Nominal and pronominal negative concord, through the lens of Belizean and Jamaican Creole

1 Introduction

In the Belizean translation of the New Testament the apostle Paul frequently stresses the importance of being honest. Examples (1) and (2) are verses from Paul’s letters to Timothy and the Corinthians.

Ah noh di tel no lai in (1) and Ai noh di tel lai in (2) mean exactly the same thing, but there is a formal difference. In the version in (1) negation is not only expressed with the clausal negator noh but also with the negative determiner no. Example (1) illustrates what Jespersen (1922: 352) called ‘concord of negatives’, Mathesius (1937: 81) ‘negation concord’ and what we now generally call ‘negative concord’, henceforth ‘NC’. Example (2) illustrates a non-NC pattern. The contrast between an NC and a non-NC pattern is also shown in (3).

The negation of the first clause has two exponents, but the second clause only has one, but differently from the non-NC pattern in (2), the one exponent here is not the clausal negator but the negative pronoun. In this article I will study the alternation of the NC and non-NC patterns – the ‘concordant’ and the ‘non-concordant’ patterns – in Belizean and Jamaican, both for nominals, as in (1) and (2), and for pronouns, as in (3). The goal is double. First, the aim is to increase our general knowledge of NC, in particular, its typology. Second, I aim to increase our understanding of the specifics of NC in Jamaican and Belizean and of the similarities and differences. The fact that I try to reach the general typological aims through a study of Creoles is in line with Ziegeler and Bao (2017) and esp. Déprez and Henri (2018).

The reason for engaging in studying and comparing Jamaican and Belizean NC has to do with two things. First, both are Caribbean English Creoles, they are similar,^[1] and in crosslinguistic research one may learn as much from the comparison of two related languages as from the comparison of less related or totally unrelated languages. Second, for both we have the same kind of data, viz. full and recent translations of the New Testament (Di Jamiekan Nyuu Testament 2012; Di Nyoo Testiment eena Bileez Kriol 2012).

Strictly speaking, the account concerns NC in the Belizean and Jamaican testaments only, and not in the languages as a whole. At least for Jamaican there is some assurance that what holds for the New Testament is representative of the basilectal and mesolectal variants of the post-creole continuum. The reason is that for the Jamaican part the study strongly relies on earlier work (van der Auwera and De Lisser 2019). This study used the New Testament translation as well the Jamaican translations of Lewis Carroll’s Alice’s adventures in Wonderland (2016) and Antoine de Saint-Exupéry’s Le petit prince (2015). It was shown that the grammar of the three data sets does not differ in any significant way and the authors take this to indicate that the Testament data represent a natural kind of Jamaican. One of the authors, viz. Tamirand Nnena De Lisser (University of Guyana), is furthermore a native speaker linguist of Jamaican as well as the translator of Alice’s adventures in Wonderland and Le petit prince, and she testifies to the naturalness of the three Jamaican texts. To the extent that I can judge and compare the Belizean texts to what is known about the grammar and the lexicon (Crosbie 2007; Decker 2009; Escure 2013), I gather that the Belizean Testament is as representative of a natural basilectal-to-mesolectal Creole as its Jamaican counterpart.

It is also important to stress that in what follows I only deal with NC, not with any other aspect of the languages nor with the adequacy of the translations vis-a-vis the source texts. The translations of the Jamaican and Belizean data are my own. Because of this strong focus on NC and also because of the fact that the two languages are lexically close to English, the glossing will be kept very simple. References to the New Testament texts will be done in the traditional way, with chapters and verses, as already put into practice in Examples (1)–(3).

Section 2 is about NC in general. Section 3 offers some general observations on NC in Belizean and Jamaican. Section 4 focusses on pronominal NC in clauses with just two exponents, viz. a negator like noh in (1)–(3) and a negative pronoun like nobadi in (3). Section 5 does the same except that the exponent other than noh is a nominal, such as no lai in (1). Throughout I not only study the concordant patterns but consider equally the various non-concordant occurrences. Section 6 is the conclusion.

2 Negative concord

The concept of NC refers to the phenomenon illustrated in (1) and (3). In addition to the clausal negator noh there is a negative determiner no or a negative pronoun nobadi. Thus, negation is expressed twice, yet, semantically, there is only one negation, as we see in the translation of (1) in Standard English. These two exponents of negation, i.e., noh and no lai are both negative, but they express different aspects of negation. The negator noh only expresses that a proposition within its ‘scope’, viz. ‘I am telling you lies’ is false. In (1) the focus is ‘no lies’ and this shows that the proposition is false because it is not lies that the speaker is telling the addressee, rather than because it is somebody other than the speaker that is telling lies or because the speaker is reminding the addressee of lies. Focus marking implies scope marking, but not the other way round. In (4) the false proposition is also ‘I am telling you lies’, but there is no focus marking; in other words, it remains vague why the proposition is false.

