**3. Determiners**

#### *3.1. Determiners in Creoles*

A classic summary description of determiners in creoles is Bickerton's (1981) claim that these languages instantiate a "bioprogram" that yields overt definite and indefinite determiners for specific meaning and zero marking for non-specific (or "generic") meaning. However, with the broader perspective on creoles possible today, decades later, this formulation is as questionably "universal" as the one stipulating SVO as fundamental.

Bickerton's characterization applies largely to Atlantic English- and French-based creoles, as well as the West African English-based creoles descended from the former such as Krio and Nigerian "Pidgin," and the Indian Ocean French creoles of Mauritius and Seychelles. However, all of the English-based creoles here are descendants of a single original language (Hancock 1987; McWhorter 1995; Baker 1999), such that they cannot be analyzed as all manifesting a trait independently. All of the French-based creoles of the Caribbean are likely descendants of a single original language (McWhorter 2000, pp. 146–94). Thus, all of these creoles could be seen as manifesting the determiner pattern just 2 times rather than in cross-creole fashion over 25 or more times.

Moreover, the English-based creoles are most imprinted substratally by African languages of the Kwa clade in Niger–Congo, which happen to display the determiner pattern Bickerton observed. Arguments that French creole substrates were similar would also be relevant, with the Gbe languages of Kwa specified by Lefebvre (1998) for Haitian and by Jennings (1995) for French Guyanais. Thus, genetic relationships and substrate influence render the prevalence of this determiner pattern less unexpected than it would seem.

Then, beyond these creoles, the bioprogram determiner pattern is barely in evidence at all. Portuguese-based creoles do not have a definite determiner, with the distal demonstrative instead "bleeding" into a role intermediate between demonstrative and determiner as needed. These languages instead have only an indefinite determiner. Tok Pisin and its sisters also have only an indefinite determiner. Creolized Chinook Jargon had only a definite one, although studies are unclear as to whether this was a demonstrative or a true determiner, given that it was derived from an original demonstrative (*ukuk*) and its phonetic erosion to *uk* cannot be taken alone as an indication that it had changed its grammatical status.

The determiner configuration that Bickerton identified is so common among creoles because the Atlantic English ones are the product of lexifier and substrate languages that all happen to have overt definite and indefinite determiners, with the substrate languages tending to zero-mark the generic. However, this combination of source languages did not always produce the Bickerton configuration (viz. the Portuguese creoles of the Gulf of Guinea, with Edo, having definite and indefinite specific determiners, as its primary substrate). The French creoles, apart from being all likely tracing to a single ancestor (McWhorter 2000, pp. 146–94 argues that even the Indian Ocean creoles trace to the same ancestor as the Atlantic French-based creoles), have always existed amidst heavy contact with French, thus making it especially likely that they would all have definite and indefinite determiners.

Beyond these creoles, among those born of different kinds of source languages, the visible tendency is that creolization most readily yields an indefinite determiner. For example, the Atlas of Pidgin and Creole Language Structures Online survey reveals but one language with a definite but not an indefinite determiner, and this is a pidgin rather than a creole (Yimas-Arafundi Pidgin). Indefinite determiners are a form of new information marking, and there is evidence that markings of this emerge in creoles before markings of given information (cf. McWhorter 2009 on new information marking in Saramaccan).

#### *3.2. Determiners in Signed Languages*

There has not ye<sup>t</sup> been as much cross-linguistic research on determiners in signed languages as on creoles. However, the work that exists suggests that the situation is rather similar.

ASL, for example, has a definite and indefinite determiner (Bahan et al. 1995; MacLaughlin 1997). However, in ASL and signed languages more generally, determiners do not mark generic (non-specific) referents (De Vriendt and Rasquinet 1989), and even specific referents are not marked as obligatorily as in many spoken languages, since they must incorporate referential information (Neidle and Nash 2012, p. 274).

Just as many creoles have definite determination only via recruitment of the distal demonstrative in especially grammaticalized meaning, the distinction between demonstrative and determiner in signed languages is also often a matter defined by a continuum (ibid. p. 271). There is evidence, on the other hand, that in signed languages as in creoles, indefinite determination is entrenched more quickly. Catalan Sign Language, for example, has a richer array of constructions for indefinite determination than for definite (Barberà 2016).

