it up to the next stress.g. Pre-head Head Tonic Tail The tone unit may also be divided into other kinds of constituents which play a very important role in determining the rhythm of the sentence.. the same name that is used for rhythmic units in literature to determine the rhythm of poems. wanted to face the problems on in the first example above: She wanted to face the problems on Tuesday. she in the first example –. e.. No. the prehead – the unstressed syllables before the first stress..g.g. Chapter 7). the tone-unit has the following parts: tail – the unstressed syllables following the tonic. He felt uneasy but the others were enjoying themselves. e. I didn't really want to come but here I am.

It means that the stressed syllables follow each other at intervals of about the same length.prob ... Tues is approximately equal although there might be different numbers of unstressed syllables between them. When connecting words into a sentence it often happens that there will be a sequence of three stressed syllables with zero or just one unstressed syllable between them.. Since this rhythmic sequence of pulses is very different from Hungarian. Recall from Chapter 7 that the special characteristic property of English rhythm is that it is stress-timed. which sounds like a pulsating rhythm. a process called Rhythmic Stress Deletion – this deletion of stress is indicated with a superscript zero in the examples: good old days → good 0old days very brave soldier → very 0brave soldier cover the big news → cover the 0big news 141 . face . foot 0 is an incomplete one as it only contains unstressed syllables – if there is a pre-head. To avoid such stress clusters the middle one of the three stresses is deleted and the syllable is pronounced as unstressed.. In such cases the rhythm becomes jerky. This means that in the sentence above the time elapsing between the stressed syllables wan. it is always an incomplete foot. it is something to be practised a lot to get used to pronouncing (sometimes many) weak syllables between the stressed ones. staccato-like...Sentence stress and intonation She foot0 wanted to face the foot1 foot2 problems on Tuesday.. foot3 foot4 It is clear that the first foot.

 142 . This is called neutral tonic placement or neutral tonicity. a process called Rhythmic StressShift: the best man asked // a stone deaf guy // a dark brown hat // → → → the best 0man asked // a stone 0deaf guy // a dark 0brown hat // As is clear from the above.Chapter 10 In cases when a word with two stresses is followed by a word stressed on the initial syllable it would result in three stressed syllables in a row. This way the stress pattern of the first word has been changed as the primary stress has shifted to the left from the last syllable. this may affect finally-stressed compounds and longer words ending in two stressed syllables in a way that they will have two slightly different stress patterns depending on whether the next word starts with a stressed or unstressed syllable. unmarked or default type: it does not express emphasis or contrast. In the following we take a look at the two major types of tonic placement. The first type of tonic placement is the neutral. Tonic on last content word Everyone was there We didn't want to talk about the details. As a result. Before words starting with an unstressed syllable nothing happens. rhythmic stress deletion will delete the stress in the middle. The neutral tonic is normally placed on the last content word but in some exceptional cases it may fall on an earlier content word or on a function word. but before words starting with a stressed one the final primary stress of the first word shifts one to the left.

Jim was there. No tonic on 2nd part of initially stressed compound. Tonic on an earlier content word (the skipped last content word italicized) He bought a new mountain bike.Sentence stress and intonation He was finally admitted to university. No tonic on common adverbs. appended remarks. Tonic on possessive pronoun. No tonic on afterthoughts. contrast a part of it with something or focus on some new information. The following two sentences demonstrate that while the first sentence with neutral tonic placement on the last content word does not emphasize or contrast any part of the sentence. That's what the book says. Tonic on a function-word No. No tonic on "obvious predicates". In the second major type of tonic placement the speaker wishes to emphasize some part of the utterance. 143 . We'll just stay here. which may be achieved by placing the tonic at a different place from where it would normally appear. you can't. Where are you from? This is mine Tonic on an auxiliary if no other stressable word. I think. Tonic on Prep in short sentences without main verb. It was nice. the second sentence with so-called dislocated tonic does. Jim was there.

. i. while the second sentence emphasizes that it was Jim who was there. or negative polarity with positive polarity. although that's not what you think. Accordingly. the first sentence has a neutral interpretation..... .g.Chapter 10 As the underlining indicates. I DID pass the test. . a certain word that we want to contrast with another one. Let us now take a look at what might be the reasons for having a dislocated tonic in a sentence. If a word is contrasted. not somebody else. negation with assertion... a name with another name.. Contrast In many cases the tonic is placed on an earlier content word to express contrast between what has been said and the word/expression bearing the tonic. I gave JACK that book on history yesterday. We visited a lot of MUSEUMS in London. there has the tonic in the first sentence while Jim has the tonic in the second. contrary to what you think/claim. and not somebody else. then the words describing this old information will be de-stressed and stress (and the tonic) will be shifted leftwards to some earlier word considered to carry new information. it is indicated in capital letters. I HAVE seen the film earlier. .. It most typically happens in answering questions repeating some words from the question – the old information skipped by tonic assignment is italicized. and not night-clubs.. 144 .e. New information If the end of the sentence contains information the speaker thinks is known by the listener as old information. e. . The two most common cases are when either it is a particular lexical item.

and. that is. Oh. pitch depends on what kind of intonation is used in the pronunciation of a particular sentence. But I already HAD lunch. there are also nonlinguistic factors that influence pitch: age – children have a higher pitch than adults –. On the other hand. Now we turn to how these units relate to intonation. It is VERY interesting. too. or pitch. Every speaker has a limit to how high or how low a pitch they may produce. The intonation or melody of a sentence is the voice-height. On the one hand. Hungarian. for instance. finally. the head and the prehead –. the melody and the melody change of a sentence. how it relates to the rest of the clause – the tail. how the head plays a role in determining the rhythm of the clause by being divided into feet. my FIANCÉE is Hungarian. the highest pitch of an average native English speaker is higher than that of an average native Hungarian speaker. This pitch range is different for each speaker but it does not influence the understandability of their speech: it is not the absolute but the relative pitch height that matters. sex – men normally have a lower pitch than women –. and the emotional state of the speaker – excited speakers tend to have a higher pitch than someone in a neutral mood. is said to have a much narrower pitch range in general than that of English. how tonic assignment may be performed in neutral cases and in dislocated cases when the speaker wishes to emphasize or contrast some part of the sentence. Pitch differences do not only occur between speakers but also between languages.Sentence stress and intonation Is the book interesting? Do you want to have lunch? I'm from Hungary. while the lowest pitch of an English speaker is generally 145 . these two are the upper and lower limits of one's pitch range. So far we have seen which part of a sentence carries the strongest stress – the tonic –.

however.e. The part of the tone-unit after the tonic. Since it is not the pitch change realized on the tonic. In the following we briefly describe the characteristic properties of the parts of the tone-unit followed by a discussion of the tones and the typical meanings or functions associated with them. so the same communicative function is not always expressed by the same tone and the same tone does not always express the same meaning. it does not count as falling intonation. which results in a lower pitch. excited and affected. Recall that the last stressed syllable of the tone-unit. that is. The reader. English speakers. If we do not only consider the pitch change realized on the tonic syllable. find the speech of the average Hungarian – whether this average Hungarian speaks English or Hungarian – too flat. the tonic always has pitch change. Hungarian speakers often report that they find English speech too emotional. Intonation is the way the pitch changes in the tone-unit. the tail – if 1 Note that this usage of tone is slightly different from tone in tone languages (cf. on the beginning of the head. This change associated with the tonic syllable is referred to as the tone1. the speakers' voice will either rise or fall on the tonic syllable. on the other hand. it is just a natural consequence of the fact that speakers are normally running out of air and this way the velocity of the outflowing air is dropping. The melody of the tone is always continued in the tail of the tone-unit. the beginning of Chapter 8).Chapter 10 lower than that of a Hungarian speaker. monotonous or boring.. which is called downdrift. i. The pitch usually gradually falls throughout the head. the tail will never contain another change in pitch. we may talk about the tune or intonation pattern of the sentence. has to be aware that this relationship is not a one-to-one relationship. For instance. This is even noticed by the untrained ear. The melody of the pre-head of the tone-unit – if there is one – normally starts at a relatively low pitch which normally jumps high up on the first stressed syllable. 146 . but rather the pitch changes occurring throughout the whole tone-unit.

implication The falling tone The falling tone is the most common. He believed that they had seen the movie earlier. gradually rise on the tail. it is most often used in plain statements. neutral tone used in English. If it falls on the tonic.Sentence stress and intonation there is one – is normally a simple continuation of the pitch change of the tonic: if the pitch rises on the tonic. meaning neutral. too. finality indifference. real. serious commands – as it expresses finality and definiteness –. then it will fall on the tail. It suggests that speakers are simply conveying information. definiteness. encouragement Yes/No questions. too. Have you seen this movie? Let us now take a look at the four types of tone and the different functions associated with them: Type of tone Falling Rising Name of tone Fall Low rise High rise Fall-rise Tone contour Function. in exclamations – expressing that the speakers are sure of 147 . it will slowly. This is demonstrated by the graphic representations below. The arrow before the tonic syllable indicates the pitch change on the tonic. inquiry old information. As a result.

and also in cases of expressing encouragement. questions starting with a question word (who. where. and in Wh-questions.e. not too good. Instead.. (Not too bad.) (Where shall we go tonight?) We may go to the cinema. it is always used to respond to something that somebody said. etc. (I don't care. This is the house where I grew up. Often it expresses indifference – an "I-don'tcare attitude" –. I think we haven't met.). greetings and when saying thanks. how. it is used in apologies. what. i.) 148 . Leave all the books on the desk! Put the gun on the ground! What a truly beautiful day! How absolutely fabulous an idea! How did you know he was my brother? What are you working on? The low rising tone The low rising tone is the most difficult for Hungarian learners of English as in Hungarian it is only used in certain types of questions while in English it is never used in this sentence type. why.Chapter 10 what they are saying –. (How was the film?) It was all right.

