<<

Applied Psycholinguistics 22 (2001), 253–267 Printed in the United States of America The acquisition of English dative constructions

AIMEE L. CAMPBELL Emory University

MICHAEL TOMASELLO Max Planck Institute for Evolutionary Anthropology

ADDRESS FOR CORRESPONDENCE Michael Tomasello, Max Planck Institute for Evolutionary Anthropology, Inselstrasse 22, D-04103 Leipzig, Germany. E-mail: [email protected]

ABSTRACT We analyzed the three main types of English dative constructions – the double- dative, the to dative, and the for dative – in the spontaneous speech of seven children from the age of 1;6 to 5;0. The main findings were as follows. First, the double-object dative was acquired by most of the children before either of the prepositional datives; this was attributed to the greater frequency with which children heard this construction with individual verbs. Second, the verbs children used with these constructions were not only the adult prototypical ones, but also a number of the less prototypi- cal ones; again, this was very likely due to their frequency and saliency in the language children heard. Third, no support was found for Ninio’s (1999) analysis of the emergence of constructions in terms of a single “pathbreaking” verb; rather, children began using the double-object dative with many different verbs and did not follow the trajectory proposed by Ninio (i.e., a single verb is used for some months before an “explosion” of new verbs is introduced in the construction). Finally, most of the verbs initially used in the three dative constructions were first used in other constructions (e.g., a simple transitive); this was even true for some obligatory datives, such as give and show. The current results provide a starting point for determining the underlying representations for the different kinds of dative constructions and for explicating how children understand the interrelations among these and other constructions.

All languages of the world have grammatical constructions for expressing the transfer of objects (and other things) between people (Newman, 1996). In English, a constellation of three related constructions is responsible for doing this: the to dative, the for dative (or benefactive), and the double-object dative (or ditransitive). These English constructions are interesting from a developmental point of view for several reasons: (a) each refers to a salient semantic situation for children and so is acquired relatively early; (b) each is relatively coherent semantically in that it is always used for some kind of trans- fer between people (either literal or metaphorical); and (c) each is cognitively

 2001 Cambridge University Press 0142-7164/01 $9.50 Applied Psycholinguistics 22:2 254 Campbell & Tomasello: Acquisition of English dative constructions complex in that it involves three participants (donor, recipient, gift) (Dixon, 1991). Although virtually all accounts of children’s early acquisition of English note the existence of these constructions, few studies have directly examined them. One exception is a study by Snyder and Stomswold (1997) in the UG frame- work; the authors used children’s acquisition of the dative constructions as part of an about parameter setting in language acquisition. Specifically, they argued from UG analyses, presenting some suggestive evidence, that chil- dren should always acquire the double-object dative before they acquire the other two constructions. Another exception is a study by Osgood and Zehler (1981); here, the authors provided evidence that children understand prototypi- cal instantiations of these constructions (in which two human participants ex- change an inanimate entity) before they understand less prototypical uses. Fi- nally, Tomasello (1998) gave a descriptive account of one child’s use of expressions of ; he found that dative constructions were among the earliest constructions to emerge. In addition, a number of proposals have been made about the relation between children’s grasp of verb meanings and constructions in general, often granting a prominent role to the dative constructions. First, Gropen, Pinker, Hollander, Goldberg, and Wilson (1989) studied children’s acquisition of the dative alterna- tion (i.e., their ability to identify verbs that do and do not participate in both double-object dative and to dative constructions). The basic question involved children’s identification of the verb classes that do and do not alternate; this was part of a larger set of questions about the relation between verbs and con- structions in general (e.g., Pinker, 1989). Second, Goldberg and Sethuraman (n.d.) argued that all of the most basic verb-argument constructions of English have one or more basic verbs – usually a “light” verb – as their central sense. In the case of the dative constructions, this verb is give. They provided some evidence that many children learn their first dative construction with give. Fi- nally, Ninio (1999) presented a related argument that children acquire their earli- est constructions on the basis of one or two “pathbreaking” verbs (again, usually a ), which pave the way for the acquisition of other verbs for use in that construction. Although Ninio did not address the dative constructions in particular, her proposals are relevant here, as these constructions form a very distinctive group. This article presents the results of a corpus study of seven children from the CHILDES database. The purpose of the study was to document the manner in which children acquired their earliest to dative, for dative, and double-object dative constructions. The following questions were posed. First, which of the three constructions did children use first? Second, which verbs were used (and which of these were used first) in the three constructions? Third, did the verbs used first in these constructions have a prior history of use in other construc- tions? Finally, what was the role of child-directed speech in the acquisition of dative constructions? In answering these questions we hoped to provide insight not only into children’s acquisition of these particular constructions, but also into the general processes of language acquisition – especially with regard to the relation between verbs and constructions. Applied Psycholinguistics 22:2 255 Campbell & Tomasello: Acquisition of English dative constructions METHOD

