Introduction
Black South African English (BSAfE) is now generally regarded as an independent variety of English rather than an interlanguage on the way to Standard English (Van Rooy, Reference Van Rooy2008: 274, 300 and in this issue). Mesthrie (Reference Mesthrie2006: 115) demonstrates that many of the characteristic features of BSAfE can be ascribed to the overarching tendency of anti-deletion. Anti-deletion is a term coined by Mesthrie (Reference Mesthrie2006: 115) to encompass three kinds of linguistic phenomena that are the opposite of deletion in generative analyses of English, namely undeletion, non-deletion and insertion. Undeletion ‘restores an element that is often assumed to be deleted or to have an empty node in generative analyses of English’ (Mesthrie, Reference Mesthrie2006: 125), e.g. She made me to go (Mesthrie, Reference Mesthrie2006: 111) in which the infinitive marker to is undeleted. Insertion entails the addition of grammatical morphemes, e.g. can be able (Mesthrie, Reference Mesthrie2006: 139–40).Footnote 1 After examination of a number of undeletion phenomenaFootnote 2 in interviews with 12 mesolectal speakers of BSAfE, Mesthrie (Reference Mesthrie2006: 129) arrives at the following principle: ‘If a grammatical feature can be deleted in [Standard English], it can be undeleted in [Black South African English] mesolect.’ He points out that such undeletions are not mandatory and adds the following corollary to the principle of undeletion: ‘If a grammatical feature can be deleted in StE, it can also be (variably) deleted in [Black South African English] mesolect, at a lower rate of frequency’ (Mesthrie, Reference Mesthrie2006: 129).
The results of a recent corpus linguistic investigation of noun phrases in BSAfE (Botha, Reference Botha2012) provide further evidence of undeletion and insertion in BSAfE. The study aimed to arrive at a ‘data-driven’ (Rayson, Reference Rayson2008), function-based description of BSAfE noun phrases making use of the Tswana Learner English Corpus (TLEC) (Van Rooy, Reference Van Rooy2008). The Louvain Corpus of Native English Essays (LOCNESS) served as control corpus (Granger et al., Reference Granger, Dagneaux, Meunier and Paquot2009: 42). Both corpora consist of essays written by university students. The TLE corpus can be regarded as representative of mesolectal BSAfE (Van Rooy, Reference Van Rooy2008: 274). Comparison with a standard variety of English and use of a native control corpus are convenient to identify potential areas of innovation and stabilization of features of BSAfE.
Keywords were taken as a starting point for mining the corpus data. Scott (Reference Scott2010: 152) defines key words as ‘those whose frequency is unusually high according to some norm’. In the study under discussion, LOCNESS is taken as the norm. Positive keywords in the TLE corpus are words that occur significantly more frequently in the TLE corpus than in LOCNESS while negative keywords occur more frequently in the control corpus. The list of key words is based on a comparison of the wordlists of the two corpora created with the WordList tool in Wordsmith 5.1. Keyness is automatically calculated in the KeyWords function (Scott, Reference Scott2010: 166).Footnote 3 Typically, if the primary and control corpus represent the same genre and register, keywords are lexical words and indicative of subject matter. For instance, the nouns soccer and poverty are positive keywords in the TLE corpus because they relate to two frequently chosen essay topics. Keywords that are grammatical function words point to possible differences in the grammatical systems of the language varieties represented by the primary and control corpora. Concordances of a selection of such key words were analysed with a view to identifying usage patterns that provide insight into the grammatical system of BSAfE. The corpus concordances of the noun people and the words some and most are relevant to this article since they provide evidence of anti-deletion phenomena. Although people is a noun, its lexical meaning is very general and its frequency in a corpus is unlikely to be explained in terms of subject matter alone. Analysis of the concordance of people in the TLE corpus yields a number of instances where this general noun is undeleted in contexts where it can be omitted. Undeletion also explains certain uses of some that have hitherto simply been regarded as non-standard or ‘uniquely characteristic’ uses of quantifiers in BSAfE literature (e.g. De Klerk, Reference De Klerk2003: 470). The TLE corpus concordances of both some and most also provide support for Mesthrie's (2006: 140) treatment of quantifying of-partitions in BSAfE as morpheme insertion.
Undeletion of the noun people instead of a fused-head construction
The noun people is the most frequent noun in both corpora. It is also a positive keyword in the TLEC, occurring significantly more frequently in the TLEC compared to LOCNESS (Keyness value: 636). The token people occurs 1029 times per 100,000 words in the TLEC and 368 times per 100,000 words in LOCNESS. Analysis of the full corpus concordances of people reveals that the proportionally higher frequency of this noun in the TLEC is partly due to its undeletion in definite noun phrases with modifiers that imply a sub-class of people and which therefore permit the omission of the noun. Payne and Huddleston (Reference Payne, Huddleston, Huddleston and Pullum2002: 410) describe noun phrases of which the head word is not a noun as fused-head constructions, maintaining that what would have been the head is fused to a dependent element in the noun phrase. The noun people can be fused with a modifier (see (1a)), or with the determiner those (see (1b)) if a postnominal modifier is present.