Sentence stress and intonation  Thank you Excuse me. then Hungarian speakers tend to have serious problems with the height: the high rise produced by a Hungarian speaker is simply not high enough for an English speaker. the first will rise.. This is also a serious difficulty. finally. It will be all right. the second last will rise. 149 . You're welcome. then depending on the number of its syllables. All in all. if there is a tail. If there is no tail. whenever there is a tail in such Hungarian sentences. i. The high rising tone The high rising tone is either a high rise on the tonic. instead. and will consequently express indifference or boredom. then the high rise continues from the tonic throughout the tail and is evenly distributed over the syllables of the tonic and the tail. or it is accomplished on the tonic and the tail. the second will fall. if the tail has two syllables.e. it will sound as a low rise. Don't you worry! It doesn't matter. Bye-bye. the whole high rise is realized on the tonic syllable. it will be a rise-fall. if not greater. the last one will fall. it will have a rise-fall and not a steady rise as it does in English. if the tail has more than two syllables. if there is one. one of the following things will happen: if the tail consists of just one syllable. In Hungarian there is only a real rise if there is no tail. In English. for Hungarians as such tonic+tail combinations may not be pronounced with a steady high rise in Hungarian as the language does not permit a rise on consecutive syllables within the same tone unit. if there is no tail. If there is a tail.

a rise from low to mid.Chapter 10 The high rise is most often used in Yes/No questions. It is one of the tones that makes English speech sound too theatrical or affected for the Hungarian ear but it is not as difficult to learn to do correctly as the high rise or the low rise. i. If something is said with a high rising intonation. (Invite me to see it.. it is always a real question. but a part of the message is only implied. Have you ever been to England? Did you do your homework? Was it your grandmother? Is Johnny coming to the party? The high rise is also found in echo-questions. I've never seen your flat. which do not start with a question word. which repeat what a speaker has previously said. it may indicate that the speaker is not telling everything. the listener has to find it out from the context.e. A fall-rise may be used for several purposes: on the one hand. Where? You met him at which station? The fall-rise tone The fall-rise is a combination of a fall from high or mid tone to low followed by a low rise.) 150 .

Sentence stress and intonation It may also occur in a sentence made up of two tone-units. 151 . In this chapter we have taken a look at two very important suprasegmental aspects of English pronunciation: the stress patterns of sentences. especially tone placement on the one hand. The new information is pronounced with a falling tone. and the basic types of intonation and their differences from Hungarian intonation on the other. In our old car  there was enough room for six people. the first expressing old information serving as background for the new information in the second part of the sentence.

On the one hand.11. English observes another principle instead. rhotic/non-rhotic. coronal. devoicing. This is made necessary by the fact that the principles of English spelling (or. and the rules regulating this relationship. Hungarian mostly represents the different pronunciation variants. On the other hand. that of morpheme identity: it prefers to keep the spelling of a morpheme unchanged regardless of whether the particular morpheme is pronounced with one allomorph or another. ház-hoz.g. Chapter 6). homorganic. clear/dark-L.. place assimilation. föld-höz where the vowels of the three different variants of this suffix are different in pronunciation and it is clearly indicated in spelling. glottalization. digraph. loanword. palato-alveolar. productive/non-productive.g. but at least much more successfully than English spelling does. kert-hez. allomorphs of a morpheme differently in spelling.e. it aims at setting up a one-to-one relationship between letters and phonemes as much as possible. want-ed //. check whether you are familiar with the following terms: allophone. This way. Yod-dropping This chapter mainly focuses on the regular correspondences between consonant letters and sounds. play-ed // (cf.. morpheme (free and bound). Letter-to-sound rules – Part 1: Consonants Before you study this chapter. suffix. orthography. allomorph. the spelling will always tell us how to pronounce the particular morpheme in question.. root/stem. aspiration. e. the correspondences between Hungarian letters and sounds are much more straightforward as spelling observes the phonemic principle more than in English. palatalization. weak/strong forms of function words. too. R-dropping. e. orthography) are quite different from those of Hungarian. kiss-ed //. i. This sometimes also happens in . tapping/flapping.

greenness //. Long consonants are only pronounced if two identical consonant sounds are put in adjacent positions at morpheme or word boundaries.. We have to note again that English lacks long or so-called geminate consonants. These and other systematic differences between RP and GA. and also at the letter-to-sound rules that regulate the connection between sounds and letters as well as the exceptions that fail to obey these rules. they are pronounced as short sounds as in letter /()/. Although doubled consonant letters do occur in English. i. In this chapter we are going to take a look at the regular pronunciation of single consonant letters and consonant digraphs one by one. recommend //.Letter-to-sound rules – Part 1 Hungarian but not as often as in English. The next chapter is going to discuss the same for vowel letters and vowel digraphs. full transcriptions are given whenever the two accents differ more significantly.e. 153 . 1 Throughout these two chapters transcriptions show RP pronunciations. unnatural //. Thus. attack //. Higgins //.1 Single consonant letters Let us take a look at single consonant letters first. mentioned in previous chapters. are not indicated separately. Keep in mind that GA is a rhotic accent (Chapter 2) with extensive Yod-dropping (Chapter 5) and frequent tapping (Chapters 2 and 7). list exceptional cases and positions in which the letter is typically silent. ballet RP // (GA //). For each consonant letter we are going to define what sound(s) it normally represents in what environments. if a word or morpheme ends in a certain consonant and the next one starts with the same as in disservice //. However. the two languages observe the principles of spelling in very different ways – although they are clearly not the two extremes on the scale.

) It is irregularly silent in certain -bt clusters as in debt //. pebble //. bomber /()/. bomb //. t It regularly represents the phoneme // and its allophonic – weakly or strongly aspirated. climbing //. [] or []. It is regularly silent in morpheme-final position after a nasal as in numb //. pirate /()/. subtle //. psychopath //. pneumatic //. climb //. b It regularly represents the phoneme // and its – devoiced or voiced – allophones as in banana RP // (GA /--/). leopard //. pen //. psychology //. numbest //. coup //. lap // [] or []. prayer 'words used in praying' /()/. supper /()/. below //. better /()/ [()] or [()]. unaspirated. unaspirated. raspberry //. late // [].as in pneumonia //. cupboard //. It is irregularly silent in corps //. receipt //. doubt //. (Cf. rotten //. debtor /()/. It is regularly silent in word-initial position in pn. It regularly represents the palatals // and // in cases of lexical palatalization (see rule at the end of Chapter 11) in words like action 154 . step // [] or []. glottalized or flapped – variants as in take //. super /()()/. tonight //. glottalized – as in plenty //. Chapter 5. label //. stab //.and ps. bombed //. rob //. fantastic //. sober /()/.Chapter 11 p It regularly represents the phoneme /p/ and all of its possible variants – weakly or strongly aspirated. psychiatrist //. rubber /()/.

buffet RP // (GA //). Christmas //. picture /()/. d It regularly represent the phoneme /d/ and its allophonic – devoiced. [()] or [()]. literature /()/. beret RP // (GA //). [()] or [()]. It is irregularly silent in words of French origin ending in -et as in ballet RP //. kissed //. see Chapter 6). motion //. rider /()/.Letter-to-sound rules – Part 1 //. Chevrolet RP // (GA //). recommend //. It regularly represents the phoneme /t/ in the past tense suffix after stem final voiceless consonants other than /t/ as in backed //. It is irregularly silent in words like grandmother /()/. tsar /()/. bouquet // or //. laughed RP // (GA //). question //. gradual //. grandeur /()/. cabaret //. flapped – variants as in damage //. nature /()/. wrestle //. madder /()/. often // (this word is pronounced by some speakers as //). forecastle //. It is irregularly silent in consonant clusters in words like boatswain // (also spelled bosun). It regularly represents the palatal // in cases of Palatalization (see below) in words like educate //. listen //. stepped // (for the pronunciation rule of the past tense suffix. sender /()/. soldier /()/. bend //. delete //. squashed //. 155 .

It is irregularly silent in Connecticut //. racing //. banking //. linguist //. knitting //. It regularly represents the phoneme /s/ as in city //. ocean //. c It regularly represents the phoneme // and its – aspirated. vicar /()/. knuckle //. acne //. endictment //. muscle //. sandwich // or //. It irregularly represents the phoneme // in words of Italian origin like cello //. It regularly represents the phoneme // in cases of Palatalization (see below) as in vicious //. unaspirated and glottalized – variants as in kettle //. social //. goulash RP // (GA //). musician //. lucid /()/. unaspirated and glottalized – variants as in cat // cover /()/. k It regularly represents the phoneme // and its allophonic – weakly or strongly aspirated.cluters as in knave //. king //. knife //. knowledge //. hug 156 . It is regularly silent in word-initial kn. czar /()/. face //. account //. poker /()/. beggar /()/. get //. dice // (see the discussion of Velar Softening below). longer /()/.Chapter 11 grandpa //. baker /()/. thank //. knock //. concerto //. bigger /()/. g It regularly represents the phoneme // and its devoiced variant as in gallop //. facial //.

singing //. forget //. café RP // (GA //).) It is regularly silent in word-initial and word-final gn clusters as in gnome //. sign //. Don Juan / /. longest //. jockey //. strong //. belong //. youngest //. strife 157 . longer /()/. garage RP // (GA //). reference //. gnu //. harbinger /()/. young //. younger /()/. ginger /()/. gym //. It is regularly silent in morpheme-final position after a nasal as in sing //. (Cf. huge //. It irregularly represents the phoneme // in some Spanish geographical names like Baja //. But it is irregularly pronounced in morpheme-final position after a nasal in the comparative and superlative forms of the following three adjectives: long //. It is irregularly pronounced as // in French loanwords as in beige //. cajole //. singer /()/. resign //. belonged //. strongest //.Letter-to-sound rules – Part 1 //. regime RP // (GA //). collage //. j It regularly represents the phoneme // and its devoiced variant as in jet //. Chapter 5. It regularly represents the phoneme // (see the discussion on Velar Softening below) and its – devoiced – variants as in engineer /()/. Note that this consonant letter is never pronounced as //! f It is regularly pronounced as // as in final //. coffee //. stronger /()/.

