The syntax of African American English borrowings in the Louisiana Creole tense-mood-aspect system

1 Introduction

This paper documents two ongoing processes of contact-induced change in Louisiana Creole (LC), a critically endangered French-lexifier creole language spoken mostly in Louisiana in the United States. Many – but not all – speakers of LC are also regularly users of local varieties of African American English (AAE). Contact between these two varieties has resulted in the ongoing incorporation of at least two new morphemes into the LC tense-mood-aspect (TMA) system. The first is a borrowing of AAE stressed BIN, which marks the remote past with a range of aspectual meanings. Some speakers additionally have borrowed the English adverb still, which is grammaticalizing as a marker of continuative aspect.

The borrowing of TMA morphemes is rare in crosslinguistic perspective (Matras 2007) and has not been thoroughly analyzed in LC. Discussion of contact-induced change in creole languages has typically been confined to examination of interactions with the lexifier, the language which contributes the majority of a creole’s vocabulary. Far fewer studies have presented detailed accounts of how creoles behave when in contact with other languages, meaning that this particular contact context remains undertheorized (Aceto 1999; Siegel 2010; Snow 2000).

This paper aims to contribute to the documentation of the first instances of contact-induced change in LC motivated by contact with AAE and, in so doing, to the theory of how creole languages behave when in contact with languages other than their lexifiers. Section 2 presents the necessary background for the study, reviewing the linguistic ecology of LC in Louisiana in historical-sociolinguistic perspective, with a special emphasis on contact with AAE. Section 3 reviews the data used in the study. The analysis itself proceeds in two parts in Section 4, examining how AAE stressed BIN and the adverb still are incorporated into LC. These changes are discussed in broader perspective in Section 5, with special emphasis on how degrees of structural similarity modulate language contact. Concluding remarks are presented in Section 6.

2 Louisiana Creole and African American English in Louisiana

Today, LC is spoken by a dwindling number of elderly people – probably around 6,000 – who are all bilingual in English and have varying levels of competence in local varieties of Louisiana Regional French (Klingler 2003; Klingler and Neumann-Holzschuh 2013; Neumann 1985; Valdman 1997; Valdman et al. 1998). The speech community today is mostly concentrated in rural south Louisiana, although smaller diaspora communities exist especially in east Texas (see Wendte 2020). These rural communities remained insulated from the gradual Americanization of Louisiana which followed the sale of the territory to the United States in 1803, although local bourgeois classes adopted English relatively early (Fortier 1891: 77; Morgan 1970). All this changed, however, in the first half of the twentieth century, when even the most isolated communities underwent a shift to English precipitated by changes in statewide educational and language policy (Dajko 2012; Dubois et al. 2018; Landry 2016).

The Americanization of Louisiana brought with it not only the dominance of English but also the imposition of a Black/White racial binary as part of the antebellum regime of “Jim Crow” racial segregation. This bifurcated rural Creole communities, where the construction of race and ethnicity had previously been negotiated as an interaction between tripartite legal status (blancs, noirs, gens de couleur libres – White, Black, Free People of Color) and a number of phenotypical descriptors (Barthé 2021; Domínguez 1986; Kein 2000; Landry 2016). Race and ethnicity therefore constitute key social constructs for understanding language variation in Louisiana in general, and in LC in particular (Dajko 2012; Klingler 2019; Neumann 1983; Wendte 2020). Previous quantitative studies (Mayeux 2017, 2019; Klingler 2019) have shown that variation and change in LC pattern according to the Anglo-American Black/White racial binary due to differing degrees of contact with French on either side of the Jim Crow divide (see Mayeux 2024).

