~ u m l l of ELSEVIER
Journal of Pragmatics 27 (1997) 781-798
How do young children reformulate assertions? A comparison with requests Hayd6e Marcos, Josie Bernicot Laboratoire PSYDEE, Universit~ de Paris 5 - CNRS, 46, rue Saint-Jacques, F-75005 Paris, France
Received August 1995; revised version August 1996
Abstract This study deals with the emergence of speech acts in young children and the conversational routines that underlie their production. More specifically, our goal was to determine whether 2;6-year-old children interacting with an adult are capable of reformulating their assertions when communication fails. Children's reformulations of assertions are studied relative to the various behaviors manifested by the adult, and are compared to reformulations of requests. Two groups of children aged 2;6 participated in the experiment. The first group carried out a request task and the second, an assertion task. The non-linguistic and linguistic productions of each group were compared in three situations: the adult accepts the child's message (satisfaction), the adult asks the child to clarify his/her message (clarification), or the adult rejects the child's message (refusal). The results show that (a) children do as much reformulating of assertions as of requests, (b) the form of the reformulations varies with the adult's behavior for both requests and assertions, and (c) these partner-dependent variations are not always the same for assertions and requests. The discussion shows that the conversational routines underlying the reformulations of 2;6-year-olds are rich and flexible.
1. Introduction The purpose o f this study was to examine the emergence o f communication skills in y o u n g children aged 2;6. Studies have already been conducted on the production and comprehension o f speech acts by adults (Blum-Kulka, 1990; Blum-Kulka et al., 1989; Garcia, 1993; Koike, ]L994; Houtkoop-Steenstra, 1990; Littman and Mey, 1991), schoolchildren (Baroni and Axia, 1989; Bernicot, 1991 ; Bernicot and Legros, 1987; Bernicot and Laval, 1996; Ervin-Tripp et al., 1990; Garton and Pratt, 1990; ~ The authors wish to thank Myriam Boussarie, Severin Pren6, and Marielle Bonnin for their contributions to this study, as well as the staff and children of the Seine and Marne (France) and Poitiers (France) day care centers, who made this experiment possible. Special thanks are extended to Viviar~ Waltz for translating this paper. 0378-2166/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved PII S0378-2166(96)00062-8
782
H. Marcos. J. Bernicot / Journal of Pragmati¢s 27 (1997) 781-798
Place and Becker, 1991; Sinclair and Gessel; 1990), and very young children (Bates et al., 1979; Dore, 1975). A number of studies have shown that at a very early age, children are capable not only of reformulating unanswered requests addressed to a partner, but also of adapting their reformulations to their partner's behavior (Anselmi et al., 1986; ErvinTripp et al., 1982; Golinkoff, 1986, 1993; Marcos, 1991; Marcos and Kornhaber-Le Chanu, 1992; Marcos and Bemicot, 1994; Shatz and Watson O'Reilly, 1990). The goal of the present study was to compare reformulations of requests to reformulations of a very different kind of speech act, assertions. To approach this question, the social knowledge underlying the reformulations must be considered. Because the production of assertive speech acts involves using the linguistic channel, assertive reformulation cannot be studied until the age of 2;6. What social knowledge underlies reformulations? Two answers - which in our minds are not contradictory - have been given to this question. Golinkoff (1986, 1993) contends that children's reformulations of requests during the second year attest to their desire to achieve a 'meeting of minds' with their partner. For Golinkoff, this interpretation is reinforced by the fact that children at this age are capable of negotiating and rejecting adults' erroneous interpretations of their messages. Shatz and Watson O'Reilly (1990) ascribe the reformulation of utterances by 2;6-year-olds to their conversational skills, i.e. to their ability to participate in conversational routines for the purpose of obtaining something (an object and/or an action) from their partner. These authors oppose children's conversational skills to their communicative competencies, which are thought to be defined by knowledge of the mental states of their addressees. Two points are essential in Shatz and Watson O'Reilly's (1990) arguments: children reformulate their requests in order to obtain an overt reaction from the partner, and their reformulations depend on conversational routines acquired during everyday experiences with individuals in the surroundings. These authors compared the reformulation of requests with the reformulation of assertions by 2;6-year-olds interacting in a natural situation. When producing an assertion, a child is making a statement about