bus //. Word-finally it regularly represents // as in hazardous //. staff RP // (GA //). but it irregularly represents // in word-final position in proper names and function words. never /()/. minus //. It never represents the phoneme //! s It regularly represents the phonemes // and // depending on the environment: Word-initially it regularly represents the phoneme // as in singer /()/. promise //.. busy //. his //. Sudan RP // (GA /-/). bosom //. close (v) //. silence //.Chapter 11 //. elves //. 158 . Denver /()/. goose //.e. house //. grave //. jive //. phase //. bison //. It is irregularly pronounced as /v/ in of RP // (GA //) (in its strong form) or // (in its weak form). i. James //. has // or //. Between vowel letters it regularly represents // as in music //. but it irregularly represents // between vowel letters. senior /()/. desert (n) //. Jones //. basic //. wives //. for instance in base //. cactus //. was RP // (GA //) or //. in words like is //. (Cf. vanity //. Charles //. cousin //. crisis //. stuff //. close (adj) //. lover /()/.) v It is regularly pronounced as /v/ and its devoiced variant as in veal //. case //. Chapter 7.

horse //. bi-sect //. message //. bass //.Letter-to-sound rules – Part 1 Between a root vowel and an affix vowel it normally represents // as in dis-integrate //. see Chapter 6. insist //. crises //. universe //. dissolve //. pre-sume /()/. l. assassin //. z It regularly represents the phoneme // and its devoiced variant as in 159 . analyses //. be-side //. possess //. pulse //. fusion //. Mercedes //. de-sign // (the hyphens indicate morpheme boundaries). in all the possible environments (for Palatalization see below) as in mission //. It regularly represents the palatalized variants of the above sounds. It regularly represents // when doubled. version RP // (GA //). tense //. dis-ease //. measure /()/. and r (silent in the non-rhotic accents) as in course //. // and //. It regularly represents // in final -es when not a regular suffix as in species //. but it irregularly represents // in words like divis-ible //. false //. mansion //. but it irregularly represents // in words like scissors //. as in kiss //. dis-agree //. ss. sure /()/. For the rules of its pronunciation. passing RP // (GA //). Hercules //. mis-understand //. It regularly represents // or // in the regular suffix -(e)s. vision //. It regularly represents // after n. dessert //.

penny //. follow //. 160 . yolk //. It is regularly made silent before consonants and a pause by the RDropping Rule (see Chapter 2) as in cart //. hammer /()/. m It regularly represents the phoneme // as in matter /()/. borrow //. meringue //. l It regularly represents the phoneme // and its – clear and dark – allophones (see Chapter 2) as in light //. notion //. fence //. salmon //. bottom //. raccoon //. banquet //.Chapter 11 zoo //. It is irregularly silent in final -mn clusters as in autumn //. building //. condemn //. razor /()/. people //.cluster in mnemonic //. Gonzo //. burial //. pin //. pink //. n It regularly represents the phoneme // as in number /()/. singing //. buzz //. fell //. It is irregularly silent before consonants in words like folk //. plumb //. language //. talk //. level //. redial //. caring //. steer /()/. walk //. flair /()/. It is irregularly silent in the word-initial mn. zeal //. sing //. r It regularly represents the phoneme // as in rifle //. solemn //. It regularly represents the phoneme // when followed by k or g (at least in spelling) as in ink //. barn //. almonds //. summer /()/. final //.

Letter-to-sound rules – Part 1 Note that it is silent in iron // (cf. footnote 1 in Chapter 4). y It regularly represents the phoneme /j/ as in yet //, yoghurt //, mayonnaise //, junkyard //. It often functions as a single vowel letter, almost like a variant of <i>, as in cry //, analysis //, bicycle // or, after a vowel letter, as a member of vowel digraphs like <ay>, <ey>, <oy> as in bay //, key //, coyote // (see Chapter 12). w It regularly represents the phoneme // and its – devoiced – variants as in want //, reward //, away //, watt //, witch //. It is regularly silent in initial wr- clusters as in writer /()/, wrong //, wretched //, wrist //. It is irregularly silent in words like who //, whom //, whose //, whole //, answer RP /()/ (GA //), sword //, two //. Note that when following a vowel letter, it often forms part of a vowel digraph as in row // or //, coward //. For details see the next chapter. For the pronunciation of the digraph wh, see below. h It regularly represents the phoneme // as in head //, hollow //, history //, ahead //, cohesion //. It is regularly silent in words like Shah //, blah-blah //, yacht //, vehicle //, annihilate //. It is irregularly silent in words like honest //, hour /()/.

161

Chapter 11 For the rule on the deletion of //, see below. x It regularly represents the sequence // and its palatalized variant (see the rule of Palatalization below) as in axe //, expand //, exit //, boxing //, tax //, anxious //, luxury //. It regularly represents the sequence // and its palatalized version // when followed by a stressed vowel as in executive //, example RP // (GA /--/), exist //, exempt //, exult //, luxurious //. It regularly represents the phoneme // when word-initial as in xerox //, xylophone //, Xavier //, xenophobia //, Xena //. q It regularly represents the phoneme // and its – weakly or strongly aspirated, unaspirated or glottalized – variants as in quotation //, quickly //, quart //, clique //, antique //, liqueur RP /()/ (GA //), liquid //, lacquer /()/. Finally, we must consider two vowel letters that may often represent the consonant // in certain environments.

u

It may regularly represent the phoneme // in the combinations qu, ngu, su in words like language //, acquaint //, aquarium //, banquet //, persuade //, dissuade //, suite //, quest //, question //.

162

Letter-to-sound rules – Part 1 o It may irregularly represent the phoneme // or the phoneme combination // in words like choir //, one //, once //, and in some words of French origin containing -oir, -ois as in reservoir /()/, bourgeois //, memoirs //. Let us now turn to those cases when two or three consonant letters represent a phoneme regularly, i.e., to digraphs and trigraphs. Consonant digraphs and trigraphs Before we start discussing consonant digraphs, we must emphasize once more that although there are a great many English words containing two identical consonant letters next to one another, these are normally pronounced as a single short consonant unless they belong to two different morphemes (see above). In the following, we only discuss cases in which the two consonant letters are different. ch It regularly represents the phoneme // and its glottalized variant as in chocolate //, bachelor /()/, beach //, chunk //, munch //, cheque/check //. It irregularly represents the phoneme // in words of French origin like machine //, moustache RP // (GA //), champagne //, chauffeur RP /()/ (GA //), chauvinism //, chic //, and also in Chicago //, Chevrolet //, Michigan //. It regularly represents the phoneme // and its allophones, mostly in words of Latin and Greek origin as in chaos //, chameleon

163

Chapter 11 //, character /()/, charisma //, chemical //, choir //, Christian //, Munich //, echo //, Czech //. tch It regularly represents the phoneme // as in catching //, fetch //, latch //, wretched //. rh It regularly represents the phoneme //, i.e., we may say that the letter <h> is regularly silent in this combination in words like rhyme //, rhythm //, rheumatism //, rhino //, myrrh /()/. sh It regularly represents the phoneme // as in shooting //, fashion //, cushion //, bushes //, crush //, hush //, Bolshevik RP // (GA /-/). ph It regularly represents the phoneme // as in phoneme //, allophone //, Humphrey //, pamphlet //, photograph RP // (GA /-/). th This digraph regularly represents the dental fricatives // and //. Unfortunately there is no rule predicting when it stands for which. However, we can say that in the majority of the cases, especially in "international" words of Greek origin, it is normally // except for rhythm //, and that in function words it is pronounced as //, e.g., they //, that //, those //. //: thinking //, bath RP // (GA //), cathedral //, healthy //, Thursday //, fifth //, length //, method //. //: bathe //, feather /()/, this //, these //, the //, 164

Anthony //. whether /()/. It irregularly represents the phoneme // in a few words. why //. qu It regularly represents the phoneme // word-finally as in cheque //. nightingale //. the former before vowels as in ghoul //.. although //. wheel //. where /()/ and wear /()/. laughing RP // (GA //). soothe //. whale //. 165 . thyme //. toughness //. e.2 For these speakers there is a difference between which // and witch //. rough //. height //. Note that in some dialects of English (especially in some American dialects and in conservative British. Scottish pronunciations) it represents a voiceless labiovelar. fought //. cough //. 2 This sound is similar to the sequence of a /h/ and a /w/. ghetto //. typically in proper names: Thomas //.g. what //. kh It regularly represents the phoneme // as in khaki RP // (GA //). //. wh It regularly represents the phoneme // as in where /()/. It is irregularly silent in many words and indicates the length of the preceding vowel as in sight //. the latter in a few words as in enough //. Thames //. gherkin //. weight //. whine //. gh It irregularly represents two phonemes. ghost //. daughter /()/. // and //.Letter-to-sound rules – Part 1 brother /()/. clique //. antique //.

dg It regularly represents the phoneme // in environments where g would represent // as in edge //. gu It regularly represents the phoneme // (for the pronunciation rules of g see below) as in guerrilla //. request //. It regularly represents // in the combination ngu as in language //. docking //. guy //. exception //. In some words gu is actually a sequence of g + u and is pronounced as // or // as in argument //.Chapter 11 It regularly represents the phoneme combination // in other positions as in queen //. bridge //. hedge //. distinguish //. liqueur RP /()/. reckon //. colleague //. i. It irregularly represents the phoneme sequence /d/ in some words as in Edgar /()/. czardas RP // (GA 166 . cz It regularly represents the phoneme // as in Czech //. excellent //. gadget //. budget //. ck It regularly represents the phoneme // as in back //. question //. xc It regularly represents the phoneme sequence // before the vowel letters e. banquet //. liquor /()/. sucker /()/. y as in excited //. quay //. guardian //. hacker /()/. Czechoslovakia //). badger /()/. It irregularly represents the phoneme // in queue //. Jaguar /()/. guest //.

