Bidirectional cross-linguistic influence in Cantonese–English bilingual children: The case of right-dislocation

Abstract

This corpus-based study demonstrates a case of bidirectional cross-linguistic influence in the acquisition of right-dislocation by Cantonese–English bilingual children and interprets the results in relation to Hulk and Müller’s hypothesis for cross-linguistic influence. Longitudinal data reveal qualitative and quantitative differences between bilingual and monolingual children in the development of right-dislocation in English and Cantonese. While right-dislocation lies at the syntax–pragmatics interface, both delay and acceleration are observed in bilingual development. The article’s findings in general support Hulk and Müller’s hypothesis for cross-linguistic influence in bilingual first language acquisition, but the bidirectional influence observed is not predicted by their formulation of the hypothesis. Moreover, the results suggest that language dominance may influence the directionality of cross-linguistic influence.

Keywords

Cantonese–English bilingual children cross-linguistic influence language dominance right-dislocation syntax–pragmatics interface

Introduction

In recent bilingual first language acquisition research, the interaction between two or more languages in children’s development has been at the center of discussion. Following Döpke (1998), Hulk and Müller (2000), Müller and Hulk (2001), and Serratrice, Sorace, and Paoli (2004), the interaction can be viewed as a form of cross-linguistic influence. Recent studies have tried to understand where cross-linguistic influence occurs and why it occurs in some domains rather than others.

One influential hypothesis regarding cross-linguistic influence in bilingual first language acquisition is proposed by Hulk and Müller (2000). They hypothesize that cross-linguistic influence occurs only when two conditions are both met: (1) a syntax–pragmatics interface is involved; and (2) the two languages overlap at the surface level. By investigating the development of object drop and root infinitives in a Dutch–French bilingual child and a German–Italian bilingual child, they found cross-linguistic influence in the domain of object drop where the two conditions hold, but not in the domain of root infinitives where the second condition is not satisfied. Hulk and Müller’s study has generated much research on defining conditions for cross-linguistic influence in bilingual first language acquisition in the past decade.

Müller and Hulk (2001) further develop the interface condition, proposing that the syntax–pragmatics interface is a vulnerable domain in acquisition and is also a prime locus of protracted delays in young bilingual children. In a similar vein, the Interface Hypothesis put forward by Sorace and colleagues (Serratrice et al., 2004; Sorace & Filiaci, 2006; Sorace & Serratrice, 2009) also suggests that language structures involving the syntax–pragmatics interface are domains that are vulnerable for cross-linguistic influence and show delay in both child and adult bilinguals. However, acceleration has also been observed in bilingual first language acquisition (Kupisch, 2005).

Hulk and Müller’s (2000) structural overlap condition predicts unidirectional cross-linguistic influence: if language A allows two or more options in a target structure, and language α allows one of them, cross-linguistic influence is likely to occur from language α to language A, but not vice versa. However, whether structural overlap alone can predict the directionality of cross-linguistic influence and whether cross-linguistic influence is unidirectional have been questioned. Some studies suggest that language dominance plays a major role in accounting for the directionality of cross-linguistic influence (Yip & Matthews, 2000, 2007). In domains such as the acquisition of dative constructions by Cantonese–English bilingual children, bidirectional cross-linguistic influence is observed (Gu, 2010).

This corpus-based study investigates cross-linguistic influence in the acquisition of right-dislocation (RD) by Cantonese–English bilingual children. It aims to test Hulk and Müller’s (2000) two conditions and contribute to the ongoing discussion about the conditions for and directionality of cross-linguistic influence in bilingual first language acquisition. Furthermore, it will enrich our understanding of the bilingual acquisition of RD from the perspective of two typologically divergent and genetically unrelated languages, namely English and Cantonese. While RD is rare in English (Notley, Van der Linden, & Hulk, 2007), it is frequent in adult Cantonese (Cheung, 2009; Luke, 2004). The frequency asymmetry makes the acquisition of RD by Cantonese–English bilingual children especially interesting. To the best of our knowledge, there have been no published studies on the acquisition of Cantonese RD.

The article is organized as follows. We first discuss previous acquisition studies on RD and outline the properties of RD in English and Cantonese. On the basis of these two sources of information, we propose three research questions, discuss the methods, and show how the corpus data are analyzed. We then present the results for the bilingual development of RD in each language. Finally, we show how our findings shed light on the factors that contribute to cross-linguistic influence.

RD in first language acquisition

RD is a clear interface phenomenon in English (and many other languages) that uses syntactic means to indicate a pragmatic function. In general, RD is used to mark a topic, which is ‘dislocated’ at the right periphery of a clause:

(1) It’s not bad, that cake.

The construction has been investigated in a number of language acquisition studies in Germanic and Romance languages (De Cat, 2007; Devlin, Folli, Henry, & Sevdali, 2015; Notley et al., 2007; Van der Linden & Sleeman, 2007). There is wide variation in the frequency of use of RD across languages. Based on the same set of corpus data, Notley et al. (2007) and Van der Linden and Sleeman (2007) found that an English monolingual child produced no RD from age 2;0 to 3;11, whereas a French monolingual child produced a large number of RDs during a similar age span. De Cat (2007) shows that dislocations are pervasive in early child French. This variation in monolingual children’s use of RD is in line with adult input, with RD being frequent in the French input, whereas it is infrequent in the English input.

In bilingual first language acquisition, cross-linguistic influence is pervasive at the syntax–pragmatics interface. Notley et al. (2007) found cross-linguistic influence from French to English manifested in the production of more English RD by French–English bilingual children. Devlin et al. (2015) investigate the development of RD in English in a trilingual Italian–English–Scots Gaelic child. They argue that more RDs are produced in English as a result of cross-linguistic influence from Italian, in which dislocations in general and clitic RD in particular are very frequent in the input.

Right-dislocation in English and Cantonese

In English RD, a Determiner Phrase (DP) is ‘dislocated’ to the right edge of the clause. There are generally two types of RD in English:¹

In Pronominal RD, the dislocated DP can be co-referential with a subject or object in the preceding clause. (2a) illustrates subject dislocation where the dislocated DP John refers to the subject he, whereas (2b) illustrates object dislocation where the dislocated element the bear refers to the object it.

