The (Non-)acquisition of the Chinese Definiteness Effect: A Usage-Based Account

Abstract

This chapter investigates the acquisition by French L1 learners of the Definiteness Effect (DE) that characterize Chinese existential–presentational constructions (EPC). Building upon a video-retelling task, oral elicited productions of 15 French advanced learners of L2 Chinese are analysed. In contrast to previous research on L2 DE, mainly conducted within generative approaches to second language acquisition, the present study adopts a functional, usage-based framework and reports on non-target-like performance at advanced levels of acquisition. It is argued that learners are aware of the DE that characterize the EPC in the target language, which is shown by the marginal use of definite pivots in referent-introducing EPCs. By contrast, what they seem not to be aware of is that the EPC should not be used in reintroduction contexts. As a consequence, learners use the EPC format when discourse-old referents are concerned. Strictly speaking, however, they do not ‘violate’ the definiteness restriction, since a different form, with a different function, is operating in the interlanguage.

1 Introduction

This paper investigates the acquisition of the Definiteness Effect (DE) that characterizes Mandarin Chinese (hereafter: Chinese) you existential–presentational constructions¹ (EPC), focusing on the production of French learners of Chinese as a second language (L2). The DE in French-speaking Chinese L2 learners is studied here within the framework of functional approaches to language learning, by way of using an elicited production task. The (non-)acquisition of the DE is considered in a holistic perspective, that is, linking the phrasal rules (i.e. DE) to the rules of contextualization (Lenart & Perdue, 2004), and considering the (in)definite expressions occurring in learners’ EPCs with respect to their general role in the interlanguage system.

DE is defined as a constraint on the occurrence of definite NPs in certain contexts, with existential sentences being the most prominent context for DE to appear (Milsark, 1977). Across languages, EPCs show a certain sensitivity to definiteness, defined as a tendency for definite NPs not to appear in these constructions (Leonetti, 2008). A large body of the relevant literature has focused on DE in English there existential sentences (1).

• (1) There is {a man/??the man} in the garden.

To account for DE, several explanations have been offered (see McNally, 2019 for a recent review). Lyons (1999: 46) claimed that the DE ‘is more likely to be a semantic or pragmatic constraint than a syntactic one. It has much in common with the constraint on indefinite subjects or topics […]; that too is a strong cross-linguistic tendency, stricter in some languages than in others, and also involving something broader than grammatical definiteness’. In McNally (2019: 1839) terms, ‘no analysis that appeals exclusively to the form of the pivot will account for the definiteness restriction’. Hence, there is general consensus on the role of the information-structure (IS) articulation over the manifestation of DE. As Leonetti (2016) puts it, ‘[t]hat IS is relevant for the appearance of DEs in existential sentences and related constructions is hardly a novel insight. It has been repeatedly pointed out that pivot DPs [determiner phrases] are typically focal and existential sentences are central cases of thetic constructions, i.e. constructions lacking an aboutness topic, typically receiving an ‘all-focus’/‘all-new’ interpretation’ (2016: 80). In this view, DE thus results from the incompatibility of definite pivotal NPs with the primary function of EPCs, which is that of introducing a novel referent into the discourse (Lambrecht, 2000).

Along with the DE, its counterexamples have long been acknowledged as well (see Rando & Napoli, 1978; Lumsden, 1988; Abbott, 1993; Ward & Birner, 1995, among others). For instance, in (2), the second occurrence of a there construction involves a demonstrative NP in the pivotal position, which refers anaphorically to the expression one flight.

• (2) I think there was one flight where

  we had one problem. It wasn’t ours,

  but there was that one flight. (Ward & Birner, 1995: 727, reported in McNally, 2019: 1834)

To account for sentences like (2), Abbott (1993) and Ward and Birner (1995) have argued that existential sentences can also serve to reintroduce or focalize referents that have already been evoked in discourse. The deictic and the so-called ‘list-reading’ existential sentences are also known to accept definite pivots (Rando & Napoli, 1978).

The literature on the existential sentences also provides extensive discussion of exceptions to the DE in different languages (Leonetti, 2008; Bentley, 2013, Bentley and colleagues, 2013, 2015). Beaver and colleagues (2006) pointed out that the cross-linguistic variation in the definiteness restriction is not to be understood as an absolute value; rather, languages will select different intervals in the continuum of ‘definiteness’, with the DE on the pivot correlating with subject canonicality, which also is a gradient phenomenon, showing relative cross-linguistic variation itself (Bentley, 2013).

This brings us back to the languages considered for the present study, i.e. French (learners’ source language) and Chinese (learners’ target language). First, French and Chinese differ with respect to the morphological marking of definiteness, the former being an article language, while the latter is a language lacking articles. As far as DE is concerned, the two languages show a contrasting behaviour as well, with Chinese EPCs manifesting a stronger DE than French (see Lena, Forthcoming (a)).² Chinese and French both have an EPC governed by the existential predicator (Creissels, 2019), yǒu ‘have, exist’ (3a) and il y a ‘there is [lit: there has]’ (3b), respectively:

While in French strong expressions can appear within the EPC (4a), the Chinese you-construction manifests DE (4b):

The sentence (4b) is considered a violation of the DE because of the use of the proper noun Yuēhàn—proper nouns being definite by definition, since they are direct labels for particular referents (Chafe, 1976: 39).

Most research has focused on the acquisition of the English DE, and to date, no studies have addressed the L2 acquisition of the Chinese DE—although some DE-relevant findings are reported in Yang and colleagues’ (2007) study on the L2 acquisition of Chinese existential patterns, which I return to in a moment. However, it should be noted that the there construction—the typical context showing DE in English—do not represent a major IS strategy in English, given that intonational devices are more often used in this language to express focus articulations (Vallduví, 1991, Lambrecht, 1994: 318, Sasse, 2006). By contrast, French speakers learning Chinese are facing an L2 which makes extensive use of EPCs as an IS device, as their source language does (Klein, 2012; Lambrecht, 1988).

