Crosslinguistic InfluenCrosslinguistic influence in two directions The acquisition of dative constructionsce in Two Directions the Acquisition of Dative Constructions

http://ijb.sagepub.com/International Journal of Bilingualism

http://ijb.sagepub.com/content/14/1/87The online version of this article can be found at:

DOI: 10.1177/1367006909356646

2010 14: 87International Journal of BilingualismChloe Chenjie Gu

English bilingual children−−in Cantonese Crosslinguistic influence in two directions: The acquisition of dative constructions

Published by:

http://www.sagepublications.com

can be found at:International Journal of BilingualismAdditional services and information for

http://ijb.sagepub.com/cgi/alertsEmail Alerts:

http://ijb.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://ijb.sagepub.com/content/14/1/87.refs.htmlCitations:

What is This?

- Mar 9, 2010Version of Record >>

at UNIVERSITY OF PERADENIYA on March 19, 2014ijb.sagepub.comDownloaded from at UNIVERSITY OF PERADENIYA on March 19, 2014ijb.sagepub.comDownloaded from

http://ijb.sagepub.com/


http://ijb.sagepub.com/content/14/1/87

http://ijb.sagepub.com/content/14/1/87



http://ijb.sagepub.com/cgi/alerts

http://ijb.sagepub.com/cgi/alerts

http://ijb.sagepub.com/subscriptions

http://ijb.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://www.sagepub.com/journalsPermissions.nav

http://ijb.sagepub.com/content/14/1/87.refs.html

http://ijb.sagepub.com/content/14/1/87.refs.html

http://ijb.sagepub.com/content/14/1/87.full.pdf

http://ijb.sagepub.com/content/14/1/87.full.pdf

http://online.sagepub.com/site/sphelp/vorhelp.xhtml

http://online.sagepub.com/site/sphelp/vorhelp.xhtml





87Gu: Crosslinguistic influence in two directions‘International Journal of Bilingualism’ • Volume 14 • Number 1 • 2010, 87–103|

Crosslinguistic influence in two directions: The acquisition of dative constructions in Cantonese–English bilingual children

Chloe Chenjie GuUniversity of Massachusetts at Amherst, USA

AbstractPrevious studies suggest that language dominance and input ambiguity are the two major determinants of crosslinguistic influence in bilingual language acquisition. This article reports a case of bi-directional crosslinguistic transfer in the acquisition of dative constructions by Cantonese–English bilingual children. Longitudinal data of five bilingual children reveal qualitative and quantitative differences between bilingual and monolingual children in the development of English prepositional datives and Cantonese inverted double object datives. Individual differences among the five bilingual children largely correspond to their language dominance patterns, and input ambiguity also helps to explain some transfer effects. It is found that crosslinguistic influence is most likely to take place at vulnerable domains in language acquisition, and a great deal of the interaction between the two languages is observed in such domains. The findings suggest that crosslinguistic influence is pervasive in both directions of bilingual acquisition.

IntroductionMany cases of crosslinguistic influence in the bilingual first language acquisition literature are accounted for by both language dominance and input ambiguity factors (e.g. Paradis & Genesee, 1996; Müller & Hulk, 2001). According to Paradis and Genesee (1996), there are three major instantiations of crosslinguistic influence: acceleration, delay and transfer. Acceleration/delay refers to precocious/postponed development in one language, while transfer is defined as ‘the incorporation of a grammatical property into one language from the other’ (Paradis & Genesee, 1996: 3).

Key words

Cantonese–English bilingual children

dative constructions

transfer

The International Journal of BilingualismCopyright © 2010 the Author/s 2010, ISSN; Vol 14 (1): 87–103; ID no 356646;DOI; 10.1177/1367006909356646 http://Ijb.sagepub.com

Address for correspondence Chloe Chenjie Gu, Department of Linguistics, University of Massachusetts, South College, Amherst, MA 01003, USA. [email: [email protected]]

Acknowledgments

The author would like to thank Yang Gu, Richard Larson, Thomas Lee, Yafei Li, Gladys Tang, Joshua Viau, Virginia Yip, audiences at BUCLD31 and ISB6, and the IBJ reviewers and editors for helpful comments and discussion. Responsibility for errors is entirely mine. This research is supported by a postgraduate studentship from the Chinese University of Hong Kong.

at UNIVERSITY OF PERADENIYA on March 19, 2014ijb.sagepub.comDownloaded from



88 INTERNATIONAL JOURNAL OF BILINGUALISM 14 (1)

The International Journal of Bilingualism

Transfer reflects the interdependence of two grammars in bilingual first language acquisition, as one of the two languages serves as a guide for the other to build certain syntactic properties on it. Two major proposals have been suggested to explain why one of the languages becomes the ‘model’ language in the development of a particular structure. The first proposal holds that language dominance patterns greatly influence the direction of crosslinguistic influence, as the ‘model’ language is very often the dominant language. Certain structures in the weak language may take longer to acquire, or reflect (target-deviant) syntactic properties that resemble the dominant language. However, there is also some evidence that suggests that certain types of crosslinguistic influence do not take place from the dominant language to the weak language (Müller & Hulk, 2001). This type of crosslinguistic influence is taken care of by the second proposal, which argues that one language becomes the ‘model’ language because it contains little input ambiguity with regard to a particular construction, whereas more ambiguity is found in the other language (Müller, 1998; Müller & Hulk, 2001). The two proposals make different predictions when the language that contains more ambiguous input is the dominant language.

In addition to the directionality issue, more details need to be sorted out about the domains in which crosslinguistic influence is likely to take place. Müller and Hulk (2001) argue that crosslinguistic influence occurs at the exact syntactic domains in which monolingual children also have trouble with—though to a less extent. Such domains can be referred to as vulnerable domains in language acquisition, and Müller (2003) defines them as domains thus: ‘children develop particular grammatical phenomena much later than others’, adding that they ‘are prone to error in acquisition in the sense that children will deduce systems for these domains which do not correspond to the target system’ (Müller, 2003: vii). Thus, predictions about the locus of crosslinguistic influence heavily depend on identification of vulnerable domains in language acquisition.