In Gapped RD in English, the subject and the copula are omitted, with an Adjective Phrase serving as the predicate, as in (3). The dislocated element (e.g., that cake in (3)) is co-referential with a gap (e.g., the subject) in the preceding clause.

(2) a. He is happy, John.

b. He likes it, the bear.

(3) Not bad, that cake. (Lambrecht, 2001, p. 1071)

Both types of RD in English are generally used for topic marking, with the dislocated element being a topic that communicates old or given information (Lambrecht, 2001).

Cantonese, like English, has Pronominal RD and Gapped RD. In both of these types, a Sentence Particle (SP) or a pause is required preceding the dislocated element. Similar to English, subject dislocation (4a) and object dislocation (4b) are both allowed in Pronominal RD in Cantonese. However, (4b) is a marked option in Cantonese, as postverbal third person pronoun keoi5佢 is commonly omitted in Cantonese.

(4) a. 佢好開心㗎, 阿明

keoi5 hou2 hoi1sam1 gaa1 Aa3Ming4

he very happy SP Aaming

‘He is very happy, Aaming.’

b 我好鍾意佢啊, 隻貓貓

Ngo5 hou2 zung1ji3 keoi5 aa1, zek3 maau1maau1

I very like it SP CL cat

‘I like it very much, the cat.’

Unlike English, which allows an inanimate DP as the dislocated part, Pronominal RD in Cantonese with an inanimate DP dislocation, either co-referring to a subject or object, is ungrammatical (5).

(5) * 佢哋幾靚喎, 啲衫

keoi5dei6 gei2 leng3 wo3, di1 saam1

they quite nice SP CL clothes

‘They are quite nice, these clothes.’

Regarding Gapped RD in Cantonese, typically a DP which is the subject of a clause appears as the dislocated element. This construction is commonly used in exclamations when the predicate is placed first for emphasis:

(6) 幾靚喎, 啲衫 (Matthews & Yip, 2011, p. 82)

gei2 leng3 wo3, di1 saam1

quite nice SP CL clothes

‘Not bad, those clothes.’

While English Gapped RD only allows DP as the dislocated element (3), a great variety of elements apart from DP, such as modal verbs (7a), adverbs (7b) or even a combination of these elements (7c), can be dislocated in Cantonese.

(7) a. 佢走咗啦, 應該

keoi5 zau2-zo2 laa1 jing1goi1

he leave-Perf SP should

‘He should have left.’

b. 佢走咗啦, 已經

keoi5 zau2-zo2 laa1 ji5ging1

he leave-Perf SP already

c. 走咗啦, 佢應該

zau2-zo2 laa1 keoi5 jing1goi1

leave-Perf SP he should

With respect to pragmatic functions, Pronominal RD in Cantonese resembles English RD, being used to mark or maintain a topic. However, the dislocated element in Cantonese Gapped RD does not have to be given information. Cheung (2009) argues that Gapped RD in Cantonese is driven by the realization of focus, with the clause preceding the dislocation containing the focus.

Similarities and differences between English and Cantonese RD constructions are summarized in Table 1. Both languages have Pronominal RD and Gapped RD, but with various differences. In Pronominal RD, while both animate and inanimate DPs are possible in English, only animate DPs are allowed in Cantonese. Another difference is that Pronominal RD with object dislocation is common in English but uncommon in Cantonese. In English Gapped RD, the dislocated element is limited to DP while Cantonese allows a great variety of dislocated elements. English RD and Cantonese RD also differ in terms of the frequency of use: in English, Pronominal RD predominates, while in Cantonese the Gapped type is more prevalent (Cheung, 1997 for Cantonese). This difference is reflected in the monolingual data for each language (Tables 5 and 8).

Table 1.

English and Cantonese RD.

	English	Cantonese
Pronominal RD	Yes - More frequent - Both animate and inanimate DPs allowed as dislocated elements	Yes - Not frequent - Limited to animate DPs as dislocated elements - Object dislocation rare
Gapped RD	Yes - Not frequent - Dislocated elements limited to DPs	Yes - Very prevalent - A variety of elements allowed as dislocated elements

Research questions

In this study, we take up the question of whether Hulk and Müller’s (2000) two conditions can successfully account for the use of RD in Cantonese–English bilingual children. By looking at the developmental rate of RD in each language, we aim to determine whether there is any delay or acceleration in bilingual children. We will also examine the role of language dominance in influencing the directionality of cross-linguistic influence. We pose three research questions as follows:

Is there any form of cross-linguistic influence in the acquisition of RD in each of the bilingual children’s target languages? If so, what is the directionality of cross-linguistic influence?

What is the developmental rate of English RD and Cantonese RD in bilingual children? Does the syntax–pragmatics interface create delay (or acceleration) in bilingual children in this domain?

Are Hulk and Müller’s two conditions adequate in predicting cross-linguistic influence? Does language dominance play a role in this respect?

Method

Participants

This study investigates the longitudinal data of seven Cantonese–English bilingual children drawn from the Hong Kong Bilingual Child Language Corpus (Yip & Matthews, 2007) available in the CHILDES database (MacWhinney, 2000). All seven bilingual children grew up in Hong Kong middle-class families where the parents followed the ‘one parent-one language’ approach, and were exposed to Cantonese and English simultaneously and regularly from birth. Parents of these children are native speakers of English or Cantonese and consciously followed the ‘one parent-one language’ approach, each using their native language to communicate with their children. The parents were bilingual to different degrees and for the most part capable of following conversations in their non-native language. In the case of six bilingual children (Timmy, Sophie, Alicia, Janet, Llywelyn, and Charlotte), the mother speaks Hong Kong Cantonese and the father speaks British English. Note that Timmy, Sophie, and Alicia are siblings in the same family. Code-mixing, which is characteristic of Hong Kong Cantonese, is frequent in the Cantonese input the children are exposed to. Despite the ‘one parent-one language’ principle, the quantity of input from the two languages is by no means balanced: Cantonese is the language of the community and also the language spoken by the children’s maternal relatives; regular input in English came solely from the father and from Filipina domestic helpers who speak Philippine English. Consequently, these six bilingual children received more Cantonese than English input in their first four years of life. But in the case of Kathryn, it is the mother who speaks British English, and the father Hong Kong Cantonese. According to her mother’s observations, Kathryn received relatively balanced Cantonese and English input during the period of recording.

