You are on page 1of 20

514588

research-article2014
ENGXXX10.1177/0075424213514588Journal of English LinguisticsWasserman and van Rooy

Article
Journal of English Linguistics
2014, Vol. 42(1) 31­–50
The Development of Modals © 2014 SAGE Publications
Reprints and permissions:
of Obligation and Necessity sagepub.com/journalsPermissions.nav
DOI: 10.1177/0075424213514588
in White South African eng.sagepub.com

English through Contact with


Afrikaans

Ronel Wasserman1 and Bertus van Rooy1

Abstract
This article draws attention to the uniqueness of modality in White South African
English (WSAfE), when compared with other native English varieties. This uniqueness
is found in the historical development of the modal category of obligation/necessity
since the nineteenth century, as well as in the factors that influence its change. A
historical corpus of written WSAfE is used to trace the development of these modals
from the 1820s to the 1990s, whereas the International Corpus of English–South
Africa (ICE-SA) helps explore contemporary written and spoken WSAfE. The main
finding is that the frequency patterns of must and should display a trend toward
semantic shifts and increasing polysemy. Must is the highest-frequency modal within
its semantic group, and it extends its semantic domain to express median obligation,
becoming partly synonymous with should. The prominent contact situation of WSAfE
with Afrikaans is suggested to play an important role in the way in which must has
developed and is currently used.

Keywords
South African English, modals, quasi-modals, diachronic corpus, language change,
language contact

The objective of this article is to compare trends in the use of modal verbs in the native
variety of English used in South Africa (henceforth WSAfE, for White South African
English) to that of other native Englishes. Previous research by Rossouw and van

1North-West University, Vanderbijlpark, South Africa

Corresponding Author:
Ronel Wasserman, North-West University, PO Box 1174, Vanderbijlpark, 1900, South Africa.
Email: ronel.rossouw3@gmail.com

Downloaded from eng.sagepub.com by guest on January 27, 2016


32 Journal of English Linguistics 42(1)

Rooy (2012) indicates that the frequencies of especially must and should in WSAfE
are different from other native varieties, which calls for an explanation. We explore
language contact between WSAfE and Afrikaans as a reason for the trends in this dif-
ference that results in the uniqueness of WSAfE compared to other native varieties.
Diachronic studies of American English (AmE) and British English (BrE) report a
general decline in modals and a general increase in quasi-modals over the course of the
twentieth century (Leech 2003, 2011; Mair & Leech 2006; Leech & Smith 2009;
Leech et al. 2009). With no evidence of interdependency between these trends, the
reasons are linked to tendencies toward colloquialization (the movement toward more
oral styles of language) and democratization (the social trend toward avoiding claims
to power or authority) (Leech 2003:235-237; Mair & Leech 2006:327). The sugges-
tion of colloquialization enjoys indirect support from Collins’s (2009a:9) analysis of
contemporary data of AmE, BrE, and AusE (Australian English) and more explicit
support from the synchronic-diachronic comparison of van der Auwera, Noël, and de
Wit (2012). The high-frequency modals suffered the least in terms of a frequency drop,
the low-frequency modals have diminished the most, and the midfrequency modals
have experienced a moderate degree of decline (Leech 2003, 2011; Mair & Leech
2006; Leech & Smith 2009; Leech et al. 2009).
Within the obligation/necessity group and even beyond, the most salient modal in
terms of a frequency drop is must. The decline has occurred to such an extent that
today must is frequently counted among the low-frequency modals in AmE, BrE, and
AusE (Myhill 1995:162; Mair & Leech 2006; Collins 2009a:43-44). This retreat of a
traditional marker for strong obligation (and often negative commands) can be seen as
an effect of democratization, attempting to avoid face-threatening language and the
creation of social distance (Myhill 1995:166-167)—notably by means of the more
“objective” quasi-modals have to and need to (Mair & Leech 2006; Collins 2009a:60).
Another reason for the rise and fall of particular quasi-modals and modals could be
a degree of interplay between changing levels of polysemy and monosemy (Myhill
1995; Smith 2003; Leech et al. 2009:70-90, 107-117). The most important observation
involves changes in the ecology of the individual modals and quasi-modals of the
obligation/necessity group, “ecology” being the niche of a modal expression, which
can be expanded, contracted, or maintained relative to competing forms (Leech et al.
2009:83-89, 114).
The midfrequency modal must declines in its expression of (strong) deontic mean-
ing to avoid such meaning (Smith 2003:242; Leech et al. 2009:114-116). Should is
found to be a beneficiary of the decline of must, in that it progressively expresses
weaker deontic meanings (Leech et al. 2009:86-88). Must has, contrastingly, been
moving toward expressing epistemic necessity, whereas should experiences a decline
in its epistemic meaning as a result (Leech et al. 2009:115-116). The movement toward
epistemic must has been partially influenced by its deontic restriction and partially
because there is no obvious alternative form with the same propensity for the expres-
sion of epistemic necessity (Leech et al. 2009:87-89). The quasi-modals have to and
need to both appear to have found their ecological niche: have to tends to express
general root modality and is becoming less face-threatening and need to is becoming