Data Participants consisted of seven children from the CHILDES database (Mac- Whinney, 1995). These children were chosen to represent a large span of early language development during the preschool years. Included in the study were all the transcripts for Eve, from ages 1;6 to 2;3 (Brown, 1973); Adam, from ages 2;3.4 to 4;10.23 (Brown, 1973); Sarah, from ages 2;3.5 to 5;1.6 (Brown, 1973); Peter, from ages 1;9.7 to 3;1.21 (Bloom, 1970); Abe, from ages 2;4.24 to 5;0.11 (Kuczaj, 1976); Naomi, from ages 1;2.29 to 4;9.3 (Sachs, 1983); and Nina, from ages 1;11.16 to 3;3.21 (Suppes, 1974).

Analytic procedure All transcripts were initially searched by hand, as a part of a larger study of the development of early verb-argument constructions. As a part of this process, all utterances identified as to datives, for datives, or double-object datives were collated. In all cases, the utterance in question had to have two postverbal argu- ments and had to convey some kind of transfer of objects or information (e.g., telling people secrets, reading books to children, and making things for people). In double-object datives, the recipient appeared postverbally before the gift (as in X gives Y the Z). In to datives, the gift appeared before the recipient, which was marked by either to (as in X gives the Z to Y) or a schwa. In for datives, the gift also appeared before the recipient, which was marked by for (as in X makes the Z for Y). For purposes of the present study, the authors analyzed and tabulated these utterances as needed for each of the specific research questions.

RESULTS AND DISCUSSION The results address three main questions. First, what was the order of first uses of the three different dative construction types? Second, which verbs were used in these constructions? Third, when were the very earliest uses of these construc- tions and verbs?

First uses of the constructions Snyder and Stomswold (1997) claimed that all of the children in their sample (some of whom overlap with the current sample) learned double-object datives before to datives. This generalization has some truth to it, but the situation is actually more complex than this if we look at both individual children and their individual verbs. First of all, there was a large range in the age of acquisition for the constructions. Eve produced her first construction when she was only 18 months of age, whereas Sarah was 33 months of age when she initially used a dative. The other five children began producing constructions somewhere in between these two ages. Although the age of initial production showed large variance among the children, patterns in the order of acquisition were apparent. As can be seen in Table 1, five of the seven children produced double-object Applied Psycholinguistics 22:2 256 Campbell & Tomasello: Acquisition of English dative constructions Table 1. Age of first use of the three dative constructions for the seven children

Double-object datives To datives For datives

Eve 1;6a 1;10 1;11 Nina 1;11.29a 2;0.17 2;1.15 Peter 2;1.21 2;0.7a 2;1.21 Naomi 2;1.7a 2;5.3 2;3.19 Adam 2;3.4a 2;11.13 2;10.30 Abe 2;6.14 2;6.18 2;5.20a Sarah 2;9.29a 3;2.23 3;0.18