#### *3.3. Implications for the Language Faculty*

A tentative conclusion from creoles and signed languages is that emergen<sup>t</sup> language develops the overt marking of new information before the overt marking of given information. This is consonant with a conception of it being central to language to transmit information, as well as to continuously justify calling upon and sustaining the interlocutor's attention. Also relevant here is Scott-Phillips' argumen<sup>t</sup> (Scott-Phillips 2015) that language would have emerged from an ostensive imperative of seeking attention for information transfer, such that pragmatics of this kind are fundamental to human language while syntax and morphology are ontogenetically secondary.

#### **4. Subordinate Clauses**

#### *4.1. Subordination in Creoles*

The overt marking of subordination is universal in creoles. They differ only in the obligatoriness of the marking (which tends to be modest) and in how wide a range of subordination constructions is marked.

All known creoles have an overt relativizer, either a pronoun or "particle." This element is almost always optional, but nevertheless robustly conventionalized, as in Saramaccan:


In Tok Pisin, the development of such marking from the pidgin to the creole stage has been observed, with one relativizing strategy grammaticalizing a pragmatic usage of *ya* "here":


Most creoles also have an overt marker of sentential complementation. This has often been grammaticalized from the verb "talk" or "say," and in many creoles beyond the literal semantics of speech; cf. *táa* (> *táki*) in Saramaccan:


Other creoles grammaticalize other words for the function, such as "how" in Santome Creole Portuguese:

(4) Ê na ta sêbê ku(ma) kwa sa pe dê fa. he NEG PAST know COMP thing be father his NEG "He didn't know that it was his father." (APiCS)

Only in many of the Atlantic French-based creoles is there no reported marker of sentential complementation, including claims that recruitments of French *que* as *ki* are borrowings rather than integral to the creole (Peleman 1978).

Thus, while pidgins indeed tend to lack overt markers of subordination, creole languages offer, for example, no support to claims that embedding is incidental rather than integral to the language faculty (Sampson 2005; Everett 2005).

This could be treated as evidence of a transfer from the source languages. However, creoles only incorporate a subset of the grammatical features their source languages offer, *even when all of the languages offer the same feature* (McWhorter 2012). For example, creoles can lack definite or indefinite determiners even if their lexifiers and/or substrate had them (cf. above). However, creoles do not eschew the overt marking of subordination in contrast to such marking in source languages. Such marking would appear to be integral to spoken languages' emergence and genesis.

#### *4.2. Subordination in Signed Languages*

In signed languages, too, there is evidence that the development of embedding is fundamental to the emergence of language (cf. Liddell 1980).

In the younges<sup>t</sup> signed languages such as certain village-based ones, embedding is absent in the first generation (cf. Sandler et al. 2011). An especially useful study is Kastner et al. (2014), documenting the emergence of subordination in the young Kafr Qasem Sign Language, which has begun at what could be called a "pidgin" stage but has developed its own type of subordination through prosodic blending of the embedded modifying expression, accompanied by non-manual signals.

In an older sign language like ASL, analysts have documented that along with a raised brow (Liddell 1980), backwards head tilt, and raised upper lip, relativization can be indicated with an overt "complementizer" sign, a postposed manifestation of "that" (ibid. p. 150):

(5) IX FEED DOG BITE CAT THAT THAT "I fed the dog that bit the cat."

Branchini and Donati (2009) also note a sentence-final relativization particle in Italian Sign Language, and manual signs for relative clauses have also been described in German Sign Language (Leuninger 2005) and Hong Kong Sign Language (Tang and Lau 2012). In older signers of Israeli Sign Language, there was no systematic marking of relative clauses. Nonmanual markers for relative clauses (Nespor and Sandler 1999) only became systematic in the second generation of the emergence of this language, and in the third generation, a manual relative pronoun emerged (Dachkovsky 2020. Both manual and nonmanual markers of relative clauses in ISL are seen in Figure 1.

