You are on page 1of 33


discussions, stats, and author profiles for this publication at:

Historical Syntax & Synchronic Morphology: An

Archeologist's Field Trip

Chapter January 1971


96 166

1 author:

T. Givn
University of Oregon


Some of the authors of this publication are also working on these related projects:

"The Story of Zero" (book; Amsterdam: J. Benjamins, 2017) View project

All content following this page was uploaded by T. Givn on 29 February 2016.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.



CHAPTER 1: Historical syntax and synchronic morphology: An archaeologist's field trip*

1. Introduction [FN 1,2]

This chapter constitutes an attempt to explain the genesis of bound--affixal--morphology,

both derivational and inflectional. I propose that in order to understand why the synchronic
morphology of a language is the way it is, one must consider the syntactic order and clausal
structure that prevailed at the time when that morphology arose. I will further attempt to show how
the synchronic syntax of a language constrains the possible pathways through which its morphology
may arise. And, throughout, I will also illustrate the converse--how one can use the synchronic
morphology of a language to reconstruct earlier diachronic stages of its syntax.[FN 3]
In approaching this topic, one must first re-consider Saussure's dogma of segregation
between synchrony and diachrony as it applies to grammar.[FN 4] In this, one of the less-surprising
observations coming out of this paper is that diachronic change in syntax quite often resembles
Chomsky's (1965, ch. 3) synchronic syntactic transformations.[FN 5]

2. Syntactic change in the verb phrase

2.1. Bantu verb affixes

In conformance with their current VO word-order, core-Bantu languages, exhibit the order
V-COMP in their verb phrase. Further, the most common Bantu verb morphology consists of the
following sequence:

(1) SP-TAM-OP-verb-VS

Whereby SP is the obligatory subject pronominal agreement prefix, TAM one or more tense-aspect-
modal prefixes, OP an optional object pronoun, and VS the various verbal suffixes.[FN 6]
In an earlier paper (Givn 1971b; see ch. 5, below), I surveyed evidence suggesting that
both the T-A-M prefixes and verbal suffixes of core-Bantu arose historically from main verbs

governing clausal complements. I further suggested that while the grammaticalization of modal-
aspectual verbs into T-A-M prefixes in Bantu is relatively recent and indeed still ongoing, the
grammaticalization of semantically-similar main verbs into verb suffixes must have occurred much
earlier, at some pre-Core-Bantu stage. Lastly, I suggested that the best explanation for the post-
verbal position of the core-Bantu verb suffixes, given the current VO, V-COMP order of Bantu
languages, lies in the hypothesis that at the time those suffixes arose from main modal-aspectual
verbs, the syntactic order of Bantu languages was OV, COMP-V. The differential position of the T-
A-M prefixes and verb suffixes in Bantu may thus be ascribed to a historical change in word-order--
OV to VO, thus COMP-V to V-COMP. That is:

(2) earlier COMP-V syntax earlier suffixal morphology


COMP ===>

(O) V V (O) V-suffix

(3) later V-COMP syntax later prefixal morphology


COMP ===>

V V (O) prefix-V (O)

Another fact that seemed at the time to support this hypothesis was the current position of
anaphoric object pronouns as prefixes in the core-Bantu verb (1), contrasting with the post-verbal
(VO) position of object nouns. A purely synchronic account would posit a syntactic transformation,
as in e.g. Swahili:

(4) synchronic nominal order (VO) synchronic pronominal order (OV):

ni-li-ona kitabu ===> ni-li-ki-ona
1s-PA-see book 1s-PA-it-see
'I saw a book' 'I saw it'

A similar suggestion may be made about Romance languages, e.g. Spanish:[FN 7]

(5) v un libro ===> lo-v

see/PA/1s a book it-see/PA/1s
'I saw a book' 'I saw it'

2.2. Amharic object pronouns

The current clausal word-order in Amharic is a rather rigid OV, COMP-V. Evidence from
Geez and other Semitic languages, however, makes it clear that the older syntactic order of pre-
Amharic, indeed of all the Ethiopian Semitic languages prior to contact with the Cushitic SOV
substratum, was VO, V-COMP. Object pronouns in Amharic, like those of Geez and other non-
Ethiopian Semitic languages, are suffixal, conforming to the old pattern of Geez, Hebrew, and
Arabic, and thus the exact converse of Bantu and Romance:

(6) synchronic nominal order (OV) synchronic pronominal order (VO)

vssu b-mkina ma ===> vssu ma-bb-(a)t
he in-car came he came-in-it
'he came in a car' 'he came in it'

Bach (1968) used this inverse position of pronominal (VO) vs. nominal (OV) objects in
Amharic to argue that Amharic still has an 'underlying', 'deep' synchronic word-order VO, a
proposal that required a 'synchronic' re-ordering transformation in order to obtain the surface OV
order. Rather than treat the discrepancy as a synchronic distortion, the frozen VO order of
pronominal suffixes in Amharic merely harkens back to the older VO word-order of Geez and
The rest of Bach's arguments reflect the ordered-rules Zeitgeist of the era, in both
phonology and syntax. Within that formal approach, one endeavored show that synchronic
syntactic rules, such as gapping, extraposition, topicalization and focusing, 'worked better' if they
were ordered before the historically-attested word-order change--now re-christened as a synchronic

transformation. The well-documented diachronic VO-to-OV change in Amharic was thus considered
as part of its synchronic syntax.[FN 8]

2.3. Romance tense-aspect affixes

A close parallel to the Bantu verb affixes, above, may be seen in Romance languages, where
the older tense-aspect conjugation is suffixal, suggesting a COMP-V, OV word-order at the time of
cliticization. Thus in Spanish, both the imperfective-past and the old future suffixal conjugations
can still be traced back to older main verbs, most likely ir 'go' and haber 'have', respectively; as in:

(7) person imperfect-past future

====== ====================== ======================
1s: compra-ba 'I used to buy' comprar- 'I will buy'
2s: compra-bas 'you used to buy' comprar-hs 'you will buy'
3s: compra-ba 's/he used to buy' comrar- 's/he will buy'
1p: comp-bamos 'we used to buy' comprar-mos 'we will buy'
3p: compra-ban 'they used to buy' comprar-n 'they will buy'

In contrast, the more-recently evolving tense-aspect markers in Spanish are pre-verbal,

conforming to the current VO, V-COMP word-order. In this way, the new perfect (from haber
'have') and future (from ir 'go') are on their way to creating a new complex prefixal conjugation,
as in:

(8) person perfect future

===== =========== ===========
1s h comprado voy a comprar
2s hs " vs " "
3s h " v " "
1p hemos " vamos a comprar
3p hn " vn " "

2.4. Negative affixes in French, English and Ute

Double negation in French involves the old negative prefix ne that has largely disappeared
in the spoken language, and a more vigorous, erstwhile emphatic, noun-derived suffix, most
generally pas 'step', but in limited contexts also personne 'person', rien 'thing' or point 'point'. The
most likely origin of the prefixal ne is a negative-implicative verb such as *negere 'deny', 'negate'.
Its prefixal position conforms to proto-Romance VO syntax, as does the suffixal position of the
noun-derived suffixes:

(9) je ne te vois pas 'I don't see you'

je ne vois rien 'I don't see anything'
je ne vois personne 'I don't see anybody'

The very same process occurred earlier in English, also during VO syntax, where the current
post-verbal negative marked 'not' is a contraction of the earlier emphatic ne-ought 'no-thing', in a
double-negation construction reminiscent of French, schematically:

(10) she ne-saw ne-ought 'she didn't see anything'

As in colloquial French, the unstressed pre-verbal ne- got zeroes out, while the stressed emphatic
ne-ought survived. When 'do', 'have', 'be' and the modals became auxiliaries, the erstwhile emphatic
ne-ought > not survived as the de-marked negation marker, much like the post-verbal pas in
French. It's pre-verbal position in English is the mere consequence of the auxiliaries' position, where
not is a relic of its earlier post-verbal (VO) position.
A mirror-image of this may be seen in Ute, until recently an OV language, where the
negative intensifier is prefixal, may still appear as a separate pre-verbal word, and conforms to the
still common OV order, while the old negative verb suffix -wa (or -'wa) is small, bound, and replete
with morpho-phonemic variation, as in:[FN 9]

(11) a. kach-u-m pu-nikya-wa '(I) don't see you' (present)

NEG-2s see-NEG

b. kachu-'u pu-nikya-na '(I) didn't see him/her' (recent past)

NEG3s see-PA/NEG

c. kach-amu- pu-nikya-pu--a '(I) did see them' (remote past)

NEG-3p see-REM-NEG

d. kachi-n pu-nikya-vaa-'wa-ni '(s/he) will not see me' (future)


While Ute is currently a flexible-order language, the order of the negative affixes around the
verb conforms to the earlier OV order--provided one grants that the prefixal kach is de-nominal
and the suffixal -wa is de-verbal. The likely verbal etymology of the old negative suffix -wa/-'wa
is the suppletive negative of -ga 'have', -'a 'lack'. This is seen in (11c) above, where the remore-pas
suffix -pu--ga changes to -pu--a in the negative.

2.5. Incorporated objects in VP nominalizations

One of the best evidence for earlier word-order are nominalized verb phrases in which the
object is incorporated into the verb. In English, currently a rigid SVO language, the most common
order in such nominalizations is OV, as in:

(12) can-opener, ditch-rider, deer-hunting, grave-digging

This OV order in English VP nominalizations harkens back to the earlier OV order in Indo-
European and Germanic (Lehmann 1969, 1971a, 1971b).
The converse case may be seen in Akkadian, a VSO Semitic language that changed to OV
due to contact with its Sumerian substratum, a case similar to that of the Ethiopian Semitic
languages. Clausal word-order in Akkadian is OV, but VP nominalizations retain the older Semitic
VO order, as in (Bucellati 1970):

(13) finite clause (OV) nominalized VP (VO)

======================= ==========================
a. ina idi amma-iya illaku alikut idi amma-niya
as side/of troops-my go/3p goers side/of troops-my
'they go alongside my troops' 'goers at the side of my troops'

b. anna sharr-im ikrub ikrib sharr-i

for king-GEN pray/3s prayer/of king-GEN
'he prays for the king' 'prayer for the king'

c. abull-am inassar massaar abul-im

gate-gate-ACC guard/3s guardian/of gate-GEN
'he guards the gate' 'guardian of the gate'

3. Syntactic change in the noun phrase

3.1. Swahili relative pronouns

Currently, core-Bantu languages have the clausal word-order of SVO, with REL-clauses
following the head noun. If they have distinct relative pronouns at all, they are most commonly
derived from demonstratives and occupy the position directly following the head noun, i.e. clause-
initial within the REL-clause. Thus, for example, in Bemba (Givn 1969):[FN 10]

(14) main clause REL-clause

===================== ==========================
a. umuana a-a-ishile umuana [uyo a-a-ishile]...
child 3s-PA-come/MB child REL 3s-PA-come/MB
'the child came' 'the child who came...'

b. n-a-mweene umuana umuana [uyo n-a-mweene]...

1s-PA-see/MB child child REL 1s-PA-see/MB
'I saw the child' 'the child that I saw...'

In Swahili, a rather complex situation is found whereby the REL-pronoun is a bound affix
that can occupy three distinct positions, depending on the tense-aspect. In the habitual -a-, the REL-
pronoun is a verb suffix, as in:

(15) main clause REL-clause

===================== =======================
a. mwana a-a-la mwana [a-a-la-ye]...
child 3s-HAB-eat child 3s-HAB-eat-REL
'the child ate' 'the child who ate...'

b. n-a-mu-ona mwana mwana [n-a-mu-ona-ye]...