Acquisition of dative constructions is of special interest to our study of vulnerable domains and crosslinguistic influence, as this is an area where both monolingual and bilingual children are prone to error or delay. Snyder and Stromswold (1997) report that monolingual English-speaking children start to produce English prepositional datives at a later stage than double object datives. In the case of monolingual Cantonese-speaking children, previous studies show that many such children find a language-specific inverted double object dative construction difficult to acquire (Chan, 2003; Yip and Matthews, 2007). This article looks at the acquisition of dative constructions in bilingual children, with emphasis on vulnerable domains and crosslinguistic influence. In particular, we examine the role language dominance and input ambiguity play regarding the direction and patterns of crosslinguistic influence. More details on the properties of English and Cantonese dative constructions are given in the next section.

Dative constructions in English and Cantonese

In English (and a number of other languages), dative verbs have two argument realization options, namely the double object dative (DOD) (1a), and the prepositional dative (PD) (1b). In prepositional datives, there is a non-benefactive vs. benefactive distinction, realized by different choices of to (1b) and for (2b).




89Gu: Crosslinguistic influence in two directions


(1) a. John gave Mary a book. (double object dative)

b. John gave a book to Mary. (prepositional to-dative)

(2) a. John baked Mary a cake. (double object dative)

b. John baked a cake for Mary. (prepositional for-dative)

Cantonese, like English, has double object datives and prepositional-like serial verb dative constructions (in which a grammaticalized dative marker bei2—originally meaning ‘give’—introduces the goal/benefactive argument as the English prepositions do). Cantonese also has an inverted double object dative (IDOD) which places the theme and the goal in the reverse order of an English double object dative. These three types of Cantonese dative constructions can be categorized as follows:

(3) a. DOD: Verb-Goal-Theme

b. SVD: Verb-Theme-Dative Marker (bei2)-Goal

c. IDOD: Verb-Theme-Goal(adapted from Tang, 1998)

Unlike English, most Cantonese dative verbs only appear in one type of dative construction.1 Tang (1998) sub-categorized these verbs into three groups: (a) teach verbs, (b) send, fry, pluck verbs and (c) give verbs. The first two classes consist of many dative verbs, but the give class only has one verb bei2 (‘give’). Teach verbs appear in the double object dative (4), send, fry, pluck verbs appear in the serial verb dative (5), and bei2 (‘give’) appears in the inverted double object dative (6).

(4) Ngo5 gaau3 keoi5 Gwong2dung1waa2 (DOD)

我教佢廣東話

I teach 3sg Cantonese

‘I teach him/her Cantonese.’ (Tang, 1998)

(5) Siu2ming4 gei3 zo2 jat1 fung1 seon3 bei2 ngo5 (SVD)

小明寄咗一封信俾我

Siuming send perf one cl letter give I

‘Siuming sent a letter to me.’ (Tang, 1998)

(6) Ngo5 bei2 zo2 jat1 zi1 bat1 keoi5 (IDOD)

我俾咗一支筆佢

I give perf one cl pen 3sg

‘I gave a pen to him/her.’ (Tang, 1998)

1 In English, a group of verbs that have Latinate origins only appear in the prepositional dative construction, but in general, non-Latinate dative verbs are compatible with both constructions.






It is suggested that the Cantonese inverted double object dative is derived from the underlying serial verb form [bei2-Theme-bei2-Goal] (Tang, 1998; Yip and Matthews, 2007). The second bei2 is deleted due to haplology effect, which penalizes repetition of the same phonological element:

(7) SVC: bei2-Theme-bei2-Goal

IDOD: bei2-Theme-Ø-Goal (adapted from Tang, 1998)

Similarities and differences between English and Cantonese dative constructions are summarized in Table 1. Both languages contain double object dative and prepositional/serial verb dative constructions, but Cantonese also has a special inverted double object dative which puts the two objects in the reverse order of double object datives. There is no dative alternation in Cantonese, as each construction is subcategorized for a specific class of verbs.

Table 1English and Cantonese dative constructions

English Cantonese

Double object datives Yes Yes (limited to ‘teach’ class verbs)Prepositional datives Yes Yes (limited to ‘send’, ‘fry’ and ‘pluck’ class verbs)Inverted double object datives No Yes (limited to ‘give’ class verbs)

Prone-to-error dative constructions

Snyder and Stromswold (1997), using age of first use as a measure, analyze the acquisition order of double object datives and to-datives of 12 English monolingual children. They find that the acquisition of the two constructions is strongly correlated with each other, but children acquire double object datives significantly earlier than to-datives. The mean age of acquisition of double object datives is 2;2.5 (years;months), while the mean age of acquisition of to-datives is 2;6.9 (years;months). There is an average gap and 4.4 months between the acquisition of double object datives and to-datives. Another study of 22 monolingual English-speaking children from the CHILDES corpus by Viau (2006) also shows a similar developmental gap between double object datives and prepositional datives.

Chan (2003) investigates the longitudinal development of Cantonese dative constructions in 8 Cantonese monolingual children and 3 Cantonese–English bilingual children. One of her major findings is that both monolingual and bilingual children have difficulty in acquiring inverted double object datives with the verb bei2 (‘give’). She reports that children’s early utterances are generally in the non-target double object order ([bei2-Goal-Theme]), and some contain a topicalized theme ([Theme-bei2-Goal]) or are in the serial verb order ([bei2-Theme-bei2-Goal]). The error rate of inverted double object constructions for monolingual children is around 64 per cent, while the rate for Cantonese–English bilingual children is as high as 88 per cent (Yip & Matthews, 2007).






However, there are still several unanswered questions concerning the development of dative constructions in Cantonese–English bilingual children. This article aims to illustrate the developmental path bilingual children follow and the interaction of two grammars in the acquisition of dative constructions. In particular, it is shown that each language—dominant or non-dominant—influences the other on certain aspects of grammar, and a great deal of interaction is found in the vulnerable domain of bilingual language acquisition.