In what follows I will use the phrase ‘NC negator’ (for ‘negative concord negator’) for elements that function like noh. I prefer it to the simpler ‘negator’, because not every negator is involved in NC. To wit, Standard English not is a negator, but not an ‘NC negator’. For constructions that function like no lai, I will use the abbreviation ‘NCI’ (for ‘negative concord Item’),^[2] not least because it forms a nice pair with ‘Negative Polarity Item’ or ‘NPI’, which is the common label for words like any. NPIs, different from NCIs, are not themselves negative, and they occur in a wide set of contexts, such as questions or conditionals.^[3]

No lai is a nominal NCI. Different from pronominal NCIs, nominal NCIs have not been prominent in the literature on NC, as a quick glance at Déprez and Espinal (2020) and Breitbarth et al. (2020) can show. A clear exception is Jäger’s analysis of nominal NC in Old and Middle High German (Jäger 2008: 201–205, 278–282).^[4] This is not to say that there are no issues that connect nominal constructions to NC that attracted a lot of research. There are two: (i) the development of nouns or noun phrases to negative pronouns, as with French personne (from ‘person’ to ‘nobody’; Déprez 2011) or English nobody (from no body) and (ii) polarity changes in nominal determiners/modifiers, as with German kein (‘no’ from ‘any’; Jäger 2008: 188–192, 260–266). Also, negative determiners have often been treated together with negative pronouns (e.g., in Haspelmath 1997) or in dedicated studies (e.g., Lindstad 2009) but neither of these focus on how negative pronouns differ from negatively ‘determined’ nominals.

As far as I can see, up to now scholarship has never posited NC as an absolutely obligatory feature of any language. In fact, definitions of NC often (as in Breitbarth et al. 2020: 19; Giannakidou 2006: 328) require NCIs to be possible without concord in one context, viz. the fragment answer context. Thus Russian (6), for instance, requires the ne NC negator in a non-elliptical clause in (6), but in its fragment answer counterpart, ne is impossible.^[5]

Also, in full clauses, negative concord is not always obligatory, as illustrated in (1) and (2) as well as in (3). If one knows the literature, one will be inclined to take (3) to illustrate that the main parameter for the presence versus absence of NC is the position of the negative pronoun relative to the verb. In that literature Russian and Italian have been prominent. Russian has NC both when the negative pronoun is in front of the verb and when it follows it, but in Italian NC only appears when the negative pronoun follows the verb.

This is the distinction generally called, after Giannakidou (1998), ‘strict’ NC, like in Russian, and ‘non-strict’ NC, like in Italian.^[6] Meanwhile, we know that there are other parameters of strictness and also that there can be degrees of ‘strictness’ (van der Auwera 2017; van der Auwera and Van Alsenoy 2016; de Swart 2010). A further parameter of crosslinguistic variation concerns the membership of the classes of NC negators and NCIs – the ‘range’ of NC. Thus, in Russian and in Croatian the privative preposition has a negative meaning, it negates, it is not an NCI, and in Russian (8), but not in Croatian (9) it functions as an NC negator.

What we also find is that what functions as an NC negator in one language functions as an NCI in another one. Jamaican, Belizean and Russian can illustrate this. In Russian the ‘never’ word is an NCI: it behaves in the same way as the pronouns for ‘nobody’ and ‘nothing’ – see (6a) and (10).

In Belizean and Jamaican, however, the ‘never’ word functions as an NC negator.^[7]

Interestingly, in Belizean and Jamaican – and in many other English Creoles and varieties of English (Kortmann and Wolk 2012: 908; Palacios 2019) – the ‘never’ word does not only have the use we know from Standard English and Russian, in which it quantifies over states of affairs. The Belizean and Jamaican ‘never’ words can also negate a single state of affairs in the past – a use that I will gloss as ‘neg.pst’, as compared to ‘never’, the gloss for the use it shares with Standard English and Russian. In the neg.pst use the Belizean and Jamaican ‘never’ word is also an NC negator, as shown in (12).