aIndicates the first construction type used by the child. datives before either to datives or for datives. After the initial production of double-object datives for these five children, there did not seem to be an advan- tage of one of the remaining constructions over the other: three children pro- duced a for dative, and two children produced a to dative. On average, 3.4 months separated the initial production of the double-object dative from that of the next dative construction, whereas, on average, only 1.2 months separated the emergence of the remaining constructions. This pattern supports the intuitive observation that to datives and for datives are more closely related to each other than either is to the double-object dative. For the two children who did not initially produce double-object datives, Abe first produced a for dative, and Peter first produced a to dative. Their first uses of the three constructions were much more closely spaced in time than were those of the other children. For the five children who initially produced double-object datives, the time from pro- ducing the double-object dative to having produced all three constructions was an average of 4.6 months; for Abe and Peter, all three constructions were ini- tially produced within one to two months. Based on these observations, it would seem that, if one construction were going to be used first by these children, it would be the double-object dative construction, followed by to datives and for datives, which would emerge ap- proximately at the same time a few months later. However, contrary to the predictions of Snyder and Stomswold (1997), there were two children who did not show a distinct preference for one construction over the others. Snyder and Stomswold did not analyze Abe’s language, but they did analyze Peter’s data; they came up with a later age of emergence for his to datives than we did. But on the transcript at 2;0.7, Peter produced the utterance “I’ll show it to you,” which is clearly a to dative construction by our criteria; this predated his first double-object dative by about 6 weeks. Thus, we believe that Peter represents an exception (as does Abe, if for datives are counted) to Snyder and Stomswold’s generalization that double-object datives always emerge first. It should be noted that, in general, some of our ages of emergence for double- object datives and to datives differ from those cited in Snyder and Stomswold (1997). These discrepancies are likely due to different criteria for identifying Applied Psycholinguistics 22:2 257 Campbell & Tomasello: Acquisition of English dative constructions the constructions. In the present study, the initial use of a dative for Eve was “Send the box a Sheila,” which was produced at the age of 1;10. Snyder and Stomswold (1997) placed Eve’s initial production two months later (at 2;0) when she uttered the word to in the phrase. Similarly, in our study, the initial production of a to dative by Nina was “Miriam gave it me.” Because of differ- ences in criteria, some of the ages of acquisition in our analyses are earlier than those presented in Snyder and Stomswold. In addition, some of the ages of acquisition cited in Snyder and Stomswold are earlier than those presented in our analyses. We double-checked by hand every instance in which Snyder and Stomswold gave an earlier date of acquisition than we did, but we could not find any evidence of such constructions until the dates presented here. The question arises as to why the double-object dative would be learned ear- lier in general, if not in every case. Snyder and Stomswold (1997) argued that it was not based on adult frequency, although their data showed that it very likely was. Their conclusion was based on one analysis. They compared the relative percentage of to datives in the parent speech with the child’s age of acquisition of to datives, but they found no correlation. However, this is a very indirect way to assess this relationship and as such does not justify their conclu- sion that there is no correlation between adult frequency and child acquisition. A closer look at their Table 2 (p. 290) tells a very different story: across all parents, children heard double-object datives an average of 73.2% of the time, whereas they heard to datives only 26.8% of the time (both calculated as a proportion of all datives heard, with the parent value based on the analysis of the verb give). Moreover, 11 of the 12 children in their sample had mothers who used more double-object datives than to datives (with the verb give), most by a margin of more than 2 to 1. This pattern of results – much more than any indirect correlational analyses – supports the hypothesis that children first ac- quire the type of dative construction they hear most often in their mother’s speech. On this hypothesis, Snyder and Stomswold’s complex UG justifications for the earlier emergence of double-object datives are superfluous. Gropen et al. (1989) also argued that frequency of adult use was not a factor in children’s early acquisition of the double-object dative. The problem in this case was that they looked at overall frequencies, combining all verbs, which may have obscured some of the lexically specific relationships. A much more sensitive analysis would be to look at each verb that was used in both the double-object dative and one of the prepositional datives and to tabulate adult frequency and child frequency for each verb separately. We performed such an analysis; the results are displayed in Table 2. When we looked at the 26 verbs that the children used in alternation (and both parent and child had one value higher than the other), we found that, in 21 of 26 cases, adult and child had the same preference ( p < .01, binomial test). Note that not all of these were prefer- ences for double-object datives. Peter and Adam each had one verb that showed the opposite pattern (i.e., both they and their mothers preferred the to dative). In addition, of the five verbs that did not match, three were cases in which the child but not the mother preferred the to dative, and two were cases in which the child but not the mother preferred the double-object dative. Thus, the overall pattern seemed to be that children learned the double-object dative first because Applied Psycholinguistics 22:2 258 Campbell & Tomasello: Acquisition of English dative constructions Table 2. Percentage of parent use and child use for the verbs the child used in both double-object datives and to datives