Counterexamples to the DE have been documented in Chinese as well (see Lena, 2020c: 208–213, Forthcoming (a) for a review). Huang (1987) showed that DE is not observed when a full lexical NP appears in the topic position (e.g. túshūguǎn yǒu nèi ben shū [library exist that CL book] ‘there is that book at the library’). Li (1996) argued that two types of you-constructions should be identified, one introducing a new entity and the other asserting the existence of an event, with only the former manifesting DE. Hu and Pan (2007) discussed the role of adverbs such as hái ‘also’ or zhǐ ‘only’: ‘what is unnoticed in the literature is that, although it is generally excluded from the post-you position, a definite NP can occur there if a focus particle is introduced into the relevant sentences’ (Hu & Pan, 2007). Cai (2000) and Xia (2009) report on the contexts that allow definite NPs within the you-construction, showing that these are subject to strong constraints and rarely found in main clauses; when this is the case, definite-pivot EPCs in Chinese are accepted in some restricted contexts, and cannot stand in isolation.

Despite these exceptions to Chinese DE, it is generally assumed that in the great majority of contexts only weak expressions such as indefinite NPs can occupy the pivotal position (3a). Therefore, native speakers (NS) of Chinese are not likely to accept EPCs including a strong pivot such as (4b). While sentences like (4a) are perfectly acceptable in French, from a statistical point of view the il y a construction does tend to include indefinite NPs, as Karssenberg’s (2018) corpus inquiry showed.³ This feature is congruent with their main pragmatic function of introducing non-topical referents into discourse, which are typically unidentifiable (‘new’) referents.⁴ There is, however, a different type of il y a construction, which is used to express a focus-background articulation, being more closely linked to the expression of strong NPs (Lambrecht, 1988: 154, Karssenberg, 2018: 63). These forms are often triggered in questions-answers contexts:

• (5) (Context: ‘What’s your favourite TV show right now?’)

  “How I Met Your Mother” c’est génial, y’a aussi “Lost” qui est bien.

  ‘“How I Met Your Mother” is great, there’s also “Lost” that is good.’ (Karssenberg, 2018: 63).

Building on a story retelling task (Sect. 5.2), the current study only reports on the first EPC type, used to introduce—and eventually reintroduce, as we shall see it later—referents into discourse and is therefore not concerned with constructions having an IS articulation like (5). It is nonetheless useful to keep in mind that EPCs in learners’ source language (i.e. French) are highly multifunctional, with the focus-background type being strongly connected to the expression of definite NPs.

In what follows, I begin by discussing how (in)definiteness is marked in Chinese.

2 (In)definiteness in Chinese

As an article-less language, Chinese lacks the grammatical category of (in)definiteness, which of course does not mean that such a distinction cannot be expressed in this language.

LaPolla (1995) describes the possible encoding of referents in Chinese on the basis of their identifiability and accessibility status (Ariel, 1990; Chafe, 1994; Gundel et al., 1993; Lambrecht, 1994: 78, 106; Prince, 1981). The [(numeral +) classifier] sequence, leaving aside its non-referential uses (Chen, 2003), is mainly used to express unanchored non-identifiable referents, that is, prototypical (quantified) new referents—the ones marked by the indefinite article in languages such English and French (e.g. yí ge lièrén ‘a hunter’ in [6]). By contrast, nouns modified by demonstrative determiners zhè ‘this’ and nà ‘that’ are used to encode identifiable referents, that can be in one of the three activation states—active, accessible or inactive (Lambrecht, 1994: 109)—according to LaPolla (1995). The NP zhè zhi gǒu ‘this dog’ in the sentence below (from Chen, 2004 : 1153) points to an activated referent thus offering an example of the anaphoric use of the demonstrative determiner:

While acknowledging that the demonstrative determiners in Chinese are ‘the closest to definite articles in other languages’ (ibid.: 1151), Chen illustrates the differences between the two. Zhè ‘this’ and nà ‘that’ serve all the typical functions of demonstratives, as they are used in situational, discourse deictic and contrastive anaphoric contexts. In addition, they can be used where demonstratives (say, in languages like English) are not allowed: non-contrastive anaphoric, shared knowledge and frame-based association (ibid.: 1151–1153). However, uses characteristics of fully grammaticalized definite articles, like those marking shared specific and general knowledge, and frame-based association, ‘are exceptional rather than the norm’ with Chinese demonstratives (ibid.: 1156). Similarly, Crosthwaite et al. (2018) show that Chinese speakers more often use bare nouns (BNs) when establishing bridging reference (i.e. frame-based anaphora).

Indeed, besides nouns that are marked (either by a numeral or by a demonstrative), Chinese speakers can make use of BNs as referring expressions (e.g. gǒu ‘dog[s]’). Their interpretation with respect to definiteness is not straightforward. Indeed, BNs as referring expressions present the highest degree of ambiguity, given that they can designate unidentifiable, inactive, accessible and active referents (LaPolla, 1995: 305, see also Lena, Forthcoming (b)). Between accessible referents, BNs can denote deictically identifiable referents and uniquely identifiable referents (Cheng & Sybesma, 1999: 510, Chen, 2004: 1165. Finally, BNs achieve an indefinite interpretation in some positional contexts (Chao, 1968: 76; Li & Thompson, 1981: 510; Xu, 1995; Hole, 2012: 61): they are generally interpreted as indefinite when appearing as postverbal subjects and within a you-construction (besides the generic-partitive readings that are not relevant for the purpose of the present study, see Lena, Forthcoming (b), for details).