Race and ethnicity have also shaped the nature of contact with English. It is important to understand that LC exists in contact with local varieties of English which themselves display contact features from LC and Louisiana Regional French. This process of leveling was first noted in publications based on field data collected from older speakers in the 1950s (Morgan 1959, 1970). Studies on English in rural Louisiana identify two distinct varieties based on race (for recent work on urban Louisiana, on the English of New Orleans, see Dajko and Carmichael 2023). Cajun English is the name given to a variety of English spoken by White, Cajun-identified Louisianians from historically French- and LC-speaking communities (Carmichael 2019). Cajun English has been the subject of a number of sociolinguistic studies, many of which comment on the presence of phonological and morphosyntactic features originating in LC and Louisiana Regional French (Carmichael 2013, 2018; Dubois 2002; Dubois and Horvath 1998, 1999, 2001, 2003a). Another variety, which has received less attention, has been termed Creole African American Vernacular English by Dubois and Horvath (2003b). This variety also exhibits various features of contact with LC and Louisiana Regional French which distinguish it as a subvariety of AAE (Dubois and Horvath 2003b; see also Oetting and Garrity 2006). AAE, and its local incarnations, has had an important presence in Louisiana since at least the early-to-mid nineteenth century. Picone (2003) provides the most detailed historical-sociolinguistic overview, demonstrating that the arrival of enslaved and formerly enslaved anglophones in the plantation societies of Louisiana exerted considerable pressure from below in encouraging shift to English. This also resulted in the introduction of contact features into local AAE, for example pronominal tagging (e.g., I got no chance, me; Picone 2003: 421).

Contact with local varieties of English has exerted extensive influence on LC. Best documented are the widespread processes of lexical borrowing and code-switching. The impact on the grammar of LC, however, has so far been subject only to preliminary investigation. Instead, much of the focus has been on the results of contact with French. This follows from an emphasis in Creole Studies on decreolization, the process of contact between a creole and its lexifier, which may result in the former becoming more similar to the latter (Bickerton 1980; DeCamp 1971; Holm 2000; Whinnom 1971). Decreolization has been roundly critiqued (see the comprehensive critical review in Siegel 2010). For our purposes, an important shortcoming of decreolization is that it excludes contexts where creoles are in contact with languages other than their lexifiers (Aceto 1999), as is the case in Louisiana. It is perhaps for this reason that this kind of contact remains under-researched (Bartens 2002; Snow 2000).

3 Data

Data are taken from a diachronic corpus of LC (Mayeux 2019), which combines historical language documentation with fieldwork recordings from 2017. Although the whole corpus was examined for the variables discussed here, no occurrences in historical texts were found. The variables were only attested in recordings made in 2017, which comprise sociolinguistic interviews from 27 speakers in St. Martin and St. James Parishes, Louisiana. Samples averaging 30 min were transcribed and analyzed for each of the speakers represented. Occurrences of the variables in this study – bɪn (Section 4.1) and stɪl (Section 4.2) – were coded. LC examples are transcribed using the International Phonetic Alphabet; English insertions are transcribed orthographically. Corpus examples are followed by 2017XX, where XX is a two-letter code identifying the speaker. Syntax trees were generated using RSyntaxTree (Hasebe 2023).

4 Analysis

Like both other French-lexifier creoles (see Syea 2017) and AAE (see Green 2002), LC encodes TMA through a series of preverbal function words usually referred to as particles or markers (Klingler 2003; Klingler and Neumann-Holzschuh 2013; Neumann 1985). This paper builds on the analysis in Rottet (1992), who finds that TMA and the negator p(a) follow strict ordering before the verb, given in (1). Rottet (1992: 275) analyzes each marker as the spell-out of a functional head, deriving the tense phrase (TP) structure in (2).

Following Henri (2016; see also Henri and Klingler 2014), this paper adopts a preliminary analysis of the negator p(a) as a clitic. The analysis also incorporates Rottet’s suggestion that LC aspectual heads can be “stacked” (1992: 275, n. 12), that is, that more than one AspP may appear in a given TP. This notion is compatible with the hierarchy of clausal functional projections in Cinque (1999). Elements of this cartographic approach are drawn upon here to extend Rottet’s account of the LC inflectional domain.

The present study treats two new borrowings into this system, which both have their origins in AAE: bɪn (< stressed BIN; Section 4.1) and stɪl (<still; Section 4.2). The sociolinguistic and linguistic factors involved in this contact-induced change are discussed in Section 5.

4.1 Remote past bɪn

The presence of the form [bɪn] in LC is attested in Neumann (1985: 211) and Klingler (2003: 262–263). Both grammars of LC classify bɪn as a borrowing of English perfective been (Klingler 2003: 262–263; Neumann 1985: 211). Rottet (1992: 270–272) follows these descriptions. In this section, bɪn is shown to originate from the phonologically identical AAE stressed BIN, which in fact can be straightforwardly incorporated into the phrase structure proposed by Rottet (1992). AAE BIN is a “camouflaged” (Spears 1982) AAE feature, that is, one which is phonologically similar to, but syntactically and semantically distinct from, a feature in other varieties of English. This may explain why LC bɪn has been mistaken as a borrowing of perfective been.

