Preferred argument structure in the narratives of Chinese-English bilinguals and their monolingual peers

Abstract

Aims and Objectives:

This paper aims to tease apart the roles of different factors that contribute to potential divergence between bilingual and monolingual speakers in the patterns of preferred argument structure (PAS) in spoken discourse.

Methodology:

Spoken narratives were elicited from two groups of Chinese-English bilinguals (early versus late) and their monolingual peers, using the wordless picture storybook, Frog, where are you?

Data and Analysis:

For each transcribed narrative, core arguments in the clauses are coded for grammatical role and information status. The data are then analyzed for conformity to four constraints that Du Bois introduced in his theory of PAS.

Findings:

The results revealed both similarities and differences in the PAS patterns in the narratives of the four participant groups. While the similarities provide support for the validity of the PAS theory, the differences were found to result from cross-linguistic influence, an age of second language acquisition effect, or a general effect of bilingualism.

Originality:

The innovative nature of this study lies in the following: (i) the examination of a new syntax–discourse interface phenomenon in bilingual development; and (ii) the intentional attempt to tease apart the effect of cross-linguistic influence and the effect of bilingualism itself on bilingual performance.

Implications:

Our findings suggest that it is necessary, and indeed empirically possible, to tease apart not only the effect of age of second language acquisition from the effect of cross-linguistic influence, but also these two effects from the general effect of bilingualism in bilingual development and use.

Keywords

Preferred argument structure narrative development Chinese-English bilinguals age of second language acquisition bilingualism effect cross-linguistic influence

Introduction

It has long been observed that the languages of bilingual speakers are quantitatively and qualitatively different from those of monolinguals (Cook, 2002; Sorace & Serratrice, 2009; Yip and Matthews, 2007). While a rich body of literature exists that documents such differences, it remains a controversial question whether such differences result from cross-linguistics influence, or from an age of second language (L2) acquisition effect (Bylund, Hyltenstam, & Abrahamsson, 2013; Johnson & Newport, 1989; Schmid, 2014), or even from the effect of just growing up with or having knowledge of two or more languages instead of one (henceforth the bilingualism effect, Sorace, Serratrice, Filiaci, & Baldo, 2009). In the present study, we set out to address this question through a detailed analysis of preferred argument structure (PAS) in the narratives of Chinese-English bilingual speakers and their monolingual peers in each of the two languages. Specifically, we evaluate the extent to which early Chinese-English bilingual speakers match that of late Chinese-English bilinguals, and how these two groups of bilinguals match their monolingual peers in each of the two languages in the production of referring expressions, such as lexical noun phrases and pronouns, to fulfill various grammatical roles and to manage information flow in their narratives.

Preferred argument structure

In his seminal work, Du Bois (1987) introduced the theory of PAS to capture how the core arguments, including the subject of a transitive verb (A), the direct object of a transitive verb (O), or the subject of an intransitive verb (S), are realized and distributed in spoken discourse. While any of the core arguments may be expressed by different forms of referring expressions such as a full NP (e.g., Jesus of Nazareth), a pronoun (e.g., he), or a zero (Ø), and the number of core arguments in a sentence as well as which of these arguments must be obligatorily expressed seem to be specified by the particular verb used, Du Bois has detected certain statistical tendencies of argument distribution in spoken discourse. Through a detailed analysis of the distribution of different types of arguments in elicited Sakapultek (Mayan) narratives, Du Bois (1985, 1987) found that Sakapultek speakers seemed to have the tendency to avoid having lexical or new referents in the transitive subject position, and to avoid expressing more than one lexical argument or more than one piece of new information in a clause. He formulated these tendencies as the PAS theory consisting of four constraints, two at the grammatical dimension and two at the pragmatic dimension. These four constraints, as shown in Table 1, define limits on lexical quantity, lexical role, information quantity, and information role (Du Bois, 1987; Du Bois, Kumpf, & Ashby, 2003).

Table 1.

Preferred argument structure constraints (adapted from Du Bois et al., 2003, p. 34).

	Grammar	Pragmatics
Quantity	One lexical argument constraint: avoid more than one lexical core argument	One new argument constraint: avoid more than one new core argument
Role	Non-lexical A constraint: avoid lexical As	Given A constraint: avoid new As

The syntactic and pragmatic generalizations as shown in the PAS theory have been attested in many typologically different languages, such as Spanish and French (Ashby & Bentivoglio, 1993), Japanese (Iwasaki, 1985), Mandarin (Tao & Thompson, 1994), Hebrew (Sutherland-Smith, 1996), Korean (Clancy, 1993, 2003), Inuktitut (Allen & Schröder, 2003), and others. The PAS has found support from studies involving different genres (written exposition as in Lin, 2009; specialized spoken discourse: O’Dowd, 1990; classroom discourse: Kumpf, 1992, 2003, and procedural discourse: Tao, 2007), and for different age or clinical populations (e.g., children aged 2;0–2;5 learning Tzeltal Mayan: Brown, 1998; child Inuktitut: Allen & Schröder, 2003; child Korean, Clancy, 2003; child Hindi, Narasimhan, Budwig, & Murty, 2005; child English and Japanese: Guerriero, Cooper, Oshima-Takane, & Kuriyama, 2001; Guerriero, Oshima-Takane, & Kuriyama 2006; children with autism: Weber, 2003; and aphasic speech: Kohn & Cragnolino, 2003).