) The vowel letter <i> is usually not 167 . // before an underlying // phoneme represented by the vowel letters <i> or <u> in certain environments.Letter-to-sound rules – Part 1 //. <c>. i. sci-fi //. 1.e. speech situation or tempo (in contrast to cross-word palatalization. hence the name CiV. <x> representing the alveolar obstruents //. C. discussed in Chapter 7). y as in science //.. Palatalization by <i> An alveolar obstruent will be palatalized before <i> if the vowel letter if the vowel letter does not represent a stressed vowel and it is followed by another vowel letter. sc It regularly represents the phoneme // before the vowel letters e. scenery //. i.e.. In the last part of this chapter we take a look at the rules that regulate some of the letter-to-sound correspondences mentioned above. //. is followed by the vowel letter <i> and another vowel letter. a lexical item.. scissors //. i. and regulates the pronunciation of the consonant letters <t>. <s>. Consonant rules Lexical palatalization Lexical palatalization is a rule that operates inside a word. This environment of palatalization is often referred to as CiV as the alveolar consonant. i. <d>. V. It is an obligatory process independent of style.e. //. It is also important that palatalization does not apply in wordinitial position (for exceptions see Palatalization by <u>). (We have seen a different effect of the same environment in CiV Laxness and CiV Tenseness in Chapter 3.

alveolar C so an mi vi an men ques sol  c c ss s x t t d unstressed <i>  i i i i i i i i V letter  a e o o o o o e l nt n n us n n r // // // // // // // /()/ Note that because of the above requirements there is no lexical palatalization if the vowel letter <i> represents a stressed vowel.g.g. construcTIVe //. e. soCIEty //.g. Palatalization by <u> does not apply in word-initial position. asSUME //. It logically follows from the above that there is no palatalization if <u> represents a stressed vowel. soCIAl // (the relevant letters of the words will be capitalized). e. The word maTURE /()/ is 168 . cacTUS //. e. constrUCT //. e.. or if it is not followed by another vowel letter. or if <u> is not followed by another vowel letter or consonant+vowel letter combination but two consonant letters or one consonant letter in word-final position. except in the words sugar /()/ and sure /()/. Palatalization by <u> An alveolar obstruent will also be palatalized before <u> if the vowel letter represents an unstressed vowel and it is followed by another vowel letter or a consonant+vowel letter combination.. 2. e..Chapter 11 pronounced at all.g..g..

b dropped climber /()/ number (adj) /()/ thumb // b pronounced timber /()/ number (n) /()/ sombrero // g dropped singer /()/ hanging // belonged // g pronounced fungus // bingo // Bangor /()/ 169 . alveolar C u ca vi ac sen  s s s t s unstressed <u>  u u u u u V letter  a a a a a C letter  r r r r l l l l l V letter  e e e e /()/ /()/ /()/ /()/ // // // // // alveolar C unstressed <u> na litera mea cen  t t s s  u u u u The dropping of <b> and <g> We have already noted in the discussion above (as well as in Chapter 5) that the consonants b and g are often dropped in certain positions. These voiced non-coronal stops are dropped if they are preceded by a homorganic nasal and are in morpheme-final position. It follows. As it will be clear from what follows.Letter-to-sound rules – Part 1 exceptional as Lexical Palatalization does apply although <u> is stressed (but it is usually // in GA). then. the two consonants are affected by the very same letter-to-sound rule. that the two stops are not dropped in morpheme-initial and internal positions.

Another difference between the two languages lies in the behaviour of h before vowels: in Hungarian h is always pronounced before vowels while in English. h silent in Hungarian //. hour /()/. honour /()/. of typically French origin. height //. Soho // In some words. potroh //.Chapter 11 The dropping of <h> The consonant h has a very restricted distribution in both English and Hungarian. that in final position it is not a glottal fricative. however. heir /()/. Sarah //. csehnek h silent in English Shah //. where it is pronounced even before unstressed vowels. except in word-initial position. // (the same sound as the so-called Ach-Laut in German). Bachtól //. h is only pronounced before stressed vowels. in both hállow and helló. as mentioned in Chapter 7. yacht // céh //. //. csehek //. that occurs in pronunciation but a voiceless velar fricative. the h is irregularly silent in initial position as in honest //. e. In both languages the h is silent in word-final position and before consonants. juh //. John //. h pronounced in Hungarian ház //. méhtől // In a great many words in Hungarian the letter h is pronounced before a vowel or in final position. and 170 .g. juhéj //. ahead //. Note. Before unstressed vowels it is always deleted in English (recall examples like véhicle vs. vehícular). méhek // h pronounced in English historical //. jacht //.. as in doh //.

Letter-to-sound rules – Part 1 before a stressed vowel in exháust, exhíbit, exhílarate, exhórt and all their derivatives. Velar Softening Velar Softening regulates the pronunciation of the consonant letters c and g, which have two regular pronunciations, a "hard" one, a velar stop, and a "soft" one, a coronal sibilant: c may be pronounced as // or // while g may represent // or //. According to the rule, c and g are pronounced soft, i.e., as a coronal sibilant, before the vowel letters <e>, <i>, <y> regardless of whether the vowel letter is pronounced and how it is pronounced, i.e., it is a purely graphic rule only based on spelling. c regularly pronounced as //
cellar /()/, facilitate //, cyber /()/, dance RP // (GA //)

g regularly pronounced as //
fragile RP // (GA //), sergeant //, stingy //, gyroscope //

There are quite a few cases when c and (especially) g fail to be pronounced soft in this environment, for instance: c irregularly pronounced as //
soccer /()/, Celtic //, sceptical // get

g irregularly pronounced as //
//, give //, hunger /()/, finger /()/, begin //, girl //

In other positions, i.e., before other vowel letters, before consonant letters and in word-final position c and g are normally pronounced hard, as a velar stop, although exceptions exist, e.g., Caesar, gaol, margarine, veg, etc. Note

171

Chapter 11 that in morpheme-final position after a nasal, g is not pronounced (see above). c regularly pronounced as /k/
catarrh /()/, function //, culinary /()/ //, cancer

g regularly pronounced as //
bogus //, language //, distinguish // //, jungle

We should also remember that root-final g is not softened if a regular, productive suffix starting with <e>, <i>, or <y> is added as in bigger /()/ and not */()/, longest // and not *//, bagged // and not *//. There are cases, though, when a non-productive suffix is added to the stem, a suffix which is normally placed after a bound and not a free stem. In such cases, if the stem ends in c or g (which is, of course, pronounced hard in final position if no suffix follows) and the non-productive suffix begins with <e>, <i>, or <y>, then the stem-final consonant changes into a coronal sibilant, i.e., into its soft pronunciation: Velar Softening as a process has taken place. In just the other way round, if a stem ends in a c or g in their soft pronounciation when followed by a suffix then in the unsuffixed form they will be present with their hard pronunciation. <c> electri[] - electri[]ity indu[]tion – indu[]e dedu[]tion – dedu[]e <g> analo[]ous – analo[]y ma[]us – ma[]ic lo[]o – lo[]ic

172

Letter-to-sound rules – Part 1 Yod-Dropping This rule was introduced in Chapter 5 as a phonotactic restriction on homorganic consonant clusters, however, it may as well be conceived of as a letter-to-sound rule. Although it refers to the deletion of a consonant sound //, it is used to distinguish between two very similar vowel pairs of English, the Plain-Tense //-// and their Broken-Tense variants //-//. The assumption underlying this distinction is that //+// or //+// sequences are not really combinations of two separate sounds but form one unit, one complex vowel, like a diphthong. The main reason for this is that the combination //+// has interesting phonotactic characteristics as it shows a very special behaviour in syllable structure. If we list all the possible two or three-member consonant clusters that may start a syllable in English, then we will find that whenever the last member of such a cluster is //, it is always followed by the vowel // or //. Of course, it cannot be a coincidence and the most obvious explanation for this state of affairs is that // and // or its Broken-Tense variant // form one unit, // and //. This complex vowel is regularly represented by <u>, <eu>, <ew>, <ue>, <ui> in spelling. However, it often happens that although one of these possible spellings occurs, in pronunciation we have // or //, i.e., // is missing. This is due to the rule of Yod-Dropping, which deletes the // of the complex vowel // in certain environments. 1. Obligatory Yod-Dropping Yod-Dropping is obligatory in RP after palato-alveolars, /, , , , / and consonant+// sequences as in the words parachute //, luxurious

173

Chapter 11 //, mature RP /()/, June //, July //, rude //, rumour /()/. Recall, however, that in GA Yod-Dropping is much more extensive as it applies after all coronal consonants – dentals, alveolars, palato-alveolars. As a result of this, many words are pronounced differently in (conservative – see below) RP and in GA. RP // // // // /()/ // // GA // // // // // // //

enthusiasm new tuna dubious super exuberant illusion 2. Optional Yod-Dropping

In RP, there is a tendency to also drop the // in some environments, especially in the speech of speakers belonging to the younger generations. Elderly speakers still often retain the Yod in these words. This version of Yod-Dropping is optional, it depends on style and speech tempo. It applies after the consonants /, , / as in super /()()/, suit /()/, assume /()/, exuberant /()/, presume /()/, illusion /()/, lukewarm /()/, lewd /()/.