Methodology

This study investigates longitudinal data of five Cantonese–English bilingual children from the Hong Kong Bilingual Child Language Corpus (Yip & Matthews, 2007). All the five children are raised in one-parent-one-language families and are exposed to Cantonese and English naturally from birth. Parents of these children are native speakers of English or Cantonese, and they use their native language to communicate with their children. The children are recorded weekly or biweekly in an age period between 1;03 (years;months) and 4;06.2

Among the 5 bilingual children, 4 are dominant in Cantonese, and 1 is dominant in English (Yip & Matthews, 2006, pers. comm.).3 This combination of different language dominance patterns allows us to investigate the role of language dominance in crosslinguistic influence. Also, note that 3 of the 4 Cantonese-dominant children (Timmy, Sophie and Alicia) are siblings in the same family. Information on age range, gender, dominant language and numbers of sessions in each language is summarized in Table 2. When comparison with monolingual children is needed, we use data from 9 out of the 12 monolingual English-speaking children Snyder and Stromswold (1997) have studied.4 For Cantonese, 8 monolingual children from the Hong Kong Cantonese Child Language Corpus are selected. All these children were born to Cantonese-speaking parents and speak Cantonese as the first and only language. Each child is recorded on a biweekly basis. Every corpus used in this study is available at CHLDES (MacWhinney, 2000).

Following Snyder and Stromswold (1997) and Viau (2006), this study chooses age of first non-imitative use as the measure of emergence.5 If children produce no non-target structures or only make a few performance errors after their first such constructions, the age of emer-gence is also considered as the age of acquisition of such constructions.6 We separate the notion of emergence from acquisition, as by acquisition we expect children to have very low

2 Previous work from Snyder and Stromswold (1997) and Viau (2006) shows that the age range 1;06–3;06 repre-sents the best time for studies on early double object and prepositional datives. Recordings of the five bilingual children fit exactly into this age period.

3 Following Yip and Matthews (2000), language dominance of the five bilingual children is determined by comparing children’s Mean Length of Utterance (MLU) in each language. Though it is still unsettled whether MLU values of different languages are fully comparable, Yip and Matthews (2006) 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, and children’s MLU differentials match patterns of transfer effects and language preference.

4 These children include: Adam, Eve, Mark, Naomi, Nathaniel, Nina, Peter, Sarah and Shem. The other 3 children (Allison, April and Ross) are excluded from our study because of infrequent recordings during an age period when first double object datives are expected to emerge.

5 To qualify as a non-imitative utterance, children must produce an utterance that is not identical to the preceding adult utterance (i.e. children’s utterance must contain at least one different element – which indicates that they have processed the preceding adult utterance and produced their own version afterwards). However, when children produce a very similar sentence after several such adult utterances, it would be regarded as imitative.

6 For an utterance to be counted as the first use of a construction, it must be non-imitative and clearly uttered.






error rates in one construction for a continuous period, while by emergence we only expect children to have some basic syntactic structures of that construction, with or without making a decision between target and non-target structures. We think that compared with order of acquisition, order of emergence is a better indicator of children’s developmental patterns in dative constructions, as at the time of first non-imitative use, children already form their initial grammar of that construction.

As previous studies report that children acquiring Cantonese dative constructions are prone to error in producing the inverted double object datives, a Cantonese dative construction will not be considered as being fully acquired until its accuracy rate reaches and remains more than 90 per cent after a certain age.7

The procedures to identify English dative constructions are similar to what Snyder and Stromswold (1997) use in their study. Snyder and Stromswold (1997) make a list of potential dativizable verbs to identify children’s double object datives.8 The CLAN program is used to extract all children’s utterances that contain any of the potential dativizable verbs, and the results are checked by hand to make sure they are not imitations of a preceding adult utter-ance. For prepositional datives, the CLAN program is used to extract all children’s utterances containing to and for, and these sentences are hand checked for prepositional datives, including ungrammatical ones. For an utterance to be counted as a prepositional dative, it must have two objects and an overt preposition to or for. The sequence of the two objects can be non-target-like, but if there is no overt preposition, it will be counted as a double object dative.

The procedures to identify Cantonese dative constructions are similar to the procedures for English ones. For inverted double object datives, as bei2 (‘give’) is the only ‘give’ class

7 It is arbitrary to set the accuracy rate at 90 per cent, but in this study, when children produce non-target structures alongside target structures, their accuracy rate is around 50 per cent. However, later most children suddenly stop producing non-target structures and their accuracy rates rise to nearly 100 per cent. Therefore, whether we take 90 per cent rather than other rates to be the indicator of acquisition does not make a significant difference to our argument.

8 These verbs included: address, admit, afford, allocate, allocating, allow, ask, assign, bake, baking, bought, bring, broadcast, brought, build, building, built, buy, buying, commend, communicate, communicating, concede, conceding, convey, demonstrate, demonstrating, denied, denies, deny, describe, describing, devote, devoting, dictate, dictating, did, dig, do, does, doing, done, dug, explain, gave, get, give, giving, got, gotten, grant, guar-antee, impart, lend, lent, made, make, making, mention, order, ordering, orders, preach, prescribe, prescribing, promise, promising, radio, read, refer, refuse, refusing, relate, relating, relay, reserve, reserving, restore, restoring, reveal, sell, send, sent, serve, serving, show, sold, submit, take, taking, taught, teach, telegraph, tell, told, took, transmit, unveil, volunteer, whisper, wire, wiring, write, writing and wrote.

Table 2Background information of children in the Hong Kong Bilingual Child Language Corpus (Yip and Matthews, 2007)

Dominant Cantonese English Child Gender language Age range sessions sessions

Alicia F Cantonese 1;03;10–3;00;24 40 40Charlotte F English 1;08;28–3;00;03 19 19Llywelyn M Cantonese 2;00;12–3;04;17 17 17Sophie F Cantonese 1;06;00–3;00;09 40 40Timmy M Cantonese 1;05;20–3;06;25 34 38






verb that is frequently used in modern Cantonese, the CLAN program is used to extract all children’s utterances containing bei2, and the results are checked by hand to select possible non-target double object datives ([bei2-Goal-Theme]) or target inverted double object datives ([bei2-Theme-Goal]). For serial verb constructions and double object datives, the CLAN program is used to extract all children’s utterances that contain potential dativizable verbs identified by Tang (1998).9 The results are hand checked to make sure they are full dative constructions.