1a The rich cannot enter the kingdom of heaven (Payne & Huddleston, Reference Payne, Huddleston, Huddleston and Pullum2002: 417)
1b Those who break the rules must expect to be punished (Payne & Huddleston, Reference Payne, Huddleston, Huddleston and Pullum2002: 414)
The fusions illustrated in (1) can also be analysed as deletion of the noun people. The TLEC concordance of this noun provides evidence of its undeletion in these contexts. The corpus examples in (2) illustrate the use of full noun phrases (i.e. with people as explicit head) in BSAfE instead of constructions in which the head is fused to the adjective, as in (3).
1a Without resources been taken and be given to the poor people, we are going to die because of AIDS. <ICLE-TS-NOUN-0133.1>
2b …because the needy people have no money to buy food. <ICLE-TS-NOUN-0186.1>
2c …for all the benefits that are received by the disabled people. <ICLE-TS-NOUN-0508.1>
3a the poor
3a the needy
3a the disabled
In Standard English varieties the noun people is deleted, as in (3), or the noun phrase is formulated as an indefinite, generic reference without the definite article, e.g. poor people. In the TLE corpus, the fused-head construction theADJECTIVE is favoured in contexts that allow it, as is the case in Standard English. For instance there are 13 nominalizations of the adjective poor (i.e. the poor) in the TLE corpus compared to two instances of the full noun phrase the poor people (see (2a) and (4)). In example (4) both undeletion (the poor people) and deletion of the head noun (the rich) occur.
4 Some end up jobless and the poor people remain poorer while the rich become richer. <ICLE-TS-NOUN-0221.1>
In the TLEC there are 22 instances of the pattern those people who-clause (exemplified by (5a) and (5b)). This pattern only occurs once in the LOCNESS data (see (5c)).
5a Even though we believe that Aids has no class it affects those people who are poor. <ICLE-TS-NOUN-0165.1>
5b According to me our government must try to help those people who came from different universities with jobs, <ICLE-TS-NOUN-0479.1>
5c Robert Rygor's words are fitting for those people who are against the idea. <ICLE-US-MRQ-0021.1>
The examples in (5) illustrate undeletion. However, there are 107 instances of the fused-head construction those who-clause in the TLEC, i.e. of the pattern with deletion. Again, the pattern with deletion occurs more frequently than the pattern with undeletion. This shows that the corollary to the principle of undeletion (Mesthrie, Reference Mesthrie2006: 129) is indeed necessary. The corpus data indicate that if the head noun can be deleted in Standard English, it is sometimes undeleted in BSAfE, but mostly deleted. Undeletion of the noun people contributes partly to its high frequency in the TLE corpus, though there are other factors outside the scope of this article that play a larger role in the keyness of this noun. What is relevant here is the evidence of undeletion of fusible head nouns.
Undeletion of non-singular indefinite marker
Per 100,000 words, the word some occurs 426 times in the TLE corpus and 161 times in LOCNESS (Keyness value: 263). The word some is used as a determiner (i.e. preceding a noun) in the majority of the instances in both corpora. Payne and Huddleston (Reference Payne, Huddleston, Huddleston and Pullum2002: 364) distinguish between the proportional and non-proportional uses of some. If some implies ‘not all’ it is interpreted proportionally, as in (6a). Non-proportional some does not have the ‘not all’ implicature (see (6b)), i.e. the speaker is not referring to a subset of a larger set of entities.
6a Some people misunderstood the question [proportional] (Payne & Huddleston, Reference Payne, Huddleston, Huddleston and Pullum2002: 364)
6b There were some children in the park [non-proportional] (Payne & Huddleston, Reference Payne, Huddleston, Huddleston and Pullum2002: 364)
An excerpt from the TLE corpus essay containing the most instances of the word some illustrates both the proportional and non-proportional uses of this word (see (7)). In the noun phrase some of the machines, some is interpreted as non-proportional. It is not clear whether some is intended proportionally or not in some schools and some big industries. In all the other noun phrases underlined in (7), some is non-proportional. Except in the noun phrase some way, some will be omitted in Standard English formulations of the underlined noun phrases in which it is used non-proportionally. In other words, non-proportional some is used in BSAfE in contexts where Standard English will use no determiner, or the so-called ‘zero determiner’, as in Ø scientists, Ø research, Ø students, Ø science subjects, Ø technology.