In sum, even if most of their functions overlap, demonstratives and numerals in Chinese do not cover the same range of functions as articles in article languages. As discussed in the following section, the acquisition of the article system is documented to be a hard task for the L2 learner—particularly if the L1 lacks articles—who is inevitably faced to the multifunctionality of the forms involved. Conversely, speakers whose L1 is an article language learning an article-less language such as Chinese will be confronted to the repertory of referring expressions available in the target language to express (in)definiteness, most of which are multifunctional as well. Chinese BNs are expected to be particularly challenging for the L2 user, not only for their morphologically unmarked form but also because the overall effect of the sentence pattern, and the discourse context, over the (in)definiteness interpretation.

3 Previous L2 Research on the DE

Previous studies have mostly focused on the acquisition of the L2 article system, where a great amount of research has been conducted on the acquisition of articles in English (e.g. Berry, 1991; Grannis, 1972; Master, 1997; Zobl, 1984). The misuse of definite and indefinite articles in a second language is often analysed as a positive or negative transfer from the source language. That is, the presence of a similar article system in the L1 should facilitate the acquisition of the article system in the target language, but when the mother tongue has no equivalents, this can result in an increased difficulty for the learner. For instance, Zobl (1984, reported in Towell & Hawkins, 1994: 9) notes that the acquisition of a and the in English L2 is faster for speakers whose first language distinguishes between definite and indefinite determiners (e.g. French, Spanish), compared to learners of article-less L1 backgrounds (e.g. Chinese, Russian). Based on an oral elicited picture description task, Sleeman (2004) compared the acquisition of definiteness distinctions by Dutch- and Japanese-guided L2 learners of French and shows that Dutch (an article language) speakers performed better in oral speech than Japanese (an article-less language) L1 speakers.

A few L2 studies have focused on the DE in relation to existential constructions. The case studies conducted by White (2003) and Lardiere (2005) analysed the production data from one advanced Turkish speaker of L2 English and one advanced Mandarin speaker of L2 English, respectively. White’s (2003) study built on spontaneous production data collected through a series of interviews and reported no DE violations, ‘even though the subject did make errors in article suppliance, in the form of omission’ (White, 2003). Lardiere (2005), in her study of a Chinese learner of English, similarly reports no DE violations. Within her dataset, she obtains 37 contexts for existential there-constructions, and no definite articles were produced in any of them, despite the fact that the speaker ‘tends to overuse definite more than indefinite articles overall’ (Lardiere, 2005).

Building on larger data, King et al. (2006) found that Chinese speakers of low intermediate proficiency in English did not distinguish between DE violations and equivalent grammatical sentences in a grammaticality judgment task (GJT). More advanced subjects, on the other hand, showed target-like sensibility to DE.

Yang et al. (2007) studied the acquisition of the Chinese existential patterns by low to intermediate learners of different L1 backgrounds (English, Japanese and Korean). To collect the data, a GJT (Fig. 1), a guided composition task (Sasaki, 1990) and a free composition task were used.

Fig. 1

Example of (natural) target item used in Yang et al. (2007) GJT (my translation)

While Yang et al.’s (2007) study was not focusing on the DE, several items included in the GJT present interesting findings in this regard. You-constructions are generally rejected by Chinese NS when including a definite pivot. However, the acceptance rate is higher when a preverbal locative is added. ‘This shows that although Chinese speakers can accept definite nouns in existential sentences, the acceptance rate of definite nouns is lower than indefinite nouns, because the main function of existential sentences is to introduce new information’. (Yang et al., 2007, my translation). Interestingly enough, English and Japanese learners are less prone to accept definite nouns appearing in EPCs than Chinese NS, thus not accepting sentences like zhuō-zi = shang yǒu nà-ben shū ‘on the table there is that book’ or fángjiā = li yǒu Xiǎo Mín ‘in the room there is Xiaomin’.⁵ While Yang et al.’s (2007) study is informative with respect to the acquisition of different features of the existential patterns in Chinese, a satisfying explanation regarding learners’ treatment of the DE is not really provided. They seem to take into account a possible L1 functional transfer, though their argument is essentially limited to English there-sentences. I suggest that uncontextualized target sentences might not be the most useful way to investigate this issue: given that the prototypical function of EPCs in Chinese is that of introducing new referents, learners might not be able to imagine contexts in which a definite pivot is allowed (see Sect. 6). Their study shows, nonetheless, that low to intermediate learners of L2 Chinese do not manifest DE violations; on the opposite, they fail to identify DE exceptions.

White (2008a) makes use of an elicited production task (based on pictures description) from 18 Turkish speakers and 15 Mandarin speakers of various proficiency levels. No DE violations are reported in her study. In another study (also based on a picture’s description elicited production task), White (2008b) show that intermediate and advanced L1 Chinese speakers of L2 English are sensitive to the DE, ‘treating there-insertion constructions in a native-like way despite non-native performance on articles in general’ (White, 2008b).

Similar conclusions are reached by Yu and Su (2011), who carried an online contextualized GJT on 50 Chinese intermediate and advanced learners of L2 English, and reported no DE violations. The study by White et al. (2012) was also based on a GJT where each test sentence was preceded by a short context (see Fig. 1 for an example), and found that advanced Russian and Turkish learners of English could perform just as well as NS on judgments of DE. The authors argue that the participants’ ability to judge English sentences appropriately cannot be explained in terms of L1 transfer.

Replicating White et al. (2012)—with minor modifications—Snape and Sekigami (2016) found that Japanese L2 learners of English at advanced levels of proficiency are able to differentiate between grammatical and ungrammatical EPCs, while the intermediate learners can correctly identify grammatical items but cannot detect DE violations. Hence, sensitivity to the English DE seems to improve along with the learner’s proficiency level. The authors claim that the presence of the DE in Japanese is likely to aid acquisition of affirmative EPCs in L2 English, despite the absence of articles in Japanese. Note, however, that their study, as White et al. (2012) did, also included negative EPCs, which can include weak or strong expressions in Japanese, but only weak expressions in English. They conclude that advanced learners’ sensibility to DE in English negative EPCs cannot be explained in terms of a facility prompted by their source language in this case.