Developmental lag of English prepositional dativesIt is found that bilingual children, like their monolingual counterparts, produce English double object datives before prepositional datives, but there is a wider gap between bilingual children’s first double object datives and first prepositional datives.

English double object datives emerge before prepositional datives in all 5 bilingual children. The age of emergence of double object datives ranges from 1;5.2 (years;months) to 2;2.9 (years; months), while the age of emergence of prepositional datives ranges from 2;4.9 to >3;4.6.10 The mean age of bilingual children’s acquisition of double object datives is 2;0.1, and their mean age of acquisition of prepositional datives is >2;9.8.11 No target to-datives are attested in the transcripts of two children, Alicia and Llywelyn, and in the transcripts of Llywelyn and Sophie, no target for-datives are attested.

We also examine children’s age of emergence of directional to as Viau (2006) reports that monolingual children produce directional to before prepositional datives, and the age of emergence of directional to can help us to figure out whether prepositional datives in bilingual children are acquired at a relatively later time. The age of emergence of directional to in bilingual children ranges from 2;1.3 to 2;5.5, and the mean age of emergence is 2;3.1. Except for Timmy, all the children produced directional to before prepositional datives and after double object datives. Timmy produces directional to first, then double object datives, and prepositional datives come last. Each child’s age of emergence of double object dative, prepositional dative, and directional to is provided in Table 3.

There is an average temporal gap of >9.7 months between the emergence of double object datives and prepositional datives and a gap of >10.4 months between the emergence of double object datives and to-datives in bilingual children. Compared with the 3.3 months gap between double object datives and prepositional datives (Viau, 2006) and the 4.4 months gap between double object datives and to-datives (Snyder and Stromswold, 1997)

9 These verbs include: sung3 ‘to give (a present)’, zoeng2 ‘to award’, bun1 ‘to move’, daai3 ‘to bring’, dai6 ‘to hand to’, deng3 ‘to pelt’, gaau1 ‘to deliver’, gaap3 ‘to lift food with chopsticks’, gei3 ‘to send’, lau4 ‘to reserve’, ling1 ‘to carry with hand/to take’, lo2 ‘to bring’, maai6 ‘to sell’, paai3 ‘to deliver’, tek3 ‘to kick’, wui6 ‘to remit’, caau2 ‘to fry’, jing2 ‘to photocopy’, pai1 ‘to cut’, sai2 ‘to wash’, tong3 ‘to iron’, waak6 ‘to draw’, zam1 ‘to pour’, zik1 ‘to knit’, zing2 ‘to make’, zak3 ‘to compose’, zyu2 ‘to cook’, coeng2 ‘to snatch’, gaan2 ‘to choose’, maai5 ‘to buy’, lo2 ‘to get’, tau1 ‘to steal’, zaak6 ‘to pluck’, ceng2gaau3 ‘to inquire’, gaau3 ‘to teach’, haau2 ‘to test’, kau4 ‘to request’ and man6 ‘to ask’.

10 ‘>3;4.6’ means greater or later than 3;4.6. The exact age of acquisition is not known because the child never produces any target construction before the recording ends.

11 Note that the mean age for prepositional datives is calculated from the earlier age of emergence of to-datives and for-datives.






in monolingual English-speaking children, Cantonese–English bilingual children exhibit a much wider temporal gap between double object datives and prepositional datives.

There is an average temporal gap of >6.7 months between the emergence of directional to and prepositional datives in bilingual children while the gap in monolingual English-speaking children is 0.9 months (Viau, 2006). In contrast, the average gap between the emergence of double object datives and directional to in bilingual children is 3.0 months, which is very similar to the 2.4 months gap in monolingual children (Viau, 2006). Table 4 summarizes the average temporal gaps between double object datives and prepositional datives, directional to and prepositional datives and double object datives and directional to in bilingual and monolingual children.

Information on the mean age of emergence of each construction in bilingual and monolin-gual children is summarized in Table 5. The mean ages of emergence of double object datives and directional to in bilingual children and monolingual children are similar while the mean age of emergence of prepositional datives in bilingual children is much greater than that of monolingual children. This finding, combined with the fact that bilingual children have wider

Table 3Emergence of English DOD, PD and directional to in bilingual children

Child DOD PD Directional to

Timmy 2;2.9 2;4.9 2;1.3 Sophie 2;1.7 2;9.8 2;5.5 Alicia 1;5.2 2;11.2 2;1.5 Llywelyn 2;2.1 >3;4.6 2;5.3 Charlotte 2;0.8 2;6.5 2;1.7 Mean 2;0.1 >2;9.8 2;3.1

Table 4Comparison of average temporal gaps of English DOD/PD/directional to between bilingual and monolingual children

Directional to–PD DOD–directional to DOD–PD (months) (months) (months)

Bilingual >9.7 >6.7 3.0Monolingual 3.3 0.9 2.4

gaps between the emergence of double object datives and prepositional datives, indicates that bilingual children are further delayed in the acquisition of prepositional datives.

Among the five bilingual children, Timmy has the smallest gap (2.0 months) between the acquisition of double object datives and the acquisition of to-datives.12 The gap in the

12 This gap is even smaller than the mean gap of 4.4 months in monolingual children (Snyder and Stromswold, 1997).






other bilingual children ranges from 7.4 months to >19.6 months. Note that Charlotte, an English-dominant child, does not progress as fast as Timmy (a Cantonese-dominant child) in the development of to-datives, and the pace of development of the four Cantonese-dominant children varies greatly. Individual bilingual children’s developmental gaps between double object datives and to-datives are summarized in Table 6.

Language dominance alone cannot answer the question why Timmy progresses exception-ally fast in his development of to-datives. Examination of Timmy’s production of to shows that he produces triadic directional to – see example (8) – at the remarkably early age of 2;04;14, (years;months;days) even before his first to-datives are produced at 2;04;28. In contrast, among the other bilingual children, only Charlotte produces such triadic directional to – example (9) – at the age of 3;03;03, which is 5.3 months after her first to-datives. No other bilingual children produce similar instances of triadic directional to in their recordings.13

(8) I take him to the hospital. (Timmy 2;04;14)

(9) Daddy can’t do take her to the park. (Charlotte 3;03;03)

Among the five bilingual children, those who exhibit a larger gap between double object datives and to-datives (Sophie, Alicia, Llywelyn and Charlotte) also fail to produce both triadic dative to and triadic directional to in a short period of time. This suggests that the ability to separate triadic dative to from triadic directional to is related to the ability to make a fast progress from double object datives to to-datives.