7 Each and every day some scientists are trying to find some way in which they can use by means of technology. They are doing some researches, and even at some schoolssome students are encouraged to take some science subjects, starting from the primary level. […] Some big industries in today's life uses some technology to operate some of the machines. <ICLE-TS-NOUN-0434.1>
The concordance data show that non-proportional some is used like the non-singular counterpart of a/an instead of the zero determiner to explicitly mark the nominal group as indefinite. Further examples of this can be seen in (8).
8a Nowadays disease known as AIDS is continuing to erode the quality of life of the people and it is very suicidal to venture out in search of some sexual partners. <ICLE-TS-NOUN-0056.1>
8b So you cannot play soccer each and every week and after that you've been paid some peanuts. <ICLE-TS-KIMC-0355.1>
In noun phrases without an explicit determiner, the ‘zero determiner’ is analysed as an empty node. The use of some as non-singular indefinite ‘article’ therefore fits Mesthrie's (Reference Mesthrie2006: 115) definition of undeletion. Since some is used merely to mark the noun phrase as indefinite in BSAfE, it is allowed in contexts where it would be disallowed in Standard English, for instance in combination with post-determiners, numerals and quantifiers (see (9)).
9a Many tourist who came to view wildlife in some other places outside North west province such as Kruga national park prefer to spend their nights at this hotel. <ICLE-TS-NOUN-0031.1>
9b In some certain circumstances, men ill-treat women. <ICLE-TS-NOUN-0501.1>
9c From my own point of view I would say the word poverty refers to people who lack some three basic needs of life which are shelter, clothing and food <ICLE-TS-NOUN-0252.1>
9d They go to some severals areas including towns where they becomes the prostitudes <ICLE-TS-NOUN-0172.1>
In the extension of non-proportional some in BSAfE, the quantifying meaning of some is diluted in the same way that the quantifying meaning of a/an was diluted in the grammaticalization of one. In the examples in 7 and 8 some functions primarily as a marker of indefiniteness and not as a quantifier. It only has a residual quantifying function in BSAfE to the same extent that the indefinite article a/an still has a quantifying function in English(es) generally. It is for this reason that some is compatible with quantifiers such as three and several as in (9c) and (9d). Careful consideration of the usage contexts of some in the form of a corpus concordance indicates that its high frequency in the TLE corpus is partly explained by determiner undeletion in non-singular (plural and uncountable) noun phrases.
Insertion resulting in quantifying of-partitions
Mesthrie (Reference Mesthrie2006: 140) mentions ‘the use of the partitive genitive more of, most of, too much of for [Standard English] “more”, “most”, “too much”’ as a characteristic of BSAfE involving insertion of extra morphemes. There is much evidence in the TLE corpus of the tendency to use partitive of-constructions rather than their non-partitive counterparts. Concordances of all words that may potentially serve as quantifiers (excluding cardinal numerals) were drawn from each of the corpora. The concordance lines were then sorted according to their functions and internal structure, distinguishing between true quantificational uses and other uses such as adverbial usage of more and most. In both corpora quantifiers typically precede the noun which they quantify, as in most people. In other words, in both corpora quantifiers occur more frequently in non-partitive constructions than in of-partitions. However, per thousand quantifiers, there are roughly twice as many quantifying of-partitions in the TLE corpus as in LOCNESS (181/1000 vs 93/1000). The raw frequencies give a clearer indication of the difference between the corpora with regard to partitive of-constructions. There are 742 quantifying of-partitions in the TLE corpus and 275 in LOCNESS.
The high frequency of partitive of-constructions in the TLE corpus can be partly attributed to the high frequency of the quantifier a lot, which cannot be used determinatively. The partitive construction a lot of occurs 112 times in the TLE corpus and 31 times in LOCNESS. The quantifiers some and most are used in of-partitions more frequently in the TLE corpus than in LOCNESS. A quarter of all the quantifying partitive of-constructions in the TLEC (186 instances) contain some as head compared to 14% (38 instances) in LOCNESS. In the TLE corpus, 22% of the instances of some occur in partitive constructions (186/854), compared to 12% in LOCNESS (38/324). The quantifier most occurs as head in 47% of the quantifying of-partitions in the TLE corpus, compared to 14% (39 instances) in LOCNESS. The partitive use of most (as in most of the people) outweighs its determinative use (as in most people) in the TLE corpus (350 vs 248), whereas in LOCNESS there are more than twice as many determinative uses of most compared to partitive uses (88 vs 39). Nearly half of all the occurrences of most in the TLE corpus (350/733 = 48%) are found in partitive of-constructions, compared to only 15% of the occurrences in LOCNESS (39/253). The partitive constructions more of and much of are rare in both corpora. In the remainder of this section, reasons for the high frequency of partitive of-constructions with the quantifiers some and most will be considered.