All of these studies on PAS have focused on monolingual discourse. To our knowledge, however, almost no study has looked at the PAS in bilingual speakers, with the outstanding exception of Kumpf (1992, 2003), who has found that while Japanese and Spanish learners of English (who either learned English through contact and did not attend school or were schooled in English) largely conformed to the PAS constraints, they produced more lexical noun phrases than native speakers of English did. Given that PAS consists of both a grammatical and a pragmatic dimension, thus PAS involves the syntax/discourse–pragmatics interface, which has been found to be a particularly vulnerable interface for L2 and bilingual speakers (Ezeizabarrena, 2012; Hulk and Müller, 2000; Müller and Hulk, 2001; Rothman, 2009). It remains to be seen whether PAS would constitute such a vulnerable interface. The purpose of this paper, therefore, is to examine the PAS patterns in the narratives of two groups of Chinese-English bilinguals who differ in the age of acquisition of L2 English (early versus late), in comparison to both their Chinese-speaking and English-speaking monolingual peers. We are interested in finding out whether narratives produced by bilingual speakers will resemble or differ from those by monolingual peers in their PAS, and if so how the similarities and differences may help us to tease apart the age of L2 acquisition effect from the effects of language-specific cross-linguistic difference and the more general effects of bilingualism.

Methods

Participants

Four groups of speakers participated in this study, a group of monolingual Chinese speakers, a group of monolingual English speakers, and two groups of Chinese-English bilingual speakers. Sociodemographic data were collected via a brief written language (learning) background questionnaire. The Chinese monolingual participants were 12 undergraduate students who studied at a university in Beijing, China, at the time of data collection and their age ranged from 18 to 22 years (mean age (years) = 20). They were born and brought up in China. They had no direct contact with English speakers and were not exposed to English until after the age of puberty when they started to take English as a foreign language (EFL) classes at school. The monolingual English group was made up of 12 natives speakers of English who were undergraduate students enrolled in a university at the greater Atlanta area and their age ranged from 18 to 22 years (mean age (years) = 20.7).

The group of early Chinese-English bilinguals was made up of 12 college students at an American university in the greater Atlanta area at the time of data collection (mean age (years) = 19.6). They were born and brought up in America, and came from Chinese-speaking families residing in the USA. They had all been exposed to Mandarin Chinese from birth. The maintenance of the Chinese language and culture is highly valued and promoted in the home of these participants. Their parents, who are from mainland China, communicate with their children in Mandarin Chinese, and have enrolled them in community-based Chinese schools to learn to read and write in Chinese. All participants were exposed to English primarily through television programs and library story times from three to five years old before they entered kindergarten. After that, their English input and output increased in their school and neighborhood environment. Through self-ratings of oral proficiency in Chinese and English on a Likert scale (0 = no proficiency, 4 = native-like proficiency), all participants were reported to be native-like in Chinese and in English. We also call the early bilingual group the ESL (English as a second language) group.

The group of late Chinese-English bilinguals included 12 speakers, aged 28–35 (mean age (years) = 25.5), whose first language (L1) was Chinese and who began to learn English after the age of puberty. They majored in English in college in China and were enrolled in a doctoral program in an American university at the time of data collection. They had been in the USA for an average of 30.2 months, with a range from 2 years and 5 months to 5 years and 1 month. For ease of presentation, the late bilingual group will be referred to as the EFL group. Through self-ratings of oral proficiency in Chinese and English on a Likert scale (0 = no proficiency, 4 = native-like proficiency), all participants were reported to be native-like in Chinese and near-native in English.

Materials

Spoken narratives were elicited from the participants using the picture book Frog, where are you? (Meyer, 1969). The book consists of 24 pictures (with no accompanying text) portraying a series of complicated events involving several animate referents. It tells a story about a boy going through a series of adventures with his dog to search for his pet frog that escaped one night. On their way to search for the frog, the boy and the dog have interesting encounters with a ground squirrel, an owl, some bees, and a deer. The interactions between these animate referents themselves and their surroundings provide a rich context for the study of the PAS in narratives.

Language elicitation and transcription

The participants were seen individually by a Chinese-English bilingual researcher who informed them that they would tell a story based on some pictures. Each participant was shown the picture book Frog, where are you? (Meyer, 1969) page by page from the beginning to the end to familiarize themselves with the story. The participants were instructed to examine the pictures as long as they wanted before beginning. To minimize the effect of mutual knowledge, they were also instructed to tell the story in as much detail as possible as if telling it to someone who would not have the privilege of seeing the pictures. When the participants were ready, they were asked to return to the first page and to tell the story from beginning to end. Half of the bilingual speakers told the story first in English and then in Chinese, and the other half told the story first in Chinese and then in English, with an interval of about one week between the two stories. In an attempt to minimize interviewer control over participant narrations, only minimal instructions, such as “this is a story about a boy and a dog,” or verbal prompting, such as “what’s next” or “what about the boy?” were given (Berman & Slobin, 1994, pp. 22–25). The task took about 10–15 minutes to complete.

Each oral narrative was audio recorded and then transcribed according to the conventions of the Child Language Data Exchange System, or CHILDES (MacWhinney, 2000). They were transcribed verbatim in clauses (i.e., units consisting of one predicate and its argument) following the guidelines given by Berman and Slobin (1994: 655–664). A native speaker of English or Chinese first transcribed the recording, and for inter-rater agreement, a Chinese-English bilingual speaker reviewed all the audiotaped samples for correspondence to the transcript. Any clauses in which there was any disagreement were reviewed and then transcribed and segmented jointly until consensus was reached. Word-by-word agreement was determined to be 100%.

Coding and analysis

The basic unit of analysis for PAS is a clause. The core argument of the predicate in the clause was coded for grammatical roles (A, O, and S), referential forms (lexical forms and non-lexical forms, including null and pronominal forms), and information status (given, accessible, and new information). Due to space limitations, we will not consider the humanness dimension (Everett, 2009).

Grammatical roles

The single arguments of intransitive verbs were coded as the S role, including the NPs in the preverbal position in existential constructions. The most agent-like arguments of transitive verbs are denoted as A, and objects as O. The NPs after the linking verbs in the there constructions are classified as members of the O category (cf. Kumpf, 1992), “because there itself is not a discourse referent: it cannot be considered as an identifiable character or object” (Kumagai, 2006, p. 680). Oblique NPs introduced by prepositions will not be included in the analysis in order to focus on the core arguments of A, S, and O.