13 Children’s utterances with the verb ‘take’ are also examined, and it is found that all five bilingual children are able to use take in Verb-Particle constructions ([V-NP-Particle] / [V-Particle-NP]) before their recordings end. So it is not difficulty in learning the individual verbs that delays production of triadic directional to in these children.

Table 5Comparison of mean age of emergence of English DOD/PD/directional to between bilingual and monolingual children

DOD PD Directional to

Bilingual 2;0.1 >2;9.8 2;3.1Monolingual 2;2.5 2;4.9 2;4.0

Table 6Bilingual children’s developmental gap between DOD and to-datives

Child Dominant language Gap between DOD and to-datives (months)

Timmy Cantonese 2.0 Sophie Cantonese 8.1 Alicia Cantonese >19.6Llywelyn Cantonese >14.5Charlotte English 7.4






Transfer of a non-target prepositional dative from Cantonese

Two bilingual children produce English for-datives in non-target [V-PPGOAl-DPTHEME] order, and no target for-datives are found in these children’s transcripts. Llywelyn produces one such for-dative in his last recording at age 3;04;17.14 Sophie produces four such for-datives in her corpus, and all these for-datives are found in the same recording at age 2;05;30. Some of the non-target for-datives are illustrated in examples (10)–(12).

(10) * will buy for Kenny that. (Llywelyn 3;04;17)

(11) * I buy for you the bear okay? (Sophie 2;05;30)

(12) * buy for me this one. (Sophie 2;05;30)

The same behavior is not found in monolingual English-speaking children. Only one child (Peter) produces one prepositional dative in the non-target [V-PP-DP] order, and this utterance is likely to be a performance error, as Peter altogether produces 17 for-datives in nine months, and except one mistake, the other 16 are all target constructions. The contrast between bilingual and monolingual children is striking: two out of five bilingual children use the non-target order for all the for-datives they have produced in their recordings, while only one out of 12 monolingual children occasionally produces one such for-dative. The use of non-target [V-PP-DP] order reveals a qualitative difference between the two groups.

An analysis of adults’ utterances containing for-datives shows that the non-target [V-PPGOAL-DPTHEME] order is never used in the input. This order is not grammatical in Cantonese, either, as Cantonese serial verb dative constructions always follow the [V-Theme-bei2-Goal] sequence. However, a search of children’s Cantonese serial verb dative constructions shows that Sophie and one monolingual Cantonese child (HHC) both produce several serial verb dative constructions with non-target [V-bei2-Goal-Theme] order. One of these non-target serial verb datives is illustrated in (13).

(13) CHI: * ngo5 dou2 bei2 lei5 seoi2 aa3.

我倒俾你水啊

I pour give you water sfp

‘I pour for you water.’ 15 (Sophie: 2;08;28)

Sophie produces three non-target [V-bei2-Goal-Theme] constructions from 2;05;02 to 2;08;22, and seven target serial verb dative constructions at the same period. Her accuracy rate of serial

14 As the construction is found in Llywelyn’s last recording, we do not know if he continues to produce the same construction or not. Also, there is a gap of 3.3 months between Llywelyn’s last two transcripts, and some early non-target for-datives may not be captured in the recording.

15 The target sentence should be: (i) ngo5 dou2 seoi2 bei2 lei5 aa3.

我倒水俾你啊

I pour water give you sfp

‘I will pour some water for you.’






verb dative constructions remains at 100 per cent after 2;9. Notice that the first occurrence of Sophie’s non-target Cantonese serial verb dative construction (2;05;02) is not far from her first use of non-target English for-datives (2;05;30). However, Sophie only produces non-target English for-datives in one recording, but the corresponding non-target construction persists longer in her Cantonese grammar.

The data suggests that bilingual children’s non-target English [V-PPGOAL-DPTHEME] construction is a transfer effect from Cantonese.16 This case shows that bilingual children’s non-target grammar in one language is able to influence parallel structures in the other language – though such non-target structures may not be found in the utterances of monolingual children (of the other language). It indicates that the two language systems in the bilingual mind are highly interactive, sharing structural properties at a time when grammars of each language are still not fully target-like.

Crosslinguistic influence in the development of English dative constructions

Crosslinguistic influence is manifested as delay in the late emergence of prepositional datives in bilingual children. Transfer effects are also identified in bilingual children’s non-target [V-PP-DP] structures, which come from their Cantonese grammar that adopts a similar non-target [V-bei2-Goal-Theme] structure.

Language dominance can account for some (but not all) differences we observe between bilingual and monolingual children. Compared with most Cantonese-dominant children, Charlotte, an English-dominant child, has smaller gaps between her first double object dative and prepositional dative constructions, but she does not progress as fast as Timmy, who is a Cantonese-dominant child. Charlotte exhibits a gap of 5.7 months between her first double object dative and first prepositional dative, while Timmy only takes 2.0 months to produce his first prepositional dative after his first double object dative.

However, according to Yip and Matthews (2006), Timmy is relatively less dominant in Cantonese than other Cantonese-dominant children. The fact that Charlotte performs better than the Cantonese-dominant children shows that her knowledge of English grammar is still more advanced than most Cantonese-dominant bilingual children. Charlotte’s dominance in English seems to help her acquire prepositional datives in a faster fashion. Timmy’s faster progress of prepositional datives, on the other hand, may be related to his early mastery of different types of dative to and directional to.

The non-target English [V-PPGOAL-DPTHEME] utterances in bilingual children exhibit transfer effects from Cantonese. Language dominance patterns mainly influence the direction of transfer in this construction, as only Cantonese-dominant children produce this non-target structure. The English-dominant child, Charlotte, does not produce any Cantonese dative constructions throughout her recordings, and her grammar of English prepositional datives is relatively unaffected.