3. The absence of Yod-Dropping It has also been mentioned in Chapter 5 above that if the complex vowel // occurs in a completely unstressed syllable, Yod-Dropping is prohibited not just in RP but also in GA, where Yod-Dropping is otherwise obligatory in a 174

175 . consulate //. In this chapter we have seen the regular and irregular pronunciation values of single consonant letters and consonant digraphs. as well as the positions in which they are silent. annual //. Then we have also seen the most important letter-tosound rules that refer to the pronunciation of consonant letters. Thus. the rule cannot apply in words like value //.Letter-to-sound rules – Part 1 much wider range of environments than in RP. menu //.

tense (plain/broken). tense vowels are further divided into two subclasses. tense and lax. This is probably the area that is the most difficult for language learners. Vowel Shift. Moreover. In English. Pre-R Breaking. Plain-Tense and Broken-Tense vowels. and each vowel letter has tense and lax pronunciations as well. laxing rules. a Plain-Lax and a BroadLax pronunciation. Trisyllabic Laxness This chapter deals with the area of the English language that has driven many language learners crazy throughout the years: the relationship between the spelling and pronunciation of English vowels. Yod-Dropping. digraph. and each vowel letter and digraph may stand for a few vowel sounds. there are two major types of full vowel in English. The main reason for this many-to-many relationship between English vowel letters and sounds originates in the fact that. Pre-R Broadening. lax (plain/broad). suffix. while lax vowels are classified into the Plain-Lax and Broad-Lax subcategories. It is these four different pronunciations that we turn to . stem. Logically. each vowel letter will have not just a tense and a lax pronunciation but a Plain-Tense. each vowel letter represents one vowel sound and each vowel sound is represented by one vowel letter. however. Hungarian letterto-sound rules for vowels are very simple. there are a few minor alternations both in length and quality but they are not significant. free U.12. especially Hungarian learners as its principles are very different from those found in Hungarian. each vowel sound may be represented by quite a few vowel letters or digraphs. orthography. Letter-to-sound rules – Part 2: Vowels Before you study this chapter. as introduced in Chapter 3. check whether you are familiar with the following terms: CiV. a Broken-Tense.

Third. containing irregularities. // does not have a regular representation in spelling – we will only find it in the last section of the chapter. it is very easy to remember the Plain-Tense value for each vowel letter as it is the pronunciation used to name the letter in the alphabet or to spell a word letter by letter.Letter-to-sound rules – Part 2 first. The regular sound values of single vowel letters are as follows: <a> // // // // <e> // // // // <i> = <y> // /()/ // // <o> // //3 // //2 <u> /(j)/ /()/ // // Plain-Tense Broken-Tense Plain-Lax Broad-Lax A few important remarks are due here concerning the table above. the vowel letter <u> has six different pronunciations instead of the expected four since it is also affected by the rule of Yod-Dropping. Fourth. Second. 177 .. in the tense values there is a yodless and yodful pronunciation (see the discussion below and in the previous chapter). namely that one of its vowel phonemes. the vowel letter <o> actually has only three different pronunciations as its Broken-Tense and Broad-Lax pronunciations are phonetically identical. First. we have to note an interesting peculiarity of English.e. i. Let us now take a look at the pronunciation values of vowel digraphs. the rules determining the pronunciation of vowel letters and finally the many different kinds of regular and irregular exceptions to leave the best for last. followed by a discussion of the pronunciation values of vowel digraphs.

some generalizations may be found in the tables above. and <u> .<w> play the same role in the digraphs.Chapter 12 <ai>=<ay> // // <oa> // //3 <oi>=<oy> // /()/ <ei>=<ey> // // <oo> // // <ea> // // <eu>=<ew> /()/ /()/ <ee> // // <ui> /()/ /()/ <au>=<aw> //1 <ie> // // <ou>=<ow> // /()/ Plain-Tense Brkn-Tense Plain-Tense Brkn-Tense Plain-Tense Brkn-Tense Plain-Lax Broad-Lax Again.<y>. Second. the letters <i> . <au>=<aw>. First. Let us now take a look at some examples for the above sound values of single vowel letters and vowel digraphs. vowel digraphs regularly represent tense vowel sounds with one exception only. <a> Plain-Tense BrokenTense Plain-Lax Broad-Lax bat // car /()/ bet // her /()/ bit // firm // bond // born // mate // <e> scene // <i> = <y> bite // fire /()/ <o> sole // sore /()/ <u> cute /j/ rude // cure /()/ sure /()/ but // burn // care /()/ here /()/ 178 . As it will be clear from the examples below there is even a tendency (although not a rule) to predict where we find which.

. powder. Bowra. The basic rule is that the five graphic representations.. one should be careful as this is not an exceptionless rule. bowl. fair. see below). and to use <y> and <w> in word-final position. our. However. <eu>. e. e. boy. //. Moira.Letter-to-sound rules – Part 2 <ai>=<ay> <ei>=<ey> Plain-Tense Brkn-Tense bay // fair /()/ <oa> boat // boar /()/ obey // heir /()/ <oo> boot // <ea> beat // fear /()/ <eu>=<ew> few // <ee> <ie> bee // believe // beer /()/ pier // <ui> suit // fruit // Muir // <ou>=<ow> house // how // our /()/ Bowra /()/ Plain-Tense BrokenTense drew // boor /()/ Europe // Jewry // Plain-Tense BrokenTense <oi>=<oy> boy // Moira /()/ Plain-Lax Broad-Lax <au>=<aw> claw // As for the digraphs ending in <i> or <y> and <u> or <w> we can claim that there is a tendency to use <i> and <u> inside a word.g. drew. typically <oo> and in some irregular cases <o> and <ou> (for irregular cases. obey. Europe. <ew>. In the following we enumerate and discuss the rules which are responsible for the letter-to-sound correspondences in English vowels. The first such rule concerns the distinction between the vowels //. <ue>. Jewry.g. <ui> normally stand for the variants starting with //. heir. // and //. that is the yodful and yodless variants. All other representations normally stand for the variants without //. bay. sauce. in many environments the vowel 179 . drown. <u>. However. how.. rather a tendency and there are quite many exceptions for it. town.g. e. claw.

The two types of graphic positions are called Free Position and Covered Position. Before actually discussing the pronunciation rules of vowel letters.Chapter 12 letters and digraphs normally representing // or // are pronounced without //. vowel letters are said to be in free graphic position if they are followed by another vowel letter. C = consonant letter. or if they are. Single vowel letters are in covered graphic position if they are followed by two consonant letters (which are either not followed by a vowel letter.) Free Position (2) -CV (3) -SLV baKE aPROn muTE caRE hoCUspoCUs As it can be seen in the table above. the consonant letter <x> normally represents the sound sequence //. L = liquid. we must introduce a pair of notions referring to the positions of vowel letters in the orthographic word. then they do not represent a stop and a liquid sound) and if they are followed by a word-final consonant letter. Note that there are some problematic cases. so it must 180 cyCLone maPLE oGRE Covered Position (5) -CC (6) -C# baNK spaM# thuNDer yaCHT fiRSt fiT# heR# spoT# (1) -V diAl goIng diE doEr (4) -# my# so# be# fly# . the hyphen indicates the position of the stressed vowel letter. a case often referred to as Yod-Dropping. On the one hand. S = stop. In the examples below the vowel letter in question is in underlined boldface and the letters of the relevant environment are in capitals. The details of this rule may be found at the end of Chapter 11. a consonant+vowel letter. (V = vowel letter. two consonant letters representing a stop and a liquid sound plus a vowel letter and also when they are word-final.

words like taXi will belong to class (5) in the table. they must be counted as one letter. words like goPHer will belong to class (2) and buSH to (6). The word yacht (//) contains a silent digraph followed by another consonant letter. the Free Position Basic Rule (FPBR) can be stated as follows: FPBR: Stressed single vowel letters in free position are normally pronounced as tense. Bear in mind that these are graphic positions for letters – whether these letters are pronounced or not does not matter. if we recall that consonant digraphs regularly represent a single consonant sound. a rule is unnecessary in their case. As a result again. There are two rules that help us decide the pronunciation of vowel letters on the basis of whether they are in a free or covered position. Thus.) Remember that these two rules only refer to stressed single vowel letters – unstressed or weak vowels behave differently. (See exceptions below.Letter-to-sound rules – Part 2 be counted as a sequence of two consonant letters. for instance.e. while vowel digraphs usually represent tense vowels. we can see that they all contain a stressed (plain or broken) tense vowel. If we take a look at the words in columns (1) to (4). On the other hand. i.. but is totally identical to bank (in class 5) in this respect. columns (5) and (6) both contain words in which the stressed vowels are (plain or broad) lax. As a result. (See exceptions below. On the basis of this we can state the Covered Position Rule (CPR) as follows: CPR: Stressed single vowel letters in covered position are normally pronounced as lax. 181 .) On the other hand. That is why dial and die. belong to the same category. Also.

soldier /()/. find //.e. gross //. manger /()/. Tenseness reversals In some cases the stressed single vowel letter is not pronounced with the tense/lax value predicted by the FPBR or the CPR. it is pronounced with a tenseness/laxness which is just the opposite of what is expected on the basis of the rules. In the following part of the chapter we take a look at exceptions.. bold //. sign //. Ruth //. danger /()/. 182 . Tenseness reversal exceptions to the Covered Position Rule bind //. most //. fight //. island //. those cases when a vowel letter is not pronounced according to the two rules above. it has a sound value which is not one of its four regular pronunciations – a kind of exception often referred to as a quality deviation. old //. i. both //. etc. kind //. i. comb //. truth //..e. range //.Chapter 12 note that the expression "stressed single vowel letter" is just a short hand for a single vowel letter representing a stressed vowel.. mild //. chamber /()/. We address the two types in this order below.e. fold //. i. told //. Exceptions may fall into two different types: those that simply do not obey the two rules but the vowel is pronounced with one of its regular pronunciations (discussed in the very first table above) – this type of exception is often called a tenseness reversal – and those that involve a vowel letter pronounced as an irregular sound.