In general, the results from previous longitudinal studies suggest that second language learners become sensitive to definiteness restriction in English as their linguistic competence becomes more target-like over time. No violations of the DE are observed at advanced levels of the acquisition. By contrast, the current research reports on DE ‘violations’ produced by adult French advanced learners of L2 Chinese. These are analysed as the results of a functional transfer of the pragmatic functions that the EPC can serve in the source language (i.e. French). Note that all the studies on the DE presented in this section—with the exception of Yang et al. (2011)—are conducted within generative approaches to second language acquisition (SLA).

Before turning to the data and the results of the current study (Sect. 5), what follows presents previous research on learners’ acquisition of the referring expressions in the target language, conducted within the framework of functional approaches to SLA.

4 Definiteness Effect in the Light of Previous L2 Research on the Acquisition of Reference

While not directly addressing learners’ ‘sensitivity to DE’, various studies conducted in a functionalist framework analysed the L2 acquisition of the linguistic forms that enable discourse construction and anaphoric linkage (Ahrenholz, 2005; Chini, 2005; Leclercq & Lenart, 2013, Ryan, 2015, Lenart, 2020). By highlighting the complexity of the form–function relationship, past research has studied the different linguistic means used to encode the noun (e.g. the use of determinants) and their relations with the expression of different referential values (e.g. introduction vs. maintenance of reference) in the productions of learners with various levels of competence (Lenart, 2006, Lenart & Perdue, 2004, Watorek, 2004, Watorek et al., 2014). Longitudinal studies have shown an early preference for distinguishing between definiteness and indefiniteness by marking only definiteness⁶ (Chaudron & Parker, 1990).

In a functionalist approach, the acquisition of the L2 article system is considered in a holistic perspective, assuming that the information expressed morpho-syntactically by the articles can be conveyed by other linguistic devices, be them lexical, morphological or positional. Past studies have shown that, even if the target forms are not mastered, nonnatives can implement distinct ways of manipulating NP forms for different discourse contexts: ‘even with less than perfect mastery of the target system, learners still use different forms of the system to make discourse distinctions’ (Chaudron & Parker, 1990). For instance, Lenart and Perdue (2004) studied the oral narrative of Polish (an article-less language) adult basic learners of L2 French and reported several problems concerning the omission of nominal determiners. Yet, these learners were fully capable of organizing their narrative according to pragmatic organizational principles, where referential continuity was ensured by nominal and pronominal anaphora or by the absence of any marking (zeros).

The complex operation of building a coherent narrative involves the interaction between the discourse level and the utterance level. Two sets of rules—phrasal rules and rules of contextualization—interact in determining the learners’ use of a particular form in a particular context (Lenart & Perdue, 2004). It has been shown that the interaction between the two may vary across languages (Lambert et al., 2008) and also within the same language depending on the speech type (Watorek et al., 2014). As Ryan (2020) puts it, ‘[t]his dual processing demand, lying as it does at the confluence between pragmatics and grammar, has proved an intriguing site for SLA research’. In the area of noun reference, learners have not only to master the internal structure of the NP but also the appropriate use of its realizations in a given situation (Lenart & Perdue, 2004).

When producing EPCs, speakers are effectuating two related yet distinct operations. The selection of a sentence pattern goes along with the choice of a referring expression which appropriately designate the NP referent—both operations being dependent on the pragmatic context. That is, the use of an EPC is motivated by the IS articulation, while the NP form is related to the discourse status of the referent involved. In Chinese, for instance, EPCs are strongly linked to the referent-introducing function, and the referents denoted by the pivot NPs tend to be brand new (Li & Thompson, 1981: 612). In French, the EPC is also frequently used to introduce referents into the discourse, but these can have an accessible status, which is reflected in the definiteness of the pivot (e.g. Jean in [4a] which belongs to the common ground but is discourse-new).

In this study, the DE is considered in relation to the discourse status of the denotatum and the function(s) that the EPC is assigned in the (inter)language. The (in)definiteness of pivotal NPs is not analysed in a strict binary fashion (i.e. non-violations vs. violations) but the frequency of the forms involved is also taken into account, as well as their status in the inventory of referring expressions used to mark (in)definiteness oppositions in learners’ productions.

5 The Current Study

5.1 Participants

The data analysed in this chapter were originally collected for a bi-directional project exploring referent (re)introductions in the narratives of both L1 French L2 Chinese and L1 Chinese L2 French learners. The oral productions of 15 NS of French, 15 NS of Chinese (the control groups) and 15 L1 French L2 Chinese learners and 15 L1 Chinese L2 French learners (the L2 groups) were collected (see Lena, 2017, 2020b). For this study, only the data produced by French learners of L2 Chinese, and the control group of Chinese NS, are considered.

The French learners’ group is composed of eight men and seven women. They have studied Chinese at the university for at least 3 years (5 years on average) and speak a third language at minimum (i.e. English). They belong to Bartning’s (1997) definition of ‘learners with a high level of education’, that is, learners that have studied the L2 in higher education and have a strong metalinguistic knowledge of the target language.

In addition, the participants in this study lived in a country where the target language is spoken (i.e. China) for at least 1,5 years (3 years on average), while 14 of them had settled in China permanently at the time of the experience, and were recorded there.

5.2 Materials and Procedure

The stimulus used to elicit film retellings is an extract from Charlie Chaplin’s silent film Modern Times developed as part of a European Science Foundation project (see Klein & Perdue, 1992). Following Turco (2008) and Sun (2008) among others, the short sequence comprising the bread robbery scene was selected, to which a final scene was added in order to study the expression of reintroduced referents which leave the stimulus for a period before returning later. The storyline can be described as follows:

Sequence one: A hungry girl steals a loaf of bread from a bakery shop. A lady nearby sees the robbery and promptly informs the baker. The baker runs after the girl. The girl bumps into Charlie Chaplin, and the two fall to the ground. A police officer arrives with the baker. Chaplin takes the blame for the stolen bread. The police officer brings him away.