16 Llywelyn does not produce any Cantonese serial verb dative construction in his corpus, and it is difficult to conclude whether his non-target English prepositional dative is influenced by his Cantonese grammar because of the 3.3-month gap between his last two transcripts. It is likely that Llywelyn’s early Cantonese serial dative constructions are not captured in the recordings.






With respect to input ambiguity, it is uncertain whether input properties of Cantonese influence children’s development in English prepositional datives. Cantonese, unlike English, adopts a different lexicalization pattern and contrasts dative to with directional to by choices of two different lexical items bei2 (‘give’ 俾) and heoi3 (‘go’ 去), and dative for with other types of for by bei2 (‘give’ 俾) and bong1 (‘help’ 幫). However, in Cantonese, no lexical distinction is made between dative to and dative for (both represented by the same dative marker bei2).17 The differences in the lexicalization patterns make it difficult to predict the direction of transfer.

Development of Cantonese dative constructionsA significant contrast is found between monolingual and bilingual children in their develop-mental patterns of Cantonese dative constructions. Monolingual children, in general, acquire serial verb dative constructions before producing other types of dative constructions, while bilingual children do not show a consistent developmental pattern in their use of dative constructions. The different developmental path is argued to play an important role in influ-encing children’s acquisition of the inverted double object datives.

Bilingual children’s order of emergence of Cantonese dative constructions is summa-rized in Table 7. It is found that non-target English double-object-like [bei2-Goal-Theme] constructions appear in several bilingual children before their first target [bei2-Theme-Goal] constructions. Most bilingual children do not produce serial verb dative constructions before their first target [bei2-Theme-Goal] and non-target [bei2-Goal-Theme] constructions. Few bilingual children produce double object datives (with teach verbs) in their recordings. Charlotte does not produce any dative constructions in her corpus. It is difficult to find a general developmental pattern in bilingual children.

Monolingual children, on the other hand, exhibit a more consistent developmental pattern. They produce target [bei2-Theme-Goal] constructions before or concurrently with non-target [bei2-Goal-Theme] constructions. Most monolingual children produce serial dative construction earlier than other constructions.18 Monolingual children’s order of emergence of Cantonese dative constructions is summarized in Table 8.

By comparing the order of emergence of Cantonese dative constructions in each bilingual and monolingual child, we find that one bilingual child (Timmy) demonstrates a developmental pattern that is more similar to that of most monolingual children, and one monolingual child (HHC) progresses in a manner that resembles most bilingual children.

Timmy differs from other bilingual children in that he produces his first target [bei2-Theme-Goal] construction before non-target [bei2-Goal-Theme] constructions, and uses a total of seven such target constructions in five sessions of his corpus (which is very frequent compared with the one or two target constructions produced by other bilingual children) and he has the highest accuracy rate in [bei2-Theme-Goal] construction in bilingual children.

HHC differs from other monolingual children in that he only produces non-target [bei2-Goal-Theme] constructions in his corpus, and he also produces non-target serial verb dative constructions. Therefore, Timmy and HHC could be considered exceptional cases in

17 Note that bilingual children never produce Cantonese-based [V-Theme-give-Goal] structure in English. This suggests that bilingual children know that the dative marker bei2 is syntactically different from the main verb bei2.

18 Except for one child, HHC, who has problems with word order of serial verb dative constructions.






each group, as their performance does not conform to the general developmental pattern of their own groups.

One major difference we observe between monolingual and bilingual children is the order of emergence between the target [bei2-Theme-Goal] construction and the non-target [bei2-Goal-Theme] construction. The mean age of first non-target [bei2-Goal-Theme] construction in bilingual children is 2;6.0, while the mean age of first target [bei2-Theme-Goal] construction is >2;9.1. If Timmy is excluded, the mean age of first non-target [bei2-Goal-Theme] construction in bilingual children is 2;5.5, while the mean age of first target [bei2-Theme-Goal] construc-tion is >2;10.5.

The mean age of first non-target [bei2-Goal-Theme] construction in monolingual children is >2;7.2, and the mean age of first target [bei2-Theme-Goal] construction is >2;9.1. If we exclude HHC, the mean age of first non-target [bei2-Goal-Theme] construction in monolingual children is >2;6.5, while the mean age of first target [bei2-Theme-Goal] construction is 2;7.2. The mean age of emergence of target [bei2-Theme-Goal] and non-target [bei2-Goal-Theme] Cantonese constructions is summarized in Table 9.

When we compare the gap between [bei2-Goal-Theme] and [bei2-Theme-Goal] construc-tions in monolingual and bilingual children, we find that bilingual children have a wider gap (>5.0 months excluding Timmy) between their first [bei2-Goal-Theme] and first [bei2-Theme-Goal] constructions than monolingual children (<0.7 excluding HHC). This suggests that

Table 7Bilingual children’s order of emergence of serial verb dative constructions, [bei2-G-T] constructions and [bei2-T-G] constructions

Child Order of emergence

Timmy [bei2-T-G] < [bei2-G-T] < serial verb dative Sophie [bei2-G-T] < serial verb dative < [bei2-T-G] Alicia Serial verb dative < [bei2-G-T] no [bei2-T-G]Llywelyn [bei2-G-T] < [bei2-T-G] no serial verb dative

e.g. [bei2-T-G]: bei2 (‘give’)—‘apple’—‘me’ (target order in Cantonese) [bei2-G-T]: bei2 (‘give’)—‘me’—‘apple’ (non-target order in Cantonese)

Table 8Monolingual children’s order of emergence of Cantonese serial verb dative constructions, [bei2-G-T] constructions and [bei2-T-G] constructions

Child Order of emergence

CCC serial verb dative, no [bei2-G-T], no [bei2-T-G]CGK serial verb dative < [bei2-T-G] = [bei2-G-T]CKT serial verb dative < [bei2-T-G] no [bei2-G-T]HHC [bei2-G-T] < serial verb dative no [bei2-T-G]LLY serial verb dative < [bei2-G-T] < [bei2-T-G]LTF serial verb dative = [bei2-G-T] < [bei2-T-G]MHZ serial verb dative no [bei2-G-T], no [bei2-T-G]WBH serial verb dative no [bei2-G-T], no [bei2-T-G]






monolingual children move faster from their first non-target [bei2-Goal-Theme] constructions to their first target [bei2-Theme-Goal] constructions.