an active phonological rule which applies if a suffix is added to the stem.e.. Before we start enumerating the laxing rules. If a suffix is added. 183 . which means that if it is a regular word then the vowel letter representing the tense vowel must be in a free graphic position. Let us repeat the most important features of the laxing rules for the convenience of the reader now focussing on the spelling of the stem and the suffix.Letter-to-sound rules – Part 2 Tenseness reversal exceptions to the Free Position Basic Rule In the case of the Free Position Basic Rule we may add one more clause to the rule: stressed single vowel letters in free graphic position are pronounced tense unless they are laxed by one of the laxing rules. In the present context. These laxing rules are those that we have already discussed in Chapter 3 in connection with the Vowel Shift. we must remember that the vowel /()/ is non-laxable in any position. The original stem vowel is tense. it is simply applied to any word even without adding suffixes. it is a regular exception to all the laxing rules below. This is a result of the Trisyllabic Laxing rule. Trisyllabic Laxness If the stressed vowel is in at least the third-last syllable of the word then it must be lax. i. Recall that in the Vowel Shift a tense vowel of a stem becomes lax if a certain kind of suffix is added. the stressed vowel may still be in free position but its pronunciation becomes lax because of one of the laxing rules.

Also. productive suffixes are not counted when determining whether a word serves as an input to the rule. e.. Free <u> is non-laxable. microphone. omega. enumerate. notify. puritan. nightingale. etc. <a> manic static habit tablet establish parish <e> metric intrepid edit level Eric perish <i> = <y> clinic timid optimistic limit critic lyric <o> polish tonic solid shovel novel comet 184 . recall from Chapter 3 that regular. isolate. this is illustrated by examples like cubicle. Laxing by ending In some words the stressed single vowel letter in free graphic position is pronounced lax despite the FPBR because one of the so-called laxing endings follows.Chapter 12 <a> Canada laminate salivate Capricorn cabaret radical <e> president hesitate regiment federate general Hemingway <i> = <y> miracle similar cinema pitiful militant typical <o> solitude domino mahogany positive solitary dominant In some words the stressed single vowel letter remains tense as required by the FPBR in spite of the fact that it is in a trisyllabic position.g.

The rule does not apply to any of the other vowel letters (cf. Its application may be witnessed in words like idiot. Cupid. navel. etc. secret. as in closure.1 Free <u> is non-laxable. which often attaches to a stem whose vowel is pronounced tense in spite of the free <u> of the suffix. polish // and Swedish // vs. 1 The word Spanish is exceptional. this is illustrated by examples like Punic.Letter-to-sound rules – Part 2 There are a few words in which the stressed vowel is followed by one of the laxing endings but still it is pronounced as tense as in basic. Laxing by free <u> The stressed single vowel letter occurring in free graphic position is regularly pronounced lax if it is followed by a free vowel letter <u> in the next syllable. this is illustrated by examples like usual. label. i. erasure. strategic. CiV tensing below). 185 . finish //. nature. Recall also that the ending -ish is only laxing when producing a noun or a verb but it is non-laxing if it makes an adjective. Free <u> itself is non-laxable. rubric. unit. hence the difference between Polish // vs. <a> gradual value <e> sensual schedule <i> = <y> visual ritual <o> soluble module A typical exception to this laxing rule is the ending -ure.e. CiV laxing The rule of CiV laxing only applies if there is a stressed vowel letter <i> or <y> which is followed by a consonant letter + another vowel letter <i> + one more vowel letter.. a <u> followed by a vowel or a consonant+vowel combination.

i. it blocks their application and applies instead resulting in tense vowels in an environment where some of the laxing rules. The rule does not have any exceptions. A very important feature of this rule is that it overrides the laxing rules. i. decision. let us note that there is a further restriction on all laxing rules. it requires that the stressed vowel letter followed by CiV be tense. Before continuing to a rule which works just in the opposite direction as compared to the ones we have just discussed. while CiV laxing enforces the opposite. revision. even if one of the laxing rules could apply. <e> nucleic spontaneity simultaneity deity <i> = <y> variety diet sobriety anxiety <o> heroic stoic echoic poet 186 .Chapter 12 idiom. still. namely. Syria. dominion.e. their stressed vowel (underlined) is tense. That is. are non-laxable. as explained in Chapter 3. that vowels occurring before another vowel. <a> prosaic archaic laity mosaic CiV tensing Recall that the rule of CiV tensing is the mirror image of CiV laxing in two senses: firstly.. should apply. they have to remain tense even before suffixes. Secondly.e.. it applies to all vowel letters except <i> or <y> – remember that CiV laxing only applies to these. prevocalic vowels. in most cases Trisyllabic Laxness. The following examples could as well be subject to either Trisyllabic Laxness or Laxing by ending. Prevocalic Tenseness is stronger than any of the laxing rules.

). for the latter we have already mentioned that it is an exception to all the laxing rules. this cannot be the result of CiV laxing as that rule only applies to the vowel letter <i/y>! Exceptions to CiV tensing include words like Daniel. it is not necessary to indicate it for a vowel letter which is always pronounced as tense in free positions anyway. RP patriot etc. and also in the second-last syllable of the word. <o> are followed by CiV in spelling but they are pronounced as lax – but remember. Slovakia. For the former we noted that it is made lax in this environment. for some reason. which is pronounced as if the final silent <e> was not present at the end.Letter-to-sound rules – Part 2 <a> maniac Australia radiate <e> serious serial senior <o> notion Gloria phobia There are no examples listed for the vowel letters <i/y> and <u>. precious. that is. national. it is still lax. fusion. 187 . The rule has a few exceptions in which the vowel letters <a>. Irregular tenseness reversals There are cases where the stressed single vowel letter occurring in free graphic position is not made lax by any of the laxing rules but. <e>. special. This typically happens in the last or only syllable of a word. union. Muriel. Since CiV tensing makes a vowel tense. in free graphic position it will always be tense even if a laxing rule could apply (cf. etc.

Isolated deviating words There are a few sporadic quality deviations which are isolated in the sense that there are very few examples for them.g. study.e. <a> = // as in any. ate RP // (cf. orange. forest <u> punish. copy. GA //).. many. RP shone2 Irregular tenseness reversal in second-last syllable <a> cabin. salad. <u> = // as in bury. 1. devil <i> city. very. i. allege <i> live (v). As a result. Latin. have <e> were. they must be memorized as exceptions. abroad (cases of broadness without r). or <oa> = // as in broad. they do not belong to any of the tenseness reversal cases discussed above.Chapter 12 Irregular tenseness reversal in last or only syllable <a> are. business. 2 GA // is regular. atom <e> lemon. public Quality deviations In a great number of words the single vowel letters or vowel digraphs are not pronounced with one of their regular pronunciation values given in the tables at the beginning of this chapter. Thames (exhaustive list!). they are often called quality deviations as the graphemes deviate from their own regular qualities and take on the quality of some other vowel grapheme. give <o> gone. e. <u> = // as in busy. consider <o> body. tenant.. pity. 188 . Since these are not simple exceptions from the FPBR or the CPR. the spelling and pronunciation of these words is not predictable by any of the rules we have seen. Dublin. burial. linen.

chauvinism. souvenir. kilo. Derby. squash war. sergeant. worth. warmth call. calf. bourgeois. heard. health. wound (n)) MEMOIR-words <oi> = // memoir. pearl. dance. née. wash. path. Groups of deviating words Some quality deviations are unpredictable but much more common than the ones above and can be classified according to some pronunciation or spelling characteristics. quarter.Letter-to-sound rules – Part 2 2. example calm. alter. the place in California called Berkeley is pronounced /()/. worth. Berkeley3 bread. earn bear. 189 . dwarf. bald. world. stalk. suede. was. café. visa crepe. clique. pizza. palm. swarm. mauve. walk. psalm work. tear (v) GA clerk // and Derby // are regular. sauté SOUP-words <ou> = // soup. swan. quantity. rouge. halve. word. swear. meant. quality. reservoir Traditional spellings with <e>. Balkans ask. jealous earth. douche (also some original English words like you. last. balm. pleasant. <ea> CLERK-words BREAD-words EARTH-words BEAR-words 3 <e> =// <ea> = // <ea> = // <ea> = // clerk. fall. talk. fiancé(e) CHAUFFEUR-words <au> = // chauffeur. past. Wordsworth Foreignisms (French or Italian loanwords with spelling imitating the original) MACHINE-words CREPE-words <i> = // <e> =// machine.) Deviations due to neighbouring sounds/letters WANT-words WAR-words CALL-words ASK-words CALM-words WORK-words <a> = // <a> = // <a> = // <a> = // <a> = // <o> = // want. route. thwart. fete. régime. (A few of these groups have been mentioned in Chapter 4. ski. group. youth. class. Also. fast.

brought. tomb look. among move. wool trouble. <oo>. London. <u> LOVE-words MOVE-words LOOK-words TROUBLE-words SOUL/BOWL-words THOUGHT-words PUT-words <o> = // <o> = // <oo> = // <ou> = // love. the majority of English words. do. country. butcher. 4 GA // – cf. onion. know <ough> = // <u> = // thought. the exceptions to the Carrot-Rule in Chapter 4. full.Chapter 12 Traditional spellings with <o>. crook. bowl. come. cushion However numerous these exceptions may seem. do conform to the basic letter-to-sound rules introduced in the first half of the chapter. good. including new coinages. nought put. prove. courage4. book. mother. young <ou/ow> = // soul. 190 . shoulder. <ou>.