Child Parent

Double-object To Double-object To

Eve Give 14 8 54 20 Read 65 4 4 Show 31 8 1 Total 23 14 66 25 Proportion 62% 38% 73% 27% Nina Give 65 22 273 91 Bring 4 3 30 34 Read 4154339 Send 2 2 30 28 Show 3 1 58 15 Total 78 43 434 207 Proportion 64% 36% 68% 32% Peter Give 85 19 16 12 Get 4111 2 Make 42 2 1 Bring 31 4 0 Show 16 2 7 1 Total 112 35 30 16 Proportion 76% 24% 65% 35% Naomi Read 36 5 2 Total 3 6 5 2 Proportion 33% 67% 60% 40% Adam Read 33 212 Show 12 5 23 12 Give 80 8 93 46 Make 5103 5 Get 14 10 2 5 Buy 44 2 1 Total 118 40 125 81 Proportion 75% 25% 61% 39% Abe Show 10 1 24 5 Get 61 26 78 21 Make 8 7 14 13 Give 10 7 50 27 Bring 2 2 13 0 Read 11 2 8 Total 92 44 181 74 Proportion 68% 32% 71% 29% Applied Psycholinguistics 22:2 259 Campbell & Tomasello: Acquisition of English dative constructions Table 2 (continued)

Child Parent

Double-object To Double-object To

Sarah Give 77 10 213 63 Read 7 1 11 1 Bring 5 2 29 3 Get 21 10 24 8 Show 12 5 40 10 Make 1 3 10 0 Total 123 31 327 85 Proportion 80% 20% 79% 21%

this was what they heard from their parents most often. Indeed, in to Gropen et al., we found a 2 to 1 advantage for double-object datives in the mother’s language. But in the case of alternating verbs, children sometimes heard one of the prepositional datives more often than the double-object dative, and in many cases that was the form they used as well.

Use of verbs in the constructions One interesting fact about the English dative constructions is that they are used with a delimited set of verbs, although the for dative has a much wider range: many different types of transitive activities can, in the right context, be con- ceived of as being done for the benefit of someone other than the actor. Of interest, then, was to see which verbs the children would initially use in the dative constructions. Table 3 lists the different verbs the children used in the dative constructions (the two prepositional datives were combined). The verbs that were used the most by the children were those that were generally most frequent in adult language. The verbs used by at least six of the seven children in one or another dative construction were give, read, bring, take, show, tell, get, buy, and make. Note that the first six verbs are classic dative-alternation verbs in which the double-object dative and to dative are both possible. Buy and make alternate the double-object dative and the for dative. Get appeared in all three constructions: children said such things as “Get that to Mommy” and “Get that for Daddy.” Indeed, at least five of the seven children used six verbs in alternation: give, get, make, show, bring, and read. However, many verbs that could potentially alternate were not used in this way by the children. For example, reasonably frequent alternating verbs such as tell, feed, hand, and pay were limited to the double-object construction by many children. Conversely, reasonably frequent alternating verbs such as fix, leave, open, and take were limited to one of the prepositional datives by many children. (All but take were used as for datives.) Applied Psycholinguistics 22:2 260 Campbell & Tomasello: Acquisition of English dative constructions Table 3. Verbs used in different constructions by more than one child up to the age of 3;0

Prepositional Both Double-object dative Neither

Get 60 10 Give 61 00 Make 60 10 Show 61 00 Read 50 20 Bring 51 01 Buy 23 11 Take 11 41 Tell 05 11 Find 13 12 Do 00 52 Send 11 23 Throw 11 23 Call 04 03 Fix 01 33 Leave 01 33 Open 00 43 Want 00 43 Ask 12 04 Draw 11 14 Feed 02 14 Have 00 34 Hold 00 34 Put 00 34 Say 00 34 Build 10 15 Write 10 15 Cut 01 15 Hand 02 05 Paint 01 15 Pay 02 05

Note: Each verb has a total of 7; each child falls somewhere on the continuum for each verb. Verbs used by one child were: bite, catch, carry, cook, forget, grate, hit, keep, lend, loan, lower, mail, move, peel, pour, push, repeat, set, sell, sing, spell, teach, tie, untie, wash, wave, whisper, wrap.