Information status

A new referent refers to the one that has never been brought up in the prior context, thus it does not presuppose mutual knowledge and can be expressed as English indefinite nominal (a dog). A given referent is the entity previously mentioned, which can be expressed as English definite nominals (the dog), pronominals (it), and null elements (e.g., He climbed over the log and Ø took a look). Accessible information is intermediate between new and given information and it comes from the expectations associated with a schema or results from deactivation from an earlier state. A piece of information is classified as accessible (a) “if it was part of a previously evoked, entity-based frame although previously unmentioned; or (b) if it had been mentioned previously, but more than 20 intonation units previously” (Du Bois, 1987, p. 816). Taking principle (a), for example, even though the head or leg of the little dog has not been mentioned before, they are still accessible because they are the body parts of the owner and the previous introduction of the owner has evoked a frame that includes the body parts as easily associable elements (Kumagai, 2006).

If direct quotes or direct character speech function as clausal complements, (e.g., the boy called out, “froggie, are you in the hole”), they are left out of the analysis in this study. Otherwise, the main clause is coded first, then the content of the subordinate clause is coded as a separate and individual clause (cf. Kumpf, 1992). Moreover, indirect objects or question words are also excluded from analysis.

Results

A total of 811 clauses were coded and analyzed for the monolingual Chinese speakers, 776 clauses for the monolingual English speakers, 674 clauses for the early Chinese-English bilingual (ESL) speakers, and 674 clauses also for the late Chinese-English bilingual (EFL) speakers. In the following analyses, we presented our results according to the four constraints of PAS put forward by Du Bois (1987).

To assess intercoder reliability, the first author and a Chinese-English bilingual research assistant who was blind to the purpose of the study each initially coded all the transcripts independently. Then the two sets of coded transcripts were compared for intercoder agreement. The intercoder agreement was calculated three times, once for grammatical role, once for surface forms of the arguments, and once for information status of the targeted structures. Initial intercoder agreement was 100% on grammatical role, 100% for surface forms, and 92.4% on information status. Differences in coding were noted and then resolved via consensus.

The one lexical argument constraint

The distribution of the clause type (that is, clauses with zero, one, or two lexical arguments) in the narratives produced by all four participant groups is described in Table 2, which provides both the total number of tokens and the percentage of the total that this number represents within each participant group. All participant groups were remarkably similar in their distribution of clauses with lexical arguments. Consistent with the one lexical argument constraint, clauses with zero or one lexical argument were the preferred structures, accounting for 88.41% of all clauses in Chinese monolinguals, 91.88% in English monolinguals, 92.29% for both early and late bilinguals in their English narratives. Preference for the clauses with different numbers of lexical argument was not equally distributed. Specifically, for all participant groups, clauses with one lexical argument or less were the preferred forms for argument realization (monolingual Chinese narratives, χ² = 478.58, p < .0001; monolingual English narratives, χ² = 544.46, p < .0001; early bilingual English narratives, χ² = 482.05, p < .0001; late bilingual English narratives, χ² = 482.05, p < .0001.

Table 2.

Overall distribution of clauses with zero, one, or two lexical arguments across participant groups.

Lexical argument	Chinese		English		ESL		EFL
	n	%	n	%	n	%	n	%
0	273	33.66	311	40.08	295	43.77	241	35.76
1	444	54.75	402	51.80	327	48.52	381	56.53
2	94	11.59	63	8.12	52	7.71	52	7.71
Total	811	100	776	100	674	100	674	100

ESL: English as a second language; EFL: English as a foreign language.

However, the distribution described in Table 2 also appears to show differences in the proportion of choice of clauses with different numbers of lexical arguments for the four groups of participants. This was confirmed by a Pearson’s chi-squared test with χ² = 25.71, df = 6, p < .0001. Multiple pairwise comparisons indicated significant differences between Chinese monolingual and English monolinguals, p = 0.007; between Chinese monolingual and ESL, p = 0.0001; and between ESL and EFL, p = 0.008. No other significant differences were found. A close examination of the standard residues of chi-square tests suggests that the early bilingual participants were more likely to produce clauses with zero lexical argument than other groups (z = 3.41), the EFL participants were more likely to produce one-lexical argument clauses than other groups (z = 2.12), and the Chinese monolingual participants were more likely to produce clauses with two lexical arguments than other groups (z = 3.17).

The one new argument constraint

Table 3 summarizes the frequency of clauses with zero, one, and two new arguments as a function of language and bilingual status. Consistent with the One New Argument constraint, which mandates no more than one argument presenting new information in each clause, all participant groups showed the same preference for clauses with zero or one new argument. For all participant groups, clauses with two new arguments are rare, accounting for less than 1% of all the clauses in the narratives. Thus, for all participant groups, clauses with zero or one new argument were the preferred forms for argument realization in monolingual Chinese narratives (χ² = 803.02, p < .0001), monolingual English narratives (χ² = 764.05, p < .0001), early bilingual English narratives (χ² = 654.15, p < .0001), and late bilingual English narratives (χ² = 658.09, p < .0001).

Table 3.

Distribution of clauses with new arguments across the four groups of participants.

New argument	Chinese		English		ESL		EFL
New argument	Token	%	Token	%	Token	%	Token	%
0	647	79.78	688	88.66	570	84.57	559	82.94
1	162	19.98	85	10.95	99	14.69	111	16.47
2	2	.25	3	.39	5	.74	4	.59
Total	811	100	776	100	674	100	674	100

ESL: English as a second language; EFL: English as a foreign language.