Downloaded from eng.sagepub.com by guest on January 27, 2016


Wasserman and van Rooy 33

even more face-saving, insofar that it is never replaceable by must (Smith 2003:242;
Leech et al. 2009:115).
For WSAfE, Rossouw and van Rooy (2012:13-14, 23) report similar general
changes to those of other native varieties, but with some exceptions as to when these
changes take place. These exceptions involve no general, steady decline of modals
across the twentieth century, but a strong decline near the end of the century (about
27 percent from the 1950s to 1990s), as well as a 41-percent rise of quasi-modals from
the 1950s to 1990s after relative stability from the beginning of the nineteenth to the
mid-twentieth century (Rossouw & van Rooy 2012:14, 23), compared to their steady
rise in AmE and BrE from 1961 to 1991/1992, as reported by Leech (2003, 2011),
Mair and Leech (2006), Leech and Smith (2009), and Leech et al. (2009:72).
Despite the broad similarity in frequencies, there are major differences between
WSAfE and other native varieties in respect of the patterns of individual modals and
quasi-modals. The high-frequency modals can and will show a strong increase in fre-
quency, whereas these modals remain relatively stable in BrE and AmE (except for
will in AmE), and a steeper decrease in frequency emerges for could and would than in
BrE and AmE (Mair & Leech 2006:327; Rossouw & van Rooy 2012:16-17).
The most striking difference from other varieties, however, is that in WSAfE the
midfrequency modal must does not decline in frequency, compared with its sharp
decline in other varieties (Leech & Smith 2009:188). Rossouw and van Rooy (2012)
report that its frequency increased until the mid-twentieth century, and remained
steady to the 1990s. If this traditional modal carrier of high social impact increases
rather than decreases in frequency, counterevidence of a democratization process
might be suggested; but it might not be as simple as this, when taking into account
Bowerman’s (2004:477) claim that must has less social impact in WSAfE than in other
varieties and that it is often used as a substitute for polite should. This raises the pos-
sibility of semantic overlap for these two modals (i.e., a partly synonymous relation-
ship). However, Bowerman does not expand on what the criteria for interpreting such
social impact would be and never overtly states what the source of his data is. Corpus
evidence of this trend is therefore yet to be provided and possible reasons are yet to be
explored—in particular, contextual reasons.
Today, native (white) speakers of English, the “Settler” or “STL” strand in
Schneider’s (2007) dynamic model, account for less than 5 percent of the entire con-
temporary South African population (Statistics South Africa 2001). This strand has
enjoyed widespread and continual contact with Afrikaans throughout the complex for-
mation period of WSAfE (Schneider 2007:173, 176). This kind of contact situation is
less present in other inner-circle varieties, except perhaps for Irish English, seeing that
Afrikaans, a transplanted variety of Dutch already spoken in the Cape for over 150
years by the time the British first arrived, first shared a complex sociopolitical situa-
tion with English due to factors like the Anglo-Boer War and the Apartheid system,
and second is genetically related to English. Both being of Germanic origin, English
and Afrikaans have a very similar modal system encapsulating many cognate modal
verbs. Such cognates of course include English must and Afrikaans moet, stemming
from the Germanic root mot (compare Dutch moeten and German müssen); both must

Downloaded from eng.sagepub.com by guest on January 27, 2016


34 Journal of English Linguistics 42(1)

and moet do not inflect for person anymore, display a high degree of similarity of
meaning, and enjoy phonological resemblance (Conradie 1976; Abraham 2001:23-30;
Traugott 2006:120).1 It is therefore possible that the obvious grammatical, phonologi-
cal, and semantic similarity of must and moet contributes toward the uniqueness of
must in WSAfE, after enjoying such close contact with Afrikaans across almost two
centuries. Afrikaans has indeed previously been found to be a strong basis of influence
on WSAfE (Jeffery & van Rooy 2004), although this possibility is challenged by Lass
and Wright (1986) as well as Mesthrie (2002), though only on the basis of particular
grammatical features. The results of this article could contribute to the debate by
revealing whether contact with Afrikaans has influenced the development of WSAfE
modality, and especially must.
We therefore aim to provide an account of contemporary WSAfE modality, which
includes spoken data, to assess the presence of colloquialization in the variety and to
evaluate the influence language contact with Afrikaans might have had on WSAfE by
comparing the frequencies of the modals in the two languages. We further aim to
investigate the diachronic evidence of a semantic-overlap phenomenon (a possible
trend toward polysemy) in the case of the obligation/necessity group, if it exists, by
tracing the semantic evolution of must alongside should.

Method
Corpora
For synchronic analyses of contemporary WSAfE, the most recent version of the
International Corpus of English–South Africa (ICE-SA) (Jeffery 2003) is used. This
corpus consists of 148,000 words in its written component and 406,810 words in its
spoken part. The use of this corpus aids in conducting comparisons with the ICE cor-
pora used in Collins (2009a, 2009b). A synchronic corpus of contemporary written
Afrikaans, Die Taalkommissie-korpus, compiled by the standardizing body of
Afrikaans, the Language Commission, is exploited to draw comparisons with the find-
ings from written ICE-SA. This corpus consists of 57 million words and is balanced
for a wide range of registers (see De Wet, de Waal & van Huyssteen 2011). For com-
parison with spoken ICE-SA, the North-West University corpus of spoken Afrikaans
comprises 72,000 words, evenly divided between informal dialogue and unscripted
formal dialogue.
The diachronic corpus used in this study consists of written texts from social and
business letters, newspaper articles, fiction, and nonfiction written at various locali-
ties, for example, the Eastern and Western Cape, Natal, Kimberley, and Johannesburg,
and amounts to 123,247 words. This is an enlarged version of the corpus used in
Rossouw and van Rooy (2012). The authors and recipients of the letters include some
literary figures of South Africa, for example, Olive Schreiner and Herman Charles
Bosman. Some extracts of these writers’ works of prose were also included, and some
newspaper texts also included discussions of their work, life, and times. Other texts
often report pivotal events in South African History, such as the discovery of gold and

Downloaded from eng.sagepub.com by guest on January 27, 2016


Wasserman and van Rooy 35

diamonds around Johannesburg and Kimberley, respectively, the Anglo-Boer War, and
the road to a democratic South Africa. This corpus is compiled from data gathered
from the National English Literary Museum in Grahamstown (Eastern Cape), as well
as other archives from university libraries in the North of the country, including the
Historical Papers collection at the University of the Witwatersrand and the Africana
collections of the North-West University campuses. Other sources include published
works of collected letters and various published works of nonfiction.
The corpus is divided into periods along the lines of Schneider’s (2007:175-185)
periodization South African English, and remains the same as the proposal by Rossouw
and van Rooy (2012). Period 0 in the diachronic corpus covers the first era of settle-
ment, that is, the 1820s (the time of input), the texts of which were obtained from the
Mesthrie and West corpus of letters from Proto WSAfE,2 amounting to 22,483 words
(see Mesthrie & West 1995). This corresponds to Schneider’s (2007:175-178) phase 1.
Period 1 (39,338 words) ranges from the 1870s to the 1900s, corresponding to
Schneider’s (2007:178-181) phase 2, whereas period 2 (61,426 words) is the 1910s to
the 1950s, covering the first part of Schneider’s (2007:181-185) phase 3. ICE-SA, the
contemporary data, is labeled as period 3 in the comparisons. As far as possible, bal-
anced amounts of data (in terms of word counts) for each of the four textual genres
were collected across the periods described above, except for period 0 (Mesthrie &
West’s 1995 letter corpus) and period 3 (written ICE-SA, which possesses some addi-
tional textual genres). As this article is not concerned with internal register variation,
these slight imbalances are not critical.