3 Negative concord in Jamaican and Belizean: general observations

For Belizean we know very little about NC: it is briefly mentioned in Escure (2013: 96–97), van der Auwera (2017: 133–134), and van der Auwera and De Lisser (2019: 13). For Jamaican the situation is different, with van der Auwera and De Lisser (2019) as a reasonably comprehensive study of NC. The current article revisits some of their findings, viz. the ones on nominal and pronominal NC. In this exercise these main findings are confirmed, but there are differences. First, the new classifications are more detailed, in part because the semi-automatic ‘search-and-find’ method of the earlier study left some types ‘under the radar’. Second, in the last resort the findings depend on the analysis of hundreds of sentences and some of these ‘micro’ analyses differ between the two studies. One reason why a ‘micro’ analysis can be tricky is that constructions can have both an NCI and a non-NCI use. Consider (13) and (14), both of which contain a negator together with the phrase fi notn ‘for nothing’.^[8]

In (13) fi notn participates in NC, as the English paraphrase with not … for anything shows, but in (14), I analyze fi notn as a set phrase meaning ‘in vain’,^[9] which is felicitous in positive environments too, as in (15).

In what follows I briefly mention four general findings from the earlier study, and I check, equally briefly, how Belizean NC fares in these respects. A study of these issues with the granularity of the analysis of (pro)nominal NC that follows in Sections 4 and 5 is far beyond the limits of this article.

First, in Jamaican the range of NC negators does not only include the direct counterparts to English not, negative auxiliaries like can’t, mustn’t etc. or never. The privative preposition is an NC negator, too, like in Russian, and unlike in Croatian.

Privatives are a complex matter. The construction in (16) does not only alternate with the non-concordant rendering illustrated in (17), but also with a ‘with’ construction preceded by a negator and followed by either a concordant NCI, as in (18), or a non-concordant nominal, as in (19).

Furthermore, what follows ‘without’ may be a nominal or a pronoun but also something more clause-like. For the latter case the Jamaican text has a construction with a doubly negative widoutn (<widout + n), arguably the privative followed by an ‘expletive’ negator (cf. Déprez 2018: 57–61 on Romance structures).

Suffice it to say here that Belizean ‘without’ is also a negator, but that a counterpart to widoutn is not attested.

If Belizean possibly does not have a counterpart to widoutn, we find the opposite situation with Jamaican possibly not having a counterpart to a Belizean ner ‘nor’ NC negator.

If the noh of noh bai hevn is an NC negator, repeating the noh of noh fi sway, then ner must be an NC negator too. It is unclear how well entrenched this use of ner is though, for the Belizean text only has two attestations, but the Jamaican one does not have any. What is also peculiar is that in the Belizean New Testament ner also functions as an NCI, a use again unparalleled in the Jamaican text. In (22) neva is the NC negator and it is in concord with needa unu and with ner unu ansesta.

Belizean ner thus illustrates a case of one word having both an NCI and an NC negator use.

This brings me to the next point: it is not just the range of NC negators that may vary between languages and between Jamaican and Belizean, in particular, but also the range of NCIs. NCIs can be pronouns and nominals, of course, and, like we have just seen in (22), the conjunction ner. Example (22) also illustrates a Belizean NCI use of the ‘neither’ word, shared by Jamaican. Both languages also use this word clause-finally. In the latter use Standard English uses ‘either’ now, though this clause-final neither existed earlier (Iyeiri 2001: 145–146; Jack 1978: 67).

Both languages also have a ‘not at all’ NCI^[10] and Jamaican has a ‘no more’ NCI.

A third point is that in Jamaican NCI (pro)nouns have developed NPI uses, as illustrated in (26)–(28).

Van der Auwera and De Lisser (2019: 12) found 101 attestations, which compared to the total of 1,180 of their straightforward NCI cases suggests that roughly one of ten attestations of nobadi, notn, and no nominals count as NPIs.^[11] Van der Auwera and De Lisser (2019: 13) offer four arguments for thinking that these NPI uses developed out of NCI uses. First, a development from NPI uses from NCI is attested in other languages (see e.g., van der Auwera and Van Alsenoy 2011; Giannollo 2018: 211–213; Hansen 2012, 2014) – so it would not be unique to Jamaican, even though the opposite development – from NPI to NCI – seems more common (Haspelmath 1997: 226–234; van Alsenoy 2014: 147–181). Second, if, as will be shown, the more prominent exponent of negation in Jamaican NC is the NC negator, then one understands how the negative sense of e.g., nobadi can weaken into a negatively polar sense. Third, there is no trace of the NPI uses in earlier Jamaican texts. Fourth, the NPI uses have not been reported in English Creoles similar to Jamaican. Van der Auwera and De Lisser (2019: 13) further mention that they engaged in a superficial perusal of the Belizean New Testament and did not find any NPI uses. I can here confirm their suspicion. The Belizean New Testament indeed does not have the NPI uses. I illustrate this with the Belizean versions of the verses shown in (26)–(28).