1s-HAB-3s-see child child 1a-HAB-3s-see-REL
'I see the child' 'the child I see...'

In the perfect -me-, the verb -amba 'say' is recruited as carrier of the suffix REL-pronoun,
and is then placed between the head noun and the REL-clause. In addition, monosyllabic verbs must
be preceded by the infinitive prefix ku-, as in:

(16) main clause REL-clause

===================== =======================
a. muana a-me-ku-la muana [amba-ye a-me-ku-la]...
child 3s-PERF-INF-eat child say-REL 3s-PERF-INF-eat
'the child has eaten' 'the child who has eaten...'

b. ni-me-mu-ona muana muana [amba-ye ni-me-mu-ona]...

1s-HAB-3s-see child child say-REL 1a-PERF-3s-see
'I have seen the child' 'the child that I have seen...'

In the past -li-, the REL-pronoun is sandwiched between the tense marker and the infinitive prefix
(in monosyllabic verbs), as in:

(17) main clause REL-clause

===================== =======================
a. mwana a-li-ku-la mwana [a-li-ye-ku-la]...
child 3s-PA-INF-eat child 3s-PA-REL-INF-eat
'the child ate' 'the child who ate...'

b. ni-li-mu-ona mwana mwana [ni-li-ye-mu-ona]...

1s-PA-3s-see child child 1s-PA-REL-3s-see
'I saw the child' 'the child that I saw...'

Finally, in the future -ta-, one finds a near-identical situation as in the past (17), with one added
detail: In the REL-clause, the future marker -ta- appears as -taka-, i.e. the verb 'want':

(18) main clause REL-clause

===================== =======================
a. mwana a-ta-ku-la mwana [a-taka-ye-ku-la]...
child 3s-FUT-INF-eat child 3s-want-REL-INF-eat
'the child ate' 'the child who ate...

b. ni-ta-mu-ona mwana mwana [ni-taka-ye-mu-ona]...

1s-FUT-3s-see child child 1s-want-REL-3s/O-see
'I will see the child' 'the child that I will see...'

The only way to make sense of this baffling synchronic situation is by reconstructing its
diachrony. First, one notes that the Swahili perfect aspect prefix -me- originates as the verb 'finish',
-maa in Swahili, -mala in Bemba, but its modified-base (MB, perfective) form, lost in Swahili but
still extant in Bemba, is -meele. Second, one notes that the past -li- originates as the defective
copular verb -li-, still found in Bemba and also used there as one of the past tenses. Finally, the verb
'want' -taka takes an infinitive verbal complement in Swahili, as in:

(19) a. n-a-taka ku-la 'I want to eat'

1s-HAB-want INF-eat

b. n-a-taka ku-mu-ona muana 'I want to see the child'

1s-HAB-want INF-3s-see child

A relatively transparent diachronic scenario may now be suggested (Givn 1969):

(a) The Swahili REL-pronoun was originally a verb suffix, as can be still seen in (15), as well as
when the 'carrier' -amba 'say' is used (16).
(b) The de-verbal past marker -li- and future marker -ta- carried that REL-pronoun with
them into its current 'sandwich' position when they grammaticalized as tense-aspect
(c) The verb-derived perfect marker -me- could not likewise carry the suffix REL-pronoun
because this is precluded by the modified-base (MB) suffixal form. That form was later
lost in Swahili, preserved only in the grammaticalized -me-.

(d) The post-verbal morphemic position of the Swahili REL-pronoun is a relic of an earlier SOV
word-order in pre-Proto-Bantu, where REL-clauses must have preceded the head noun.[FN
11] The simplified innovative pattern in (16) is now spreading, first as a stylistic option, to the
other tense-aspect paradigms in Swahili, slowly displacing the old V-suffix REL-pronoun
pattern, thus perhaps supporting the idea that the most natural position for REL-pronouns
(or REL-subordinators) is between the head noun and the REL-clause.[FN 12]
(f) The infinitive prefix ku- is preserved in monosyllabic verbs because of the Swahili lexical
stress pattern, which requires a penultimate stress on the verb stem.

3.2. Amharic adpositions

Amharic is a rigid SOV language with a MOD-N order, conforming to Greenberg's (1966)
suggested correlation between clausal and noun-phrase order.

(20) demonstrative adjective REL-clause

============= ========= ====================
yvh svra vru lvjj y-sbbr-ku-t wmbr
this work good boy GEN-broke-1s-3s chair
'this work' 'a good boy' 'the chair that I broke'

As noted earlier, comparative evidence from Geez and other Semitic languages suggests that
Amharic changed from VO to OV and N-MOD to MOD-N order, most likely due to contact with
its Cushitic SOV substratum. In this section I will show how that change is etched indelibly in the
petrified morphology of the language.
Amharic has both pre-positions and post-positions. The most common pre-positions, ka-,
-ba- and la-, are monosyllabic, semantically depleted, and have cognates in Geez and other Semitic
languages. Thus:

(21) b-mkina ma l-gudday ma

in-car came/3s to-business came/3s
'he came by car' 'he came for business'

Other, larger prepositions have a more complex provenance. Thus, svla 'because' probably
incorporated la- 'to', 'for'; wd 'toward' may have come from a verbal source, possibly the Arabic

wdy 'lead to'; and v nd 'like' may have been borrowed from the Cushitic Agaw v n-ta 'like-this' (R.
Hetzron, ipc).
In addition, Amharic has also developed a number of post-positions that supplement and
enrich the old semantically-depleted prepositions. This yields complex adpositional phrases such

(22) b-meda wvst ma b-bet lay all

in-field inside came/3s in-house top be/3s
'he came through the field' 'he is on top of the house'

k-bet tach all b-vssu mvknvyat nw

at-house bottom be/3s in-he reason be/3s
'he is under the house' 'it is because of him'

k-vssu gar all

at-he place be/3s
'it is near him'