First analyzed as a remote past marker by Labov (1972) and Rickford (1975), AAE stressed BIN indicates that “the time period referred to is longer than normal for an activity, or … that a state has indeed held for a long [period] of time” (Green 2002: 54). Incidentally, this resembles the description for LC bɪn in Klingler (2003: 263) as encoding “an action or state that began before, and continued up to, a subsequent point in time, which may be the moment of utterance or a point prior to it”. The form bɪn is distributed categorically across speakers in the corpus and in other available data (Klingler 2003; Neumann 1985), attested only amongst Black speakers who are proficient in AAE.

AAE BIN is normally transcribed in all capitals, a convention which denotes that it typically bears prosodic stress, unlike its perfective counterpart been (Green 2002: 54). When it is used, bɪn bears prosodic stress as in (3) (⟨ˈˈ⟩ indicates prosodic stress). No counterexample was heard during fieldwork.^[1]

(3)

⟨Vitamin C⟩	ˈˈbɪn	∅	la!
Vitamin C	remp	cop	there
‘Vitamin C has been around for a long time (and still is)!’
(2017EJ)

AAE BIN has varied aspectual readings distinct from those of perfective been. Green (1998a; see also Green 2002: 55–59) finds BIN can encode stative aspect, as in (4a); habitual aspect, as in (5a); and completive aspect, as in (6a). All three are attested for contemporary LC bɪn, as seen in (4b)–(6b).

(4)

Stative

a.

He BIN running.

‘He’s been running for a long time and still is.’

(Green 1998a: 117)

b.

li	bɪn	muri	ɛ̃?
3s	remp.stat	dead	eh
‘He has been dead for a long time (and still is), eh?’
(2017EO)

(5)

Habitual

a.

Bruce BIN running.

‘Bruce started running some time ago and he still runs from time to time.’

(Green 2002: 57)

b.

mo	bɪn	ole	⟨retire⟩
1s	remp.hab	want	retire
‘I’ve been wanting to retire for a long time (and still want to).’
(2017ML)

(6)

Completive

a.

Bruce BIN ate those potato chips.

‘Bruce ate those potato chips a long time ago (and is no longer doing so).’

(Green 1998a: 130)

b.

n	t	ape	kite,	nu	bɪn	ape	⟨pack up⟩	no	zafɛ
1pl	ant	prog	leave	1pl	remp.cmpl	prog	pack up	1pl.poss	thing
‘We were leaving, we had been packing up our things for a long time (and no longer were).’
(2017BM)

Further similarities are found in the domain of negation. As in AAE (Green 1995), bɪn cannot take sentence-level negation, as shown in (7a). AAE negative auxiliaries ain’t or haven’t can be used in metalinguistic negation and ellipsis contexts (e.g., He ain’t BIN running; he only just started; see Green et al. 2022: 964). This may be possible in LC, as seen in (7b), though no examples of this kind of metalinguistic negation are attested in the corpus. When the negator p(a) is placed after bɪn in (7c), another constructed example, the result is a change of scope.

(7)

a.

*li pa=	bɪn	muri
3s neg	remp.stat	die
Intended: ‘He has not been dead for a long time.’

b.

li	pa=	bɪn	muri;	li	ʒɪʃ	sɔr	muri
3s	neg	remp.stat	die	3s	just	aux	die.
‘He hasn’t been dead for a long time; he only just died.’
cf. AAE He ain’t BIN dead; he only just died. (after Green et al. 2022: 964)

c.

li	bɪn	pa=	muri
3s	remp.stat	neg	die
‘He has been not-dead for a long time (i.e., he has been alive for a long time and still is).’

Finally, English perfective been is not rendered in LC using bɪn. Example (8a) contains an extended code-switch into English where the perfective been is used (we’ve been there). This is rendered in LC with a plain verb kuri ‘go’. With bɪn, the sentence would produce a different interpretation, denoting the remote past with stative aspect rather than a perfective aspect (i.e., ‘we’ve been there and still are there’). A similar case can be found in Neumann (1985), reproduced in (8b). The example shows an insertion of been in a calque on the English have been to V construction associated with motion events (see Gesuato 2012).