Sequence two: Chaplin is freed from the police station. He founds the girl outside waiting for him. They hug.

The researcher met each participant individually in a quiet room. The subjects were instructed to watch the 2-min video on a computer screen. They were told that they could watch the video as many times as they needed, and that they could take time before starting their retelling. When they felt ready, they were then asked to retell the story to a fictional naïve listener—sharing no mutual knowledge—while the experimenter (the author of this paper) was recording them. The subjects were also told that they should try to speak in a spontaneous way, as they would tell the story to a close friend. To emphasize this point, the researcher highlighted that there was not a ‘correct’ or ‘wrong’ way of saying things, and that they should just try to speak as they would normally do.

The author of this paper then transcribed the recordings collected and coded them according to the types of referring expressions (e.g. quantified nouns, BNs, DEM + N, etc.—see Tables 3 and 4). Further, EPCs were identified in each dataset (Tables 1 and 2).

Table 1 The overall use of EPCs in Chinese NS’ and French learners’ narratives

Table 2 Nominal expressions occurring within the EPC in the narratives of Chinese NS and French learners of L2 Chinese

Table 3 The overall use of bare nouns and demonstrative phrases in the narratives of Chinese NS and French learners of L2 Chinese

The bread robbery scene includes the introductions of five main characters (the thief, the baker, the snitch, Chaplin, and the police officer), offering typical contexts for EPCs to be used. While in a previous study (Lena, 2020b), I have focused on the encoding of those characters, for the purpose of the current research the syntactic unit (i.e. the EPC) is taken as the starting point for the analysis. That is, the introduction of peripheral individuals, inanimate entities as well as introductions in the reported speech were all considered, since any occurrence of EPC provided useful data for the analysis of learners’ sensitivity to Chinese DE and possible DE ‘violations’. Hence, all the percentages are calculated over the total acts of reference (Ryan, 2015) observed in each corpus (Chinese L1 and Chinese L2).

Table 4 The overall use of lexical referring expressions in the narratives of Chinese NS and French learners of L2 Chinese

5.3 Results

The total number of EPCs found in the oral narratives produced by the two groups of speakers (French learners of Chinese and Chinese NS) are presented in Table 1.

Note that, even if frequently used for referents introductions in Chinese, you-constructions are marked forms, since the introductions of new referents are statistically rare when compared to the operation of establishing anaphoric relations in discourse.⁷ Overall, the relatively increased use of you-constructions in the French learners’ narratives when compared to the NS group (7.2% contra 4.8%) is discussed in detail elsewhere (see Lena, 2020b, 2020c: 427–429) as the result of a ‘unicity of functions’ effect (see Bartning & Kirchmeyer, 2003). That is, learners lean on a smaller inventory of presentational constructions to introduce new referents into the discourse while underusing bridging operations.

Table 2 presents in detail the nominal expressions appearing in the EPCs produced by Chinese NS and French learners of Chinese.

A note regarding the (in)definiteness characterization of the referring expressions presented in Table 2: quantified nouns are classified as ‘indefinite’, and so do BNs that typically acquire an indefinite reading in this context (see Sect. 2). Nouns modified by a demonstrative determiner, as well as proper nouns, are considered ‘definite’. Bare genitive nouns such as miànbāodiàn (de) lǎobǎn ‘[the] bakery owner’ stand somewhere in the middle. According to Chen (1986: 16–17, cited in LaPolla, 1995: 307), all NPs marked by a genitive phrase or a relative clause are ‘definite’.⁸ Consider the examples below, from an NS (7) and a learner (8):

In both extracts, the NPs included in the you-constructions present a new referent which is connected to a discourse-old entity (i.e. miànbāodiàn ‘bakery’) that has been introduced as an indefinite quantified object a few propositions earlier. Such nouns can be considered pragmatically ‘anchored’ (Prince, 1981) since the nominal head is indefinite (i.e. discourse-new), while the genitive phrase is definite (i.e. discourse-old or inferable, depending on the context). From a morphological point of view, such forms are unmarked since they commute with quantified genitive phrases (e.g. yí ge miànbāodiàn de lǎobǎn ‘a bakery owner’)—i.e. ‘indefinite’—and with genitive phrases modified by a demonstrative determiner (e.g. nà ge miànbāodiàn de lǎobǎn ‘that bakery owner’)—i.e. ‘definite’. Bare genitive NPs are indeed marginal in our corpora, but they do raise the question of what precisely should be considered as a DE violation in Chinese. Given that such forms are [-marked] (i.e. not modified either by a quantifying expression or by a demonstrative, as just said), bare genitive NPs will be classified as ‘indefinite’ (just as BNs) for the purpose of this study.

With all this in mind, let us now consider the DE in the data (Table 2). To begin with, the you-constructions in the two groups tend to include indefinite quantified nouns, in equal proportions (66.6%). Note that learners’ absolute number of indefinite quantified nouns in EPCs is not sufficiently informative per se, as it might result from an overuse of nouns modified by an indefinite marker (notably yí-ge ‘one-CL’) in their corpus. As just said, bare genitive nouns are marginal in both groups. The use of a proper noun within the EPC is observed only in French learners’ productions but remains marginal as well (2 occurrences). Overall, two main tendencies set apart the use of EPCs observed in learners’ narratives from the ones produced by NS: the reduced use of BNs and the increased use of nouns modified by a demonstrative phrase. Examples of each case are provided below⁹:

Note that, in (10), the first zhè is used in a typical anaphoric context (miànbāodiàn ‘bakery’ has been evoked in previous discourse), while, in its second occurrence, the proximal demonstrative is used in a way that could be considered cataphoric, since it modifies a new entity of high thematic importance with continuing presence in the following discourse.