Analysis
After the corpora were tagged with the Lancaster CLAWS Part of Speech Tagger, both
the synchronic (WSAfE and Afrikaans) and diachronic analyses were conducted using
Oxford WordSmith Tools (version 5), by means of which concordance lists for all
modal and quasi-modal verbs were compiled. All examples within these lists were
then manually checked for accuracy (especially in the cases of the quasi-modals), and
raw frequencies were normalized per 100,000 words. Where appropriate, a log likeli-
hood calculation (indicated by the Greek capital letter lambda [Λ]) was performed on
the raw (nonnormalized data) to assess whether differences were statistically signifi-
cant, using the cutoff points of Λ > 3.84 for p < .05 and Λ > 10.83 for p < .001 (Rayson
& Garside 2000). Within the diachronic corpus, samples of one hundred concordances
were selected to conduct the semantic analyses. In the cases of low-frequency modals
or quasi-modals, the semantic frequency results were normed to match the samples of
one hundred concordances so as to produce comparable data sets.
For the semantic analysis of the obligation/necessity group within the historical
corpus, clear classifications of distinctions between levels of deontic obligation were
needed. Such degrees of strength in obligation are categorized in terms of high
(required), median (supposed), and low (allowed) degrees by Halliday and Matthiessen
(2004:620), with must and have to being generally regarded to represent a high or
strong degree of obligation, should a median or weaker degree, and may a low degree

Downloaded from eng.sagepub.com by guest on January 27, 2016


36 Journal of English Linguistics 42(1)

Table 1.  Parameters for Strong and Median Obligation.

Parameter Strong obligation Median obligation


1. Subject • Animate • Inanimate
•  Indication of status •  No difference in status
•  Personal pronouns •  Impersonal pronoun it
2. Adjuncts/expressions •  High probability •  Low or median probability
• Counterexpectancy •  Low or median usuality
•  High usuality •  Low intensity
•  Total or high intensity • Weak comment/mood
adverbials
• Strong comment/mood  
adverbials
• Polarity  
3.  Negation temporality •  Negative polarity/negation •  Positive polarity
•  Present or future time •  Past time
4.  Pragmatic functions •  Motivated by social norms •  Unmotivated by social norms
•  Motivated by emotions •  Unmotivated by emotions
•  Habitual obligation •  Intermittent obligation
•  Face-threatening situation •  Polite or friendly situation
• Hedging
5.  Broad textual analysis • Interpersonal • Hypothetical/philosophical

(Huddleston 2002:177; Collins 2009a:26; Leech et al. 2009:86-88, 114-116).


Huddleston (2002:205-206), however, distinguishes between strong must, which usu-
ally has a subjective source of obligation, and have to, which usually has an objective
source, causing must to be slightly stronger in this respect. Have to is therefore consid-
ered less face-threatening than must (Leech et al. 2009:115). Recognizing that these
semantic judgments involve gradient phenomena dependent on context, which can
sometimes be inconclusive, we set out to compile a set of parameters as a basis for a
more accurate interpretation involving strong and median obligation in must and
should, which is presented in Table 1.
The first parameter is the subject (external force of obligation) of the clause. If a
subject was animate and revealed a degree of status and the clause further contained
personal pronouns like I and you, the instance was marked for strong obligation. On
the other hand, if the subject was inanimate (e.g., the impersonal pronoun it) or, in the
instance of an animate subject, portrayed no status difference to that of the addressee,
the case was marked for median obligation.
Certain collocating adjuncts or expressions served as the second parameter. Halliday
and Matthiessen (2004:125, 128) first refer to certain adverbs serving as mood adjuncts
of modality, all of which either denote a low, median, or strong degree of probability
or usuality. We interpreted those adjuncts classified as low or median as indicative of
median obligation and naturally high-degree adjuncts were flagged for strong obliga-
tion.3 Second, the presence of mood adjuncts involving a total, high, or low degree of

Downloaded from eng.sagepub.com by guest on January 27, 2016


Wasserman and van Rooy 37

intensity4 (total and high intensity flagging strong and low flagging median obligation)
or an exceeding or limiting level of counterexpectancy,5 where both levels denote
strong obligation, aided in interpretation (Halliday & Matthiessen 2004:129). Third,
modal adjuncts that serve as comment or mood adverbials were considered. Comment
adverbials denoting weak or median obligation may include please and kindly (linking
with politeness in Parameter 5), and mood adverbials in this category include proba-
bly, perhaps, and maybe. On the other hand, comment adjuncts like frankly and mood
adjuncts like certainly and really usually denote a strong degree of obligation (Halliday
& Matthiessen 2004:354-355). Modal adjuncts of polarity can aid in interpreting for
degree of obligation as well, with for instance yes and no pointing toward a strong,
obligating answer to a matching question, and not and never pointing toward restric-
tiveness and therefore also strong obligation. Certain fixed or metaphorical expres-
sions equivalent to many of these adjuncts were also helpful for interpretation, for
example, expressions denoting degrees of probability, for example, It is
certain/probable/possible that . . . and polarity, for example, I am certain/sure that
. . ., It is true that . . ., as well as expressions of personal opinion, conviction, or stance,
for example, It is obvious that . . ., Nobody tries to deny that . . ., There can be no doubt
that . . ., and so on (Halliday & Matthiessen 2004:148-149, 616-617).
The third parameter is related to Collins’s (2009a:24-25) suggestion that the propo-
sitional “value” of a modal utterance can be linked to negation and temporality. This
gives the impression that negated modals or quasi-modals are more likely to convey a
stronger degree of permission or obligation (negative polarity), for example, must not,
while positive polarity is simply unmarked in respect of this parameter. Modal expres-
sions grounded in present and future time (posterior proposition) may indicate stron-
ger degrees of obligation than those situated in the past time (anterior). This of course
does not mean that positive statements (positive polarity) cannot denote a high degree
of obligation (cf. Coates 1983:39), which is why this parameter cannot be relied on in
isolation from other parameters.
The fourth parameter relies on Myhill’s (1995) reference to pragmatic functions in
subcategorizing the force of obligation, as well as the ideas of politeness and face
threatening acts (cf. Brown & Levinson 1987:65). Myhill (1995:163) suggests that
there are three subfunctions of strong obligation, which include obligation motivated
by (a) social norms and (b) emotion, as well as (c) habitual obligation. He exemplifies
these subfunctions via various modals and quasi-modals as follows: (a) Since you
broke it, you must pay for it, (b) You’ve just got to help me!, and (c) I have to take the
bus to work every day. If the reverse of these three subfunctions was noticed, the
instance was marked for median obligation. We also found strong obligation to apply
in cases where the social setting could be seen as face threatening, but where it was
obviously friendly, median obligation was identified. Instances of polite or formulaic
hedging aided in this process.
The final parameter is that of broad textual analysis and is perhaps the most subjec-
tive parameter in this case, which is why it is last on the list. It was found that, espe-
cially in the cases where other parameters were nonconclusive, the degree of obligation
was recoverable from the broad thematic analysis of the text. Median obligation

Downloaded from eng.sagepub.com by guest on January 27, 2016


38 Journal of English Linguistics 42(1)

Table 2.  Normalized Frequency of Modals in WSAfE.