Spontaneous speech data were collected by two research assistants at the children’s homes at weekly or bi-weekly intervals from age 1;03 (years; months) to 4;06. In each recording, the children were encouraged to speak in Cantonese for half an hour and in English for half an hour when they were engaged in daily activities such as playing, reading, and role play. The research assistants were native speakers of Cantonese and spoke English as their second language. On most occasions, only the research assistants and the children were involved in the recordings, while parents and adult caregiver(s) (e.g., grandparents and domestic helpers) were not present. As a result, we are not able to undertake systematic analysis of adult input in this study. The input data available in the corpus were mainly from research assistants, which was not considered representative of the input the bilingual children were exposed to for two reasons. First, the research assistants’ English was qualitatively different from the native English input the children were regularly exposed to, showing some influence from Cantonese. Second, the amount of input from the research assistants was relatively small. It is unlikely that an hour’s input from the research assistants every week would dramatically change the children’s developmental patterns.

Following Yip and Matthews (2007), language dominance of the seven bilingual children is determined by comparing children’s Mean Length of Utterance (MLU) in each language. Among the seven bilingual children, five are dominant in Cantonese, one is balanced in both languages, and one is dominant in English. Yip and Matthews (2007) show that MLU differential (the difference between MLU scores in each language) is a reliable measure of language dominance for children in the Hong Kong Bilingual Child Language Corpus. The bilingual children’s MLU differentials match patterns of cross-linguistic influence and language preference.

In addition to the corpus data, diary data with relevant contextual information were kept for Timmy (1;03–6;00), Sophie (1;06–5;06), and Alicia (1;00–5;04). The diary data for the three siblings fill some gaps left by the corpus data, since RD in English is relatively infrequent and may not be captured by corpus data. Diary data do not lend themselves to quantitative comparison, and will be used only for qualitative analysis.

Corpus data for seven English-speaking children were analyzed as the baseline for child English: Adam, Eve, and Sarah from Brown (1973), Jane from Cruttenden (1978), Thomas from Lieven, Salomo, and Tomasello (2009), Peter from L. Bloom (1973), and Trevor from Demetras (1989). The choice of comparing bilingual children’s development to these seven English monolingual children is based on three considerations: first, the Brown and Bloom corpora are well known and widely studied in the literature; second, most of the children are from middle-class families (except for Sarah, who was the child of a working-class family); third, and most importantly, their recordings cover a similar age span to that of the bilinguals in the present study. In the cases of Adam, Sarah, and Thomas, we selected 20–25 sessions for each child in order to be consistent with the bilingual and the other English monolingual corpora.

As the baseline for child Cantonese, eight monolingual children from the Hong Kong Cantonese Child Language Corpus (Lee et al., 1996) were selected. All the Cantonese monolingual children, aged from 1;05 to 3;08, were recorded on a bi-weekly basis. Six out of the eight children are from middle-class families with the exception that CGK and WBH are from working-class families.

Information on dominant language, age range, number of sessions, and total number of utterances (code-mixing utterances excluded) in each language in bilingual and monolingual children are summarized in Tables 2 –4.

Table 2.

Background of the seven Cantonese–English bilingual children in the Hong Kong Bilingual Child Language Corpus (based on Yip & Matthews, 2007).

Child	Dominant language	Age range^a	English		Cantonese
Child	Dominant language	Age range^a	Files	No. of analyzed utterances	Files	No. of analyzed utterances
Alicia	Cantonese	1;03.10–3;00.24	40	3521	40	5660
Sophie	Cantonese	1;06.00–3;00.09	40	5735	40	11,408
Timmy	Cantonese	2;01.22–3;06.25	38	6205	34	9669
Janet	Cantonese	2;10.16–3;11.11	22	3468	22	5584
Llywelyn	Cantonese	2;00.12–3;04.17	17	3884	17	3822
Kathryn	Balanced	3;01.05–4;06.07	14	3394	17	4229
Charlotte	English	1;08.28–3;00.03	19	3934	19	1989
Total			190	30,141	189	42,361

Age is indicated in years;months.days.

Table 3.

Background of the seven English monolingual children.

Child	Age range	English files	No. of analyzed utterances
Adam	2;03.04–3;00.11	20	16,874
Eve	1;06.00–2;03.00	20	10,826
Sarah	2;03.05–2;07.05	20	3636
Jane	1;05.17–3;11.17	21	1323
Thomas	2;00.12–3;00.07	25	12,494
Peter	1;09.08–3;01.20	20	23,003
Trevor	2;00.27–3;11.27	28	6587
Total		154	74,743

Table 4.

Background of the eight Cantonese monolingual children.

Child	Age range	Cantonese files	No. of analyzed utterances
CCC	1;10.08–2;10.27	22	9251
CGK	1;11.01–2;09.09	19	4837
CKT	1;05.22–2;07.02	25	10,502
HHC	2;04.08–3;04.14	16	12,062
LLY	2;08.10–3;08.09	20	7479
LTF	2;02.10–3;02.18	16	9661
MHZ	1;07.00–2;08.06	26	5026
WBH	2;04.15–3;02.20	16	3082
Total		160	61900

Data analysis

All of the children’s utterances were first extracted by CLAN from the CHILDES database (MacWhinney, 2000). Interjudge coding reliability was assessed: all of the children’s utterances were checked by the first author and a research assistant who received extensive training in linguistics. Only those that are identified as RD by both assessors are included in the present study. Imitations, recitations or unclear utterances are excluded. To be qualified as a non-imitative, the utterance produced by the child must not be identical to the preceding adult utterance: there must be at least one element different from the preceding adult utterance, showing that the child has already produced his or her own version of the structure. At the same time, the non-imitative utterance must be clear and recognizable by ear. If the dislocated element is a vocative phrase (e.g., You come in, Daddy.), the utterance is not counted as RD. To determine the age of emergence of RD, we choose age of first non-imitative use as the measure of first emergence, following Snyder and Stromswold (1997).