However, the distribution in Table 3 also appears to show differences in the choice of clauses with different numbers of new arguments in the narratives produced by the four groups of participants. Pearson’s chi-squared test confirmed this initial observation, χ² = 27.48, df = 6, p= 0.0001. Multiple pairwise comparisons with a Bonferroni adjusted alpha level of .008 per test (.05/6) indicated significant differences between Chinese monolingual and English monolinguals, p = 0.000; and between English monolingual and EFL, p = 0.007. No other significant differences were found.

In order to gain more insight into the observed differences, we examined the standard residue of chi-square tests, as well as the expected frequencies and chi-square values for individual cells. The results show that monolingual English speakers produced fewer clauses with one new argument but more with zero new arguments than expected in their narratives (χ² = 10.62, z = −4.14), whereas Chinese monolingual speakers produced more clauses with one new argument and fewer clauses with zero new arguments in their narratives than expected (χ² = 10.10, z = 4.07).

The non-lexical A constraint

The non-lexical A constraint claims that lexical arguments are avoided at the A position. Table 4 shows the frequency of lexical and non-lexical referents across the grammatical roles in the narratives produced by the four groups of participants.

Table 4.

Distribution of lexical versus non-lexical arguments across grammatical roles by group.

Grammatical roles	Chinese		English		ESL		EFL
Grammatical roles	Lexical	Non-lexical	Lexical	Non-lexical	Lexical	Non-lexical	Lexical	Non-lexical
A	134(33.8%)	262 (66.2%)	86 (34.3%)	165 (65.7%)	69 (25%)	207 (75%)	70 (25.5%)	205 (74.5%)
S	193 (54%)	165 (46%)	259 (50.9%)	250 (49.1%)	171 (44%)	217 (56%)	193 (53.6%)	167 (46.4%)
O	305 (89%)	39 (11%)	183 (84.3%)	34 (15.7%)	190 (82%)	42 (18%)	222 (90.3%)	24 (9.7%)
Total	632	466	528	449	430	466	485	396

ESL: English as a second language; EFL: English as a foreign language.

Table 4 shows that across all participant groups, arguments that fulfill the A role are predominantly non-lexical, which is consistent with the non-lexical A constraint. By contrast, lexical and non-lexical arguments that fulfill the S role are almost equally distributed, whereas lexical arguments that fulfill the O role predominate. These results indicate that all four groups obey the non-lexical A constraint in their narratives.

However, the distribution described in Table 4 appears to show differences in the proportion of A roles that are fulfilled by lexical arguments in the narratives produced by the four groups of participants. Pearson’s chi-squared test confirmed this initial observation, χ² = 47.17, df = 6, p = 0.0000. Multiple pairwise comparisons with a Bonferroni adjusted alpha level of .008 per test (.05/6) indicated significant differences between Chinese monolinguals and ESL, p = 0.004; between English monolinguals and ESL, p = 0.004; between Chinese monolinguals and EFL, p = 0.006; and between English monolinguals and EFL, p = 0.005. No significant difference was found between Chinese and English monolingual speakers, p = 0.77, nor was there any significant difference between the ESL and EFL, p = 0.56. These patterns are shown more clearly in Figure 1.

Figure 1.

Proportions of non-lexical arguments across grammatical roles and groups. ESL: English as a second language; EFL: English as a foreign language.

These differences between groups seem to result from the unexpected distribution in the narratives of ESL and EFL participants. In particular, the ESL (z = −5.4) and EFL participants (z = −4.68) produced fewer lexical arguments in S roles, but more lexical arguments in A roles (z = 3.11 for the monolingual Chinese participants, and z = 5.30 for the EFL participants) than expected.

The given A constraint

The given A constraint states that new information is avoided at the A position. Table 5 displays the distribution of new information versus given and accessible information across the grammatical roles in the narratives of the four participant groups. We also provided the proportions of given arguments for each grammatical role within each participant group. The proportions were calculated by dividing the number of given As by the total number of As (new, given, accessible). The table shows that arguments that encode new information rarely occurred in the A position, and arguments in the A position predominantly encoded given information. No significant difference was observed among the participant groups, χ² = 7.00, df = 6, p = 0.32. Thus, argument realization in the narratives of all participant groups conformed to the given A constraint.

Table 5.

Distribution of grammatical roles as a function of group and information status.

	Chinese			English			ESL			EFL
	New	Acc	Given	New	Acc	Given	New	Acc	Given	New	Acc	Given
A	8	4	384 (97.9%)	4	2	245 (97.6%)	7	0	265 (97.4%)	6	3	275 (96.8%)
S	33	13	312 (87.2%)	18	10	481 (94.5%)	10	10	366 (94.8%)	24	8	328 (91.1%)
O	125	36	183 (53.2%)	69	35	113 (52.1%)	92	21	120 (51.5%)	89	22	132 (54.3%)
Total	166	53	879	91	47	839	109	31	751	119	33	735

ESL: English as a second language; EFL: English as a foreign language.

This is shown more clearly in Figure 2, representing the proportion of given arguments that fulfill the three grammatical roles of A, S, and O (out of all the arguments that fulfill each of three particular grammatical roles).

Figure 2.

Proportions of given arguments across grammatical roles and groups. ESL: English as a second language; EFL: English as a foreign language.

Figure 2 also shows that across all four groups, there exists a distributional similarity between A and S roles that suggests the A/S alignment under this constraint. This provides some evidence for a correspondence between S and A arguments (i.e., for the relevance of a “subject” category) in both Chinese and English.

Discussion

The results revealed both similarities and differences in the PAS patterns in the narratives of the four participant groups. The similarities provide further support for Du Bois’ (1987) PAS hypothesis, while the differences presents a complex picture regarding the effects of bilingual status, language status, or age of L2 acquisition on patterns of argument realization. These points are elaborated below.