The nominal origin of the Amharic post-positions is fairly transparent, given their use in
nominal constructions, much like other nouns:

(23) b-lay-u b-tach-u b-wvst-u b-bet-u

in-top-the in-bottom-the in-inside-the in-house-the
'at the top' 'at the bottom' 'in the middle' 'at/in the house'

And while there is a constraint against one preposition following another, the origin of the post-
position in genitive constructions is still hinted in expressions such as:

(24) y-bet lay y-bet tach

of-house top of-house bottom
'the top of the house' 'the bottom of the house'

The three semantically bleached prepositions b- 'in/at', l- 'to/for' and k- 'from/at' are of
course reminiscent of their short English equivalents 'at', 'on', 'to' and 'in', which are augmented

in the same way by locational nouns. Except that the English pattern conforms to the current SVO
syntax of the language, with a post-nominal genitive modifier:

(25) 'on top of the house', 'in-side (of) the house', 'at the bottom of the ocean',
'in front of the gate', 'in-stead of running', 'in the back of the theater'

The same pattern of augmentation of semantically-depleted old prepositions is found in

Bantu languages, where the three old prepositions pa- 'at', ku- 'to/at' and mu- 'in' have been more
recently augmented by locational nouns in a post-nominal genitive construction, again conforming
to the current VO syntax of core-Bantu. Thus, from Bemba:

(26) a. pa-isaamba lya nuumba 'under the house' (i-saamba 'bottom', cl. 5)
at-bottom 5/of house

b. ku-muulu wa nuumba 'on top of the house' (umu-ulu 'sky', cl. 3)

at-sky 3/of house

c. pa-kati ka nuumba 'in the middle of the house' (aka-ti 'center', cl. 12)
at-center 12/of house

d. pa-nnuma ya nuumba 'behind the house' (in-numa 'back'. cl. 9)

at-back 9/of house

e. mu-nse ya nuumba 'outside the house' (in-se 'the outside', cl. 9)

in-out 9/of house

f. pa-ntaanshi ya nuumba 'in front of the house' (in-taanshi 'face', cl. 9)

at-face 9/of house

g. ku-nshi ya nuumba 'down from the house' (in-shi 'earth', cl. 9)

to-earth 9/of house

The only thing that makes Amharic different from English and Bemba is that while the
Amharic old preposition arose during a period of VO syntax, its new post-positions have been
innovated more recently, during the current SOV, MOD-N syntax.[FN 13]

3.3. Germanic noun compounds

One of Greenberg's (1966) observations was that OV languages tend to have pre-nominal
genitive modifiers, as in, e.g.:

(27) a Amharic: y-vssu bet 'his house'

of-he house

b. Ute: mamachi kani 'the woman's house'

woman/GEN house

Such genitive constructions are often compressed into nominal compounds, preserving the original
MOD-N order, as in:

(28) a. Amharic: bet mhaf 'school' (frozen old N-MOD order)

house/GEN book

b. Ute: nu-gani 'tipi'


English is currently a rigid SVO language, but its productive noun compounding harkens
back to the old OV, MOD-N pattern of Germanic (Lehmann 1969). A group of older compounds
still preserve the old genitive -s suffix. Thus contrast:

(29) a. current pattern: roadhouse, barnyard, mailman, skylark, watermark

b. older compounds: swordsman, marksman, woodsman, craftsman, huntsman

This old GEN-N pattern also survives in German, in expressions such as (T. Vennemann, ipc):

(30) des Vaters wegen im Himmels willen an Vaters statt

the/GEN father's reasaon in heave's will at father's place
'because of (the) father' 'for Heaven's sake' 'instead of father'

The synchronic MOD-N order of Germanic noun-compounds is thus explained by the diachrony
of word-order change, from the earlier OV to the current VO, coupled with the genitive origin of
modifying nouns.

3.4. Independent pronouns and pronominal agreement

Pronominal subject and object agreement morphemes on the verb are widespread (see ch.
8, below). Often, the prefixal or suffixal slot of the pronominal agreement morphemes correspond
to the current nominal word-order, as in the Bemba (Bantu; SVO) subject agreement paradigm:

(31) n-a-ile 'I left' tu-a-ile 'we left'

1s-PA-go/MB 2p-PA-go/MB

u-a-ile 'you left' mu-a-ile 'y'all left'

2s-PA-go/MB 2p-PA-go/MB

a-a-ile 's/he left' ba-a-ile 'they left'

3s-PA-go/MB 3p-PA-go/MB

In Semitic, Indo-European and other languages, different tense-aspects have different subject
pronominal agreement paradigms, and some paradigms have a mixed prefixal-suffixal agreement.
Thus in Hebrew (VSO > SVO), the irrealis paradigm is mostly prefixal, with some suffixal
augmentation for feminine and plural:

(32) irrealis-SG irrealis-PL(suffixal)

=================== ======================
'e-lekh 'I will go' ne-lekh 'we will go'
1s-go/IRR 1p-go/IRR

te-lekh 'you (m.) will go' te-lkh-u 'y'all (m.) will go'
2-go/IRR 2-go/IRR-pm

te-lkh-i 'you (f.) will go' te-lekh-na 'y'all (f.) will go'
2-go/IRR-sf 2-go/IRR-pf

ye-lekh 'he will go' ye-lkh-u 'they (m.) will go'

3-go/IRR 3-go/IRR-pm

te-lekh 'she will go' te-lekh-na 'they (f.) will go

3f-go/IRR 3f-go/IRR-pf

The perfect/past paradigm, on the other hand, is strictly suffixal:

(33) perfect/past-SG perfect/past-PL

=================== ======================
halakh-ti 'I went' halakh-nu 'we went'
go/PERF-1s go/PERF-1p

halakh-ta 'you (m.) went' halakh-tem 'y'all (m.) went'

go/PERF-1sm go/PERF-2pm

halakh-t 'you (f.) went' halakh-ten 'y'all (f.) went'

go/PERF-2sf go/PERF-2pf

halakh 'he went' halkh-u 'they went'

go/PERF/3sm go/PERF-3pm

halkh-a 'she went' ------------------


A similar mix of predominantly prefixal subject agreement with some suffixal

augmentation is found in Topotha (E. Nilotic; VSO):

(34) present-SG present (PL)

======================= ===========================
a-lozi ayong 'I am going' ki-lozi iswa 'we are going'
1s-go/PR I 1p-go/PR we

i-lozi iyong 'you are going' i-lozi-te ees 'y'all are going'
2-go/PR you 2-go/PR-PL y'all

e-lozi inges 's/he is going' e-lozi-te ikes 'they are going'

3-go/PR s/he 3-go/PR-PL they

The presence of both prefixal and suffixal pronominal-agreement conjugations in the same
language, or within the very same paradigm, begs for a diachronic explanation. Such an explanation
relies on the fact that in many languages the pronominal-agreement morphemes are transparently
derived from independent pronouns. Put together as a tentative hypothesis:[FN 14]

(a) Pronominal agreement morphemes arise via the cliticization of independent pronouns.
(b) When independent pronouns cliticized, they do so at their current word-order position.
(c) Word-order change may later occur.
(d) The new generation of pronominal-agreement morphemes then cliticize in conformance
with the new word-order.

4. A wild safari through the jungle of English derivational morphology

In the preceding sections, I tried to show how bound grammatical (inflectional) morphology,
in particular its morphotactic position, is a rich repository of data by which earlier diachronic
syntactic states, and subsequent changes, can be reconstructed, mostly through the method of
Internal Reconstruction. In this section, I will suggest that the same diachronic approach can also
be applied to derivational morphology. I will use primarily English data.

4.1. Causative suffixes

English has a plethora of causative derivational suffixes that apply to adjectives, nouns or
verbs, converting them into transitive verbs of causation. Some of these suffixes harken back to the
older Germanic base of English; others came in with the later influx of Romance vocabulary. Some
of these causative suffixes have known etymologies that suggest a causative verb such as 'cause',
'make' or 'get' governing a verbal complement. Thus, for example, the Latin-derived causative suffix
-fy probably came from the verb facere 'make', and applied originally to adjectives, as in:

(35) liquid > liqui-fy pure > puri-fy

solid > solid-fy rare > rarify
horrid > horrify just > justify
*pace > faci-fy simple > simpli-fy
intense > intensi-fy *rect > recti-fy
*dign > digni-fy ample > ampli-fy

Likewise the suffix -ize, with no firm etymology:

(36) public > public-ize real > real-ize

material > material-ize clitic > clitic-ize
critic > critic-ize emphatic > emphas-ize
politic > poilic-ize normal > normal-ize

The Germanic causative suffix -en, with no firm etymology, seems to apply to both nouns and
adjectives, as in:

(37) strength > strength-en weak > weak-en

length > length-en wide > wid-en
sharp > sharp-en
dark > dark-en
black > black-en
white > whit-en
short > short-en

The suffixal position of such causative suffixes, if one assumes they came from main verbs,
harkens back to the earlier OV, COMP-V word-order of both Germanic and Romance (Lehmann

4.2. Nominalized verb-phrase compounds

English is currently a rigid SVO language. But nominalized VPs yield N-V compounds,
clearly pointing back to the older Germanic OV syntax, as in:

(38) house-cleaning house-cleaner

lion-taming lion-tamer
baby-sitting baby-sitter
man-eating man-eater

A V-N compounds pattern was initiated in Middle-English in conformance with the new VO
order, as in:

(39) pick-pocket, cut-throat, Make-peace, kill-joy, Catch-pole, spoil-sport

But the old N-V pattern has prevailed by sheer force of numbers.

4.3. Pre-verbal incorporated prepositions

The pre-verbal incorporation of prepositions in both Romance and Germanic languages

strongly hints at an earlier OV word-order (Lehmann 1971a, 1971b), although the exact mechanism
that gave rise to this pattern is yet to be elucidated. Thus consider:[FN 15]

(40) Romance Germanic

========== ============
im-press in-put
ex-press out-put
com-press up-keep
de-press down-grade
re-press in-come
ex-port out-come
im-port in-take
re-port out-take
com-port up-take
de-port up-shot

4.4. OV nominalizations and the GEN-N noun-phrase order

As suggested earlier (section 4.2.), the OV order of nominalized verb phrases in English is
a relic of the old Germanic OV syntax. While currently this pattern requires no genitive marking of
the object, genitive objects and subjects are one of the hallmarks of clausal and VP nominalization,
as in (see vol. III, chs 25, 26):

(41) finite nominalized

=========== ===================
she knew math her knowledge of math
they murdered him his murder
he died his death

As noted earlier, N-N compounds in Germanic involved, originally, the genitive marking of
the modifier noun, i.e. a GEN-N construction, which Greenberg (1966) suggested correlated to OV
clausal order. But how did this correlation come about? Lehmann (1973) suggested an abstract
principle of "harmony" between the head-modifier relation in the VP and NP. But this is neither an
explanation nor a mechanism, merely a formal-sounding recapitulation of the facts.[FN 16] A
more plausible diachronic explanation--cum mechanism--would be:

(a) In VP nominalizations, the erstwhile verb becomes the head of the derived NP, with
the erstwhile object becoming its genitive modifier.
(b) Possessive modifiers that do not arise via nominalization adopted this pattern by analogy.
(c) VP nominalizations are thus the venue through which the prevailing VP word-order is
imported into the NP as a MOD-N (OV) or N-MOD (VO) order.