Pronominal RD in English and Cantonese must have the following properties: the construction contains a subject or object pronoun in the canonical position and a co-referring DP that appears to be dislocated to the right periphery of the utterance; the dislocated DP can take the form of a bare noun or a demonstrative. Examples of Pronominal RD in English and Cantonese are given as follows:

(8) a. He likes us, Prince. (Sophie 3;00.00)

b. I like it, skipping ropes. (Janet 3;02.06)

c. 佢去買嘢呀, 呢個媽咪 (Janet 3;10.27)

keoi5 heoi3 maai5 je5 aa3 nei1 go3 maa1mi4

she go buy thing SP that CL mommy

‘She wentshopping, that mommy.’

The following criteria are applied in defining Gapped RD: while the dislocated element is restricted to DP in English, the dislocated element in Cantonese can take the form of Verb Phrase, Adverbial Phrase or a combination of constituents; the dislocated part must be interpreted as if it is in the gap in the preceding clause, as in (9):

(9) a. So big this castle. (Alicia 2;10.15)

b. 救返佢呀佢想 (Kathryn 4;03.15)

gau3 faan1 keoi5 aa3 keoi5 soeng2

save back him SP he want

‘He wants to save him.’

The identification of RD in the child data is complicated by the fact that children have grammars that initially license null subjects, both in Cantonese, a pro-drop language, and English, a non-pro-drop language (P. Bloom, 1990; Wang, Lillo-Martin, Best, & Levitt, 1992). If the dislocated element co-refers with a null subject, we consider it as an instance of Gapped RD (10); if the dislocated element is co-referential with the overt object, the utterance is considered as Pronominal RD (11).

(10) a. Have two legs, this turtle. (Timmy 2;11.12)

b. 係咩嚟㗎, 嗰個? (Sophie 2;06.12)

hai6 me1 lei4 gaa3 go2 go3

be what SP SP that CL

‘What is it, that one?’

(11) Saw it, ball. (Adam 2;04.03)

Moreover, we need to take object omission into consideration. English-speaking children omit objects much less frequently than subjects (Wang et al., 1992). As the omission rate of object is high in child Cantonese (Yip & Matthews, 2007), utterances with a dislocated part co-referring with the null object of the clause are regarded as Gapped RD, as given below:

(12) a. Buy for me, this one. (Sophie 2;05.30)

b 我食晒呀糖. (Sophie 2;09.05)

Ngo5 sik6-saai3 aa3 tong2

I eat-all SP candy

‘I have eaten up the candy.’

Regarding RD in Cantonese, we restrict our analysis to utterances with a SP preceding the dislocated element. Though a SP is not obligatory in Cantonese RD, utterances without a SP may allow different interpretations. For example, gam1ziu1今朝 ‘this morning’ in (13) can be interpreted as a misplacement of adverbial rather than an instance of RD. Such ambiguous examples were discarded.

(13) 食咗今朝 (Janet 3;11.11)

sik6-zo2 gam1ziu1

eat-Perf this morning

‘I’ve finished eating this morning.’

Results

Acquisition of RD in English

Regarding the frequency of Pronominal RD in English, the bilingual group show a higher mean frequency (0.103%) than the monolinguals (mean 0.034%) but this difference does not reach significance (U = 10, p = .073).² However, bilingual children produced significantly more Gapped RD than English monolingual children, who produced no exemplars of this type (U = 0, p = .001) (see Table 5).

Table 5.

Frequency of English RD in bilingual and English monolingual children.

Group	Child	Pronominal RD	Frequency of Pronominal RD (%)	Gapped RD	Frequency of Gapped RD (%)
Bilingual(N = 7)	Alicia	3	0.085	9	0.256
	Sophie	12	0.209	5	0.087
	Timmy	10	0.161	4	0.064
	Janet	3	0.087	2	0.058
	Llywelyn	6	0.154	4	0.103
	Kathryn	0	0	2	0.059
	Charlotte	1	0.025	3	0.076
	Mean	5	0.103	4.1	0.1
English monolingual(N = 7)	Adam	12	0.071	0	0
	Eve	3	0.028	0	0
	Sarah	3	0.083	0	0
	Peter	9	0.039	0	0
	Trevor	0	0	0	0
	Jane	0	0	0	0
	Thomas	2	0.016	0	0
	Mean	4.1	0.034	0	0

By comparing the frequency of Pronominal RD and Gapped RD between Cantonese-dominant (shaded rows in Table 5) and non-Cantonese-dominant (i.e., balanced and English-dominant) bilingual children, we have observed higher frequency in the five Cantonese-dominant bilingual children (Pronominal RD: mean 0.139%; Gapped RD: mean 0.113%) than the two non-Cantonese-dominant bilingual children Kathryn and Charlotte (Pronominal RD: mean 0.013%; Gapped RD: mean 0.068%).

Another contrast is found between bilingual and English monolingual children in the grammatical functions of the dislocated part of RD, as summarized in Table 6: English monolingual children, in general, dislocate a DP that co-refers with the object of a clause in Pronominal RD (mean 89.7% for object dislocation vs. mean 10.3% for subject dislocation). Bilingual children dislocate more subjects to the right periphery of the clause than English monolinguals in both types of RD (Pronominal RD: mean 48.6% for bilingual vs. mean 10.3% for monolingual; Gapped RD: mean 75.9% for bilingual vs. 0 for monolingual).

Table 6.

Distribution of dislocated elements in bilingual and English monolingual children’s English RD.

Group	Pronominal RD		Gapped RD
Group	Subject(%)	Object(%)	Subject(%)	Object(%)	Others(%)
Bilingual(N = 7)	17(48.6)	18(51.4)	22(75.9)	5(17.2)	2(6.9)
English monolingual (N = 7)	3(10.3)	26(89.7)	0(0)	0(0)	0(0)

In particular, Gapped RDs, exemplified below, are only found in the bilingual data:

(14) Wash her hand, Barbie. (Charlotte 2;07.23)

(15) Drink something in where, we? We drink something in where?