Similarities in the statistical tendencies of argument distribution in the narratives

The results of this study are in accord with a number of previous studies that have indicated the validity of PAS in spoken discourse. In the narratives of all four groups of participants, there are tendencies to (a) avoid more than one lexical core argument, (b) avoid lexical A, (c) avoid more than one new core argument, and (d) avoid new A. In other words, the results in the present study show that the four constraints in PAS hold regardless of language type, bilingual status, and age of acquisition. These similarities suggest that both monolingual and bilingual speakers are guided by universal strategies of argument distribution in spoken discourse. These language universal strategies may form part of a discourse grammar that provides all language users with mental processing instructions (Givón, 1990, 1992) for the construction of narrative discourse. All in all, the similarities across the four groups present additional evidence for the universality of PAS.

Differences in the actual form-function mapping of argument distribution in the narratives

This study suggests some important subtle differences in the actual form-function mapping in the realization of arguments in the narratives of the four groups of participants. Such differences point to complex interactions between bilingual status, language status, age of acquisition, and patterns of argument realization. To start with, there are significant differences between Chinese and English in several aspects that may have led to differences between the two groups of bilinguals and the monolingual English speakers. For example, in the choice of clauses with different numbers of lexical arguments, the Chinese monolingual participants produced fewer clauses with zero lexical arguments (33.66%) than English monolingual participants (40.08%). The late bilinguals’ English patterns (35.76%) resembled the monolingual Chinese participants, whereas the early bilinguals’ English patterns (43.77%) resembled the monolingual English participants. This example seems to reveal a cross-linguistic influence (from Chinese to English) on late bilinguals, but the cross-linguistic effect is modulated by an age of L2 acquisition effect such that the two groups of bilinguals differed from each other. Similar patterns were observed in the choice of clauses with different numbers of new arguments. Hohenstein, Eisenberg, and Naigles (2006) found similar interactions between cross-linguistic effect and age of L2 acquisition effect. They used videos to elicit descriptions of motion events in L1 English and L2 Spanish from two groups of English-Spanish bilinguals: early versus late. Results showed that when describing in Spanish, bilinguals produced more manner verbs than Spanish monolinguals did; when describing in English, bilinguals produced fewer manner verbs than English monolinguals did. This was interpreted to support the effect of cross-linguistic influence. However, this pattern was qualified by an age of acquisition effect. Specifically, the authors found an effect of L2 on L1 only in early bilinguals (their Spanish had fewer path verbs), whereas in late bilinguals they found the influence of each language over the other, a so-called bidirectional influence. A notable limitation of the present study is that we focused on the L2 of the bilingual speakers and did not examine the PAS patterns in their Chinese narratives. Thus, we were not able to explore the potential influence of L2 on L1, as reported by Hohenstein et al. (2006). We are currently examining the narratives in Chinese collected from the same bilingual participants in comparison with their Chinese-speaking monolingual peers. If parallel influences of English on Chinese and Chinese on English were found, we would have another case of bidirectional cross-linguistic influence (Aveledo & Athanasopoulos, 2016; Jarvis & Pavlenko, 2008).

The second difference relates to some interesting patterns in the choice of lexical arguments to fulfill different grammatical roles. The early bilinguals and monolingual English speakers behaved similarly, and differed from both the late bilinguals and the monolingual Chinese speakers. These patterns seem to provide some evidence for the age of L2 acquisition effect. Interestingly, we also observed that the late bilinguals differed from the monolingual Chinese speakers too. In particular, the monolingual Chinese participants produced fewer lexical arguments in S roles, but more lexical arguments in A roles and in O roles than expected. The late bilingual participants, however, produced more lexical arguments in S roles but fewer lexical arguments in O roles than expected in their English narratives. Therefore, as cross-linguistic effect is ruled out, what may be the source of the unique patterns found in the English narratives of the late bilinguals? Pirvulescu, Roberge, Thomas, Pérez-Leroux, and Strik (2014) found similar patterns in their study of the acquisition of pronominal direct objects in French and English (clitics in French and strong pronouns in English). Among other findings, they observed that rate of object omissions in French differed across the various bilingual groups with varying degree of proficiency in their two languages. The English dominant and balanced bilinguals produced more object omissions (67% and 69%, respectively) than the French dominant bilinguals (51%), which in turn produced more than the French monolingual children (42%). They attributed this to the effect of language dominance: the more dominant in one language, the fewer omissions in that language. While this makes some sense, the effect of age of L2 acquisition and the effect of language dominance are not necessarily mutually exclusive.