Written Spoken
can 287 420
could 139 145
may 125 47
might 30 40
must 128 132
shall 13 6
should 104 105
need 1 2
ought 1 2
will 330 231
would 207 254
Total 1,365 1,384

Table 3.  Normalized Frequency of Quasi-modals in WSAfE.

Written Spoken
(be) able to 43 39
have to 75 138
have got to 2 65
need to 18 33
(be) supposed to 3 10
(be) going to 26 150
want to 32 111
Total 199 546

usually accompanied philosophical or hypothetical content, and shades into the epis-
temic meanings. If the context, whether clausal or broadly textual, revealed interper-
sonal contact or interaction of any sort, strong obligation was marked.
Unsurprisingly, such behavior is not always clear-cut in written texts.
Methodologically, if one or more of the parameters proved to apply to the case, it was
marked for force of obligation. But, naturally, the more parameters applied, the higher
the likelihood of accurate interpretation, so the instance was finally marked only after
all the parameters were properly considered. Such analyses were conducted within
each of the one-hundred-concordance samples of must and should for each period in
the historical corpus.6

Results
Quantitative Findings
Tables 2 and 3 present the general quantitative, synchronic findings from ICE-SA.7

Downloaded from eng.sagepub.com by guest on January 27, 2016


Wasserman and van Rooy 39

These frequencies show that the modals maintain similar frequencies across spoken
and written WSAfE (difference is not statistically significant; Λ = 0.24, p > .05),
whereas the quasi-modals are significantly more frequent in speech (Λ = 380.38, p <
.001), as is expected. The ratio of modals to quasi-modals is 6.5:1 in written and 2.3:1
in spoken WSAfE. The ratio of quasi-modals in speech to writing is 2.85:1. A compari-
son to the findings of Collins (2009a:160) reveals that the latter ratio is the closest to
AusE (2.59:1), somewhere between the ratios of AmE (4.33:1) and BrE (2.01:1).
According to Collins (2009a), a relatively higher proportion of quasi-modals in speech
is a sign of more advanced language change, and therefore WSAfE can be regarded as
more advanced than its parent variety, BrE, in this respect, but more conservative than
AmE. To the extent that this quasi-modal differential between speech and writing pre-
dicts language change in other English varieties (see Collins 2009a; van der Auwera,
Noël & de Wit 2012), it appears that WSAfE might also be changing due to colloqui-
alization, seeing that quasi-modals are more popular in speech. On a more specific
level, such an explanation, based on the assumption that diachronic predictions can be
made by considering synchronic findings, would be strengthened if quasi-modals were
found to replace modals on an individual level, especially in spoken WSAfE.
A different picture arises, however, when we look at the individual frequency
results. There are, in fact, modals that are notably more frequent in speech than in writ-
ing, namely, can and would, and many modals for which the spoken and written fre-
quencies are very close (even slightly higher in speech), for example, could, might,
must, and should. Only may, shall, and will are noticeably more frequent in writing.
The preference of may and shall for the written register may point toward their increas-
ing semantic monosemy and restriction to certain (formal or archaic) usage contexts,
as is also noted for AmE and BrE by Leech et al. (2009:80-89), which no doubt con-
tributes to the overall low frequency of especially shall.
It is apparent that can is by far the highest frequency modal in WSAfE (even before
will and would, especially in speech). This runs contrary to the findings of Collins
(2009a:92, 126), which show that will and would are the highest frequency modals
across BrE, AmE, and AusE, whereas can has a slightly lower or near equal frequency
compared to these. The possibility exists that the high frequency of can might be attrib-
uted to Afrikaans influence (see discussion in the next section), promoting the extended
use of the deontic permission meaning (but with a weaker force than may), as in (1).

(1) You know drawing pins stick in it without leaving a mark so you can pin.
(ICE-SA; S1B-072)

May is virtually restricted to written WSAfE; this, together with the overall very low
frequency of might (in both writing and speech), might account for the soaring frequency
of can (especially in speech)—with may losing ground to the expression of deontic per-
mission with can, and with might losing ground to can’s less tentative expression of
dynamic possibility. The expression of dynamic possibility via can is seen in (2):

(2) These pieces can really be classed as violin concertos with string trio accompa-
niment. (ICE-SA; W1A-002)

Downloaded from eng.sagepub.com by guest on January 27, 2016


40 Journal of English Linguistics 42(1)

100%

90%

80%

70%

60%
have to
50%
should
40% must
30%

20%

10%

0%
SAfE AusE BrE AmE

Figure 1.  Relative proportion of must, should, and have to across regional varieties, pooled
across written and spoken data.

In view of the high frequency of can and will, it is perhaps easy to overlook must as a
midfrequency modal in WSAfE. It is not its higher frequency relative to other native
varieties that makes it significant, however, as will be shown below. As the obligation/
necessity group (of which must is a member) is the focus of this article, we explore the
dynamics of these modals (in particular must) in greater detail. Have to stands out at
first glance in its high frequency in spoken WSAfE, in contrast to the lower frequency
in writing—almost half of its frequency in speech. A very similar, if slightly more
extreme pattern arises in all the other quasi-modals, (be) supposed to, need to, and
(have) got to (as well as the modals need and ought), but this is not surprising consid-
ering the much higher overall frequency of quasi-modals in spoken WSAfE.
An altogether different picture emerges when considering the modals must and
should, whose frequencies in spoken and written WSAfE are close to equivalent,
unlike in AmE, BrE, and AusE (see Collins 2009a:44, 52, this volume ). Collins
(2009a) reports that must in written language is about 50 percent more frequent on
average than in spoken language and should in written language is about 20 percent
more frequent on average than in spoken language for the three varieties he investi-
gates; that is, there is a clear differentiation in medium (spoken vs. written) for must.
Should is reported to be twice as frequent in speech as must (even more than double
for both AmE and AusE), and on average one-and-a-half times more frequent in writ-
ing than must, which reveals, once again, a clear distinction in medium between these
modals in these varieties. The data from Collins (2009a:34) can be compared to
WSAfE as in Figure 1.