(Sophie 3;11.09 from diary data)

(16) So blue, this one. (Alicia 2;10.15)

(17) Come back you can. (Alicia 2;01.01)

(18) A monster they are. (Sophie 2;11.05)

In the Discussion section, we argue that these forms of Gapped RD are the results of cross-linguistic influence from Cantonese.

The age of first emergence of the two types of English RD in the two groups of children is shown in Table 7.

Table 7.

First emergence of English RD in bilingual and English monolingual children.

Group	Child	Pronominal RD	Gapped RD
Bilingual(N = 7)	Alicia	2;03.02	1;10.16
	Sophie	2;02.16	2;04.18
	Timmy	2;10.01	2;05.05
	Llywelyn	2;06.20	2;02.03
	Janet	3;02.06	3;07.21
	Kathryn	Not attested	4;02.16
	Charlotte	2;07.23	2;04.20
	Mean^a	2;05	2;02
English monolingual(N = 7)	Adam	2;04.03	Not attested
	Eve	1;11.00	Not attested
	Sarah	2;04.19	Not attested
	Peter	1;10.11	Not attested
	Trevor	Not attested	Not attested
	Jane	Not attested	Not attested
	Thomas	2;10.21	Not attested
	Mean	2;03	Not attested

Janet and Kathryn are excluded for calculation of the mean age of first emergence as they were recorded mainly after age 3.

English monolingual children began to produce Pronominal RD as early as 1;10 and the mean age of first emergence is 2;03. The age of first emergence of Pronominal RD in bilingual children ranges from 2;02 to 3;02, and the mean age of first emergence is 2;05. In five out of seven bilingual children (Alicia, Timmy, Llywelyn, Kathryn, and Charlotte), Gapped RD emerged before Pronominal RD in English.

Acquisition of RD in Cantonese

Bilingual children produced significantly more Pronominal RD (mean 0.064%) than Cantonese monolingual children (0.009%) (U = 3, p = .002). No significant difference was observed between bilingual and Cantonese monolingual children in terms of Gapped RD (U = 21, p = .463), though bilingual children produced Gapped RD even more frequently (mean frequency 0.835%) than Cantonese monolinguals (mean frequency 0.366%) (see Table 8). Comparing one English-dominant and six non-English-dominant (i.e., Cantonese-dominant and balanced) bilinguals, the non-English-dominant bilinguals (shaded rows in Table 8) produced Gapped RD in Cantonese much more frequently (mean 0.958%) than the English-dominant bilingual child Charlotte (0.101%).

Table 8.

Frequency of Cantonese RD in bilingual and Cantonese monolingual children.

Group	Child	Pronominal RD	Frequency of Pronominal RD (%)	Gapped RD	Frequency of Gapped RD (%)
Bilingual(N = 7)	Alicia	4	0.071	26	0.459
	Sophie	2	0.018	57	0.500
	Timmy	6	0.062	61	0.631
	Llywelyn	1	0.018	9	0.161
	Janet	6	0.157	122	3.192
	Kathryn	3	0.071	34	0.804
	Charlotte	1	0.050	2	0.101
	Mean	3.3	0.064	44.4	0.835
Cantonese monolingual(N = 8)	CCC	1	0.011	15	0.162
	CGK	0	0	5	0.103
	CKT	0	0	19	0.181
	HHC	0	0	34	0.282
	LLY	4	0.053	75	1.003
	LTF	1	0.010	49	0.507
	MHZ	0	0	4	0.071
	WBH	0	0	19	0.616
	Mean	0.75	0.009	27.5	0.366

Regarding the dislocated part (see Table 9), both bilingual and Cantonese monolingual children primarily dislocate subjects to the right periphery of a clause in Pronominal RD. Object dislocations in Pronominal RD are observed in bilinguals (4 tokens) but absent in monolinguals. Subject dislocation is also the most frequently produced type in both bilingual and monolingual children’s production of Gapped RD. Like Cantonese adults, bilingual children and Cantonese monolingual children are able to dislocate different constituents (‘Others’ in Table 9) to the end of a clause (making up 45% in the bilingual group and 28.6% in the Cantonese monolingual group).

Table 9.

Distribution of dislocated elements in bilingual and Cantonese monolingual children’s Cantonese RD.

Group	Pronominal RD			Gapped RD
Group	Subject (%)	Object (%)	Others (%)	Subject (%)	Object (%)	Others (%)
Bilingual(N = 7)	17 (73.9)	4(17.4)	2(8.7)	159(51.1)	12(3.9)	140 (45.0)
Cantonese monolingual(N = 8)	5(83.3)	0	1(16.7)	144(65.5)	13(5.9)	63(28.6)

To determine whether there is any delay or acceleration in this domain of bilingual development, we also have examined the age of the first emergence of Cantonese RD in the bilingual and monolingual groups (see Table 10).

Table 10.

First emergence of Cantonese RD in bilingual and Cantonese monolingual children.

Group	Child	Pronominal RD	Gapped RD
Bilingual(N = 7)	Alicia	1;10.16	1;09.10
	Sophie	2;06.12	1;06.28
	Timmy	2;02.27	2;01.29
	Llywelyn	2;06.20	2;05.10
	Janet	3;02.13	⩽ 2;10.16^b
	Kathryn	3;08.27	⩽ 3;01.05
	Charlotte	1;08.28	2;03.17
	Mean^a	2;02	2;00
Cantonese monolingual(N = 8)	CCC	2;09.07	2;01.10
	CGK	Not attested	1;11.29
	CKT	2;02.05	1;09.14
	HHC	Not attested	2;05.03
	LLY	3;00.22	2;09.09
	LTF	2;04.27	2;03.30
	MHZ	Not attested	2;00.03
	WBH	Not attested	⩽ 2;04.15
	Mean	2;07	2;02

Janet and Kathryn are excluded for calculation of the mean age of first emergence as they were recorded mainly after age 3.