In order to link these facts—i.e. the underuse of BNs and the overuse of DEM + N in learners’ EPCs—to the role that these forms play as referring expressions in the interlanguage, Table 3 shows the relative frequency of BNs and DEM + N in the narratives of Chinese NS and French learners of L2 Chinese. The percentages are calculated over the total acts of reference observed in each corpus.

The only notable difference between the two groups of speakers is a decreased use of BNs in French learners’ productions (12.8% contra NS 20.1%) (see also Liu & Huang, 2015). Table 4 offers a more detailed view of the overall use of lexical referring expressions in the oral narratives of the two groups, by distinguishing between marked and unmarked forms.

Several observations are noteworthy. First, NS’ and learners’ overall use of indefinite quantified nouns is roughly the same (13% and 13.7%, respectively). In addition, no inappropriate use of the indefinite marker (i.e. cases where a definite should be expected) is observed in learners’ corpus. The (unlikely) hypothesis of overgeneralization of indefinite quantified NPs in learners’ productions influencing the use of such forms within their EPCs (Table 2) thus can be excluded.

Second, French learners use BNs to a lesser extent (12.8%) when compared to the proportion observed in the NS’ productions (20.1%). The difference is slightly higher when one considers any bare or bare-head NP, that is including bare-head genitive phrases and bare-head relative phrases¹⁰ (17.3 vs. 26.8%).

As previously noted, the acquisition of BNs can be triggering for the L2 learner whose L1 is an article language, given the absence of pre-nominal modifiers in these forms by definition. Thus, the relative under-representation of BNs in French learners’ narratives could be seen as the influence of the L1’s system (an article language) preventing the use of an unmarked form (i.e. BNs) in the IL. The low frequency of BNs in learners’ EPCs (Table 2) can be linked to the reduced use of these forms more generally observed in their narratives.

Table 4 also shows that learners do not overuse DEM + N in general, since these forms represent 21% of the referring expressions observed in their productions, contra 20.2% in NS’ narratives. Hence, the under-representation of BNs does not result in an increased use of DEM + N in learners’ discourse.

It has been shown that article-language learners tend to introduce definite new referents (i.e. bridging reference) with DEM + N, instead of ‘simply’ using felicitous BNs (Crosthwaite et al., 2018, Lena 2020b). It could be the case for such a marked strategy to be mobilized for referent introductions, while in acts of reference tracking learners do not overuse deictic demonstratives (even though not relying on BNs as much as natives do). To test this hypothesis, one should analyse the distribution of referring expressions with regard to the activation state of the referents involved, that is, including pronouns and zero anaphora.¹¹

For the purpose of the current study, it is sufficient to note that French learners of L2 Chinese do not overuse demonstratives in order to ‘replace’ the definite article available in their source language. What they do, however, is produce you-constructions including definite NPs, namely nouns marked by the demonstrative determiner, thus ‘violating’ the Chinese DE. Why is it so?

As it has been pointed out, in most cases, you-constructions in French learners’ productions include an indefinite quantified NP pragmatically denoting a brand-new unanchored referent (in Lambrecht’s 1994: 165 terms):

When French learners’ you-constructions include bare nouns (9) and bare genitive nouns (8), such forms point to referents that are uniquely identifiable or anchored, respectively. As opposed to cases like (11–12), these are accessible referents. Yet the EPC is used to introduce them into discourse for the first time.

As far as the maintenance and the reintroduction of referents are concerned, the you-construction should not be used if speakers understand the function of this device, which is to introduce (brand-)new referents into discourse. Indeed, only one occurrence of you-construction including a maintained referent is observed in the Chinese NS’ data (whereas the existential verb you is preceded by the adverb hai ‘also’ triggering a ‘list’ interpretation—see Hu & Pan, 2007), and no occurrences of you-construction including a reactivated referent are found in the L1 corpus.

With such issues in mind, Table 5 considers now French learners’ ‘violations’ of the target language DE in the light of the pragmatic status of the referent denoted by the pivot NP.

Table 5 Distribution of definite and indefinite NPs within French learners’ you-constructions in relation to the pragmatic status of the pivot referent

In French learners’ narratives, only four occurrences (6.6%) of a you-construction including a strong determiner are used to introduce new referents into the discourse. In other words, referent-introducing you-constructions in most cases include indefinite pivots.

When EPCs include a DEM + N in the context of referent introductions, they are used to encode the characters of the baker (13) and the police officer (14), that is, the same entities that are likely to be introduced into discourse as definite preverbal subjects, by virtue of a frame-based association and unique reference, respectively (Lena, 2020b).

EPCs like (13–14) are still used to put forward discourse-new referents, just as the corresponding forms including bare nominals discussed above (8–9). These sentences, however, represent DE violations. Note that the issue here is two-fold: at the utterance level, learners produce an EPC where the canonical sentence structure (bridging) is an option; at the NP level, they select a demonstrative ([+anaphoric]) to mark referents with a discourse-old status. In Chinese, BNs can—and more often do—felicitously encode such accessible referents.

In (15-16) the context is radically different. The EPC is used to reintroduce referents into the narrative and the entities involved are by consequence discourse-old. It is not the nominal marking with a demonstrative [+anaphoric] which is inappropriate, but the selection of an EPC in these contexts. (Note that in Chinese L1, such reintroduced referents are encoded by nouns modified by a demonstrative, but not appearing within EPCs.)

In sum, DE violations are not frequent in general, and are quite marginal very rare in the case of referent introductions. Recall that in Chinese L1, the DE does not result in a categorical constraint, given that you-including a strong determiner are marginally possible (Sect. 1). One could speculate that Chinese L2 French learners display a native-like behaviour, where the you-construction is mainly linked to the expression of prototypical unidentifiable referents but can marginally include identifiable referents. However, granted that EPC patterns per se are marked forms, definite-pivot EPCs are even more marked in terms of their frequency. It is very unlikely that learners are exposed to sufficient input to integrate those exceptions into their treatment of definiteness in EPCs, assuming that ‘a large and representative sample of language is required for the learner to abstract a rational model that is a good fit to the language data’ (Wulff & Ellis, 2018: 75).