As is the case with some, the word most can have proportional and non-proportional uses. The proportional use of most also has the ‘not all’ implicature. The non-proportional use of most encompasses cases where most is used as superlative of many and much (Payne & Huddleston, Reference Payne, Huddleston, Huddleston and Pullum2002: 365). In standard English, partitive of-constructions convey the proportional sense of some and most and the complement of the preposition of is a definite full noun phrase, as in most of the people, or a pronoun, as in most of them. In other words, in Standard English, some of and most of constructions are used to refer to an identifiable (definite) subset of a larger set of entities. The concordances of some and most in the TLE corpus were examined to establish whether the partitive uses of these words were motivated by a need to refer to proportions of identifiable sets of entities or to some other function. The structural types of complement of the preposition of are set out in the table below.
Complements of of in quantifying partitions in the TLE corpus
As in Standard English, when the thing being quantified is identified by a demonstrative or possessive determiner or is conveyed by an anaphoric pronoun, the partitive of construction is required and a non-partitive formulation is not possible (*most these people, *most them). In the TLE corpus, the use of partitive of before noun phrases with the definite article is not always motivated by referent identification. In (10) the writer refers to students in general and not an identifiable set of students. As can be seen from the second noun phrase underlined in (10), the determiner does not always signal referent identifiability in BSAfE. In fact, it is doubtful if a subset of students is intended.
10 And some of the students can be motivated to work hard so that they can also get a chance to be short listed for the vacant post in certain companies or at different departments of the government. <ICLE-TS-NOUN-0482.1>
A closer look at the 178 TLE concordance lines with the pattern most of the reveals that in more than half of the cases, the use of the definite article is not motivated by a necessity to mark the quantified noun as identifiable. In a Standard English reformulation of (11), of the will be omitted from the underlined noun phrases.
11a Most of the South Africans love soccer and they are using it in order to gain a better living. <ICLE-TS-NOUN-0504.1>
11b Most of the teenagers and Adults are suffering from this desease. <ICLE-TS-NOUN-0116.1>
The strongest support for the claim that of-partitions in BSAfE are often the result of conventionalized morpheme insertion come from the instances where the noun phrase that complements of (and denotes the thing being quantified) has no determiner, as illustrated in (12).
12a In the country that I came from which is Botswana most of people drink too much. <ICLE-TS-NOUN-0119.1>
12b Some of young people become prostitutes because they want to earn some living. <ICLE-TS-NOUN-0210.1>
The propositional content of the clause in (12a) (as well as that of (11a)) suggests that, as is the case with most as determiner, most in of-partitions does not necessarily mean ‘the majority’, but is just as likely to mean ‘many’. These examples, as well as (10), indicate that the of-partition is not used to mark the quantifier as proportional. The insertion of the morphemes of or of the after quantifiers like some and most are not functionally motivated, but conventionalized.
Conclusion
Analyses of the concordances of two keywords in the TLE corpus (people and some) – originally with the aim of explaining their relatively high frequencies – uncovered further evidence of the undeletion tendency in BSAfE in constructions not mentioned by Mesthrie (Reference Mesthrie2006). Corpus data suggest that the noun people is often undeleted in contexts that allow for a fused-head construction. However, it should be noted that the constructions with deletion are more frequent than their undeleted versions in the data. Upon examining the concordance lines of some, an extension of the non-proportional sense of the word to an article-like function was discovered. The use of non-quantifying some as an overt marker of indefiniteness with non-singular nouns is a form of undeletion since it restores an element (the determiner in this case) that would have an empty node in a generative analysis of its Standard English counterpart. In order to explain the high frequency of quantifying partitive of-constructions the concordance lines with some of and most of were examined to determine whether the partitive constructions are motivated by a need to mark the quantified thing as identifiable (by means of the definite article) or to make explicit that the quantifier is used in its proportional sense. It was found that referent identification and differentiation between proportional and non-proportional senses of the quantifier do not motivate the use of quantifying of-partitions. The corpus data confirm that this is a case of morpheme insertion that is conventionalized rather than functionally motivated.
The evidence of undeletion and insertion discussed in this article was uncovered in an attempt to explain high frequency phenomena in the TLE corpus. The BSAfE corpora currently available (see Van Rooy in this issue) allow for future corpus-based investigation of all of the anti-deletion phenomena identified thus far (Mesthrie, Reference Mesthrie2006) as well as potential frequency-based discovery of new anti-deletion phenomena in both spoken and written registers covering a wider range of sociolects.
YOLANDE BOTHA is a lecturer in English at the Potchefstroom Campus of North-West University. Her doctoral thesis was on the grammar of noun phrases in Black South African English and was based on corpus data. Email: Lande.Botha@nwu.ac.za