(8)

a.

paske	⟨we’ve been there⟩	[.]	no	kuri	la
because	we’ve been there		1pl	go	there
‘Because we’ve been there. We’ve been there (and no longer are).’
(2017LD)

b.

m o	te	pa=	⟨been⟩	wa	twa	lɔ̃tɔ̃	e	tɔ̃t?
1s	pst	neg	been	see	2s	long.time	eh	aunt
‘I have not been to see you in a long time, eh Auntie?’
(Neumann 1985: 211, n. 1)

Given its syntactic and semantic properties, I posit that remote past bɪn can be incorporated into Rottet’s (1992) phrase structure as a functional head which lexicalizes stative, habitual, or completive aspects, as in (4)–(6). This is the same position it occupies in AAE (Green 1998b). This Asp_{STAT/HAB/CMPL} is higher than the progressive Asp_PROG, since bɪn must precede the progressive marker where the two co-occur, as in (6b) (though see footnote 5 for discussion of another potential solution for future investigation). This structure is schematized in (9).

(9)

​

This analysis allows for tense markers to appear under T. This structure is not found in my data from St. Martin or St. James Parishes. In the LC of Pointe Coupee Parish, however, two examples where anterior te precedes bɪn are recorded by Klingler (2003: 263). This possible regional variation, which should be subject to further investigation, may result from differing sociolinguistic dynamics in that region. By way of summary, I present one of these examples from Klingler (2003: 263) in (10a) and (10b), which shows a more fully elaborated structure.

(10)

a.

nu	te	bɪn	e	vole
1pl	ant	remp.hab	prog	steal
‘We had been stealing (peaches) for a long time.’
(Klingler 2003: 263)[2]

b.

​

4.2 Continuative stɪl

In another incipient process, an English-origin temporal adverb may be undergoing grammaticalization as preverbal marker for the continuative aspect. The adverb [stɪ(l)] ‘still’ (< English still) is recorded as an alternative to the adverb tuʒu ‘always; still’ (Klingler 2003: 343; Neumann 1985: 27). Today, speakers who also know AAE have replaced tuʒu with stɪl for the sense ‘still’, as in (11). For the sense ‘always’, tuʒu is used, as in (12). I submit that this change has occurred due to reanalysis of this adverb as an aspect marker on the basis of structural similarities between preverbal TMA elements in both AAE and LC.

(11)

no	stɪl	mãʒ	pul
1pl	cont	eat	chicken
‘We still eat chicken.’
(2017BB)

(12)

to	kɔ̃nɛ̃	to	tuʒu	bɛzɛ̃n	kɛkʃɔʒ
2s	know	2s	always	need	something
‘You know, you always need something.’
(2017BB)

Since adverbs can encode aspectual information and may act as the specifiers of functional heads (see Cinque 1999: 95), they represent good candidates for grammaticalization as aspect markers. In languages where adverbials share a similar distribution to TMA markers – as is the case in both LC and AAE – their reanalysis and grammaticalization is robustly attested (Abunya and Amfo 2013; Bybee et al. 1994: 240; Marchese 1986). In such cases, another sign of the grammaticalization of adverbs as TMA markers is phonological reduction (cf. Bybee et al. 1994), as seen, for example, in Nyabo kɛ́ɛ̄té ‘today’ > kɛ̀ɛ̄ prs (Marchese 1986: 258) and Kaakyi ɔ̀kɛ́ ‘tomorrow’ > kɛ̀ fut (Abunya and Amfo 2013). LC continuative stɪl also occurs in a phonologically reduced form [stɪ ∼ sti], even where followed by a vowel, as in (13).

(13)

me	le	piti	astœr	je	sti	ole!
but	pl	child	now	3pl	cont	want
‘But the children nowadays, they keep wanting [things].’
(2017MY)

The first stage in this process of grammaticalization was the substitution of the borrowed adverb stɪl for tuʒu in cases where the latter adverb had a continuative reading (‘still’). Here, stɪl took on the special distribution of tuʒu, which belongs to a closed class of adverbials which can appear preceding the verb (Neumann 1985: 296), as in (14) (cf. example [12]). Other adverbials, for example manner adverbs such as dusmã ‘slowly’, cannot, as is shown in (15a)–(15b).

