«Abstract We report an experiment that examined 3- and 4-year-old children’s representation of the passive structure. Early studies of ...»
Children’s Early Acquisition of the Passive:
Evidence from Syntactic Priming
Katherine Thatcher, Holly Branigan, Janet McLean and
University of Edinburgh
We report an experiment that examined 3- and 4-year-old children’s representation of
the passive structure. Early studies of typically-developing children’s acquisition of
the passive suggest that this construction is acquired late and – or that its acquisition
is semantically constrained: children comprehend actives much earlier than passives and comprehend actional verb passives earlier than non-actional verb passives (Maratsos, Fox, Becker & Chalkley, 1985). Conversely, some production studies have shown the passive is acquired earlier than thought: 3-4 year old children produce passives following training (Brooks & Tomasello, 1999) and priming (Huttenlocher, Vasilyeva & Shimpi, 2004), however, such studies have not examined whether the passive is constrained to actional verbs early on. We report a syntactic priming study that manipulated Prime Structure (active vs. passive) and Verb Type (actional vs. non-actional). We found a strong and reliable effect of Prime Structure for children (27%) and adult controls (19%). There was, however, no effect of Verb Type (Fs 2). Participants were more likely to produce passive targets following passive primes than active primes, irrespective of the verb. We conclude that children do acquire an
syntactic representation for the passive early on (by 3-4 years) that is not constrained by verb type.
1.2 Studies of Children’s Acquisition of the Passive Previous research into children’s comprehension of the passive has repeatedly found that they comprehend actional verb passives (1a) better than non-actional verb passives (1b), the latter not being reliably understood until beyond the age of five years (Maratsos et al., 1985; Sudhalter & Braine, 1985; Borer & Wexler, 1987;
Gordon & Chafetz, 1990; Marchman, Bates, Burkardt, & Good, 1991; Fox & Grodzinksy, 1998; Hirsch & Wexler, 2004).
1) a. The boy was hit by the girl b. The boy was loved by the girl One explanation to account for this oft-replicated result is that the acquisition of the passive is semantically constrained such that children generalize the passive structure to highly transitive verbs first, such as physical action verbs or verbs of © 2008 Thatcher, Branigan, McLean and Sorace Child Language Seminar 2007 University of Reading result, before generalizing it to less transitive, non-actional verbs such as psychological or experiential verbs at a later age (Maratsos et al. 1985).
An alternative argument is that aspects of the passive construction are acquired late and that children use some other strategy at a younger age which results in them comprehending or appearing to comprehend actional but not non-actional verb passives. For example, Borer & Wexler (1987) argue that the ability to form argument chains, required to move a verb’s object into subject position in constructions such as the passive, is not acquired before 5 years of age and that before then the strategy children use is to analyse verbal passive sentences as adjectival passives. Their explanation for the discrepancy in results of comprehension tests is that such an analysis may be felicitous with actional verb past participles but not with non-actional verb past participles, hence children’s inability to comprehend these passives.
Alternatively, Fox & Grodzinsky (1998) propose that the ability to transmit the passive verb’s external thematic role to the oblique noun phrase is acquired late and that the strategy young children use to interpret full passives is to assign an agentive thematic role from the preposition by, a strategy that works for actional passives (hence children’s comprehension of these) which tend to have an agent subject role but not for non-actional verb passives (hence children’s poor performance with these) whose external argument tends to be a theme or experiencer and therefore is not compatible with the role assigned by the preposition by.
A wider review of the literature however reveals a great deal of evidence suggesting the passive is not acquired late. For example, the passive appears in children’s spontaneous speech from around the age of three (Budwig, 1990) and elicited production studies have also shown that children can produce full passive sentences by four years. Huttenlocher et al. (2004) used passive primes to elicit passive sentence descriptions of pictures from four year olds and Tomasello, Brooks & Stern (1998) found that three and a half year olds who heard the structure modelled in sentences with novel verbs produced more passives with different novel verbs than a control group. Furthermore, such studies using novel verb experiments also show early productive use of the passive structure suggesting that children of this age can generalize the syntactic construction to new items (see also Pinker, Lebeaux & Frost, 1987; Brooks & Tomasello, 1999). Finally, it has also been shown that when placed in a discourse context in which a full passive utterance was appropriate, that is, one where the discourse focus was on the patient but where there was also more than one possible agent present, children as young as three were able to produce full passive utterances (Crain, Thornton & Murasugi, 1987).
This language production research suggests that the passive is neither acquired as late as previously suggested nor that children use alternative strategies for producing passive-like utterances before being able to produce full passive sentences.
Though this research shows that alternative experimental methods may show earlier competence with the passive than has been demonstrated in comprehension tests, these studies have, however, only tested children’s production of actional verb passives and as such the question of whether children’s early knowledge of the passive is restricted to highly transitive verbs before non-actional, psychological verbs remains unaddressed by language production research. The present study addresses the question of whether there is a semantic constraint on English-speaking children’s early passives using a method, syntactic priming, that has successfully shown that children have a syntactic representation for the passive at a young age.