4.5. The noun-to-adjective derivational suffix -ly

The German post-position -lich, related to gleich 'like' (T. Vennemann, ipc), is used to
derive adjectives from nouns in English, as in:

(42) freund-lich 'friendly', wierk-lich 'truthfull', heim-lich 'homely'

English has retained this pattern in many lexical items with the equivalent suffix -ly, as in:

(43) friend-ly, home-ly, man-ly, woman-ly, prince-ly, tiger-ly

In addition, English has reproduced the same pattern with -like, as in

(44) child-like, man-like, horse-like, goat-like

This pattern, with a post-position following the noun, no doubt harkens back to the earlier OV
word-order of Germanic.

4.6. The noun-to-verb derivational prefix en-

This Romance-derived pattern, using the preposition en- 'in', converts nouns into transitive
verbs meaning roughly 'put X into Y', with the Y being the original noun. Thus:

(45) original noun derived verb

============ ============
tomb en-tomb
circle en-circle
close en-close
balm em-balm
shrine en-shrine
rage en-rage
case en-case
danger en-danger

The pre-verbal position of en- harkens back to its original pre-nominal position, but it could have
also been reinforced by analogy with the verb-incorporated prepositions (see sec. 4.3, above).

4.7. Noun-to-noun derivational suffixes

Several old Germanic derivational suffixes in English derive new nouns from existing
nouns, as in:

(46) source noun derived noun

========== ===========
king king-dom
child child-hood

Etymologically, the suffix -dom is the old Germanic noun tuom 'judgement', still preserved in
'deem' and 'doom'. And -hood is the old Germanic noun heituz 'quality' (T. Vennemann, ipc). The
derivational pattern thus arose from genitive-noun compounds with the initial derived meanings
in (46) 'the king's judgement' and 'a child's quality'. The head noun then contracted into a
derivational suffix. And the N-suffix order harkens back to the older Germanic MOD-N, OV word-
In Bantu languages, with VO and N-MOD order, the opposite configuration can be seen:

(47) unu-ana 'child'

umu-kashi 'woman'
umu-kashi-ana 'girl' (lit. 'little-woman')

And in Ute, until recently an OV language and still retaining the GEN-N order, the
derivational pattern echoes that of Germanic (46):

(48) nu-chi 'person' kava 'horse'

tua-chi 'child' tua-chi 'child'
nu-rua-chi kava-raa-chi 'colt', 'little horse'

5. Discussion

5.1. The morphogenesis cycle and the attrition of bound morphemes

Bound morphemes are invariably much smaller than the lexical stems they are affixed to.
With time depth data, one can observe them shrink, acquire wild morpho-phonemic variation, and
eventually zero out altogether, whereby a new cycle of morphogenesis will eventually start from
scratch. But why do bound morphemes atrophy, undergoing phonetic bleaching? One tentative
answer is that in becoming bound morphemes they also undergo a parallel change of semantic
bleaching,[FN 17] ridding themselves of many specific lexical-semantic features but retaining a
few classificatory, generic ones. Thus, when the verb 'go' becomes a future marker, its spatial-
motion sense is bleached out, but its implied future-goal sense remains. When 'want' becomes a
future marker, the lexical sense of volition is bleached out but the implied sense of yet-unrealized
futurity remained. When 'have' is turned into a perfective, then perfect, then past marker, the lexical
sense of possession is bleached out but the implication of accomplishment, thus perfectivity, is
retained. And when 'fail' becomes a negative marker, the lexical sense of having tried hard is
bleached out but the negative implication vis-a-vis the complement verb is retained.
The morphogenesis cycle is specific to particular morphemes and their rise-and-attrition
time-table. This makes it unlikely that a language could be, at any given time, purely isolating,
purely agglutinating or purely polysynthetic. The most obvious exception to this are languages that
have recently started the cycle from a pidgin phase, i.e. Creoles. At least in the early stages of their
development, all their entire morphological systems spring into being at roughly the same time; that
is, in phase.[FN 18]

5.2. Do all bound morphemes come from lexical words?

In the overwhelming majority of the cases where etymological information is available,

bound morphemes turn out to have arisen from lexical words. But how would one know this in
spoken languages with no written records, where etymological information is harder to come by?
Or in a written language whose old morphology is highly-eroded with no etymology unavailable?
My methodological preference here is grounded in the normal practices of empirical science:[FN
!Find clear cases where the etymological information is available.
!Establish the pattern and elaborate it in a theoretically grounded, testable hypothesis.
!Argue the less-clear cases on the basis of the established general pattern.
!Keep looking for more etymological information that may either falsify or uphold the
How about morphology that has decayed to the point of leaving only faint traces on the
lexical stem, such as stem-vocalization patterns in Germanic or Semitic verbs? The Germanic
evidence suggest that these patterns arose from vowel harmony with now-eroded bound suffixes.
The Semitic pattern is much older and may be argued, tentatively, by analogy with the Germanic
pattern--till more evidence turns up.

5.3. Historical syntax and synchronic morphology

From the discussion thus far, it may seem that the relation between morphology and syntax
involves only linear order. Words that become bound morphemes cliticize to lexical words that are
adjacent to them, either preceding or following. This impression is misleading. Morpho-genesis is
defined, indeed constrained, by the hierarchic syntactic structure--syntactic construction--within
which a morpheme rises. A few simple examples will illustrate this point.
Tense-aspect-modal markers arise, universally, from main verbs that govern verbal
complements. The relevant syntactic construction that constrains it is that of a verb phrase, as in:

(49) go > FUT:


going to-work gonna-work


(50) be > PROG:


is work-ing is-work-ing

(51) have > PERF:


have do-ne have-do-ne

Likewise, in the genesis of causative affixes:

(52) make > CAUS:



make John sleep make-sleep John

And likewise in the derivation of nominalized object-verb compounds:

(53) VP NOUN

deer hunt deer-hunt-er


The relevant syntactic construction for the genesis of N-N compound is a noun phrase, as

(54) NP NOUN

dog/GEN house dog-house

Finally, the relevant syntactic construction for the genesis of pronominal agreement on the
verb is the entire clause:

(55) S S


he works he-works

5.4. Moral

To paraphrase an old master: "If today's bound morphemes are yesterday's lexical words,
then today's morphology is yesterday's syntax".