Moreover, bilingual children demonstrate a strong preference for the non-target [bei2-Goal-Theme] construction. The average accuracy rate of [bei2-Theme-Goal] constructions in bilingual children is 20 per cent. Among the four children who produce these constructions, Alicia has the lowest accuracy rate (0%), and Timmy has the highest accuracy rate (41.2%). Also, when both target and non-target constructions appear in the same transcript, the non-target constructions always outnumber (or appear as often as) the target ones. Timmy’s accuracy rate of [bei2-Theme-Goal] reaches 100 per cent only in the last recording at 3;06;25, but we still need more information (such as accuracy rates in several continuous recordings after 3;06;25) to determine whether he has fully acquired the target [bei2-Theme-Goal] construction at that age.

Monolingual children show higher accuracy rates and more frequent uses of [bei2-Theme-Goal] constructions throughout their recording sessions. The average accuracy rate is 50 per cent. One child (CKT) solely uses the target [bei2-Theme-Goal] form in the corpus.19 Three children (CGK, LLY and LTF) produce both target [bei2-Theme-Goal] and non-target [bei2-Goal-Theme] constructions, and among them, LTF has the highest accuracy rate (63.6%), while CGK has the lowest accuracy rate (40%).

Except for HHC, every monolingual child demonstrates a higher accuracy rate in [bei2-Theme-Goal] constructions than bilingual children (excluding Timmy). The highest accuracy rate in monolingual children is 100 per cent, while the lowest accuracy rate in monolingual children (CGK: 40%) is close to the rate of Timmy (41.2%), and much higher than rates of Alicia (0%), Llywelyn (20%) and Sophie (10.5%).

In addition to higher accuracy rates, monolingual children also use the target construction more frequently than bilingual children. Among the three monolingual children who produce both target [bei2-Theme-Goal] and non-target [bei2-Goal-Theme] constructions, the number of target constructions generally exceeds (or equals) the number of non-target ones.

Inverted double object datives and serial verb dative constructions

As Cantonese inverted double object datives are fundamentally serial verb dative constructions with the dative marker deleted (see [7]), acquiring the serial verb dative construction may be a necessary step for children to discover the underlying [bei2-Theme-bei2-Goal] structure

19 Note that CKT only produces two target [bei2-Theme-Goal] constructions in one transcript. It is possible that later he will also use the non-target [bei2-Goal-Theme] form.

Table 9Mean age of emergence of target [bei2-Theme-Goal] and non-target [bei2-Goal-Theme] Cantonese constructions

Target [bei2-Theme-Goal] Non-target [bei2-Goal-Theme]

Bilingual 2;9.1 (>2;10.5 excluding Timmy) 2;6.0 (2;5.5 excluding Timmy)Monolingual >2;9.1 (2;7.2 excluding HHC) >2;7.2 (>2;6.5 excluding HHC)






and the bei2-deletion rule, which generates the surface inverted [bei2-Theme-Goal] order. If children do not produce any serial verb dative before their first inverted double object datives, it is questionable whether the underlying structure is in place, and it follows that children may not apply the bei2-deletion analysis to form the inverted bei2-double object dative.

Analysis of the order of emergence of serial verb datives (including bei2-Theme-bei2-Goal constructions) and inverted double object datives shows that it is very unlikely that the bilingual children we study have formed a syntactic connection between serial verb datives and inverted double object datives. When they start to produce the [bei2-Theme-Goal] construction, they have little (or still uncertain) knowledge of the structure of serial verb dative constructions, and even though one child (Alicia) acquired the serial verb dative construction at an earlier age, no evidence shows that she linked the two structures together, as she produced no target inverted double object datives in her corpus.

Monolingual children, on the other hand, acquire serial verb dative constructions before they produce their first [bei2-Theme-Goal] constructions. Four monolingual children have produced both serial verb dative constructions and [bei2-Theme-Goal] constructions in their transcripts, and their first serial verb dative constructions all appear before their first [bei2-Theme-Goal] constructions.

In general, only monolingual children demonstrate some evidence that they have linked the structure of serial verb dative construction to the inverted double object dative in their development of Cantonese dative constructions. Bilingual children, on the other hand, may not have the underlying structure and try to form the surface [bei2-Theme-Goal] order by other means.20

Notice that though Timmy produces target [bei2-Theme-Goal] constructions more frequently than other bilingual children and has an accuracy rate that is close to the average rate of monolingual children, his first serial verb datives are produced after his first [bei2-Theme-Goal] constructions. This developmental pattern indicates that Timmy may not form a connection between the surface [bei2-Theme-Goal] sequence and its underlying structure. Though quantitatively Timmy’s accuracy rates are closer to that of the monolingual children, qualitatively, he is not different from other bilingual children with regard to the availability of the underlying structure.

Crosslinguistic influence in the acquisition of Cantonese dative constructions

The earlier onset of the non-target double-object-dative-like [bei2-Goal-Theme] structure in bilingual children suggests that bilingual children are influenced by their English grammar and start to use non-target [bei2-Goal-Theme] constructions at a relatively earlier developmental stage.

As Charlotte’s dominant language is English, she progresses slower than other Cantonese-dominant children in her development of Cantonese dative constructions. Charlotte does not produce any full Cantonese dative constructions in her corpus. She only uses the verb bei2 in its bare form or in null-object constructions.

20 We are open on how bilingual children form their early [bei2-Theme-Goal] constructions. From their develop-mental pattern, we can only conclude that they do not seem to use the target bei2-deletion analysis when they start to produce their first [bei2-Theme-Goal] constructions.






However, language dominance does not explain why Cantonese-dominant children differ from monolingual Cantonese-speaking children in their developmental patterns. None of the four Cantonese-dominant children show evidence that they have made a connection between target inverted double object datives and serial verb dative constructions.