Downloaded from eng.sagepub.com by guest on January 27, 2016


Wasserman and van Rooy 41

Table 4.  Normalized Frequency of Afrikaans Modals and Quasi-modals with English
Semantic Equivalents.

Written Spoken
kan (can) 609 747
kon (could) 86 65
mag (may/might)a 38 44
moet/moes (must)b 432 515
behoort te (should) 104 3
sal (will/shall) 403 326
sou (would) 89 139
wil/wou (want to/wanted to)c 149 302
Total 1,910 2,141

a. The archaic past tense of mag (i.e., mog), which is equivalent to English might, has fallen out of use in
Afrikaans, and therefore the meaning of might is conveyed by the present mag (Ponelis 1979).
b. As there is no contemporary grammatical past tense for must in English, both moet and its past tense
moes express the meaning of must (Ponelis 1979:246-252).
c. Want to and wanted to were analyzed simultaneously, seeing that the latter has regular inflection for
past tense, and is hence automatically included in the concordance lists (by means of searching for
want*). For this reason wil and its past tense wou are grouped together.

The strong preference for must in WSAfE is plain here, when compared to the other
first-language varieties. This modal is around 17 percentage points more common in
WSAfE than in AusE, 15 percentage points more common than in BrE, and a substan-
tial 22 percentage points more common than in AmE—at the expense of should and
have to. As mentioned above, should enjoys nearly double the frequency of must in the
other varieties, but here it can be seen that should is about 29 percentage points less
frequent than must in WSAfE, which means that this semantic group evidently devel-
ops in a different direction than BrE, AmE, and even AusE. Within this group, the
quasi-modal have to does not increase its frequency (as shown below), and thus devi-
ates from the general pattern in the other native varieties where quasi-modals increase
their frequency. Overall, WSAfE remains quite conservative in not undergoing changes
similar to the other native varieties as far as have to, must, and should are concerned.
Nevertheless, the most important difference between WSAfE and the other native
varieties studied by Collins (2009a) is not frequency, but the semantic change it has
undergone.

Comparison with Afrikaans


The constant contact between WSAfE and Afrikaans can yield a possible explanation
for the quantitative anomaly of must, and also can (as mentioned earlier). Table 4
presents the frequencies of the Afrikaans equivalents of some English modals and
quasi-modals, according to semantic and not always structural correspondence
(Ponelis 1979:246-247).

Downloaded from eng.sagepub.com by guest on January 27, 2016


42 Journal of English Linguistics 42(1)

100%

90%

80%

70%

60%
have to
50%
should
40% must

30%

20%

10%

0%
Period 0 Period 1 Period 2 Period 3

Figure 2.  Relative proportion of must, should, and have to in WSAfE across the historical
periods.

Across both written and spoken Afrikaans, the highest-frequency modals (in order
of frequency) are kan, moet/moes, and sal. Wil/wou are also quite frequent in the spo-
ken register. For the purpose of this article, the focus will be on only kan and moet/
moes. In Table 4 it is evident that the high frequency and indeed the spoken–written
distribution pattern of kan is mirrored by that of can in WSAfE.
The second most frequent modal group in Afrikaans, and the most important for
this article, moet/moes, also counts heavily among the Afrikaans modal expressions,
especially in the spoken register. Since must also has a higher frequency in WSAfE
speech this can be seen as direct influence from the use of moet/moes on must, since
the latter is more common in spoken language. As mentioned above, this conflicts with
Collins’s (2009a:44) finding that written must is on average almost twice as frequent
as spoken must in all the varieties he considered. Contact with Afrikaans is therefore a
plausible factor in WSAfE modal usage. It would count as further evidence of this if
indeed WSAfE must has taken on a meaning similar to that of Afrikaans moet/moes.
Moet in Afrikaans expresses not only strong, but also often weak or median obligation
meanings,8 whereas moes expresses only weak obligation (Ponelis 1979:251-252).
Therefore, if must could be found to increasingly express weaker meanings, evidence
for Afrikaans influence would certainly be strong.

Diachronic Findings
Figure 2 shows the proportional frequency change of the three main carriers of obliga-
tion and necessity meanings from period 0 to period 3.

Downloaded from eng.sagepub.com by guest on January 27, 2016


Wasserman and van Rooy 43

Table 5.  Semantic Change per One-Hundred-Concordance Sample.

Median obligation Strong obligation Epistemic


Period 0 must 0 42 58
should 29 0 71
Period 1 must 44 42 14
should 54 7 39
Period 2 must 63 10 27
should 50 8 42
Period 3 must 66 23 11
should 70 18 12

Figure 2 shows a clear diachronic shift in the relative weight of must, should, and
have to in their semantic group. Have to displays the most variable pattern within the
group, but it has clearly not overtaken must and should in the expression of obligation/
necessity, as is the case in other varieties, which could be a result of more stable pat-
terns in must and should. The most obvious development is indeed the general rise of
must within this group, which increases its use by about 10 percentage points from
period 0 to period 1, from 28 to 38 percent, then dropping ever so slightly (by 3 per-
centage points) to around 35 percent during period 2 and rising again (by 7 points) to
42 percent in period 3. By contrast, the decline of should is very clear and stable across
the periods. It drops from 49 percent in period 0 to 47 percent in period 1, then to 38
percent in period 2 and to 34 percent in period 3. Must therefore displays an overall
rise of 14 percentage points from the 1820s to the 1990s, whereas should decreases by
15 percentage points. This suggests that must is rising mostly at the expense of should.
This is indeed very different from other native Englishes, where must was still a mid-
frequency modal in the first half of the twentieth century, but declined more than
should, could, and might as the century progressed (Smith 2003): the use of must
decreased by 29 percent in BrE and 34.4 percent in AmE from 1961 to 1991/1992
(Mair & Leech 2006:331). The reason behind the different trend in WSAfE can be
found in semantics. The one-hundred-concordance samples from the diachronic cor-
pus yielded the results shown in Table 5.
Table 5 illustrates the gradual decline of epistemic meanings of must as well as
should, but with should overall slightly more frequent in expressing these meanings.
Strong obligation remains stable for must from period 0 to period 1, dropping by
thirty-two instances in period 2 and then rising by thirteen in period 3—which still
shows an overall decline. Should, on the other hand, displays a modest rise in its
expression of strong obligation, increasing from zero to eighteen instances from period
0 to period 3. Thus, by the 1990s, there is a difference of only five instances out of one
hundred between strong obligation must and should. These are indeed interesting
developments considering the fact that must is reported to increasingly express epis-
temic meanings and should weaker deontic meanings in BrE and AmE during the
twentieth century (cf. Leech et al. 2009:86-87, 115-116). Furthermore, median