(14)

no	tuʒu	mãʒ	pul
1pl	still	eat	chicken
‘We still eat chicken.’

(15)

a.

no	pal	dusmã
1pl	speak	slowly
‘We speak slowly.’

b.

*no	dusmã	pal
1pl	slowly	speak
Intended: ‘We speak slowly.’

The second stage in the process concerns how the borrowed adverb stɪl was grammaticalized. I propose that stɪl goes from being the AdvP specifier of Asp_CONT to being the head, as in (16). This change proceeds by spec-to-head reanalysis in accordance with the Head Preference Principle, which holds that grammars will prefer heads rather than a phrases (van Gelderen 2004).

(16)

Within this analysis, stɪl conforms to the principles of ordering outlined in Rottet (1992) and must appear following tense elements such as te, as in (17), or the future a, in (18). Example (18) also shows that, as for bɪn, more than one aspect marker may co-occur, though with ordering constraints. In (18), stɪl precedes the progressive marker ape and may not follow it. The architecture of examples such as (18) can be schematized as in (19) according to Rottet’s analysis.

(17)

li	te	stɪl	dɑ̃	lɛkɔl
3s	ant	still	in	school
‘She was still in school.’
(2017MM)

(18)

n	a	stɪl	ape	pale	krejɔl
1s	fut	cont	prog	speak	Creole
‘We will still be speaking Creole.’
(2017GB)

(19)

​

Finally, as for bɪn, the presence of multiple aspectual heads can also allow for contrastive negative scope (via different merging positions of Neg as per Cinque 1999: 140–141). In (20a), the negator p(a) intervenes between continuative stɪl and progressive ape. Sentences such as (20b) are also judged as acceptable, albeit with a different reading. Sentences such as (20c) may represent the adverbial adjunction of stɪl to the TP (Rottet 1992: 285–286), possibly for focus purposes.

(20)

a.

li	te	stɪl	p=	ape	pale	frãse
3s	pst	cont	neg	prog	speak	French
‘She still was not speaking French.’
= ‘It remained the case that she was not speaking French.’
(2017GB)

b.

li	te	pa=	stɪl	ape	pale	frãse
3s	pst	neg	cont	prog	speak	French
‘She was not still speaking French.’
= ‘It did not remain the case that she was speaking French.’

c.

?li	stɪl	te	p=	ape	pale	frãse
3s	cont	pst	neg	prog	speak	French
Intended: ‘Still, she was not speaking French.’

It is my hypothesis that this reanalysis of stɪl has been facilitated in particular by contact with AAE, explaining why nearly all speakers who employ this feature are proficient in that variety. This is because AAE, like LC, features a distributionally similar set of preverbal aspectual heads. LC-AAE bilinguals are primed to interpret the aspectual semantics on adverbials preceding the verb; since stɪl shares both the distribution and semantics of a TMA marker, it is vulnerable to grammaticalization as part of this system by means of spec-to-head reanalysis in accordance with the Head Preference Principle (van Gelderen 2004).

5 Discussion

To understand the changes-in-progress examined in this study, we must draw on a framework of contact-induced change which acknowledges the interplay between sociolinguistic context and linguistic change, including language-internal mechanisms of reanalysis (Chamoreau and Léglise 2012; Farrar and Jones 2002; Thomason and Kaufman 1988; Weinreich 1953).

The results of the analysis can be understood from a historical-sociolinguistic point of view as the result of differing levels of contact with different superstrates on either side of the Jim Crow divide in the areas where I have conducted fieldwork. In communities which were racialized as Black, LC took on an important role as a marker of ethnolinguistic identity (Maguire 1987). In White communities, though, LC speakers in St. Martin Parish were under pressure to dissociate themselves from Blackness and therefore to accommodate to Louisiana Regional French (Neumann 1983; Tentchoff 1977; see Mayeux 2024). Therefore, although both Black and White communities were in intensive contact with English, it is in Black communities that LC was maintained for longest in St. Martin Parish. One result of this maintenance is the increased presence of local English features – and, in particular, local AAE features – in the speech of Black speakers of this variety of LC.^[3]