Oxford: Blackwell. Charles W. Ladefoged. Péter (1986/1989) Angolra hangolva. Peter (1993) A course in phonetics. (1989) The pronunciation of English: A coursebook in phonology. Budapest: Nemzeti Tankönyvkiadó. 3rd edition. Cook. (1980) An introduction to the pronunciation of English. (1992) English phonology: An introduction. London: Hodder Arnold. Ladefoged. János and Siptár. Kreidler. Cambridge: CUP. New York: Harcourt Brace Jovanovich. John (1994) English sound structure. Gimson. Giegerich. János and Siptár. Budapest: Nemzeti Tankönyvkiadó. Philip (1999) English phonetics and phonology. Carr. Vivian (2004) The English writing system.C. Odden. Nádasdy. Oxford: Blackwell. 191 . Budapest: Tankönyvkiadó. Nádasdy. Oxford: Blackwell. Kovács. An introduction.Bibliography Selected readings in English phonetics and phonology Carney. Heinz J. 3rd edition. Ádám (2006) Background to English pronunciation. London: Edward Arnold. David (2005) Introducing phonology. A. Cambridge: CUP. Peter (2001) Vowels and consonants. Harris. Oxford: Blackwell. Edward (1994) A survey of English spelling. London: Routledge. Budapest: Helikon. An introduction to the sounds of languages. Kovács. Ádám (2003) Practice book in English phonetics and phonology. Péter (2000) Újra angolra hangolva.

(1982) Accents of English. Oxford: Blackwell. John C. 192 . Volume 1. Linda (2003) Sound patterns of spoken English.Bibliography Roach. Wells. Yavaş. Cambridge: CUP. Shockey. Peter (1983/1991/2000) English phonetics and phonology: A practical course. Oxford: Blackwell. Cambridge: CUP. Mehmet (2006) Applied English phonology.

közelítő hang 2:16 articulation // artikuláció. hasonulás 7:95. váltakozás 1:8 alveolar ridge / / fogmeder 2:16 approximant // approximáns. 11:170 allophone // allofón 1:7 Alternating Stress Rule Hangsúlyváltakozás 9:131 alternation alternáció. hangképzés 1:6 ASK-words 4:59. 6:86 assimilation // asszimiláció. voice assimilation asterisk csillag 1:8 auditory phonetics / / percepciós fonetika 1:7 . hehezet 2:22.Subject index This index lists all the technical terms introduced and/or highlighted as significant in the book. // 3:34 ambisyllabic // ambiszillabikus 2:26 accent // akcentus 1:3. 10:139 accidental gap véletlen hiány 5:61 acoustics // akusztika 1:7 Adam's apple ádámcsutka 6:78 affricate // affrikáta. the number(s) following the colon indicate(s) the page(s) where the term is (first) mentioned. see place assimilation. The first number refers to the chapter. Whenever thought to be relevant and useful. zárréshang 2:15 Aitch-Dropping /h/-elhagyás 7:97. the RP pronunciation of the term is given in slashes and the Hungarian equivalent in italics. They appear in boldface in the text. 12:189 aspiration // aspiráció.

12:186 clear-L világos /l/ 2:20 CLERK-words 12:189 close vowel zárt magánhangzó 3:35 closed syllable zárt szótag 5:69 closing diphthong 3:36 cluster // mássalhangzókapcsolat/torlódás 2:24 Cockney // 1:3 12:189 Black English fekete angol 1:3 brackets. types of zárójelek típusai 1:10 BREAD-words 12:189 compensatory lengthening // pótlónyúlás 4:49. folyamatos 2:14 Broadness without r Szélesedés r nélkül 4:58 CALL-words CALM-words 4:59. 8:121 consonant chart mássalhangzós tábla 2:16 consonant deletion 7:97 content words see lexical continuant // kontinuáns. 12:189 Carrot-Rule Sárgarépa-szabály 4:56 cartilage // porc 6:78 Celtic languages // kelta nyelvek 1:2 central vowel centrális magánhangzó 3:35 centring diphthong 3:36 CHAUFFEUR-words 12:189 CiV Laxing // CiV Lazítás 3:42. 12:189 4:59. 54 complementary distribution // kiegészítő eloszlás 1:8 completely unstressed see zero stress compound stress összetett szavak hangsúlyozása 7:99. 12:185 194 .Subject index auto-stressed // önhangsúlyos 8:119 back vowel hátulképzett magánhangzó 3:35 backing diphthong 3:36 b-dropping b-elhagyás 11:169 BEAR-words CiV Tensing CiV Feszesedés 3:42.

6:81 digraph // digráf. kettőshangzó 3:31 12:189 egressive airstream / / egresszív/kilégzett/ exspirációs levegőáram 1:6 extrametrical // extrametrikus 9:125 falling tone ereszkedő/eső dallam 10:147 fall-rise tone ereszkedő-emelkedő dallam 10:150 195 . időtartam 3:31 Early Stress Requirement Korai hangsúly követelménye 8:113 EARTH-words 12:189 cross-word palatalization 7:96 cultural relatedness kulturális kapcsolat 1:2 dark-L sötét /l/ 2:20 degrees of stress hangsúly fokozatai 8:109 Derivational Secondary Stress Derivációs Másodlagos Hangsúly 8:114 devoicing // zöngétlenedés 2:24. kettősbetű 11:163 diphthong // diftongus.Subject index contraction // összevonás 7:102 contrast in intonation 10:144 controversy of length marking 3:31 coronal // koronális 2:13 covered position fedett helyzet 12:180 Covered Position Rule 12:181 CPR see Covered Position Rule CREPE-words diphthongs of RP RP diftongusai 3:35 dislocated tonic elmozdított tonik 10:143 downdrift lecsúszás 10:146 DRAMA-words 4:59 dropping of <b> b-elhagyás 11:169 dropping of <g> g-elhagyás 11:169 dropping of <h> h-elhagyás 11:170 duration // hosszúság.

kettőzött mássalhangzó 1:11.Subject index Finno-Ugric // finnugor 1:1 flap see tap flapping see tapping foot láb 7:99. rokonság 1:1 Gimsonian system // Gimson-féle átírás 1:9 glide félmagánhangzó. 10:140 fortis // fortisz képzésű 6:81 FPBR see Free Position Basic Rule free position szabad helyzet 12:180 Free Position Basic Rule 12:181 fricative // frikatíva. réshang. viszonyszó 10:139 graphic position grafikus helyzet 12:180 groups of deviating words 12:189 . siklóhang 2:15 glottal // glottális. 11:153 General American 1:3 196 half-close vowel félig zárt magánhangzó 3:35 genetic relatedness genetikai kapcsolat. 6:87 glottis // hang(szalag)rés 2:16. hangszalaghang 2:13 glottal reinforcement glottális megerősítés 6:87 glottal replacement glottális behelyettesítés 6:87 glottalization glottalizáció 2:24. spiráns 2:15 Fricative Devoicing Frikatívazöngétlenedés 7:96 front vowel // elölképzett magánhangzó 3:35 fronting diphthong 3:36 full vowel teljes magánhangzó 3:33 function word see grammatical function word see strong forms function word see weak forms GA see General American g-dropping g-elhagyás 11:169 geminate // gemináta. 6:79 grammatical differences between British and American English 1:4 grammatical/function word funkciószó.

8:108. hanglejtés. ajakhang 2:13 labio-velar // labioveláris 2:17 larynx gégefő 6:78 lateral laterális 2:14 Latinate see stress-fixing lax laza 3:33 197 Jonesian notation // Jones-féle átírás 1:9 information types determining stress placement 9:123 integrated suffix integrált szuffixum 9:131 International Phonetic Alphabet nemzetközi fonetikai ábécé 1:9 intonation intonáció. hanghézag 3:43 hiatus-filler hiátustöltő 7:95 high rising tone magas emelkedő dallam 10:149 high vowel magas nyelvállású magánhangzó 3:35 homorganic // homorganikus 5:71 Hungarian terminology for vowels 3:36 Hungarian vowels 3:37 iambic reversal see stress shift iambic rhythm // jambikus ritmus 8:113 Iambic Secondary Stress Jambikus Másodlagos Hangsúly 8:115 ill-formed rosszul formált 1:8 Indo-European // indoeurópai 1:1 labial // labiális. 7:93 IPA see International Phonetic Alphabet . beszéddallam 1:6. 10:138 intonation pattern see tune intonation phrase intonációs frázis 10:139 Intrusive-R // intrúzív/betoldott r 2:19.Subject index half-open vowel félig nyílt magánhangzó 3:35 hard palate / / kemény szájpadlás 2:16 head fej 10:140 heavy syllable see light hiatus // hiátus.

12:184 laxing processes lazítási folyamatok 3:39. tartalmas szó 10:139 lexicon // lexikon. 12:185 laxing ending lazító végződés 3:41. hangkötés 7:95 light/heavy syllable könnyű/nehéz szótag 9:124 Linking-R kötő r 2:19. szókincs 1:12 liaison // liaison. 11. 12:183 L-darkening l-sötétedés 2:21. folyékony hang 2:15 LOOK-words MACHINE-words low vowel alacsony nyelvállású magánhangzó 3:35 low-starting diphthong 3:36 L-vocalization l-vokalizáció 2:22 12:189 magas (hangrendű) 3:36 main stress főhangsúly 8:110 Main Stress Rule Főhangsúlyszabály 9:125 Main Stress Rule for adjectives Főhangsúlyszabály melléknevekre 9:130 Main Stress Rule for nouns Főhangsúlyszabály főnevekre 9:128 Main Stress Rule for verbs Főhangsúlyszabály igékre 9:126 major stress erős hangsúly 7:98. 7:93 liquid likvida. hangosság 8:109 LOVE-words 12:190 low rising tone alacsony emelkedő dallam 10:148 198 . 3:41. 7:91 lenis // lénisz képzésű 6:81 letter-to-sound rules betű-hang szabályok 1:11. 8:109 Major Stress Deletion erős hangsúly törlése 8:115 manner of articulation képzési mód 1:6 manners of articulation of consonants mássalhangzós képzési módok 2:14 12:190 loudness hangerő. 12 lexical palatalization lexikális palatalizáció 11:167 lexical/content word fogalomszó.Subject index Laxing by free U Lazítás szabad helyzetű U előtt 3:42.