At the same time, if French EPCs generally do not manifest the DE (4a), it has been shown that the il y a construction statistically tends to include indefinite NPs (Sect. 1). But in French L1, the il y a construction can also serve the function of reintroducing referents (as also noted by Lenart & Perdue, 2004). Below is an example from a retelling produced by a French NS produced in the same context as (15–16):

Overall, French learners’ producing in Chinese L2 adopt a pragmatic organization of the information typical of the interlanguage, where the source language influence seems to be playing only a minor role (as discussed in Lena, 2020b). However, the DE ‘violations’ observed here reasonably seem to follow the IS principles operating in learners’ L1. Crucially, such an L1 influence is related to the inventory of functions that the EPC can convey in the source language, and which are transposed in the interlanguage.

The presence in the target language of an EPC which has a functional equivalent in one’s L1 can trigger its use in the L2, but may also lead to an overgeneralization: French learners are known to overuse EPCs in the L2 (see Turco, 2008; Leclercq, 2008). The proximity between French and Chinese EPCs (Sect. 1) might have the paradoxical effect of making the need for readjustments less critical (a similar speculation is reported in Lambert and colleagues, 2008).

6 The Need for an Integrated Approach to Study the DE Acquisition

Snape and Sekigami (2016), reporting White and colleagues’ (2012) concern, note that ‘[p]ast findings were dependent on spontaneous production data and as a result there were rather infrequent productions of there constructions. There may not have been many contexts where a there construction was required, thus spontaneous spoken production may not be the most appropriate method to look for DE violations’. (Snape & Sekigami, 2016). To date, researches on DE seems to have been conducted mostly on the basis of GJTs (King et al., 2006; Snape & Sekigami, 2016; White et al., 2012; Yang et al., 2007; Yu & Su, 2011; Zielke, 2016). That being said, I do agree with Snape and Sekigami (2016) that the number of EPCs in elicited production data is indeed small. Instead of ‘dismiss’ spoken productions data altogether, the solution is perhaps to integrate different kinds of datasets. While GJTs can provide additional information, they can be misleading when considered alone. It is delicate, for instance, to leave the learner with the task of figuring out the right discourse context for a sentence, even if indications are provided. To illustrate my point, the following are attempts to create informative judgment task items on the basis of White and colleagues’ (2012) model (see Fig. 2). Figure 3 presents a natural (occurring in L1 data) target item. Such sentences typically occur at the beginning of the story, that is, without context. Note that the canonical subject–verb (SV) equivalent is possible in Chinese, thus without using an EPC (as shown in Fig. 4). For instance, it might be the case that subjects situate the target sentence as the beginning of a novel instead of an oral narrative (also prompted by the written nature of the stimulus), which could influence their judgment. Though not affecting DE sensitivity altogether, subjects’ correction of the target item in favour of an SV sentence would not be informative with regards to their treatment of definiteness in EPCs.

Fig. 2

Example of (unnatural) target item used in White et al. (2012) GJT on DE

Fig. 3

Possible natural target item in GJT on Chinese DE

Figure 4 is an example of a not-target-like use of the you-construction (from learners’ data) which includes a definite NP used to initiate a cataphoric chain. The following context (in brackets) is necessary here to specify the salience of the referent denoted by the pivot, and the referring expression selected to encode the target entity in the contextualizing text might bias the subject’s response.

Fig. 4

Possible unnatural target item (cataphoric definite NP) in GJT on Chinese DE

Figure 5 provides an example of non-native-like use of the you-construction (also from learners’ data) which includes a definite NP in a reintroduction context. In order to provide a sufficient background, the contextualizing text should be as precise as possible—facing space constraints. It can be questioned whether a short-text context can possibly supply the extra-linguistic background like the one provided by the filmed stimulus used in this study. Then, the question arises as to which forms should be selected to denote discourse-new and discourse-old referents (in brackets), to avoid bias. Plus, as presented in Fig. 5, the context might not be enough centred on the character of the girl to justify its reactivation by using a marked syntactic pattern (i.e. the you-construction). This being the case, French learners could reject the unnatural target item even if the same forms are found in their spoken productions.

Fig. 5

Possible unnatural target item (reintroduced referent) in GJT on Chinese DE

Finally, in some cases, learners could identify a natural target item correctly (e.g. Fig. 3) without providing any information about their preference (that is, one form is accepted but another one would be preferred, or the opposite situation around). Multiple choice items such as Fig. 6 could perhaps provide this kind of information.¹²

Fig. 6

Possible multiple-choice item in GJT on Chinese DE

While agreeing with White et al. (2012) that traditional GJTs with uncontextualized sentence items are misleading when one analyses speakers’ sensitivity to the DE—and IS-motivated phenomena in general—in what precedes I raised some doubts about the possibility of recreating natural occurring contexts by using this kind of tasks. My intention is by no way that of discredit GJTs as a method to collect learners’ data. Rather, what I wish to highlight is the amount of information that might be lost due to IS values that cannot be fully evoked by means of this elicitation technique, and the need for an integrated methodological approach to study the DE acquisition. Though my remarks specifically concern French L1 learners of L2 Chinese, it might be the case for similar problems to arise when considering different L1/L2 combinations.

7 Discussion and Conclusions

In contrast to previous findings on the acquisition of definiteness restrictions in a second language (Lardiere, 2005; Snape & Sekigami, 2016; White et al., 2012; White, 2003, 2008a, 2008b), the current study does report on DE ‘violations’, that is, French learners of L2 Chinese participating in this study produce definite pivots, contrasting with the DE existing in the EPCs of the target language (i.e. Chinese). This is all the more interesting because the learners that produced the retellings were advanced learners.