The third finding relates to the non-lexical A constraint. It was observed the two bilingual groups had almost identical performance, but they differed from the two groups of monolingual speakers who in turn behaved very similarly. Unless evidence is found to support the contention that the critical period for the acquisition of this aspect of argument realization is around the time when the early Chinese-English bilinguals was first exposed to English as their L2, we may rule out the age of L2 acquisition effect on the observed monolingual versus bilingual difference. In addition, since the two monolinguals had almost identical performance, we may rule out the effect of cross-linguistic influence as a possible explanation as well. Then, how can we account for the divergence between the bilingual speakers and their monolingual peers? We suggest, following Sorace (2016), the monolingual–bilingual differences that we observed in the present study regarding the non-lexical A constraint is due to an effect of bilingualism and should be ascribed to the bilingual speakers’ experience with two languages. Bylund (2011a, 2011b) reported a similar scenario, in which they compared the patterns of event segmentation in the retellings of a short film (Modern Times) by adult L1 Spanish–L2 Swedish bilinguals and their monolingual peers living in Sweden. The bilingual speakers were divided into two groups according to their acquisition age (prepubescent versus postpubescent). Results showed that there were no significant differences between prepubescent and postpubescent groups, but both groups of bilinguals’ patterns of event segmentation differed from each of the monolingual peers in both languages. In addition, no difference was found between the bilinguals’ performance in their typologically different L1 and L2. This suggested that the monolingual–bilingual differences in event segmentation did not result from an age of second acquisition effect. Overall, Bylund suggested that the monolingual–bilingual differences in his studies were due to bilingualism effects and should be ascribed to the bilingual speakers’ knowledge and use of two languages. Other recent studies (e.g., Alferink & Gullberg, 2013; Ameel, Malt, Storms, & Van Assche, 2009; Brown, 2015; Brown & Gullberg, 2011, 2013) that have suggested evidence for the bilingual convergence, that is, increased similarity between some elements of a bilingual’s two languages in various aspects of bilingual language use, also point to the need to distinguish between the language-specific effects of a particular L1 and the more general effects of bilingualism. Brown (2015), for example, examined the construal of manner of motion in speech and gesture in the narrative descriptions of motion in the L1 and L2 that were elicited from bilingual Mandarin-English (n = 12) and Japanese-English (n = 15) speakers at an intermediate, CEFR–B level of L2 proficiency, and from monolingual speakers of Mandarin (n = 14), Japanese (n = 16), and English (n = 13). Among other findings, she first found that construal of manner in speech in L2 English was similar between the two bilingual groups in spite of different performance in their L1s (i.e., Chinese and Japanese, respectively), but the L2 performance of the two bilingual groups differed from monolingual speakers of both the source and target languages. The bilingual versus monolingual difference in the construal of manner in speech in L2 was interpreted as possibly a developmental pattern resulting from the CEFR–B level of L2 proficiency of the participants. Secondly, Brown found a convergence between the L1 and L2 construal of motion in gesture. Specifically in the use of manner-highlighting gestures, the two bilingual groups did not entirely pattern with their monolingual counterparts in both L1 and L2, despite considerable differentiation in the monolingual baseline. For example, the mean proportion of manner-highlighting gestures in the English oral descriptions of bilingual Mandarin-English speakers was similar to that in their Mandarin descriptions, even though monolingual English speakers produced significantly more manner-highlighting gestures than monolingual Mandarin speakers. Similarly, the mean proportion of manner-highlighting gestures in the English oral descriptions of bilingual Japanese-English speakers was similar to that in their Japanese descriptions, even though monolingual English speakers produced significantly fewer manner-highlighting gestures than monolingual Japanese speakers. These patterns cannot be explained by the language-specific effects of a particular L1 on a L2 or the effect of a L2 on a L1. Rather, these patterns highlight bidirectional interactions between specific properties of the source and target languages involved and point to the more general effect of bilingualism.

Implications for the study of L2 ultimate attainment

Results from the present study have some important implications for the explanation of non-convergence in bilingual and L2 acquisition in general, and the variable degrees of ultimate attainment in the L2 in particular. Researchers have typically resorted to age of L2 or language-specific cross-linguistic influence to explain the bilingual speakers’ lack of adherence to monolingual or native norms (Bylund et al., 2013). While an increasing number of studies have been conducted to separate the effect of cross-linguistic influence and the effect of age of acquisition on the unique patterns of bilingual language use (e.g., Hohenstein et al., 2006), the results of the present study, as well as an increasing number of other studies reported in the literature (e.g., Bylund 2011a, 2011b), seem to suggest it is time to tease apart not only the effect of age of L2 acquisition from the effect of cross-linguistic influence, but also these two effects from the general effect of bilingualism on the varying degree to which the end state of L2 knowledge resembles that of a native speaker of the target language. In order to identify the major sources of the uniqueness of bilingual language use, it is necessary for future studies to compare different bilingual populations with both typologically distant and typologically close language pairs on the same target form-function mappings in narratives. Explorations along this direction would give a more solid foundation from which to evaluate the uniqueness of bilingual language use, and furthermore to tease apart the effects of language-specific cross-linguistic difference from the more general effects of growing up with or having knowledge of two languages rather than just one (Chen & Pan, 2009).

A related issue is which group should be chosen as the comparison group for the study of L2 ultimate attainment. In other words, should bilingual or L2 speakers ever be expected to behave like or be compared to (monolingual) native speakers of the target language (Birdsong, 2009; Brown, 2016; Bylund et al., 2013; Cook, 2003)? Results from our study provide support for the proposal that it is necessary to compare late-onset bilingual speakers with (monolingual) native speakers of the target language, but to compare them also with early-onset bilingual speakers (Muñoz and Singleton, 2011).

Limitation of the study

It should be noted that our monolingual Chinese participants are not completely “monolingual,” as they did have years of formal English education. As Bylund et al. (2013) rightly pointed out, future studies need to “scrutinize the alleged monolingualism among native speakers, and then the alleged bilingualism among L2 speakers” (p.83) and take into account the participants’ language proficiency and frequency of use of the target languages. A related limitation is that we were unable to include a group of simultaneous Chinese-English bilinguals for comparison. If the simultaneous bilingual group were to converge with the two groups of bilinguals in the present study, and then the source for the divergence between monolinguals and bilinguals might be more clearly identified. Last but not least, the process of second and bilingual language development is highly complex, and various other factors, such as cognitive, neurological, and social/affective factors, quality and quantity of input, L2 use, and so on might all play important roles in predicting different patterns of language use between monolingual and bilingual speakers as well as between different populations of bilingual speakers. In spite of these limitations, the present study contributes significantly to our understanding of the PAS in the narratives of monolingual and bilingual speakers, and sheds insight into the unique contribution of cross-linguistic influence, age of L2 acquisition effect, and bilingualism effect to the ultimate attainment of form-function mapping in narratives of bilingual speakers.

Footnotes

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by one grant from the MOE (Ministry of Education in China) Project of Humanities and Social Sciences (Youth Foundation Project, Grant No. 17YJC740032) and one grant from the Teaching and Research Program of Harbin Institute of Technology at Weihai (Youth Special Program, Grant No. ITIA 10002010).