There were, however, many individual differences. Eve, Sarah, Nina, and Adam acquired some verbs that alternated and some that were confined to a specific construction. On the other hand, Naomi used only one verb in alternating form, and Peter had no verbs that he used solely in double-object datives. Also of interest were the semantic classes into which the children’s early dative verbs fell. According to Goldberg (1995), the double-object dative con- Applied Psycholinguistics 22:2 261 Campbell & Tomasello: Acquisition of English dative constructions Table 4. Semantics of the children’s dative verbs that could be a double-object dative in adult language

A. Central sense Giving: give, hand, lend, loan, sell, feed Motion: hit, push, throw Deictics: bring, take B. Non-prototypical Promising: Refusing: Permission: Future transfer: leave Creation: make, build, cook, cut, draw, fix, grate, paint, pour, tie, untie Obtaining: get, buy, catch, find, hold, pay C. Metaphors Sending: send, mail Communicative message: ask, read, show, sing, tell, spell, teach, write Instrument of communication:

Source: Based on Goldberg (1995). struction in adult language has a prototypical meaning (or central sense) in which an successfully causes the transfer of a to a recipient (in Dixon’s [1991] terms, the donor causes the recipient to receive a gift). The verbs that most faithfully convey the central sense of the construction can be divided into three types: (a) verbs that inherently signify acts of giving (give, pass, hand, serve, feed, etc.), (b) verbs of instantaneous causation of ballistic motion (throw, toss, kick, poke, fling, etc.), and (c) verbs of continuous causation in a deictically specified direction (bring, take, etc.). Goldberg (1995) noted some other, less prototypical senses as well: verbs of giving but with satisfaction conditions (guarantee, promise, etc.), verbs of refusal (refuse, deny, etc.), verbs of permission ( permit, allow, etc.), verbs of creation (make, bake, build, cook, etc.), verbs of obtaining (get, grab, win, earn, etc.), and verbs of future transfer (leave, bequeath, allocate, grant, etc.). Finally, she claimed that some senses are derived metaphorically from the central sense in that they do not designate the physical transfer of an object, as would the central sense and most of the less prototypical senses. These include verbs of sending (send, mail, ship, etc.), verbs of communicated message (tell, show, ask, teach, read, etc.), and verbs of instrument of communication (radio, e-mail, telephone, fax, etc.). Table 4 presents the verbs that the children used before age 3 that fall into these semantic categories. Contrary to what we might expect, the children did not begin with the prototypical senses of the dative verbs. They used many giving verbs (e.g., give, hand, etc.), but they used just as many obtaining verbs (e.g., get, find, buy, etc.) and even more verbs in which someone creates some- thing for someone (e.g., make, cook, draw, etc.) or someone conveys informa- tion to someone (e.g., read, tell, ask, etc.). The latter two verb classes in particu- Applied Psycholinguistics 22:2 262 Campbell & Tomasello: Acquisition of English dative constructions lar are non-prototypical, the last being designated by Goldberg (1995) as a metaphorical extension. Returning to Table 3, we can see that, of the six verbs used by at least five children, one is a giving verb, one is an obtaining verb, one is a deictic motion verb, one is a verb of creation, and two are verbs of communicated message. Apparently, children do not begin to use this construc- tion with verbs that linguistic analysis has designated as prototypical in adult language, and verbs of communicated message are not, developmentally, meta- phorical extensions (although historically they may be so). A more likely expla- nation for the children’s choice is that these particular verbs are used most often by their parents in talking to them about activities that are of interest to them (for evidence of a correlation between parent use of particular verbs and child use of those same verbs, see De Villiers, 1985; Naigles & Hoff-Ginsburg, 1998; Tomasello & Kruger, 1992).