how things are in the world, and consequently, the reformulation of an assertion following a partner's refusal to agree with the statement made, cannot be linked to the pursuit of some concrete benefit. For Shatz and Watson O'Reilly (1990) the reformulation of an assertion is indicative of the child's desire during communication to achieve a meeting of minds. The results of their study showed that in cases where the adult asks for clarification, children reformulate their requests significantly more often than they do their assertions. They concluded that 2;6-year-olds reformulate in order to obtain some concrete benefit, and that such reformulations are dependent upon the child's capacity for participating in conversational routines. The positions expressed by Golinkoff (1986, 1993) and Shatz and Watson O'Reilly (1990) can be discussed along two lines: (a) the contribution of the study of assertions relative to that of the study of requests, and (b) the types of conversational routines involved in reformulation. With respect to requests, what do assertions have to offer to our understanding of reformulations by young children? According to speech act theory (see Searle and
H. Marcos, J. Bernicot / Journal of Pragmatics 27 (1997) 781-798
783
Vanderveken, 1985), via a request, a speaker produces a statement aimed at making the partner accomplish some future action. The satisfaction of the request implies the accomplishment of the action by the partner. Note that we are dealing here with cases where the partner can choose to accomplish or not to accomplish the action, by opposition to cases where the situation constrains the partner's behavior (see Vanderveken, 1990a,b). Via an assertion, on the other hand, a speaker makes a statement for the purpose of representing the way things are in the world. Satisfying an assertion implies agreeing that the statement made is true. In dyadic, asymmetrical child-adult communication, the. child's goal is to produce a true statement with which the adult will agree (see Vanderveken, 1990a,b). The assertion is satisfied if the adult explicitly recognizes the truth of the statement made. From the child's standpoint, the satisfaction of requests, like that of assertions, is manifested by an overt reaction: the accomplishment of an action for the former and the explicit approval of an utterance for the latter. The fundamental distinction between requests and assertions lies in the direction of the adjustment: with requests, the world is made to conform to the words, whereas with assertions, the words are made to conform to the world. Although requests can be achieved solely via gestural messages (Bates et al., 1979; Bruner et al., 1982; Marcos, 1991; Marcos and Kornhaber-Le Chanu, 1992), asserting cannot be done without the linguistic messages needed to describe the world. For the child, then, it can be assumed that the experience of assertion satisfaction (adult approval of the truth value of the message) begins when the production of linguistic messages begins; this comes after the satisfaction of requests, which can be produced as soon as gestural messages are possible. What types of conversational routines underlie reformulations? Regarding requests, prior studies have shown that very young children are able to vary their reformulations according to the content of the request (Garvey, 1977; Bloom et al., 1977; Johnson, 1980) and to the characteristics of their conversational partner (Ervin-Tripp et al., 1982). The partner's behavior is an important factor. Marcos (1991) and Marcos and Kornhaber-Le Chanu (1992) showed that in cases where the adult ignores the initial message or asks the child to clarify, children as early as 1 ;6 adapt their request reformulations to the behavior of the adult. Marcos and Bernicot (1994) found similar results with children between the ages of 1 ;6 and 2;6 in situations where the adu]tt either requested a clarification or explicitly refused to satisfy a request; the linguistic productions of 2;6-year-old children obviously outnumber those of 1;6 year-olds. As a whole, these results support Golinkoff's (1986, 1993) position. However, the highest level of understanding of the other person's mental state is obviously not reached by the age of 2;6. This level exists outside the communication situation, and corresponds to the child's explanation of an intention explicitly attributed to a partner. A number of studies which have tested this level using a preliminary or revised version of the false belief paradigm 1 The false belief paradigm, which requires only simple materials (dolls and everyday objects), can be used to determine whether a child can attribute a mental state that differs from objective reality, to a character who has a different understanding of reality. It can also be used to see if the child can attribute differing mental states to various characters with different understandings of reality.