**Figure 1.** 'The girl who is eating ice cream is swinging'. The relative clause is marked nonmanually by squint and head movement forward to the end of the clause, and manually by the clause-final relative pronoun pointing sign. There is a prosodic break between the relative clause (GIRL EAT-ICE-CREAM IX) and the rest of the sentence (SWING). (Dachkovsky 2020). Pictures courtesy of the Sign Language Research Lab, University of Haifa.

The overt marking of sentential complementation is documented in many signed languages. Padden (1988) notes that ASL marks sentential complementation with a final pronoun copy that refers to the first, matrix subject in embedded structures (6a), but must refer to the subject of the second clause in coordinate structures, so that (6b) is ungrammatical.


 \*1HITi, iINDEX "Ihithimandhe toldhismother,Idid."

Strategies for marking sentential complementation are multifarious in signed languages; however, the ASL construction is in no sense a default. In Israeli Sign Language, a relativizer has developed in a fashion familiar in spoken language: from a locative deictic sign (Dachkovsky 2020). But in Dutch Sign Language, direct speech complements can be marked with the sign for "attract attention" (Dutch *roepen*) (Van Gijn 2004, p. 36):

(7) rightASKsigner ATTRACT-ATTENTIONsigner IXaddressee WANT COFFEE "He/she asks me 'Do you want any coffee?'"

In Hong Kong Sign Language, embeddedness reveals itself in otherwise unexplainable ungrammaticalities. Examples can be found in direct argumen<sup>t</sup> questions, in which whwords can be sentence-initial or sentence-final, but when embedded can only be sentencefinal (Tang and Lau 2012, p. 353):

(8) FATHER WONDER \*(WHO) HELP KENNY WHO "Father wondered who helped Kenny."

In Hong Kong Sign Language, as in others, sentential complement subordination can also be accompanied by non-manual signs such as shakes of the head, leaning of the body.

#### *4.3. Implications for the Language Faculty*

If, as Everett (2005) famously argued, the Amazonian language Pirahã lacks embedding, his proposition that this invalidates the idea of embedding (and thus recursion in general) as a feature of Universal Grammar is not supported by emergen<sup>t</sup> languages of either the spoken or manual modality. Creole languages develop overt markers of subordination quite readily in the transition from pidgin to creole: no creole could be recruited as support for a claim that Universal Grammar lacks recursion. Meanwhile, signed languages also quickly develop markers of both relativization and sentential complementation, with only the younges<sup>t</sup> sign languages lacking these (just as pidgins do in contrast to creoles).

#### **5. Tense and Aspect Marking**

*5.1. Tense and Aspect Marking in Creoles*

Bickerton (1981) claimed that creoles universally share a three-way contrast between three preposed "particles" marking:


Predictably, creoles' marking of tense and aspect has been shown to be less uniform than Bickerton implied. Not all creole past markers are "anterior" ones that express a past-before-past with dynamic verbs (Mauritian Creole French's *ti* dose not, for example). Saramaccan has dedicated habitual markers, one of them well-entrenched in the grammar (*ló* > *lóbi* "love"), rather than expressing the habitual via context with the progressive marker *tá*. Many creoles do not have a single irrealis marker but distinct ones for future and potential (cf. Saramaccan's *gó* vs. *sá*), etc.

However, missed amidst the critiques but worthy of remark is the fact that creoles do share a core of three markers of, roughly, past, progressive, and future. As ordinary as this may seem superficially, the questions are why:


 same basic trio of markers, including Nubi Creole Arabic.

If creoles actually were not distinguishable as a class, then they would presumably diverge considerably in their choices of tense and aspect markers, just as older languages do.

A riposte could be that creoles were too recently imprinted by their source languages to have morphed grammar-internally to this extent. But if source languages determined which tense and aspect markers each creole had, we would still expect much more variation. Why does there not exist an English-based creole in which the habitual is dedicatedly marked with zero as it is in English, or in which the progressive marker is extended beyond the continuative to the present as it is in English? To extend the analysis to mood, why has no creole emerged with a monomorphemic marker of the conditional as there is in western European languages, except ones highly decreolized towards, for example, English?