1.2 Syntactic Priming The term ‘syntactic priming’ refers to the tendency amongst speakers to repeat the syntactic structure of an utterance used in previous discourse and has been both observed in conversational contexts (Weiner & Labov, 1983) and used as an experimental method for manipulating participants’ speech (Bock, 1986). Through prior processing of a syntactic structure a speaker becomes more likely to repeat that structure in a subsequent utterance, thus a speaker is more likely to produce a passive sentence after hearing a passive sentence prime, such as (2a), than after hearing an active sentence prime such as (2b). This effect of structural repetition has been variously attributed to repetition of the processing mechanisms of language production (Bock, 1986); residual activation of linguistic representations (Branigan, Pickering, Liversedge, Stewart & Urbach, 1995); and implicit learning (Bock, Dell, Chang & Onishi, 2007).
2) a. The pig is being washed by the farmer b. The farmer is washing the pig How can this be related to children’s acquisition of the passive? It follows that participants may be primed to reproduce a syntactic structure only if that structure has been acquired already. To date, there have been a few priming experiments that have tested children’s acquisition of the passive. These studies have generally shown that young children were more likely to produce passive descriptions following passive primes than following active primes – this structural repetition in the absence of repeated lexical items between prime and target suggests that children have indeed acquired an abstract representation for the passive by four (Huttenlocher et al., 2004; Whitehurst et al., 1974) or even three (Bencini & Valian, 2006), though see also Savage, Lieven, Theakston & Tomasello (2003) for different results and interpretation. However, these studies used actional verbs for both the prime and target items, none of these studies have employed this method to investigate the issue of semantic constraints to children’s early passives. The present study therefore investigated priming of passive sentences with children aged three and four years old and specifically looked at whether any priming effect varied according to the verb type of the prime.
We hypothesized that if children’s early development of the passive construction was based on a core class of verbs, actional verbs, the children would have an early syntactic representation for the passive that was only linked to verbs from that verb class and would therefore only be primed by same-class verbs also linked to the passive representation (see Figure 1: hit and push are both action verbs and so are linked to the passive structure node, either should prime the other, however scare and love are not action verbs and so the passive construction is not yet generalized (linked) to them – they should not therefore prime or be primed by passives). The alternative, null, hypothesis is that children acquire the passive with non-actional verbs as early as action verbs (see Figure 2). In this case we would expect the same priming effect from both actional and non-actional primes.
1.3 Questions and Predictions This study addressed the following questions therefore: do children acquire an abstract syntactic representation for the passive before 5 years old? It was predicted that if they do not, then no priming effect would be observed. It was furthermore predicted that if children’s early knowledge of the passive construction is restricted to item-based (verb-based) representations (Tomasello, 2000) then no priming would occur since there was no overlap between the lexical items in the primes and targets.
The second question addressed in this study was: is children’s early acquisition of the passive constrained to actional verbs and only generalized to non-actional verbs at a later age? It was predicted that if children do have a semantically constrained representation for the passive early on (Figure 1) then priming would only be observed in the actional prime condition but if not (Figure 2) then priming would occur in both verb type conditions.
2.1 Priming Task The priming task was a picture description task embedded in a children’s card game, ‘Snap’ (Branigan, McLean & Jones, 2005). In Snap, two players each have a set of picture cards placed face-down in front of them. They take it in turns to turn over their top card and reveal its picture, when both players reveal the same picture on their cards it is the first player to shout ‘snap’ who wins the cards in play. The game continues until one player has won all the cards.
In the experiment the only variations to the game were that the players described their picture as they turned it over and the game ended when all cards had been turned over once. A game was used to mask the priming and to make the task easy and more appealing to children; they were not under pressure to provide ‘correct’ answers.
The participants were 20 pre-school children (10 girls), ranging in age from 3;1 to 4;11 (mean age 4;2). All children were acquiring English as their first and only language and none were reported to have any language or developmental difficulties.
A control group of 20 adult, native speakers of English (15 female), were recruited from the University of Edinburgh student population and paid for their participation.
There were two priming conditions: Prime Structure (Active vs. Passive) and Verb Type (Actional vs. Non-Actional) which combined created the four priming conditions.
The experiment was based on a repeated measures design: all participants experienced all levels of all priming conditions and so the experiment measured whether the children alternated their response structure after hearing alternative prime structures with alternative verb types.
The materials consisted of a set of Snap picture cards depicting transitive events with human patients and animal agents. Where possible, nouns and verbs were chosen that had been used in previous experiments of children’s acquisition of the passive and – or that had a suitable age of acquisition rating for the children participating in the study (Morrison, Chappell & Ellis, 1997). Those required to describe a target item were different to those used in its primes (see Figure 3); the repetition of lexical items between the prime and target was avoided to ensure any priming effect observed was related to abstract structural representations and not attributable to item-based representations (Tomasello, 2000).
Figure 3. Actional verb and non-actional verb primes with actional verb target
Twelve actional verbs (shake, wash, tickle, push, kiss, punch, lick, hug, chase, kick, scratch, pinch) were used twice each to create 24 target items. Each target item had an actional verb and non-actional verb prime, both depicted with the same agent and patient, though a participant only saw either the actional or non-actional prime for a given target picture. There were 24 actional primes made up of six different actional verbs (hit, pat, bite, pull, squash, carry) used four times each and 24 non-actional primes created from six non-actional verbs (frighten, shock, annoy, upset, surprise, scare) used four times each.
Further to the prime and target items, a set of eight filler items formed the Snap items of the game; these items had the same picture so here the same verb was used for the prime and target: four new actional verbs were used once in the passive and once in the active each. Finally a set of four practice items was also created using different verbs and nouns to the experiment and filler items.