Author biographies

Xiangyu Jiang received her M.A. degree in Linguistics from University College London in the U.K. in 2009, and Ph.D. in Linguistics from the University of Georgia, USA in 2015. She began to work as an English teacher at Harbin Institute of Technology at Weihai in China since 2010. Her research interests include second language acquisition, psycholinguistics, and foreign language teaching.

Liang Chen, Ph.D., is an Associate Professor of Communication Sciences and Special Education at the University of Georgia, USA. He conducts research in Chinese linguistics, and language development and disorders in monolingual and bilingual populations.

References

Alferink

Gullberg

(2013). French–Dutch bilinguals do not maintain obligatory semantic distinctions: Evidence from placement verbs. Bilingualism: Language and Cognition, 17, 22–37.

Allen

S. E.

Schröder

(2003). Preferred argument structure in early Inuktitut spontaneous speech data. In DuBois

J. W.

Kumpf

L. E.

Ashby

W. J.

(Eds.), Preferred argument structure: Grammar as architecture for function (pp. 301–338). Amsterdam, The Netherlands: John Benjamins.

Ameel

Malt

B. C.

Storms

Van Assche

(2009). Semantic convergence in the bilingual lexicon. Journal of Memory and Language, 60, 270–290.

Ashby

W. J.

Bentivoglio

(1993). Preferred argument structure in spoken French and Spanish. Language Variation and Change, 5(1), 61–76.

Aveledo

Athanasopoulos

(2016). Second language influence on first language motion event encoding and categorization in Spanish-speaking children learning L2 English. International Journal of Bilingualism, 20, 403–420.

Berman

R. A.

Slobin

D. I.

(1994). Relating events in narrative: A crosslinguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum.

Birdsong

(2009). Age and the end state of second language acquisition. In Ritchie

W. C.

Bhatia

T. K

(Eds.), The new handbook of second language acquisition (pp. 402–424). Bingley, UK: Emerald.

Brown

(2015). Universal Development and L1–L2 Convergence in bilingual construal of manner in speech and gesture in Mandarin, Japanese, and English. The Modern Language Journal, 99(1), 66–82.

Brown

Gullberg

(2011). Bidirectional cross-linguistic influence in event conceptualization? Expressions of path among Japanese learners of English. Bilingualism: Language and Cognition, 14, 79–94.

10.

Brown

Gullberg

(2013). L1–L2 convergence in clausal packaging in Japanese and English. Bilingualism: Language and Cognition, 16, 477–494.

11.

Brown

(1998). Early Tzeltal verbs: Argument structure and argument realization. In Clark

E. V.

(Ed.), Proceedings of the 29th annual child language research forum (pp. 129–140). Stanford, CA: Center for Studies in Language and Information (CSLI).

12.

Bylund

(2011a). Segmentation and temporal structuring of events in early Spanish-Swedish bilinguals. International Journal of Bilingualism, 15, 54–86.

13.

Bylund

(2011b). Ultimate attainment of event construal patterns in speakers of L2 Swedish. Vigo International Journal of Applied Linguistics, 8, 29–53.

14.

Bylund

Hyltenstam

Abrahamsson

(2013). Age of acquisition effects or effects of bilingualism in second language ultimate attainment? In Granena

Long

(Eds.), Sensitive periods, language aptitude, and ultimate L2 attainment (pp. 69–101). Amsterdam, The Netherlands: John Benjamins.

15.

Chen

Pan

(2009). Development of English referring expressions in the narratives of Chinese–English bilinguals. Bilingualism: Language and Cognition, 12(4), 429–445.

16.

Chui

K.-W.

(1992). Preferred argument structure for discourse understanding. Paper presented at the Proceedings of the 14th Conference on Computational linguistics-Volume 4, Nantes, France.

17.

Clancy

(1993). Preferred argument structure in Korean acquisition. In Clark

(Ed.), Proceedings of the 25th Annual Child Language Research Forum (pp. 307–316). Stanford, CA: Stanford University: Center for the Study of Language and Information.

18.

Clancy

P. M.

(2003). The Lexicon in interaction: Developmental origins of preferred argument structure in Korean. In Du Bois

J. W.

Kumpf

L. E.

Ashby

W. J.

(Eds.), Preferred argument structure: Grammar as architecture for function (pp. 81–108). Amsterdam, The Netherlands: Benjamins.

19.

Contemori

Dussias

P. E.

(2016). Referential choice in a second language: Evidence for a listener-oriented approach. Language, Cognition and Neuroscience, 31(10), 1257–1272.

20.

Cook

V. J.

(Ed.). (2002). Portraits of the L2 user. Clevedon, UK: Multilingual Matters.

21.

Cook

V. J.

(Ed.). (2003). Effects of the second language on the first. Clevedon, UK: Multilingual Matters.

22.

Du Bois

J. W

. (1985). Competing motivations. In Haiman

(Ed.), Iconicity in syntax (pp. 343–365). Amsterdam, The Netherlands: Benjamins.

23.

Du Bois

J. W

. (1987). The discourse basis of ergativity. Language, 63, 805–855.

24.

Du Bois

J. W.

Kumpf

L. E.

Ashby

W. J

. (2003). Preferred argument structure: Grammar as architecture for function. Amsterdam, The Netherlands: John Benjamins.

25.

Everett

(2009). A reconsideration of the motivations for preferred argument structure. Studies in language, 33(1), 1–24.

26.

Ezeizabarrena

(2012). Overt subjects in early Basque and other null subject languages. International Journal of Bilingualism, 17(3), 309–336.

27.

Givón

(1990). Syntax: A functional-typological introduction. Amsterdam, The Netherlands: John Benjamins Publishing.

28.

Givón

(1992). The grammar of referential coherence as mental processing instructions. Linguistics, 30, 5–55.

29.

Guerriero

Cooper

Oshima-Takane

Kuriyama

(2001). A discourse–pragmatic explanation for argument realization and omission in English and Japanese children’s speech. In Do

Dominguez

Johansen

(Eds.), Proceedings of the 25th annual Boston University Conference on Language Development (pp. 319–330). Somerville, MA: Cascadilla Press.