784
H. Marcos, J. Bernicot / Journal of Pra~matics 27 (1997) 781-798
(Hogrefe et al., 1986; Wellman and Bartsch, 1988; Wellman and Wolley, 1990; Wimmer and Perner, 1983) have established that the age of mental state comprehension is approximately 4;0 years. The way young children reformulate requests, a testimony to their ability to finely adapt to the situation, cannot be explained by the existence of rigid conversational routines either. This is the case described by Shatz (1983) in which children regard their partner (e.g. their mother) as a magical instrument activated by their message and acting on their behalf. The notion of conversational routine nevertheless appears to be an interesting concept for studying reformulation in young children. Indeed, conversational routines are typical exemplars of interaction situations, and nearly all reformulation situations belong to this category. In daily life, an adult interacting with a child says yes, says no, misunderstands, is reluctant, etc. The study by Halle and Shatz (1994) describes and quantifies the different linguistic behaviors exhibited by mothers addressing children between the ages of two and five. The routines appear to vary in complexity and richness. As Beruicot (1994) suggests for 2-year-olds, conversational routines are initially quite rudimentary and rigid, then gradually become richer through the addition of new elements which allow children to adapt with increasing flexibility to the situations of communication with which they are faced. The information provided by studies on these routines primarily concerns requests. In order to assess the richness of the routines available to children, it is important to find out whether and to what extent this kind of subtle adaptation to the partner's behavior also exists for reformulations of a speech act with features unlike those of requests. As already stressed, assertions differ from requests as to the direction of the adjustment and to the necessary use of the linguistic channel. Accordingly, the study of how 2;6-year-olds reformulate assertions should provide us with some arguments supporting our three-part hypothesis concerning routine flexibility: (1) Children should not reformulate any less for assertions than for requests. (2) The form of the reformulations should vary with the partner's behavior for both requests and assertions. (3) The variations observed for assertions should differ, at least in part, from those observed for requests. To this end, we conducted a finer comparison of the reformulation of requests and assertions. In order to control the type of clarification request produced by the adult, as well as add a refusal situation, an experimental rather than natural situation was used. Unlike Shatz and Watson O'Reilly (1990), who combined all reformulations regardless of the type of request for clarification formulated by the adult, only one kind of request for clarification was possible in our experiment: an unspecific request for repetition.
2. Method 2.1. Subjects
Two groups of twelve native French-speaking children aged 30 months (range: 29 to 31 months) participated in the experiment. The first group, composed of 5 boys
785
H. Marcos, J. Bernicot / Journal of Pragmatics 27 (1997) 781-798
and 7 girls (mean age: 30 months and 23 days), performed a task involving requests. The second group, composed of 6 boys and 6 girls (mean age: 30 months and 18 days), performed a task involving assertions. Using independent groups allowed us to avoid an important risk in ve:ry young children, that of the influence of one task on another. Moreover, this reduced the amount of testing time for each child. All of the children were from middle-class families. They were recruited and observed in day-care centers in greater Paris and Poitiers (France). The second selection criterion (after age) was that the subjects had to fluently produce utterances of at least two words. 2 The experimenter (the child',; conversational partner) was a young female psychologist from outside the day-care center. She spent two half-days at the center before the experiment in order to become acquainted with the children, and consequently, was neither entirely unknown nor highly familiar. The experimenter checked each child's language level during this acquaintance period. 2.2. Materials
For the requests, a wooden puzzle depicting common farm animals was used. The puzzle, shown in Fig. 1, consisted of nine items (pieces to insert).
dog
duck and flog
pig
scarecrow
sheep goat
horse
hen and chicks
COW
Fig. 1. Nine-piece insertion puzzle (the exact shape of the pieces is not reproduced). 2 The utterancecollected in the present experimentexhibited that children possessed a minimum of linguistic skills: mean MLU was 2.82.