⩽ indicates that the use of the structure might appear before or at the age given because it is the time when the first recording was available for analysis.

There is evidence of acceleration in the bilingual group for both types of RD in terms of the mean age of emergence. Moreover, the mean temporal gap between the emergence of Pronominal RD and Gapped RD was around 2 months in bilingual group, much shorter than 5 months in the Cantonese monolingual group. The English-dominant bilingual child, Charlotte, behaved differently from the non-English-dominant bilingual children: she produced the first Pronominal RD at the age of 1;08.28, much earlier than her first production of Gapped RD, while the use of Gapped RD occurred well before the appearance of Pronominal RD in the non-English-dominant bilinguals. This is consistent with the bilingual children’s language dominance as discussed below.

Discussion

Cross-linguistic influence in bilingual acquisition of RD

In the acquisition of RD in English, cross-linguistic influence manifests itself in the use of significantly more Gapped RD in bilingual children than in English monolingual children. As Gapped RD is not commonly observed in English monolingual acquisition, it is not surprising that Gapped RD is not attested in the English monolingual data. In Cantonese, Gapped RD is much more frequent, as seen in the Cantonese monolingual data (see Table 8). The higher rate of Gapped RD in bilingual children’s English may be attributed to the presence of extensive dislocation in Cantonese. This finding matches those of Notley et al. (2007) and Devlin et al. (2015) where the frequency of RD in English in bilingual children’s data is elevated under the influence of French and Italian respectively.

Cross-linguistic influence from Cantonese is also seen in bilingual children’s lower rate of object dislocation in both types of RD in English (Table 6). In the cases of Pronominal RD produced by English monolingual children, a dislocated DP is commonly co-referential with the object pronoun it, whereas bilingual children show a higher rate of subject than object dislocation. The differences between bilingual and English monolingual children may be attributed to the influence of Cantonese, in which object dislocation is infrequent in both Pronominal RD and Gapped RD (see Table 9). Moreover, Pronominal RD with object dislocation co-referential with an overt third person pronoun keoi5佢 is less prominent, as in (4b). Therefore, bilingual children tend not to produce English Pronominal RD with object dislocation, as in ‘I like it, the cat.’

The use of Gapped RD reveals a qualitative difference between the two groups, exhibiting cross-linguistic influence from Cantonese to English. The example in (14) can be ‘reconstructed’ using the SVO word order like ‘Barbie wash her hand.’ (15) is taken from the diary data of Sophie. The dislocated pronoun we co-refers with the subject of the clause, which is confirmed by the following SVO sentence. These utterances produced by bilingual children are very likely based on the parallel Cantonese Gapped RD with subject dislocation. Before age 3, bilingual children produced a number of comparable Cantonese Gapped RD constructions which follow VOS word order:

(19) 飲水呀, 呢個 (Sophie 2;03.01)

jam2 seoi2 aa3 nei1go3

drink water SP this CL

‘It drinks water, this one.’

Bilingual children’s development reflects the possibility of transferring Cantonese Gapped RD to their English. As the majority of the bilingual children are Cantonese-dominant, their Cantonese is ahead of their English in general. Cantonese RD, in particular, is more frequent and develops earlier than English RD.

Subject and copula are both omitted in Gapped RDs such as (16). As these constructions are not observed in the monolingual children’s English, it appears that they are the result of cross-linguistic influence from the child’s Cantonese. Such dislocations are possible in colloquial English and overlap with Cantonese Gapped RD at the surface level, which are commonly observed in adult and child Cantonese:

(20) a. 好叻喎, 你 (Matthews & Yip, 2011, p. 82)

hou2 lek1 wo3, lei5

very smart SP you

‘You are so smart.’

b. 好污糟呀巴士 (CKT 2;07.02)

hou2 wu1zou1 aa3 baa1si2

very dirty SP bus

‘The bus is very dirty.’

Furthermore, the diary data show that Alicia was still producing utterances like ‘Very nice, this one.’ at the age of 4;02.20. The duration for which these constructions occurred clearly suggests that the constructions are representative of a grammatical option in the bilingual children’s grammar.

There are two tokens of more complex cases in which more than a single constituent of the utterance is dislocated, as exemplified in (17) and (18). The dislocation of strings like ‘you can’ and ‘they are’ is not grammatical in adult or child English, but allowed in Cantonese as in (7a–c). It appears that bilingual children transferred Cantonese grammars to their English.

As far as the acquisition of RD in Cantonese is concerned, cross-linguistic influence from English takes the form of (a) significantly higher frequency of Pronominal RD (Table 8) and (b) earlier emergence of both types of RD (Table 10) in bilingual children than their Cantonese monolingual counterparts. Pronominal RD is not a prominent type compared to Gapped RD in adult Cantonese. Based on Cheung (1997), the majority of RDs in adult Cantonese belong to the Gapped RD type (91.6%), whereas Pronominal RD occurs only rarely (0.6%). Similar to Cantonese adults, Cantonese monolingual children did not produce Pronominal RD frequently (Table 8). As Pronominal RD is more prominent than Gapped RD in English and occurs comparatively early in monolingual acquisition, it is not surprising to observe the influence of English in bilingual children’s production of Pronominal RD in Cantonese.

Moreover, four tokens of object dislocation in Cantonese Pronominal RD produced by bilingual children, which is absent in Cantonese monolingual children’s production, can be interpreted as cross-linguistic influence from English. An example is given below:

(21) 捉咗佢啦, 蝙蝠 (Timmy 3;06.25)

zuk1-zo2 keoi5 laa1 pin1fuk1

catch-Perf it SP bat

‘I have caught it, the bat.’

Cantonese monolingual children commonly omit keoi5佢 in postverbal position, as Pronominal RD with overt postverbal keoi5佢 co-referring to the dislocated element is uncommon in Cantonese. Specifically, example (21) resembles English Pronominal RD ‘I have caught it, the bat.’ where an overt object pronoun is required.