Facing an article-less target language, French learners’ ‘violation’ of the Chinese DE manifests itself in the production of you-constructions including nouns marked by a demonstrative determiner and marginally proper nouns. Two main tendencies were identified in the learners’ data. The increased use of nouns modified by a demonstrative determiner goes along with the reduced use of BNs in the you-constructions produced by French learners. Further, an investigation of the overall distribution of the lexical referring expressions in their narratives showed that learners do not produce more nouns modified by a demonstrative determiner in general, nor do they use quantified nouns to a greater extent. Thus, the hypothesis of a general overgeneralization of the indefinite marker and demonstratives—which could bias the occurrence of natural and unnatural EPCs, respectively—is not proven true. That is, the frequency in which these forms are found in EPCs appears to be correlated with this syntactic context. BNs, on the other hand, are indeed less represented in French learners’ overall productions: in EPCs, the reduced use of BNs thus does not solely correlate with the sentence pattern. A study of the general use of Chinese BNs with respect to the other—not exclusively lexical—referring expressions in French learners’ interlanguage was beyond the scope of the present contribution. As far as EPCs are concerned, the few instances including a BN are used by learners to introduce uniquely identifiable and inferable referents (i.e. NS’ sentences including brand-new indefinite-referent BNs such as yǒu rén tōu-le nǐde miànbāo ‘there’s someone who stole your bread’, yǒu miànbāo zài chē = shang ‘there’s [some] bread on the truck’ are not found in the L2 corpus).

Turning to DE violations stricto sensu, it has been shown that these are marginal for referent introductions. In these cases, the demonstrative [+anaphoric] is used to introduce new—though inferable—referents into discourse. The NP-level marking (i.e. the demonstrative determiner) is inappropriate to denote these referents and incompatible with the selection of a marked sentence structure (i.e. the EPC pattern). Though not preferred by natives, the EPC is still possible in these contexts, as far as new referents are introduced into discourse. In most cases, however, French learners produce definite pivots for referent reintroductions. Here, the nominal marking with a demonstrative [+anaphoric] is appropriate, yet incompatible with the selection of an EPC pattern. This can be explained by considering that in the learners’ interlanguage, the you-construction is assigned a function that is not available in Chinese L1. In other words, learners are aware of the DE that characterize the EPC in the target language, which is shown by the marginal use of definite expressions in referent-introducing EPCs. By contrast, what they seem not to be aware of is that the EPC should not be used in reintroduction contexts. As a consequence, they use the EPC format when discourse-old referents are concerned. Strictly speaking, however, they do not ‘violate’ the definiteness restriction of the target language, since a different form, with a different function, is operating in the interlanguage.

Observations show that the acquisition of the DE in a second language cannot be acknowledged by having recourse uniquely to a sentence-level approach. Nor it can be accounted for by a single-level analysis. For instance, a reduced number of BNs within learners’ EPCs is observed. Even if this results in no DE violation, it has been shown that frequency is informative on how definiteness is treated by learners. In a functionalist perspective of L2 acquisition, it seems more useful to connect the sentence level (‘learners’ sensitivity to DE’) to both the system of referring expressions available in the L2 and the pragmatic function(s) that the sentence pattern can convey. Learners give (at least) two functions to the you-construction: that of introducing and reintroducing referents. In Chinese, the you-construction cannot be associated to the expression of reactivated referents. In the interlanguage, learners do assign this function to the you-construction, which is reasonably interpreted as the result of a negative transfer from the L1. Given the additional function acquired by the you-construction, as a consequence, learners produce definite pivots, since reactivated referents are definite (i.e. discourse-old) by definition.

8 Limitations and Suggestions for Future Research

Given the nature of the elicited task used to collect data, speakers were not ‘forced’ to choose one particular linguistic form (i.e. the EPC) to encode (new) reference. As a consequence, referents could be introduced—or reintroduced—by means of other devices, which accounts for the reduced number of EPCs found in the data. It is therefore necessary to point out the limitations of the current study, which included a limited set of EPCs available for analysis and cross-linguistic comparison. Hence, the hypotheses formulated though this paper should be further confirmed by larger samples of data.

As said earlier, the elicitation task was first conceived in order to explore more generally the linguistic strategies for referent (re)introductions in Chinese and French as second languages (Lena, 2017, 2020b, 2020c: 126–137). In hindsight, it would have been useful to clearly space out the scene including the reintroductions of referents by a sequence of distractors, instead of relying on the speakers’ understanding of the logical progression of the narrative. That is, the avoidance of an EPC to encode reactivated referents might also originate from the missed perception of an interruption from the preceding point of the story. In other words, if speakers do not conceive a break in the narrative, the referents will be treated as maintained referents, not as reactivated ones. However, sentences such as yǒu zhè ge nǚhái děng tā ‘there is this girl waiting for him’, as in (15), were submitted to three Chinese NS, giving them the appropriate context, and were systematically rejected as unnatural. Finally, in the oral narratives collected using the same stimulus from L1 Chinese L2 French speakers (Lena, 2020b)—which were not considered for the current study—no occurrences of EPCs in the context of reintroduced referents were found. These issues nonetheless demand for a study based on a stimulus that address systematically the referent reintroducing function in L2 Chinese.

Finally, it is worth emphasizing that in production tests triggered by a video sequence like the one adopted in this study, the cognitive load is important (Chini, 2005). Speakers have to introduce and track referents in discourse while managing the narrative cohesion and avoiding ambiguity, all of this in a short time. Many introductions (and reintroductions) are condensed in the two-minute film used as a stimulus. As Ryan (2020) recently suggested, learners’ referent tracking in a second language can be influenced by the extra-linguistic context, with informal and unpressured contexts leading to more target-like performances. While elicited production tasks have the undeniable benefit of providing comparable data, they should be integrated with more diversified sources of learners’ productions. The combination of different elicitation techniques with spontaneous corpus data is promising, especially when studying IS aspects of the acquisition process.