A final point that appears from the earlier study is that Jamaican constellations with two or more NCIs, as in (32), overwhelmingly have a negative predicate.

Example (33) shows a rare case in which two NCIs are arguably not accompanied by an NC negator – although one could venture the hypothesis that niida is not an NCI here but an NC negator. Under the latter hypothesis one would have to accept that Jamaican niida can function both as negator and as an NCI, a variable status possibly needed for Belizean ner.

The very strong preference for two or more NCIs to be accompanied by a negative verb is found in Belizean too. To rephrase this finding in the jargon of the field: both Jamaican and Belizean exhibit ‘Negative Spread’ in the wide definition of e.g., Breitbarth et al. (2020: 20), allowing two or more NCIs to be accompanied by an NC negator, but hardly in the narrow definition of e.g., de Swart (2010: 46), in which the NCIs are not accompanied by an NC negator.^[12]

More common than the structure shown in (33), in both New Testaments, is a constellation with a negative predicate in which one would expect two or more NCIs, but in which one or more NCIs is replaced by a non-concordant word or phrase, as illustrated in (34).

So much for some elements of a general characterization of Jamaican and Belizean NC. I now study pronominal and nominal NC in non-Spread configurations in detail.

4 Pronominal negative concord

As mentioned in Section 2 the main parameter for variation in the study of NC has been the position of the NCI relative to the verb. However, at least for Jamaican and Belizean, the point of reference is not the position of the verb, but that of the predicate, verbal or not. A ‘prepredicative’ NC with a non-verbal predicate is shown in (35).

In what follows I study prepredicative and postpredicative NC separately, both for pronouns and nominals. I abstain from what could be called ‘unpredicative’ uses, i.e., elliptic uses in which there is no overt predicate, as in (36).

The question in (36) received two answers: only the second one has a predicate, and we see that only the second one has NC. The unpredicative so-called ‘fragment answer’ as in Nobadi! has received a lot of attention in the NC literature, and I pointed out just before Example (6) that the behavior of NCIs in fragment answers has even been made part of the definition of NCIs. It is of interest to point out that unpredicative uses are not restricted to fragment answers. Examples (37) and (38) show unpredicative uses that are not in fragment answers, sometimes called ‘resumptive negation’ structures (after Jespersen 1917: 72–75). Note also that there is NC in (37) but not in (38).

Despite the fact that unpredicative uses are interesting, I shelve them for later research.

5 Nominal negative concord

6 Conclusion and further perspectives

This article aims at contributing to our understanding of NC in general through the lens of Jamaican and Belizean NC, as it is represented in the two translations of the New Testament. Let me start by outlining what I take to be the contribution to the general understanding.

First and foremost, this article demonstrates the need for distinguishing between pronominal and nominal NC.^[21] The present study confirms – with a more detailed analysis – the finding of van der Auwera and De Lisser (2019) that Jamaican has much less NC for nouns than for pronouns. This also holds for Belizean. Pace van der Auwera and De Lisser (2019) the absence of NC for nouns is found with roughly the same proportion in the prepredicative and the postpredicative domain. This finding partially confirms also what Jäger (2008: 201–205, 278–282) claims for Old and Middle High German, but her claim only concerns the postpredicative domain. Interestingly, the proportions for the various strategies found for Old and Middle High German postpredicative nominals may be different from the ones found for Jamaican and Belizean. Jäger (2008: 205, 278) finds bare nominals to be more frequent than NCIs, while I find the opposite, both for Jamaican and Belizean. However, Jäger (2008) includes attestations in Negative Spread constructions, which I do not. It is also interesting that Jäger (2008: 282) finds that the proportion of NC for nominals increases as we go from Old High German to Middle German, even though the overall picture is one of a decrease in NC (Jäger 2008: 286–287). This brief and superficial comparison between Old High German, Middle German, Jamaican and Belizean illustrates that the field should devote as much attention to nominal NC as it has done for pronominal NC. This plea for the study of non-pronominal NC should also further the study of the NC of adverbials, like ‘no longer’ and ‘not at all’ and conjunctions like ‘neither’, as well as the study of privative constructions, all touched upon briefly in Section 3.