The linguistic consequences of these sociolinguistic dynamics are evidenced in the borrowing of bɪn and stɪl and their incorporation into LC as aspect markers. This represents a case of the borrowing of morphosyntactic matter, resulting from the prolonged intensive contact described above. However, the borrowing of TMA morphemes is rare even in situations of intensive contact (Matras 2007: 44). I suggest that the borrowing of bɪn and stɪl has been facilitated not only by the intensive contact between LC and AAE, but by structural congruity between these two systems. Alongside the intensity of contact, structural and lexical congruity have long been known to modulate borrowing (Thomason and Kaufman 1988; Weinreich 1953). For instance, Jones (2015) compares contact between Mainland Norman and French, two very similar varieties, to that between Insular Norman and English, which are more distantly related. She finds the close relationship between Mainland Norman and French facilitates the transfer of grammatical material. The result is that changes in Mainland Norman driven by contact with French are more fully “worked through” than those in Insular Norman driven by contact with English (Jones 2015: 142).

LC and AAE share similar syntactic structures in their inflectional domains, with both marking aspect on functional heads. The borrowings bɪn and stɪl share near-identical distributions in both LC and AAE. Compare the LC sentences in (21a) and (22a) with their AAE renderings in (21b) and (22b).^[4]

In the case of bɪn, this represents a more straightforward case of the borrowing of structurally congruent material. The case of stɪl involves multiple stages, beginning with the borrowing of the adverb still (which has no special distribution in AAE). Once borrowed into LC, stɪl was subsequently grammaticalized via reanalysis from adverb specifier to head of Asp_CONTP in accordance with the Head Preference Principle (van Gelderen 2004; see Section 4.2). In both cases, LC-AAE bilinguals employ similar structures across both linguistic systems which facilitates transfer of material and its grammaticalization.

Finally, I propose a revision to the hierarchy in (1) and (2) (see Section 4) proposed by Rottet (1992; cited in Cinque 1999: 164). The present analysis takes up Rottet’s suggestion that aspectual heads may be “stacked” (1992: 275, n. 12), a notion compatible with the approach in Cinque (1999). The first task is to account for the ordering of aspect markers. It has been shown that bɪn and stɪl must both precede, not follow, the progressive ape. Since no co-occurrences of bɪn and stɪl are attested, their relative ordering constraints are unclear. I provisionally place them within a single point on the hierarchy, pending further research.^[5] Cinque (1999: 140–141) proposes that negation may merge in various positions depending on language- and context-specific considerations. In LC, Henri’s (2016) analysis of p(a) as a clitic allows its insertion in various positions to obtain different scopal readings, as in (20) in Section 4.2. Nevertheless, it must occur below mood (Rottet 1992). These observations are summarized in (23).

6 Conclusions

This study has documented two novel aspect markers, bɪn and stɪl, which have emerged in LC due to that language’s intensive contact with AAE. This relatively rare case of morphosyntactic borrowing has been facilitated in linguistic terms by the syntactic congruity between LC and AAE, both of which employ preverbal aspect markers subject to similar ordering constraints. Now that these novel aspect markers have been documented, future research should seek to uncover the finer details of the LC inflectional domain, notably negation, and situate these findings in crosslinguistic perspective. Future work should also seek to investigate variation in different varieties of LC, notably that of Pointe Coupee Parish, as well as the extent to which AAE contact features are found in Louisiana Regional French.

The borrowing of bɪn and stɪl constitutes a case of contact-induced change which involves the reanalysis and grammaticalization of borrowed material, emphasizing the need to consider multiple causation in language contact (Farrar and Jones 2002; Mougeon and Beniak 1991; Thomason and Kaufman 1988). The results of this study confirm a need for further examination of contact between creoles and languages other than their lexifiers (Aceto 1999; Snow 2000), which may present valuable data for the comparative study of language contact. Creoles and their lexifiers hold a particular sociohistorical relationship and, by definition, share most of their lexical material. On the other hand, creoles and non-lexifier languages share much less lexical material and diverge structurally. The comparison of creole/lexifier and creole/non-lexifier contact may thus yield important insights into the borrowability of material as a function of linguistic similarity however defined (see also Mayeux 2019; Thomason and Kaufman 1988; Yakpo 2021).