ACC accusative 1s 1st person singular

FUT future 1p 1st person plural
GEN genitive 2s 2nd person singular
HAB habitual 3s 3rd person singular
INF infinitive 3p 3rd person plural
IRR irrealis s singular
MB modified base (Bantu verb form) p plural
NEG negative m masculine
O object f feminine
PA past 1 1st person
PERF perfect 2 2nd person
PL plural 3 3rd person
REL relative (subordinator, pronoun)
REM remote (past)
S subject

This paper was originally published in the proceedings of the CLS #7 meeting (Givn 1971a). My
aim at the time was rather modest--explaining Greenberg's (1966) typological correlations between
clausal word-order (VO vs. OV) and bound morphology (prefixal vs. suffixal, respectively). At the
time, I had read neither F. Bopp (1833) nor H. Paul (1880) nor A. Meillet (1921) nor Jespersen
(1917), and thus set out rather innocently to re-invented grammaticalization from scratch. The
paper argued the grand theme that synchrony is largely explained by diachrony. The other major
explanatory parameters of grammar--communicative function, the neuro-cognitive processor and
language evolution--were grafted onto my evolving perspective a few years later (Givn 1979; see
chs 2,3,4, below). While the main thrust of the original paper seems sound, I have removed several
examples that seem, in retrospect, either dead wrong or irrelevant. I also dispensed with some of the
more embarrassing formal discussion of ordered transformational rules and overly abstract syntactic
The original footnote, perhaps forgiven in view of the Zeitgeist, went as follows: This trip was
inspired in part by an Analect variously ascribed to Confucius but most likely emanating from that
greatest of all time-trippers, Lao Tse; who is reported to have remarked, on the occasion of being
informed that Chinese was an isolating language: "Weep not, my children, for today's syntax is
tomorrow's morphology".
I remain indebted to Theo Vennemann, W.P. Lehmann, Robert Hertzron,, A.E. Meeussen, Benji
Wald and Larry Hyman for many comments, suggestions and criticisms of an earlier draft of the
original paper.
Primarily by the oft-maligned method of Internal Reconstruction (see vol. III, ch. 29).
See discussion in vo. III, ch. 31. At the time the original paper was written (1971), I didn't fully
appreciate how my thesis constituted a flat-out rejection of Saussuere's (1915) second idealization
dogma. Nor did I realize how muddled and self-contradictory his arguments were for a strict
segregation between synchrony and diachrony. Given the posthumous provenance of the Course,
it is not entirely clear how much of the muddle was due to Saussure himself.

The reigning syntactic formalism of the era, ordered syntactic transformations (Chomsky 1965,
ch. 3), struck me at the time as isomorphic to some of the diachronic morpho-syntactic changes I
was investigating (Givn 1967). Around the same time, Wally Chafe (1968) noted that Chomsky
and Halle's (1968) proposed ordered rules of 'synchronic' English phonology largely recapitulated
the diachrony of English sound change.
The larger Bantu verb suffixes mark either de-transitive voice (passive, middle-voice, reciprocal),
transitivity changes (causative, applicative) and the 'reversive', a species of lexicalized negation.
But smaller--thus presumably older--suffixes, primarily vowels or morpho-phonemic changes with
the i/e, u/o vowel harmony, mark an old perfective, a subjunctive and a negative. For more details
see ch. 5, below.
The whole argument has turned out to be either fallacious or highly controversial, requiring a later
recantation (Givn 1977a).
A similar approach was suggested by McCawley (1968), arguing that the 'deep', 'underlying'
word-order of English was SOV rather than the attested 'surface' SVO. At the time of the original
paper, Robert Hetzron (i.p.c.) suggested to me that Semitic languages presented a similar case,
whereby the clausal word-order of both Biblical Hebrew and Koranic Arabic was VSO but the
pronominal order in at least one conjugation was SVO, perhaps harkening to an older SVO syntactic
order. However attractive the argument seemed at the time, I was soon impelled to reject it on both
empirical and theoretical grounds (see fn. 7).
This example did not appear in the original paper, which predated the beginning of my Ute work
by 5 years. For negation in Ute see Givn (2011).
There are also tonal changes in the REL-clause, both to the subject pronoun and verbal stem.
For the SOV reconstruction of Proto-Niger-Congo, see Givn (1975), also ch. 7, below.

In a subsequent paper (Givn 1972) I argued that there was a general functional principle, in
Bantu and elsewhere, of REL-pronoun (or REL-subordinator) 'attraction' to the position between
the head noun and the REL clause, thus making their boundary clear. In retrospect, the explanation
for such 'attraction' is probably grounded in the most common diachronic sources of REL-clauses,
such as emphatic Y-moved pronoun or WH-pronouns (Givn 2009; see also vol. III, ch. 26).
Robert Hetzron (1970) gave an elaborate synchronic transformational account of the Amharic
mixed case-marking system, using the then-current ordered-rules format. His treatment was
reminiscent of Bach's (1968) and McCawley's (1968) 'deep' order cum ordered rules proposals.
Five years later, part of this hypothesis--the independent-pronoun origin of pronominal
agreement--was fleshed out in considerably more detail (Givn 1976; see ch. 8, below). The
hypothesized correlation between word-order and prefixal vs. suffixal pronominal agreement
proved much less durable; first because the order of independent--contrastive, emphatic, topic-
changing--pronouns often varies from the nominal word-order (Givn 1977a). Second, because
pronominal agreement paradigms are often imported with main verbs that grammaticalize as new
tense-aspect-modal markers, so that their morphemic position depends on where the new de-verbal
T-A-M markers cliticize (Givn 1977a). And third, because the phenomenon of second position
clitics may interact with this process independently (Givn 2012; see also vol. II, ch. 20).
More recent data from Rama, an SOV Chibchang language from Nicaragua (see vol. III, ch. 24),
suggest a possible mechanism, the elision of the core noun in prepositional phrases in either
zero anaphora or zero non-referring contexts, most likely the latter (Tibbitts 1990).
In retrospect, Lehmann's (1973) 'explanation' anticipated Chomsky's X-bar formalism.
The term 'semantic bleaching', coined by analogy with phonological bleaching, was not used
till eight years later (Givn 1979).
This argument has been fleshed out more recently in Givn (2014); see also vol. II, ch. 21.
At the time (1971), I was only seven years removed from having quit my first graduate career in
molecular biology, though my understanding of the methodology and philosophy of science was
largely implicit. For a more explicit treatment, see Givn (2005, ch. 9), as well as vol. III, ch. 29.