These qualitative as well as quantitative differences between Cantonese-dominant children and monolingual Cantonese-speaking children can be attributed to input ambiguity. As Cantonese input does not show overt evidence for the derivation of inverted double object datives, and contains many null-object [bei2-Goal] constructions which exhibit a different surface word order from the target [bei2-Theme-Goal] construction, bilingual children seem to be influenced by their English grammar to use the English double-object order, as English input does not contain any ambiguity with this double object dative structure.

The fact that bilingual children start to produce [bei2-Goal-Theme] constructions at a relatively earlier developmental stage than monolingual children also makes it more chal-lenging for them to build the target syntactic connection between the target [bei2-Theme-Goal] structure and the underlying serial verb dative form. As bilingual children adopt a non-target double-object grammar before they construct the link, they might encounter more difficulty analyzing the structure of [bei2-Theme-Goal] constructions, and as a result have lower accuracy rates and longer periods of non-target uses.

As we can see from the different developmental patterns in bilingual and monolingual children in the development of Cantonese dative constructions, it is crucial to form the correct analysis of the target inverted bei2-double object dative at the beginning of acquisition, otherwise the non-target [bei2-Goal-Theme] structures will emerge at a relatively earlier developmental stage, and postpone the acquisition of inverted double object datives. Input ambiguity appears to play an important role in the acquisition of [bei2-Theme-Goal] construc-tions, because this construction, compared with other dative constructions, is more difficult to acquire and more likely to be affected by the English grammar.21

Vulnerable domains and crosslinguistic influence

The delay of English prepositional datives, as well as children’s non-target [V-PP-DP] utter-ances, fit into Müller’s (2003) definition of vulnerable domains. However, there is a major quantitative difference between monolingual and bilingual children in the length of the developmental lag of prepositional datives, as well as a qualitative difference in the use of non-target prepositional datives.

The inverted double object datives form a vulnerable domain in the acquisition of Cantonese dative constructions. Both monolingual and bilingual children have high error

21 The inverted bei2-double object datives are more difficult to acquire than other dative constructions, because no negative evidence is provided to children in the input, and with null-object [bei2-Goal] constructions being frequent in the input (Chan, 2003 reports that 48.46% of adult input containing the verb bei2 is in this form), there are few straightforward mechanisms for children to unlearn the [bei2-Goal-Theme] grammar.

The most useful analysis children can rely on, as we have proposed, is the syntactic connection between [bei2-Theme-Goal] and serial verb dative constructions/double bei2 constructions ([bei2-Theme-bei2-Goal]), and we find that this is the path monolingual children follow in their acquisition of inverted bei2-double object datives.

However, because of the influence from English that allows a double object dative structure for the verb give, bilingual children are not so capable of converging on the correct analysis for the inverted bei2-double object datives.






rates in producing the non-target [bei2-Goal-Theme] rather than the target [bei2-Theme-Goal] construction, and both groups have not fully acquired the target structure when their record-ings end around 3;0–3;6. However, monolingual children demonstrate qualitative differences from bilingual children in their developmental patterns and syntactic knowledge of various Cantonese dative constructions.

The crosslinguistic influences we observe in this study conform to the proposal of Müller and Hulk (2001). English prepositional datives and Cantonese inverted double object datives are two areas where transfer and delay are identified for both monolingual and bilingual children. Crosslinguistic influence takes place exactly at these domains, and quantitative differences between the two groups are observed. This shows that the two language systems in a bilingual child are more likely to interact with each other when certain grammatical domains present challenges to them, as such problems do to monolingual children as well.

We also report some qualitative differences between bilingual and monolingual children in these vulnerable domains. The order between PP and DP in prepositional datives is not prone to error for monolingual English children, but bilingual children still receive influence on this construction from their non-target Cantonese grammar. As the interaction of the two grammars is highly active in vulnerable domains, bilingual children may have a stronger tendency to adopt non-target analysis from the other language and produce structures mono-lingual children do not use.

ReferencesCHAN, W.-S. (2003). The development of bei2 dative constructions in early child Cantonese. Unpublished

MPhil Thesis. The Chinese University of Hong Kong.MACWHINNEY, B. (2000). The CHILDES project: Tools for analyzing talk (third edition). Mahwah,

NJ: Lawrence Erlbaum Associates.MÜLLER, N. (1998). Transfer in bilingual first language acquisition. Bilingualism: Language and

cognition, 1(3), 151–171.MÜLLER, N. (Ed.), (2003). (In)vulnerable domains in multilingualism. Amsterdam: Benjamins.MÜLLER, N. & HULK, A. (2001). Crosslinguistic influence in bilingual language acquisition: Italian

and French as recipient languages. Bilingualism: Language and Cognition, 4(1), 1–21.PARADIS, J. & GENESEE, F. (1996). Syntactic acquisition in bilingual children: Autonomous or

interdependent? Studies in Second Language Acquisition, 18(1), 1–25.SNYDER, W. & STROMSWOLD, K. (1997). The structure and acquisition of English dative construc-

tions. Linguistic Inquiry, 28(2), 281–317.TANG, S.-W. (1998). On the ‘inverted’ double object construction. In Stephen Matthews (Ed.) Studies

in Cantonese linguistics. Hong Kong: Linguistic society of Hong Kong.VIAU, J. (2006). Give = CAUSE + HAVE/GO: Evidence for early semantic decomposition of dative

verbs in English child corpora. In David Bamman, Tatiana Magnitskaia and Colleen Zaller (Eds.), Proceedings of the 30th annual Boston University Conference on Language Development (pp. 665–676). Somerville, MA: Cascadilla Press.

YIP, V. & MATTHEWS, S. (2000). Syntactic transfer in a Cantonese–English bilingual child. Bilingualism: Language and Cognition, 3(3), 193–208.

YIP, V. & MATTHEWS, S. (2006). Assessing language dominance in bilingual acquisition: A case for Mean Length Utterance differentials. Language Assessment Quarterly, 3(2), 97–116.

YIP, V. & MATTHEWS, S. (2007). The bilingual child: Early development and language contact. Cambridge: Cambridge University Press.




Crosslinguistic InfluenCrosslinguistic influence in two directions The acquisition of dative constructionsce in Two Directions the Acquisition of Dative Constructions

Documents