Downloaded from eng.sagepub.com by guest on January 27, 2016


44 Journal of English Linguistics 42(1)

obligation must rises steadily across all periods from zero instances in period 0 to a
very significant sixty-six instances in period 3. Should also yields an increase in
expressing median obligation, with instances rising by twenty-five from period 0 to
period 1, dropping slightly by four instances from period 1 to period 2, but increasing
again to seventy in period 3. Hence, there is only a difference of four instances out of
one hundred between median obligation must and should. The remainder of this sec-
tion considers the semantics of must and should individually for all periods and illus-
trates the application of different combinations of the parameters described in the
methodology, and moreover provides some examples of the unique use of median
obligation must, as well as strong obligation should, which is absent in other native
varieties.
In period 0 must conveys epistemic meanings in 58 percent of all its instances, with
strong obligation occupying the remaining 42 percent. It is important to note that must
is not used at all to convey median obligation during this period. In the cases of must
expressing strong obligation, the following example illustrates the use of the modal
expression be compelled to, which intensifies the degree of obligation, together with
ideas of social norms, a sense of urgency conveyed by the adverb speedily, and a future
temporality.

(3) . . . unless something be speedily done for them by Government they must be
compelled to abandon the settlement . . . (1820s; letter)

In the same period should, like must, mostly expresses epistemic meanings—71 per-
cent of all its instances. It further expresses median obligation in 29 percent of the
cases and reveals no expressions of strong obligation whatsoever during this period. In
the cases of median obligation should, polite hedging is often noted, which naturally
weakens the force of the obligation. In (4), expressions of politeness hedge the force
of a request to an addressee of higher status.

(4) Your Exellancy will be pleased to name what rent it is expected your Memorialist
should pay for the same. (1820s; letter)

It is therefore clear that the semantic domains of must and should are still distinct at
the start of the nineteenth century in that, where obligation is concerned, must always
portrays strong and should median obligation. This is, in a sense, the traditional picture
of how must and should are used in more conservative BrE (e.g., Collins 2009a).
A very different picture develops in period 1. The epistemic meaning of must
diminishes strongly from 58 percent to 14 percent during this period—leaving a large
44 percent of the ecology (in the terms of Leech et al. 2009:114) vacant, which appears
to be filled with a new meaning—that of median obligation. This meaning immedi-
ately overtook strong obligation at 42 percent. Strong obligation was interpreted to be
expressed in (5).

(5) When the first shot is fired and we know that all is over, we can “trek los.” You
however, and the men with you must sit fast and quiet, to the end: even after all

Downloaded from eng.sagepub.com by guest on January 27, 2016


Wasserman and van Rooy 45

the country is in flames. You must hold fast and give no opportunity to the
enemy displacing you. (1899; letter)

In this example, the broad textual analysis is that of interpersonal communication writ-
ten in the context of the first year of the Anglo-Boer War: a politically and socially
tense time frame. Although parameters for strong obligation are not omnipresent, the
weight of the parameters leans toward strong obligation, mostly because of the con-
text, where wartime intensifies a sister’s concerned (emotional) requests.
Examples (6) and (7) contain two of the earliest instances of WSAfE median obli-
gation must, which in period 1 are mainly found in letters.

(6) Some day you must come up with me. You can have a horse. (1892; letter)

In (6), even though the obligation is set in a future time, the adverbial expression some
day (contemporary someday) renders it a hypothetical textual analysis, being an
unspecified invitation, together with low or median probability and usuality. The sub-
sequent proposal concerning the horse can be seen in the light of friendliness, adding
to the median obligation expressed by must. In (7), median obligation must is also used
as part of the description of a hypothetical situation (induced by if), but here with some
formulaic expressions: for the rest and at all, expressing resignation in uncontrollable,
unpromising circumstances, as well as the notion of thanking God if these circum-
stances end well.

(7) B will see you all right, if Percy suffers loss, and for the rest we must thank God
if we come out at all well from the fearful mess into which Jameson plunged us
all. (1896; letter)

During period 1 epistemic should drops by 32 percentage points at the expense of


median obligation meanings, an example of which is (8).

(8) What should be the basis of representation? (1908; news)

Here, an unspecified, inanimate subject marks median obligation. During this period,
as well, strong obligation should emerges to cover a large 54 percent of the instances,
as in (9).

(9) The settlement should be such as to kill the Bond absolutely and for ever.
(1899; letter)

In (9) the strong obligation is marked for intensity and temporality by the adjuncts
absolutely and for ever, the intensity of which is strengthened by the verb kill.
In period 2, apart from slightly rising epistemic meanings, median obligation must
rises by 19 percentage points at the expense of strong obligation, resulting in median

Downloaded from eng.sagepub.com by guest on January 27, 2016


46 Journal of English Linguistics 42(1)

obligation becoming the most frequent meaning of must (63 percent). Strong obliga-
tion drops by 31 percentage points during this period, but was still found to occur in
extracts like (10).

(10) I spoke to your house master about a number of things. He said, “You must
conform but you need not agree.” [. . .] As part of the school and as a member
of your own house you must toe the line in conformity with the rest of the
boys, even though you disagree in principle with some of the rules and regula-
tions. (1950s; letter)

Here the subject of the reported speech is of a higher status than the son addressed by
his father, by which the son is compelled to do what is expected of everyone (always
obey rules) in the school, despite the fact that he does not agree with them. The situa-
tion is therefore governed by social norms, refers to habitual action, and threatens the
face of the son.
Should, however, remains relatively stable during period 2, with only a very slight
rise in epistemic meaning (3 percentage points) and strong obligation (1 percentage
point), and a slight drop in median obligation (4 percentage points). Strong obligation
should can be seen in the following example, where the same father as in (10) states
his expectations for his son’s performance. In (11) but activates the contrast between
never and the implied always: the son is expected to always be in the first five.

(11) I don’t care if you never come top of the class, but you should be in the first
five. (1950s; letter)

Period 3 marks the rise of must (by 13 percentage points) to express strong obligation,
which happens at the expense of epistemic meanings (dropping by 16 percentage
points), as median obligation continues to rise by another 3 percentage points to finally
occupy a large 66 percent of its semantic ecology. Strong obligation was marked in
(12), where the expression no doubt about it denotes high probability, the subjects of
the clause are perceived criminals (low status), and where the act of preventative pun-
ishment (social norms) occurs in a future temporality and is linked to emotions.