Bach, E (1968) "Is Amharic an SOV language?" (unpublished ms.)

Bopp, F. (1833) Vergleichennde Grammatik des Sanskrit, Send, Armenischen, Griechischen,
Lateinischen, Litauischen, Altslawischen, Gotischen und Deutschen, Berlin
Bucellati, G. (1970) A Grammar of Babylonian, Los Angeles, CA: CLA (unpublished ms.)
Chafe, W. (1968) "The Ordering of Phonological Rules", IJAL 34
Chomsky, N. (1965) Aspects of the Theory of Syntax, Cambridge, MA: MIT Press
Chomsky, N. and M. Halle (1968) The Sound Pattern of English, Cambridge, MA: MIT Press
Givn, T. (1967) "Transformations of ellipsis, sense development and rules of lexical
derivation", Technical Report, Santa Monica, CA: Systems Development Corporation
Givn, T. (1969) Studies in ChiBemba and Bantu Gramma, PhD dissertation, Los Angeles:
Givn, T. (1971a) "Historical syntax and synchronic morphology: An archaeologist's field trip",
CLS #7, University of Chicago: Chicago Linguistics Society
Givn, T. (1971b) "On the verbal origin of the Bantu verb suffixes", Studies in African
Linguistics, 2.2
Givn, T. (1972) "Pronoun attraction and subject post-posing in Bantu relativization", CLS #8,
University of Chicago: Chicago Linguistics Society
Givn, T. (1975) "Serial verbs and syntactic change: Niger-Congo", in C. Li (ed. 1975)
Givn, T. (1976) "Topic, pronoun and grammatical agreement", in C. Li (ed. 1976)
Givn, T. (1977a) "On the SOV origin of the suffixal agreement conjugation in Indo-European
and Semitic", in A. Juliand (ed.) Linguistic Studies Offered to Joseph Greenberg on the
Occasion of his Sixtieth Birthday, Saratoga, CA: Anma Libri
Givn, T. (1977b) "The drift from VSO to SVO in Biblical Hebrew: The pragmatics of tense-
aspect", in C. Li (ed. 1977)
Givn, T. (1979) On Understanding Grammar, New York: Academic Press
Givn, T. (2005) Context as Other Minds, Amsterdam: J. Benjamins
Givn, T. (2009) The Genesis of Syntactic Complexity, Amsterdam: J. Benjamins
Givn, T. (2011) Ute Reference Grammar, Amsterdam: J. Benjamins
Givn, T. (2012) "The diachrony of pronominal agreement: In Ute and maybe elsewhere",
Seminario de Complexidad Syntctica, Hermosillo, Son.: University of Sonora (ms.)

Givn, T. (2014) "Is polysynthesis a valid theoretical notion? The diachrony of complex verbs
in Ute", in M. Fortescue, M. Mithun and N. Evans (eds) A Hadbook of Polysythesis,
Oxford: Oxford University Press
Greenberg, J. (1966) "Some universals of grammar with particular reference to the order of
meaningful elements", in J. Greenberg (ed.) Universals of Language, Cambridge, MA:
MIT Press
Hetzron, R. (1970) "Toward an Amharic case grammar", Studies in African Linguistics, 1.3
Jespersen, O. (1917) Negation in English and Other Languages, Copenhagen: A.F. Host
Lehmann, W.P. (1969) "Proto Indo-European compounds in relation to other Proto-Indo-
European syntactic patterns", Acta Linguistics Hafniensia, XII
Lehmann, W.P. (1971a) "Contemporary linguistics and Indo-European studies" (unpublished
Lehmann, W.P. (1971b) "Proto-Germanic syntax", The Grammar of Proto-Germanic, Ithaca:
Cornell University Press
Lehmann, W.P. (1973) "A structural principle for language and its implications", Language, 49
Li, C.N. (ed. 1975) Word Order and Word Order Change, Austin, TX: University of Texas Press
Li, C.N. (ed 1976) Subject and Topic, NY: Academic Press
Li, C.N. (ed. 1977) Mechanisms of Syntactic Change, Austin, TX: University of Texas Press
McCawley, J. (1968) "Lexical insertion in a transformational grammar without deep structure",
CLS #4, University of Chicago: Chicago Linguistics Society
Meillet, A. (1921) Linguistique Historique et Linguistiqie Gnrale, Paris: Societ Linguistique
de Paris, 8
Paul, H. (1880) Principien der Sprachgeschichte, Halle (English translation 1890)
Saussure, F. de (1915) Course in General Linguistics, edited by Charles Bally and Albert
Sechehaye, English translation by Wade Baskin, New York: Philosophical Library
Tibbitts, B. (1990) "Pre-verbal incorporated prepositions in Rama", Eugene, OR: University of
Oregon (ms.)

View publication stats