As to the reason why NC is less strong for nominals, I endorse the view taken by van der Auwera and De Lisser (2019). The bulk of the non-concordant structure employ polarity-neutral constructions. At least in the Jamaican and Belizean texts studied here, nominals can be polarity-neutral and thus be felicitous is negative contexts. Jamaican and Belizean pronouns corresponding to somebody, anybody and nobody are not polarity-neutral: they are either negative, negatively polar or positively polar. This is not to claim that there can’t be polarity-neutral indefinite pronouns. Judging from Example (76) the ‘somebody’ word in Saramaccan could be one.

That a polarity-neutral strategy is important is something we know from the general typology of negative indefiniteness: this strategy is the most frequent strategy world-wide in the analysis of van der Auwera and Van Alsenoy (2016), but it is important to stress that there are languages that do not have indefinite pronouns corresponding to the English somebody series and that these resort to nominal strategies. Van der Auwera and Van Alsenoy (2016: 483) do not distinguish between pronominal and nominal strategies. Judging from what we find in Jamaican, Belizean, Old and Middle High German it is my conjecture that the reason why the polarity-neutral strategy is world-wide the most frequent strategy to express negative indefiniteness is that many languages resort to nominal strategies.

The anarthrous nouns that occur with a negative predicate, it appears from at least the Jamaican and Belizean New Testament texts, have a different profile depending on their position relative to the predicate. In the prepredicative position we exclusively or predominantly find the plural generic noun ‘people’. In the postpredicative position ‘people’ is still the most frequent noun, but it accounts for only one fourth to one fifth of all cases, and there is much variation.

A second general point is that this study has advanced a new take on the ‘Neg Early’ principle, by arguing it to be relevant both for NC negators and NCIs. This is necessary because an undifferentiated Neg Early principle does not explain why a non-strict pattern with an early NC negator and a late NCI, as in the first part of (3), makes sense.

In the first clause noh is early, so on a simple version of the Neg Early Principle, one has to admit that it is satisfied. However, it is only the scope that is expressed and not the focus. This contrasts with the second clause, which, through marking focus, marks scope as well.

A third general point is that studies of NC have not, as far as I know, recognized (sufficiently) that there could be NC with negative definite nouns, as in Example (63). There is no question that King Solomon in (63) is a definite nominal. Of course, it is a very special use, a scalar superlative use implying that it is not only King Solomon that doesn’t look as good as the flowers, but everyone less rich too. It is not just ‘not even’ but also ‘neither … nor’ that makes NC tolerate definite nominals (van der Auwera 2021).

Finally, general studies of NC have also not dealt with the possibility that one and the ‘same’ marker may be an NCI in one language but an NC negator in a second language, and that this variable membership may be found even in one language (as with Belizean ner and Jamaican niida, and also, mentioned in Note 7, London English never). Of course, that words or constructions can be multifunctional is trivial, but the claim that multifunctionality can involve NCI and NC negator uses is not. In this article I put this issue on the agenda for further research.

For Jamaican the current study largely confirms the findings of the earlier study but corrects it on one important point. I found more variation in the prepredicative domain, both for pronouns and nominals. For pronouns, the earlier study missed out on the strategies with sumadi and enibadi. For nouns, the earlier study did not find as many polarity-neutral cases as the present one. For prepredicative non-NCI pronouns I noticed that they occur with restrictive relative clauses or in conditional clauses – a matter that demands further scrutiny.

For Belizean all the findings are new. It is interesting that at least for pronouns and nouns NC seems to work in roughly the same way in both languages. There are, of course, frequency differences, but most are not statistically relevant.

The main statistically relevant difference between the Jamaican and Belizean findings is that the Jamaican text has more NC for prepredicative pronouns. There are also substantial differences that I did not focus on this article. The most important one is that only in Jamaican did the NCI (pro)nouns develop NPI uses. What made this possible, I propose, is that in Jamaican the main exponent for negation is the NC negator, not the NCI. But this is true for Belizean as well. So, the potential for the reinterpretation of NCIs and NPIs is present in Belizean as well. I have not detected any reason why the potential was realized in Jamaican but not in Belizean.

Abbreviations

1

first person

a

answer

Ant

anterior

cop

copula

fut

future

nc

negative concord

nci

negative concord item

npi

negative polarity item

neg

negator

pl

plural

ppi

positive polarity item

prosp

prospective

prs.progr

present progressive

pst

past

pst.compl

past completive

q

question

sg

singular

Supplement: The data underlying this article may be accessed on the Zenodo repository: https://doi.org/10.5281/zenodo.5792083.