First uses of the constructions Ninio (1999) proposed that children begin learning a new syntactic construction with a “pathbreaking” verb, which then paves the way for the relatively rapid acquisition of subsequent verbs in this construction. She described these verbs as follows: “The ‘pathbreaking’ verbs that begin the acquisition of a novel syn- tactic rule tend to be generic verbs expressing the relevant combinatorial prop- erty in a relatively pure fashion” (p. 619). Although her analysis was confined to basic in English and Hebrew, her hypothesis may be extended to the dative constructions – perhaps especially the double-object construction in that it has unique syntactic properties (i.e., there is a core argument intervening between the verb and the direct object). In this case, Ninio would predict that the prototypical verb of giving (give) should be the pathbreaking verb for most children in that it is the lightest verb that can be used with the double-object dative. (It also expresses a “fundamental object relation of incorporation into and ejection from the personal” [p. 620], which Ninio considered essential for pathbreaking verbs.) Table 5 presents the first five verbs that were used with the double-object dative and the prepositional datives, along with the age of first use. The first verbs used in the double-object construction were give (4), show (2), bring, feed, send, read, get, make (some children used more than one of these verbs in the initial sample). Thus, not all of the children began with give; three of the four who did begin with give began with another verb in the same month as well. Moreover, many of these verbs did not appear to be pathbreaking verbs and/or did not seem to be the best at “expressing the relevant combinatorial property in a relatively pure fashion” (Ninio, 1999, p. 619). Feed, send, show, and read are fairly heavy verbs; get and make have more frequent uses in other construc- tions, and so they would not seem to be great candidates for expressing the dative relation. Again, in our view, the most important variable seems to be whether the parents used these verbs frequently and in salient contexts (for some evidence, see Goldberg & Sethuraman, in press). The fact that children acquired many light verbs early in development is probably the result of these verbs being used widely and frequently in many different contexts. Applied Psycholinguistics 22:2 263 Campbell & Tomasello: Acquisition of English dative constructions Table 5. First five verbs used by each child in each construction type (the two prepositional datives were combined)

Double-object dative Prepositional dative

Verb Age Verb Age

Naomi Give 25 Say 28 Bring 25 Get 29 Call 27 Read 29 Read 35 Open 31 — Put 35 Nina Feed 23 Give 24 Send 23 Bring 25 Make 25 Take 25 Give 25 Get 26 Get 25 Read 27 Eve Show 18 Send 22 Give 19 Bring 22 Make 20 Fix 23 Read 21 Read 23 Bring 21 Hold 23 Sarah Give 33 Make 36 Pour 35 Get 37 Get 36 Give 38 Tell 37 Want 38 — Forget 39 Adam Give 27 Get 34 Read 27 Show 35 Hand 30 Throw 35 Tell 36 Give 36 Buy 38 Put 37 Peter Get 25 Show 24 Make 25 Get 25 Give 25 Give 26 Bring 27 Throw 27 Show 27 Loan 28 Abe Show 30 Make 29 Get 31 Give 30 Send 33 Get 30 Give 34 Take 30 Call 34 Show 32 Applied Psycholinguistics 22:2 264 Campbell & Tomasello: Acquisition of English dative constructions

Figure 1. Cumulative number of verbs used with double-object data construction as a func- tion of age in months.