30.

Guerriero

Oshima-Takane

Kuriyama

(2006). The development of referential choice in English and Japanese: A discourse–pragmatic perspective. Journal of Child Language, 33, 823–857.

31.

Hohenstein

Eisenberg

Naigles

(2006). Is he floating across or crossing afloat? Cross-influence of L1 and L2 in Spanish–English bilingual adults. Bilingualism: Language and Cognition, 9, 249–261.

32.

Hulk

Müller

(2000). Bilingual first language acquisition at the interface between syntax and pragmatics. Bilingualism: Language and Cognition, 3, 227–244.

33.

Iwasaki

(1985). The ‘Given A Constraint’ and the Japanese Particle ga. In DeLancey

Tomlin

R. S.

(Eds.), Proceedings of the first annual pacific linguistics conference (pp. 152–162). Eugene, OR: Department of Linguistics, University of Oregon.

34.

Jarvis

Pavlenko

(2008). Crosslinguistic influence in language and cognition. New York: Routledge.

35.

Johnson

Newport

(1989). Critical period effects in second language learning: The influence of maturational state on the acquisition of English as a second language. Cognitive Psychology, 21, 60–99.

36.

Kohn

S. E.

Cragnolino

(2003). The role of preferred argument structure for understanding aphasic sentence planning. In Du Bois

J. W.

Kumpf

L. E.

Ashby

W. J.

(Eds.), Preferred argument structure: Grammar as architecture for function (pp. 339–351). Amsterdam, The Netherlands: John Benjamins.

37.

Kumagai

(2006). Information management in intransitive subjects: Some implications for the preferred argument structure theory. Journal of Pragmatics, 38(5), 670–694.

38.

Kumpf

L. E.

(1992). Preferred argument in second language discourse: A preliminary study. Studies in Language, 16(2), 369–403.

39.

Kumpf

L. E.

(2003). Genre and preferred argument structure: Sources of argument structure in classroom discourse. In Du Bois

J. W.

Kumpf

L. E.

Ashby

W. J.

(Eds.), Preferred argument structure: Grammar as architecture for function (pp. 81–108). Amsterdam, The Netherlands: John Benjamins.

40.

C. N.

Thompson

S. A.

(1989). Mandarin Chinese: A functional reference grammar. Berkeley, CA: University of California Press.

41.

Lin

W.-H.

(2009). Preferred argument structure in Chinese: A comparison among conversations, narratives and written texts. In Xiao

(Ed.), The Proceedings of the 21st North American Conference on Chinese Linguistics (NACCL-21), (pp. 341–357). Rhode Island: Bryant University, Smithfield.

42.

MacWhinney

(2000). The CHILDES project: Tools for analyzing talk: Volume I: Transcription format and programs, volume II: The database. Computational Linguistics, 26(4), 657–657.

43.

Meyer

(1969). Frog, where are you. New York: Dial Books.

44.

Müller

Hulk

(2001). Crosslinguistic influence in bilingual language acquisition: Italian and French as recipient languages. Bilingualism: Language and Cognition, 4, 1–21.

45.

Muñoz

Singleton

(2011). A critical review of age-related research on L2 ultimate attainment. Language Teaching, 44(1), 1–35.

46.

Narasimhan

Budwig

Murty

(2005). Argument realization in Hindi caregiver-child discourse. Journal of Pragmatics, 37, 461–495.

47.

O’Dowd

(1990). Discourse pressure, genre and grammatical alignment—after Du Bois. Studies in Language, 14(2), 365–403.

48.

Rothman

. (2009). Pragmatic deficits with syntactic consequences: L2 pronominal subjects and the syntaxpragmatics interface. Journal of Pragmatics, 41, 951–973.

49.

Pirvulescu

Roberge

Thomas

Pérez-Leroux

A. T.

Strik

(2014). Bilingual effects: Exploring object omission in pronominal languages. Bilingualism: Language and Cognition, 17(3), 495–510.

50.

Schmid

M. S.

(2014). The debate on maturational constraints in bilingual development: A perspective from first-language attrition. Language Acquisition, 21(4), 386–410. doi:10.1080/10489223.2014.892947.

51.

Sutherland-Smith

(1996). Spoken narrative and preferred clause structure: Evidence from Modern Hebrew Discourse. Studies in Language, 20(1), 163–189.

52.

Sorace

(2016). Referring expressions and executive functions in bilingualism. Linguistic Approaches to Bilingualism, 6(5), 669–684.

53.

Sorace

Serratrice

(2009). Internal and external interfaces in bilingual language development: Beyond structural overlap. International Journal of Bilingualism, 13(2), 195–210.

54.

Sorace

Serratrice

Filiaci

Baldo

(2009). Discourse conditions on subject pronoun realization: Testing the linguistic intuitions of older bilingual children. Lingua, 119, 460–477.

55.

Tao

Thompson

S. A.

(1994). The Discourse and grammar interface: Preferred clause structure in Mandarin conversation. Journal of Chinese Language Teachers Association, 29(3), 1–34.

56.

Tao

(2007). Caozuo Yuti zhong Dongci Lunyuan Jiegou de Shixian Ji Yuyong Yuanze [The pragmatics of argument structure in Chinese procedural discourse], Zhongguo Yuwen [Chinese Language], 1, 3–13.

57.

Weber

E. G.

(2003). Nominal information flow in the talk of two boys with autism. In Du Bois

J. W.

Kumpf

L. E.

Ashby

W. J.

(Eds.), Preferred argument structure: Grammar as architecture for function (pp. 353–383). Amsterdam, The Netherlands: John Benjamins.

58.

Yip

Matthews

(2007). The bilingual child: Early development and language contact. Cambridge, UK: Cambridge University Press.