Creoles have instead regularly selected from their source languages three features, while omitting to incorporate the many others. Bickerton's characterization of these three elements was too specific, but the heart of his observation was correct.

#### *5.2. Tense and Aspect Marking in Signed Languages*

In signed languages, the expression of aspect is universal, but many do not obligatorily express tense overtly (Friedman 1975), instead using adverbials where necessary (such as Israeli Sign Language (Meir and Sandler 2008, p. 89)). Where some signed languages are claimed to express tense, it is on subsets of verbs rather than all of them. ASL has what Aarons et al. (1995) term lexical tense markers, distinguishable from adverbials in occurring inside the VP, while Jacobowitz and Stokoe (1988) identified flexion as marking past and

extension as marking future in a subset of ASL verbs. British Sign Language has some verbs that have distinct signs in the present and past (Sutton-Spence and Woll 1999, p. 116).

However, signed languages have richer arrays of ways to mark aspect than tense, often combining manual and non-manual signs (Pfau et al. 2012a, p. 196). For example, Israeli Sign Language, while leaving tense unmarked, has three aspectual markers (Meir and Sandler 2008, p. 91). In addition to the common marking of the continuous and the habitual, for example, with repetitive movements and different path shapes (e.g., Klima and Bellugi 1979; Meir and Sandler 2008), one of Israeli Sign Language's aspectual markers is a perfect marker glossed ALREADY (Meir 1999), somewhat similar to the ASL marker FINISH. Whereas signed languages typically restrict tense marking, if present, to a subset of verbs, Sign Language of the Netherlands has a habitual affix for verbs whose signs cannot by iterated physically (Hoiting and Slobin 2001). Signed languages seem as prone to develop aspectual marking as creoles are to developing a trio of markers of past, progressive, and future. Examples of temporal aspect marking in ASL are shown in Figure 2.

**Figure 2.** Temporal aspects in American Sign Language (Klima and Bellugi 1979) (**a**) LOOK (citation form), (**b**) LOOK Habitual, (**c**) LOOK Durational, (**d**) LOOK Continuative. (With permission from Urusula Bellugi).

#### *5.3. Implications for the Language Faculty*

Signed languages, in particular, sugges<sup>t</sup> that despite the traditional centrality of tense to the analysis and pedagogy of Indo-European languages, aspect is more fundamental to the human language capacity. There is typological support for this as well, in that while there exist languages that mark aspect but not tense (Chinese and many East Asian languages), there seem to be few to none that mark tense but not aspect.

Creole languages regularly mark tense as well as aspect. However, the bias in extant creoles' source languages may play a part here similar to the part it plays in word order and the presence of overt determiners. In any known language emerging from a pidgin-level variety, all or at least most of the source languages have both past and future markers (one exception being that Niger-Congo's Ewe lacks a past marker, but no creole is known to have had this language as its main substrate as opposed to one of many). For example, there is no creole based on languages like Chinese and Vietnamese, which might show us a creole emerging with no marking of tense.

The closest example is varieties of Malay/Indonesian born of second-language acquisition by various peoples of Indonesia and contiguous areas. Some of these have been classified as creoles. Baba Malay, for example, is Malay acquired incompletely and affected by transfer from Hokkien Chinese. Baba Malay has a marker of perfective aspect rather than past tense, and no grammaticalized marker of futurity as a category (Lee forthcoming). However, the incomplete acquisition in Malay/Indonesian cases like these was not as extreme as pidginization, and both the perfective marker and the adverbially marked future are Malay/Indonesian features.

Still, there are suggestions even within creoles of the centrality of aspect. In creolized Chinook Jargon, as well as the pidgin variety, amidst the highly limited amount of grammatical machinery there was an aspect marker but no tense markers. Also, creoles develop new aspectual constructions more readily than new tenses.

In Saramaccan, beyond the past, future, and progressive markers, the ones in this table have emerged as well, via grammaticalization of verbs, as shown in Table 1:

**Table 1.** Aspect and mood markers in Saramaccan beyond the "big three".


Of the six, four are aspectual. Saramaccan has not developed a pluperfect, remote future, or narrative-present construction. Aspect would seem to have been felt more urgen<sup>t</sup> to express.