(12) No doubt about it, killers and rapists must be punished in a way that prevents
them again becoming dangers to the public. (ICE-SA; W2E-003)

This contrasts with the occurrences of median obligation, where a hypothetical textual
analysis played a role in identifying an instance such as (13).

(13) This elected person must serve the whole community, he must listen to the
needs and wants of the people. (ICE-SA; W1A-016)

The semantic domains of should appear to be arranged much like must in their rela-
tive frequency. It expresses 10 percentage points more strong obligation and 30

Downloaded from eng.sagepub.com by guest on January 27, 2016


Wasserman and van Rooy 47

percentage points less epistemic meanings, but 20 percentage points more median
obligation meanings than in period 2. (14) is an example of median obligation should.

(14) Vaguely aware of what I perhaps should call a pricking of my conscience, it


never crossed my mind. . . (ICE-SA; W2B-013)

Here the adverbs vaguely and perhaps diminish the probability and the intensity of
the awareness. Strong obligation should is illustrated in example (15), in which the
adverb immediately conveys a strong sense of urgency, and the consequences, should
the obligation not be fulfilled, would be serious.

(15) If you do not already have a will, you should make one immediately—dying
without one always leaves complications . . . (ICE-SA; W2D-004)

Data from the above interpretations therefore reveal that there is a very strong,
gradual increase in the use of must to express median obligation. It has gradually
moved into the semantic domain of should, making it an increasingly polysemous
modal within the obligation/necessity group. We attribute this to the infiltration, into
WSAfE, of the semantics of Afrikaans moet/moes, which covers the entire semantic
space of median and strong obligation. Should has similarly moved into the semantic
space of must, increasingly expressing stronger obligation. As to the question of
whether these modals are on a path toward synonymy, this article cannot confirm this.
There may still be a slight tendency of politeness adhering more to should than to
must, but this will have to be investigated in the future.

Conclusion
This article has three main conclusions. First, to the extent that the more frequent use
of quasi-modals in speech is consistent with the interpretation of colloquialization (the
tendency toward a more speech-like style of language; see, e.g., Collins 2009a),
the likelihood exists that WSAfE also experiences colloquialization. However, while
the quasi-modals generally have higher frequencies in speech than in writing, have to
was found to be an exception, having a higher frequency in written language than
other quasi-modals, but not showing a much higher frequency in the spoken
language.
The high-frequency of can can be credited to the high-frequency modal kan, its
cognate in Afrikaans. Similarly, the high frequency of must is also found to be due to
Afrikaans influence via moet/moes. The second conclusion is therefore that Afrikaans
has been found to influence the grammar of WSAfE, confirming the suggestions of
Jeffery and van Rooy (2004).
The third and principal conclusion is that must and should are both increasingly
polysemous, and in consequence progressively overlapping in their expressions of
both strong and median obligation. This is clearly different from the trend toward
monosemy in other native varieties. WSAfE must expresses median obligation far

Downloaded from eng.sagepub.com by guest on January 27, 2016


48 Journal of English Linguistics 42(1)

more than strong obligation, confirming Bowerman’s (2004) notion of less “social
force” carried by must in WSAfE. The influence of Afrikaans is, once again, found to
be at work here, due to the corresponding semantic and structural mapping of must
with moet/moes. The fact that should has no semantically corresponding cognate in
Afrikaans can contribute to it having an overall lower frequency than must.
From these conclusions it is clear that WSAfE is unique in its expression of modal-
ity among other native English varieties, largely due to the impact of language contact
on the obligation and necessity group. The full extent of the uniqueness of this variety
in terms of grammar remains to be further investigated. Future research on modality
can first expand on the semantic interpretations by refining the parameters for seman-
tic analysis and then adding comparisons with other quasi-modals in the obligation
and necessity group, for example, have to and need to. Further explorations into the
role of Afrikaans influence on WSAfE will also add scope to the research.

Authors' Note
All opinions are those of the authors and should not be attributed to the NRF.

Declaration of Conflicting Interests


The author(s) declared no potential conflicts of interest with respect to the research, authorship,
and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship,
and/or publication of this article: The authors acknowledge the support of the National Research
Foundation (NRF) of South Africa, in the form of a project grant to the second author entitled
“The development of South African English.”

Notes
1. Within the obligation/necessity group, should, in contrast to must, does not have semantic
correspondence with its Afrikaans cognate, sou, which, being the past tense of sal (cognate
with English shall), expresses a meaning closer to that of would. The meaning of should is
expressed via the Afrikaans construction behoort te, which acts as a quasi-modal (Ponelis
1979:247). Have to, on the other hand, has no cognate form in Afrikaans.
2. This corpus was generously made available to us by Rajend Mesthrie.
3. High probability is expressed by adverbs like certainly and definitely, and low or median
by, for instance, possibly, perhaps, and maybe (Halliday & Matthiessen 2004:128). High
usuality is expressed by, for instance, always and never, and low or median by adverbs like
usually, sometimes, and rarely (Halliday & Matthiessen 2004:129).
4. Examples of intensity degrees include totally and utterly (total), quite (high), and scarcely
and hardly (low) (Halliday & Matthiessen 2004:129).
5. Exceeding levels of counterexpectancy are expressed by, e.g., actually and in fact, and
limiting levels by, e.g., just and merely (Halliday & Matthiessen 2004:129).
6. A very small number of gray areas remained in interpretation. The five-parameter system
produced certainty to a large extent, but in a few (only one or two per sample of one hun-
dred) cases unclear occurrences were not analyzed and a random substitute sample from

Downloaded from eng.sagepub.com by guest on January 27, 2016


Wasserman and van Rooy 49

the same text was chosen to fill the one-hundred-concordance sample. The point of the
analysis was to determine the relative changes between median and strong obligation, and
therefore the extent to which ambiguous or unclear examples are present in the data is
inconsequential to our analysis.
7. The selection of quasi-modals in Table 3 includes that of Collins (2009b), with the addition
of (be) supposed to from Rossouw and van Rooy (2012) and (be) able to from Leech et al.
(2009). We excluded (be) to and (had) better (as found in Leech et al. 2009), because of
their minute frequencies.
8. The combination of the modal sou with moet to form the construction sou moet can further
weaken the force of moet in Afrikaans.