The quantitative and qualitative differences between bilingual children and their monolingual peers in the acquisition of RD in English and Cantonese indicate that cross-linguistic influence is bidirectional, from Cantonese to English and vice versa.

Delay and acceleration

In the acquisition of Pronominal RD in English, the data suggest a slight delay (2 months) in bilingual children’s development in terms of the mean age of emergence. Since RD is a syntax–pragmatics interface phenomenon which is predicted to be a locus for delay in bilingual development (Müller & Hulk, 2001), the delay relative to monolingual children found in our study is not unexpected.

Gapped RD is not attested in the English monolingual children’s data, but emerges earlier than Pronominal RD in five out of seven bilingual children. Acceleration is found in bilingual children’s acquisition of Pronominal RD and Gapped RD in Cantonese on the basis of the mean age of first emergence attested in the corpus data. In addition, the smaller temporal gap between bilingual children’s first Gapped RD and first Pronominal RD also suggests that bilingual children take less time to acquire Pronominal RD after the emergence of Gapped RD than their Cantonese monolingual counterparts.

We now consider how to account for the acceleration effects observed in the present study. Patuto, Repetto, and Müller (2011) suggest that acceleration effects result from processing preferences, which are interpreted not in terms of cross-linguistic influence but in terms of bilingualism effects in general. If acceleration in the bilingual development is due to bilingualism effects, we would expect similar acceleration patterns in English Pronominal RD, whereas in fact a slight delay was observed. We argue that the overlap between English and Cantonese in the construction of RD may create the possibility for bilingual bootstrapping and thus facilitate bilingual development in this domain. The development of English RD may bootstrap the development of a parallel construction in bilingual children’s Cantonese, and vice versa.

Conditions for cross-linguistic influence in the domain of RD

In what follows, we evaluate Hulk and Müller’s (2000) two conditions and language dominance in accounting for cross-linguistic influence in light of our findings. In accordance with their two conditions, cross-linguistic influence is expected to occur in the acquisition of RD by Cantonese–English bilinguals, as RD is a syntax–pragmatics interface phenomenon. Our findings support the relevance of Hulk and Müller’s conditions for cross-linguistic influence, as cross-linguistic influence manifests itself in the form of quantitative and qualitative differences as well as differential developmental rates between the bilingual and monolingual children.

As shown in Table 1, structural overlap exists in English and Cantonese in the domain of RD: regarding Pronominal RD, English allows both animate and inanimate DP dislocation but Cantonese allows animate DP dislocation; for Gapped RD, different varieties of dislocated elements are possible in Cantonese but only one of them is possible in English. Following the formulation of the structural overlap condition, cross-linguistic influence is predicted to occur from Cantonese to English in Pronominal RD and from English to Cantonese in Gapped RD, but not vice versa.

However, we found cross-linguistic influence in two directions in this domain of bilingual development. On the one hand, we do observe some influence from Cantonese to English in the domain of English Pronominal RD in the form of more subject dislocation in bilinguals than in monolinguals. On the other hand, in the acquisition of Pronominal RD in Cantonese, the influence goes from English to Cantonese; in the domain of the Gapped RD in English, the influence goes from Cantonese to English. For Gapped RD in Cantonese, the directionality of influence, if there is any, is less clear. Our findings suggest that cross-linguistic influence occurs in a domain of overlap but the bidirectional influence observed is not predicted by Hulk and Müller’s (2000) formulation of the hypothesis.

With respect to the role of language dominance, it has been questioned whether language dominance is necessary in accounting for cross-linguistic influence (Hulk & Müller, 2000). We observe clear and consistent differences between Cantonese-dominant and other bilinguals in the acquisition of RD: for English RD, the frequency of both types is higher in Cantonese-dominant bilinguals than in non-Cantonese-dominant bilinguals; in Cantonese, the English-dominant bilingual child Charlotte produced Cantonese RD less frequently than the non-English-dominant bilinguals. Moreover, the developmental patterns of Cantonese RD among the seven bilingual children largely correspond to their language dominance patterns: Pronominal RD emerged earlier than Gapped RD in the English-dominant bilingual child, whereas Gapped RD emerged before Pronominal RD in the Cantonese-dominant bilingual children (Table 10). Our findings suggest that language dominance patterns influence the directionality of cross-linguistic influence in the domain of RD. As we lack a sufficient number of English-dominant bilingual children in the data, more research is needed to further investigate the role of language dominance in bilingual first language acquisition.

Conclusion

This corpus-based study has investigated the acquisition of right-dislocation by seven Cantonese–English bilingual children. Quantitative and qualitative differences are shown between bilingual children and their monolingual counterparts in the development of RD in each language. Cross-linguistic influence manifests itself in the bilingual data primarily in the form of (a) the use of more Gapped RD in English and Pronominal RD in Cantonese than in the monolingual data; (b) the lower rate of object dislocation in both Pronominal RD and Gapped RD in English; (c) the occurrence of RD type in English which are absent from the monolingual children’s production. Unlike previous research which widely reported delay in bilingual development of constructions at the syntax–pragmatics interface, we observed both delay and acceleration in bilingual children’s development of RD. The acceleration effects may be related to the structural overlap between English and Cantonese, which facilitates bootstrapping in bilingual development of RD.

Our study in general supports Hulk and Müller’s (2000) hypothesis for cross-linguistic influence in bilingual first language acquisition, from the perspective of two typologically divergent and genetically unrelated languages. However, the bidirectional influence observed in our study cannot be explained by their formulation of the structural overlap condition. In particular, our study suggests that cross-linguistic influence and its directionality are not independent of language dominance. Cross-linguistic influence cannot be fully accounted for on the basis of one or two factors, but may be attributed to a number of factors and their interaction. The challenge for future research will be to identify mechanisms of cross-linguistic influence in a variety of language combinations and to tease apart the role of the different factors.