The other part of Ninio’s (1999) claim is that a new construction is initially used by a child for a relatively extended period with the pathbreaking verb (in her study of transitives, about 1 month) before it is used with other verbs; from the acquisition of the second verb onward, the construction expands to include new verbs quickly. We examined this hypothesis for the double-object dative construction by looking at each child’s addition of verbs to the construction as a function of age (see Figure 1, modeled on Ninio’s Figures 1 and 2). Overall, we found that only one child fit this hypothesis of a relatively long period with a pathbreaking verb before the rapid extension to other verbs. The results may be summarized as follows (in descending order from those that fit best the hypothesis; see Table 5 for data). Sarah started with give in the double-object dative, only added a second verb two months later, and then added more verbs fairly rapidly over the next months. She fit the hypothesis reasonably well. Eve, at the first recording, was already using show, which may have had a history before observation began; show, however, did not seem to be a path- breaking verb. Nina was using feed and send at her first recording, but neither was a pathbreaking verb. Adam was using give and read at his first recording; give may have had a history prior to our observation. Adam added only four new verbs during the next thirteen months. Abe started with show some months after his first recording and then rapidly added new verbs (five new verbs in the next four months); there was no protracted period with show alone. Naomi started with two verbs, give and bring, some months after her first recording; she then added only three more verbs during the next fifteen months. Peter suddenly began using three verbs with the double-object dative some months after his first recording. A related issue is that many dative verbs may also be used as simple transi- tives (e.g., “Give it” or “I throw this”) or in other structures. In order to identify where the dative verbs came from, we looked at all verbs that appeared in more Applied Psycholinguistics 22:2 265 Campbell & Tomasello: Acquisition of English dative constructions Table 6. Construction in which verbs were first used

Dative Ditransitive Transitive Intransitive Other

Bring 5 Call 21 Fix 31 Get 11 32 Give 12 4 Make 51 1 Put 21 Read 51 2 Send 31 Show 241 Take 41 Tell 33 Throw 21 1 Totals 7 3 43 7 10

Note: Verbs are those that appear in more than one child’s first five verbs as either a prepositional dative or a double-object dative. Numbers in the cells are number of children

than one child’s first five productions of either the double-object dative or the prepositional dative construction. As can be seen in Table 6, the major finding was that, for most children, verbs generally began as transitives. Many were requests in which the implied recipient was the self (e.g., “Send that,” “Read story,” or “Give that”) or comments on an activity for which, on some later occasion, the child would specify a recipient (e.g., “Make a cake” or “I may give some”). With respect to Ninio’s (1999) hypothesis, this suggests that the dative constructions are, in some sense, not primary constructions in that the verbs used in them have previous histories; perhaps this is why these construc- tions do not fit very well into her pathbreaking verbs hypothesis. The most surprising finding in this analysis was that even the verbs that were obligatorily dative (e.g., give, send, show, tell, and call – in the sense of “call someone a name”) quite often began their careers without both postverbal arguments being expressed, typically for very good pragmatic reasons in the context of adult– child conversation. This suggests that even the double-object dative is not a primary construction in that it may come from simpler constructions.

CONCLUSION In the present article, we provided some basic information about young chil- dren’s use of the three main dative constructions of English: the double-object dative, the to dative, and the for dative. We used children’s acquisition of these constructions to test some of the more general hypotheses about how young children acquire the basic verb-argument structures of the English language. Applied Psycholinguistics 22:2 266 Campbell & Tomasello: Acquisition of English dative constructions Following other researchers, we found that the double-object dative was most often the first to appear in children’s language, but this was not invariably the case. Indeed, a lexically specific analysis of the verbs used in alternation be- tween constructions showed that the children basically used each verb in the way their mothers used it. This kind of analysis vitiates the need for a formal linguistic analysis that explains why double-object datives are used first by most children, as proposed by Snyder and Stomswold (1997) and Gropen et al. (1989). Frequency effects were also apparent in the verbs the children initially used with these constructions – especially with the double-object dative – which accords with De Villiers (1985), Tomasello and Kruger (1992), and Naigles and Hoff-Ginsburg (1998). Contrary to the notion that children begin with light verbs and/or verbs that are prototypical for the construction, we found that some children used many verbs that, from the point of view of adult analyses, are not prototypical for the double-object dative: for example, verbs of creation (e.g., make, cook, etc.) and verbs of communicated message (tell, show, etc.), the latter being considered a metaphorical extension of the construction in adult language. Our analysis found no support for Ninio’s (1999) hypothesis that children’s initial entry into a syntactic construction is with a single pathbreaking verb that is semantically light and central to the prototypical constructional meaning, and that they use this verb for some months before learning other verbs in that construction. This pattern was only found with the double-object dative for one of the seven children – this construction being the most distinctive of the three. At the same time, we found that most verbs that were used early in the various dative constructions had been previously used in simpler constructions (espe- cially simple transitives). This was even true of several verbs (e.g., give and show) that in adult language require both postverbal arguments to be expressed; children typically used reduced constructions (e.g., give it or show it). It may be that the dative constructions are not primary constructions, and so they are not strictly appropriate for Ninio’s hypothesis. This raises the question as to the nature of this very special set of constructions – some of the only ones to express three core arguments – from the child’s point of view. There is no evidence in the present study to elucidate children’s understand- ing of the abstractness of the three dative constructions or their understanding of the relationship among the three constructions. However, other data from both naturalistic and experimental studies suggest that most English-speaking children do not really have an abstract understanding of the transitive construc- tion until about 3 to 31⁄2 years of age (see Tomasello, 2000, for a review). Thus, it is likely that children’s initial use of the three dative constructions is based on the individual verbs involved and not on an abstract schema or construction. But, given the dependence of the dative constructions on the general transitive construction (both conceptually and developmentally), it may be that the dative constructions begin life with a head start on the process of becoming abstract. Toward that end, experimental studies of children’s productivity with these con- structions during the early phases of their use would be useful in helping to decide among the competing hypotheses. Applied Psycholinguistics 22:2 267 Campbell & Tomasello: Acquisition of English dative constructions ACKNOWLEDGMENT This research was supported by a grant from the National Institutes of Health (Number R01 HD 35854-01) to the second author.