References
Abraham, Werner. 2001. How far does semantic bleaching go? About grammaticalization that
does not terminate in functional categories. In Jan Terie Faarlund (ed.), Grammatical rela-
tions in change, 15-64. Amsterdam: John Benjamins.
Bowerman, Sean. 2004. White South African English: Morphology and syntax. In Bernd
Kortmann, Edgar W. Schneider, Kate Burrage, Rajend Mesthrie & Clive Upton (eds.),
A handbook of varieties of English: Morphology and syntax, 472-478. Berlin: Mouton de
Gruyter.
Brown, Penelope & Stephen C. Levinson. 1987. Politeness: Some universals in language usage.
Cambridge: Cambridge University Press.
Coates, Jennifer. 1983. The semantics of the modal auxiliaries. Beckenham: Croom Helm.
Collins, Peter. 2009a. Modals and quasi-modals in English. Amsterdam: Rodopi.
Collins, Peter. 2009b. Modals and quasi-modals in world Englishes. World Englishes 28(3).
281-292.
Conradie, C. J. 1976. Die modale werkwoord in Nederlands en Afrikaans. Taalfasette 20(4).
62-68.
De Wet, Febe, Alta de Waal & Gerhard B. van Huyssteen. 2011. Developing a broadband auto-
matic speech recognition system for Afrikaans. Florence: 12th Annual Conference of the
International Speech Communication Association paper.
Halliday, M. A. K. & Christian M. I. M. Matthiessen. 2004. An introduction to functional gram-
mar. 3rd edn. London: Arnold.
Huddleston, Rodney. 2002. The verb. In Rodney Huddleston & Geoffrey K. Pullum (eds.), The
Cambridge grammar of the English language, 71-212. Cambridge: Cambridge University
Press.
Jeffery, Chris. 2003. On compiling a corpus of South African English. Southern African
Linguistics and Applied Language Studies 21(4). 341-344.
Jeffery, Chris & Bertus van Rooy. 2004. Emphasiser now in colloquial South African English.
World Englishes 23(2). 269-280.
Lass, Roger & Susan Wright. 1986. Endogeny versus contact: Afrikaans influence on South
African English. English World-Wide 7(2). 201-223.
Leech, Geoffrey. 2003. Modality on the move: The English modal auxiliaries 1961-1992. In
Roberta Facchinetti, Manfred Krug & Frank Palmer (eds.), Modality in contemporary
English, 223-240. Berlin: Mouton de Gruyter.
Leech, Geoffrey. 2011. The modals ARE declining: Reply to Neil Millar’s “Modal verbs in
TIME: Frequency changes 1923–2006”, International Journal of Corpus Linguistics 14:2
(2009), 191–220. International Journal of Corpus Linguistics 16(4). 547-564.

Downloaded from eng.sagepub.com by guest on January 27, 2016


50 Journal of English Linguistics 42(1)

Leech, Geoffrey, Marianne Hundt, Christian Mair & Nicholas Smith. 2009. Change in contem-
porary English: A grammatical study. Cambridge: Cambridge University Press.
Leech, Geoffrey & Nicolas Smith. 2009. Change and constancy in linguistic change: How gram-
matical usage in written English evolved in the period 1931-1991. In Antoinette Renouf &
Andrew Kehoe (eds.), Corpus linguistics: Refinements and reassessments, 185-204. New
York: Rodopi.
Mair, Christian & Geoffrey Leech. 2006. Current changes in English syntax. In Bas Aarts &
April McMahon (eds.), The handbook of English linguistics, 318-342. Oxford: Blackwell.
Mesthrie, Rajend. 2002. Endogeny versus contact revisited: Aspectual busy in South African
English. Language Sciences 24(3-4). 345-358.
Mesthrie, Rajend & Paula West.1995. Towards a grammar of Proto South African English.
English World-Wide 16(1). 105-133.
Myhill, John. 1995. Change and continuity in the function of American English modals.
Linguistics 33(2). 157-211.
Ponelis, F. A. 1979. Afrikaanse Sintaksis. Pretoria: J. L. van Schaik.
Rayson, Paul & Roger Garside. 2000. Comparing corpora using frequency profiling. In
Adam Kilgarriff & Tony Berber Sardinha (eds.), Proceedings of the workshop on com-
paring corpora: Held in conjunction with the 38th annual meeting of the Association for
Computational Linguistics (ACL2000), 1-6. Hong Kong: Hong Kong University of Science
and Technology.
Rossouw, Ronel & Bertus van Rooy. 2012. Diachronic changes in modality in South African
English. English World-Wide 33(1). 1-26.
Schneider, Edgar W. 2007. Postcolonial English: Varieties around the world. Cambridge:
Cambridge University Press.
Smith, Nicolas. 2003. Changes in modals and semi-modals of strong obligation and epistemic
necessity in recent British English. In Roberta Facchinetti, Manfred Krug & Frank Palmer
(eds.), Modality in contemporary English, 241-266. Berlin: Mouton de Gruyter.
Statistics South Africa. 2001. Census 2001: Census in brief. Pretoria: Statistics South Africa.
Traugott, Elizabeth. 2006. Historical aspects of modality. In William Frawley, Erin Eschenroeder,
Sarah Mills & Thao Nguyen (eds.), The expression of modality, vol. 1, The expression of
cognitive categories, 107-139. The Hague: Mouton de Gruyter.
van der Auwera, Johan, Dirk Noël & Astrid de Wit. 2012. The diverging need (to)’s of Asian
Englishes. In Marianne Hundt & Ulrike Gut (eds.), Mapping unity and diversity world-
wide: Corpus-based studies of New Englishes, 55-75. Amsterdam: John Benjamins.

Author Biographies
Ronel Wasserman is a PhD student at the Vaal Triangle Campus of the North-West University
in Vanderbijlpark, South Africa, specializing in the field of South African English, with a focus
on the native variety. Her main research areas are corpus linguistics, sociohistorical linguistics,
and the grammar of the verb, and her current work involves the compilation of a historical cor-
pus of South African English.
Bertus van Rooy is professor in English language studies at the Vaal Triangle Campus of the
North-West University in South Africa, and is a former president of the International Association
for World Englishes. His main interest is the varieties of English, from a corpus perspective. He
examines both native and nonnative varieties in their contemporary state and earlier forms, with
a particular focus on the emergence of new grammatical conventions in nonnative varieties.

Downloaded from eng.sagepub.com by guest on January 27, 2016

You might also like