Subject index mély (hangrendű) 3:36 MEMOIR-words neutral prefixes semleges prefixumok 9:126 neutral tonicity semleges tonicitás 10:142 new information in intonation 10:140 noncontinuant nem kontinuáns/folyamatos 2:14 non-laxable nem lazítható 3:38 non-neutral prefixes nem semleges prefixumok 9:127 non-neutral see stress-fixing non-productive nem produktív 3:39. 8:117 non-rhotic // nem rotikus. r-elhagyó 2:17 noun-verb pairs főnév-ige párok 9:130 nucleus // (dallam)csúcs 10:139 obstruent obstruens. tiszta magánhangzó 3:30 monophthongization monoftongizáció 4:49 monophthongs of RP RP monoftongusai 3:35 morpheme identity szóelemzés elve11:152 morpheme structure condition morfémaszerkezeti megkötés 3:40 morphology morfológia. félhangsúly 8:111 monophthong // monoftongus. zörejhang 2:15 occasional cross-word palatalization 7:96 open syllable nyílt szótag 5:69 open vowel nyílt magánhangzó 3:35 12:189 mid vowel középső nyelvállású magánhangzó 3:35 minimal pair minimális pár 1:7 minor stress gyenge hangsúly. alaktan 1:11 MOVE-words 12:190 MSR see Main Stress Rule narrow diphthong szűk diftongus 3:36 nasal // nazális 2:14 nasals nazális mássalhangzók 2:15 199 .

helyesírás 1:11 palatalization palatalizáció 7:96. hangelrendezés 5:63 phrasal stress szócsoportok/ szószerkezetek hangsúlyozása 7:99 pitch // hangmagasság 8:108. 11:167 palatalization by <i> 11:167 palatalization by <u> 11:168 palate see hard palate phoneme // fonéma 1:7 phonemic principle // kiejtés/hangjelölés elve 11:152 phonetic classes of vowels magánhangzók fonetikai osztályai 3:30 phonetics // fonetika 1:7 phonological classification of vowels // magánhangzók fonológiai osztályozása 3:32 phonology // fonológia 1:7 phonotactic restrictions fonotaktikai megkötések 5:69 phonotactics // fonotaktika.Subject index optional consonant deletion 7:97 oral // orális 2:14 oral stop orális zárhang 2:15 orthography // ortográfia. 10:145 pitch range hangmagasságterjedelem 10:145 place assimilation képzési hely szerinti hasonulás 7:96 place of articulation képzési hely 1:6 places of articulation of consonants mássalhangzós képzési helyek 2:13 places of articulation of vowels magánhangzók képzési helyei 3:34 plosive // explozíva 2:15 post-stress syncope hangsúlyt követő szinkópa 5:76 pre.and postvoicing elő.és utózönge 6:82 Pre-cluster Laxness Lazítás mássalhangzócsoport előtt 3:42 200 .

12:179 schwa // svá 1:9 201 . hátraható 6:83 rhotic accent // rotikus/ rotacizáló/R-ejtő akcentus 2:17 rhyme // rím 5:74 rhyme weight rímsúly 9:124 rhythm rule see stress shift rhythmic beat ritmikus ütem 7:98 Rhythmic Stress Deletion Ritmikus Hangsúlytörlés 10:141 Rhythmic Stress Shift Ritmikus Hangsúlyeltolódás 10:142 rounded vowel ajakkerekítéses magánhangzó 3:35 RP see Received Pronunciation rule szabály 1:8 SAUCE-words 12:190 quality deviation 12:182. főhangsúly 8:110 productive // produktív 3:39. 7:93 Received Pronunciation elfogadott kiejtés 1:3 reduced vowel // redukált/ sorvadt magánhangzó 3:32 regressive // regresszív. 188 4:59. előreható 6:83 prominence of the left edge 8:113 prominence of the right edge 8:113 prosodic // see suprasegmental pulmonic airstream // pulmonikus levegőáram 1:6 PUT-words quantity-sensitive hosszúságérzékeny 9:124 R-dropping r-elhagyás 2:17. 8:117 progressive // progresszív. 12:186 primary stress elsődleges hangsúly.Subject index Pre-fortis Clipping Rövidülés fortisz előtt 6:88 pre-head elő-fej 10:140 Pre-R Breaking Törés r előtt 4:46 Pre-R Broadening Szélesedés r előtt 4:52 pre-stress syncope hangsúlyt megelőző szinkópa 5:76 pre-stressed elő-hangsúlyos 8:119 Prevocalic Tenseness// Prevokalikus Feszesedés 3:43.

7:98. 12:177 smoothing // simulás 4:49 soft palate lágy szájpadlás 2:16 sonorant // szonoráns.Subject index secondary stress másodlagos hangsúly. zengőhang 2:15 sonority // szonoritás. 118 stress-marking conventions 8:112 stress-neutral suffix hangsúlyra nézve semleges szuffixum 8:117. hangzósság 2:15 sonority peak szonoritási csúcs 5:64 Sonority Principle Szonoritási Elv 5:65 202 sonority scale szonoritási skála 2:15 SOUL/BOWL-words 12:190 sound correspondence hangmegfelelés 1:3 sound value hangérték 12:177 SOUP-words 12:189 spelling helyesírás 1:10 spelling differences between British and American English 1:4 standard varieties standard/ köznyelvi változatok 1:2 stem szótő 1:12 stop zárhang 2:14 stress hangsúly 1:6. sziszegő-susogó hang 2:17 single consonant letters 11:153 single vowel letters 1:11. mellékhangsúly 8:110 secondary stress assignment másodlagos hangsúlykijelölés 8:114 segment szegmentum 1:6 segmental feature szegmentális jegy 1:6 self-stressed see auto-stressed sentence stress see nucleus short-long vowel pairs rövidhosszú magánhangzó-párok 3:32 sibilant // szibiláns. 8:109 stress clash hangsúlyütközés 8:113 stress shift hangsúlyeltolódás 8:116 stress-fixing hangsúlyrögzítő 8:117. 118 stress-sensitive affixes hangsúlyérzékeny affixumok/toldalékok 9:136 .

toldalék 1:12 suprasegmental feature szupraszegmentális/prozódiai jegy 1:6. hármashangzó 3:31 Trisyllabic Laxness // Harmadszótagi Rövidülés 3:39. hármasbetű 11:163 triphthong // triftongus. félhangsúly 8:111. 5:66 Syllabic Consonant Formation 5:75 syllabification // szillabifikáció. mondattan 1:11 tail farok 10:140 tap. 10:146 tone group dallamhordozó 10:139 tone-unit dallamhordozó 10:139 tonic placement 10:142 transcription transkripció. 105 suffix // szuffixum. szótagolás 9:133 syllable boundary szótaghatár 2:25 syllable weight szótagsúly 9:124 syllable-timed szótagmértékes 7:99 syncope // szinkópa 5:76 syntax // szintaxis. érintőhang 2:23 tapping. 12:190 tone // tónus 8:109. flap egyperdületű/legyintett r. 120 thirteen men rule see stress shift THOUGHT-words 4:59. fonetikus átírás 1:9 trigraph // trigráf. 7:92 tense feszes 3:33 tenseness reversal 12:182 tertiary stress // harmadlagos hangsúly.Subject index stress-timed hangsúlymértékes 7:100 strong forms of function words funkciószavak erős alakjai 7:100. flapping 2:24. 12:183 TROUBLE-words 12:190 tune dallam 10:146 typological relatedness nyelvtipológiai kapcsolat 1:2 203 . 8:108 syllabic consonant // szótagalkotó mássalhangzó 2:22. 7:98.

12:177 vowel reduction magánhangzóredukció/sorvadás 7:100 vowel shift magánhangzóeltolódás 3:37 vowels magánhangzók 3:30 WANT-words WAR-words 12:189 12:189 weak forms of function words funkciószavak gyenge alakjai 7:100. 105 West Germanic // nyugat-germán 1:1 wide diphthong széles diftongus 3:36 WORK-words 12:189 . 11:173 voicing zöngésség.Subject index underived secondary stress deriválatlan másodlagos hangsúly 8:114 ungrammatical agrammatikus 1:8 unrounded vowel réses magánhangzó 3:35 Uralic // uráli 1:1 utterance megnyilatkozás 2:28 velar // veláris 2:13 Velar Softening Veláris Puhulás 11:171 velarized velarizált 2:20 velum see soft palate verbal prefixes igei prefixumok 9:126 vibration (of the vocal cords) hangszalagok rezgése 6:79 vocabulary differences between British and American English 1:4 vocal cords / / hangszalagok 2:16. 7:95 voiced // zöngés 1:6. 102. 6:78 vocal folds see vocal cords voice assimilation zöngésségi hasonulás 6:83. 6:80 204 zero stress hangsúlytalan 8:111 yod-dropping j-elhagyás 5:64. zönge 2:13 vowel digraphs magánhangzódigráf 1:11. 71. 6:80 voiceless zöngétlen 1:6.