Footnotes

Acknowledgements

We have benefited from the comments and suggestions of a number of colleagues, Ziyin Mai, Deng Xiangjun, Zhou Jiangling, Hinny Wong, Jacky Kwan, Szeto Pui Yiu, and the anonymous reviewers, as well as the technical support of Tracy Au and Sophia Yu. Thanks are also due to the participants who attended the talks by Ge, Cheung, Matthews, and Yip (2012, ) at two international conferences which formed the basis of the present article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by funding from the Chinese University of Hong Kong (CUHK) for the CUHK-Peking University-University System of Taiwan Joint Research Centre for Language and Human Complexity; General Research Fund from the Hong Kong Research Grants Council (Project no. 14413514 and 14632016) and a start-up grant to set up the Bilingualism and Language Disorders Laboratory at Shenzhen Research Institute of the Chinese University of Hong Kong.

Notes

References

Bloom

(1973). One word at a time: The use of single-word utterances before syntax. The Hague, The Netherlands: Mouton.

Bloom

(1990). Subjectless sentences in child language. Linguistic Inquiry, 21, 491–504.

Brown

(1973). A first language: The early stages. Cambridge, MA: Harvard University Press.

Cheung

Y. L.

(1997). A study of right dislocation in Cantonese (Unpublished MPhil thesis). Chinese University of Hong Kong, Hong Kong.

Cheung

Y. L.

(2009). Dislocation focus construction in Chinese. Journal of East Asian Linguistics, 18, 197–232.

Cheung

Y. L.

(2015). Bi-clausal sluicing approach to dislocation copying in Cantonese. International Journal of Chinese Linguistics, 2, 227–272.

Cruttenden

(1978). Assimilation in child language and elsewhere. Journal of Child Language, 5, 373–378.

De Cat

. (2007). French dislocation: Interpretation, syntax, acquisition. Oxford, UK: Oxford University Press.

Demetras

(1989). Working parents’ conversational responses to their two-year-old sons (Unpublished doctoral dissertation). University of Arizona

10.

Devlin

Folli

Henry

Sevdali

(2015). Clitic right dislocation in English: Cross-linguistic influence in multilingual acquisition. Lingua, 161, 101–124.

11.

Döpke

(1998). Competing language structures: The acquisition of verb placement by bilingual German–English children. Journal of Child Language, 25, 555–584.

12.

Cheung

Matthews

Yip

(2012, May 15–16). Syntactic transfer of right dislocation in Cantonese-English bilingual children. Paper presented at the International Conference on Bilingualism and Comparative Linguistics, Chinese University of Hong Kong, Hong Kong.

13.

Cheung

Matthews

Yip

(2013, June 10–13). Cross-linguistic influence in Cantonese-English bilingual children: The case of English right-dislocation. Paper presented at the 9th International Symposium of Bilingualism, Nanyang Technological University, Singapore.

14.

C. C.

(2010). Crosslinguistic influence in two directions: The acquisition of dative constructions in Cantonese-English bilingual children. International Journal of Bilingualism, 14, 87–103.

15.

Hulk

A. C. J.

Müller

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

16.

Kupisch

(2005). Acceleration in bilingual first language acquisition. In Gaerts

Jacobs

(Eds.), Romance languages and linguistic theory 2003: Selected papers from ‘Going Romance’ 2003 (pp. 143–160). Amsterdam, The Netherlands: John Benjamins.

17.

Lambrecht

(2001). Dislocation. In Haspelmath

König

Oesterreicher

Raible

(Eds.), Language typology and language universals: An international handbook (Vol. 2, pp. 1050–1078). Berlin, Germany: Walter de Gruyter.

18.

Lee

T. H. T.

Wong

C. H.

Leung

Man

Cheung

Szeto

Wong

C. S. P.

(1996). The development of grammatical competence in Cantonese-speaking children (Report of RGC earmarked grant 1991–94).

19.

Lieven

Salomo

Tomasello

(2009). Two-year-old children’s production of multiword utterances: A usage-based analysis. Cognitive Linguistics, 20, 481–508.

20.

Luke

K. K.

(2004). Yanshenju de kua yuyan duibi [A cross-linguistic study of incremental sentences]. Yuyan jiaoxue yu yanjiu [Language Teaching and Linguistic Studies], 6, 1–9.

21.

MacWhinney

(2000). The CHILDES Project: Tools for analyzing talk (3rd ed.). Mahwah, NJ: Lawrence Erlbaum.

22.

Matthews

Yip

(2011). Cantonese: A comprehensive grammar. London, England: Routledge.

23.

Müller

Hulk

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

24.

Notley

van der Linden

Hulk

(2007). Cross-linguistic influence in bilingual children: The case of dislocation. In Baauw

Drijkoningen

Pinto

(Eds.), Romance languages and linguistic theory 2005: Selected papers from ‘Going Romance’, Utrecht, 8–10 December 2005 (pp. 229–259). Amsterdam, The Netherlands: John Benjamins.

25.

Patuto

Repetto

Müller

(2011). Delay and acceleration in bilingual first language acquisition. The Development of Grammar: Language Acquisition and Diachronic Change (In honour of Jürgen M. Meisel), 11, 231.

26.

Serratrice

Sorace

Paoli

(2004). Crosslinguistic influence at the syntax–pragmatics interface: Subjects and objects in English–Italian bilingual and monolingual acquisition. Bilingualism: Language and cognition, 7, 183–205.

27.

Snyder

Stromswold

(1997). The structure and acquisition of English dative constructions. Linguistic Inquiry, 28, 281–317.

28.

Sorace

Filiaci

(2006). Anaphora resolution in near-native speakers of Italian. Second Language Research, 22, 339–368.

29.

Sorace

Serratrice

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

30.

Van der Linden

Sleeman

. (2007). Clitic dislocation: Evidence for a low topic position. In Los

van Koppen

(Eds.), Linguistics in the Netherlands 2007 (pp. 173–187). Amsterdam, The Netherlands: John Benjamins.

31.

Wang

Lillo-Martin

Best

C. T.

Levitt

(1992). Null subject versus null object: Some evidence from the acquisition of Chinese and English. Language Acquisition, 2, 221–254.

32.

Yip

Matthews

(2000). Syntactic transfer in a Cantonese-English bilingual child. Bilingualism: Language and Cognition, 3, 193–208.

33.

Yip

Matthews

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