REFERENCES Bloom, L. (1970). Language development: Form and function in emerging grammars. Cambridge, MA: MIT Press. Brown, R. (1973). A first language: The early stages. Cambridge, MA: Harvard University Press. De Villiers, J. (1985). Learning how to use verbs: Lexical coding and the influence of input. Journal of Child Language, 12, 587–596. Dixon, R. (1991). A new approach to English grammar, on semantic principles. Oxford: Oxford University Press. Goldberg, A. (1995). Constructions: A construction grammar approach to argument structure. Chi- cago: University of Chicago Press. Goldberg, A., & Sethuraman, N. (n.d.). Learning argument structure generalizations. Manuscript. Gropen, J., Pinker, S., Hollander, M., Goldberg, R., & Wilson, R. (1989). The learnability and acquisition of the dative alternation in English. Language, 65, 203–257. Kaczaj, S. A. (1976). Arguments against Hurferd’s “Aux copying rule.’’ Journal of Child Language, 3, 423–427. MacWhinney, B. (1995). The CHILDES Project: Tools for analyzing talk. Hillsdale, NJ: Erlbaum. Naigles, L., & Hoff-Ginsburg, E. (1998). Why are some verbs learned before other verbs? Effects of input frequency and structure on children’s early verb use. Journal of Child Language, 25, 95–120. Newman, J. (1996). Give: A cognitive linguistic study. Berlin: Mouton deGruyter. Ninio, A. (1999). Pathbreaking verbs in syntactic development and the question of prototypical transitivity. Journal of Child Language, 26, 619–654. Osgood, C., & Zehler, A. (1981). Acquisition of bitransitive sentences: Prelinguistic determinants of language acquisition. Journal of Child Language, 8, 367–384. Pinker, S. (1989). Learnability and cognition: The acquisition of verb-argument structure. Cam- bridge, MA: Harvard University Press. Sachs, J. (1983). Talking about the there and then: The emergence of displaced reference in parent– child discourse. In K. E. Nelson (Ed.), Children’s language (Vol. 4). Hillsdale, NJ: Erlbaum. Snyder, W., & Stomswold, K. (1997). The structure and acquisition of English dative constructions. Linguistic Inquiry, 28, 281–317. Suppes, P. (1974). The semantics of children’s language. American Psychologist, 29, 103–114. Tomasello, M. (1998). One child’s early talk about possession. In J. Newman (Ed.), The linguistics of giving (pp. 58–76). Amsterdam: John Benjamins. (2000). Do young children have adult syntactic competence? Cognition, 74, 209–253. Tomasello, M., & Kruger, A. (1992). Joint attention on actions: Acquiring verbs in ostensive and non-ostensive contexts. Journal of Child Language, 19, 311–333.