786
H. Mareos. J. Bernicot / Journal of Pragmatics 27 (1997) 781-798
F o r the assertions, three b o o k s about the adventures o f the same character (Titi the m o u s e ) were used. Each b o o k had a different theme (illness, vacation, and painting) in which Titi was interacting with one o f his parents. Three items (pictures) from each b o o k were presented to the children, m a k i n g nine items in all. A given item covered the left- and right-hand pages o f the o p e n e d b o o k and d e p i c t e d several objects and two characters (Titi the m o u s e and one o f his parents). The basic structure o f the items is shown in Table 1. F o r each picture, there were two potential agents (Titi and his partner) and several potential actions, c o r r e s p o n d i n g to the various objects shown in the picture. Table 1 Structure of each item in the book and list of 9 items (3 items per theme) Left-hand page
Right-hand page
Character A and/or B accomplishing action A
Character A and/or B accomplishing action B
Theme: Illness
Titi is lying in bed Titi is lying in bed, his mother is giving him something to drink Titi is lying in bed, his mother is playing cards with him
Titi's mother is bringing him a pillow Titi is lying in bed, his mother is telling him a story Titi is lying in bed, his mother is showing him the moon
Theme: Vacation
Titi is standing up with a knapsack Titi and his mother are checking what's inside the knapsack Titi finishes filling up his knapsack (with a canteen and some candy)
His mother is preparing his clothes Titi is filling up his knapsack (with his boots) Titi is lifting up his knapsack and Titi's father is lifting up a big suitcase
Theme: Painting
Titi is starting a picture Titi is finishing the picture with other colors, he spills the glass of water Titi is finishing his picture
Titi is continuing the picture with his mother who is holding out a box of paints Titi is choosing the color in the box his mother is holding It's Titi's mother who is holding the paintbrush, Titi is holding the box of paints
2.3. Procedure
Subjects were tested individually in a separate r o o m in the d a y - c a r e center. The adult c o n d u c t e d a p r e l i m i n a r y trial to show the child h o w to p l a y the game. G i v e n the y o u n g age o f these children, the instructions were given v e r b a l l y and non-verbally. A t the same time as it initiated the e x p e r i m e n t a l phase proper, this trial served to familiarize the child with the situation as a whole, including the c a m e r a on a tripod located nearby.
H. Marcos, J. Bernicoi' / Journal of Pragmatics 27 (1997) 781-798
787
For the requests, the task was. to complete the puzzle, the adult and child were seated face-to-face in front of a table where the puzzle was placed. Although the child was able to touch the puzzle board, the adult had the pieces, so that in order to play, the child had to 'ask' the adult for them. For the assertions, the experimenter presented the characters in the story and had the child find the theme; the adult and the child were seated side-by-side in front of a table 3 where the book was placed. Then for each picture, the child had to answer the question, 'What are they doing?' By responding to this question, the child produced an assertion. In each case, the adult responded to the child's request (assertion) by exhibiting one of the following three behaviors. The adult's behavior defined the three experimental situations. Situation of satisfaction. For the request task, the adult immediately satisfied the request. For the assertion task, the adult agreed with the child's utterance by saying 'Yes, he did that' or 'Yes, they did that'. Situation of clarification. For the request task, the adult did not immediately satisfy the request and asked the clarification question: 'What do you want?'. For the assertion task, the adult did not immediately agree with the assertions and asked the clarification question: 'What did you say?'. Situation of refusal. For the request task, the adult refused to satisfy the request and said: 'No no, I won't give it to you ... I'm keeping it for myself'. For the assertion task, the adult rejected the child's assertion and said: 'No, he's not doing that' or 'No, they're not doing that'. The satisfaction situation fulfilled a dual function. Firstly, it provided a setting for ecological interaction between the experimenter and the child. Secondly, the behaviors noted in this situation were regarded as representative of the child's initial productions, for both requests and assertions. During the testing, the criteria for considering the child's behavior as a request were whether it corresponded to a conventional way of expressing a request, and how quickly the child took the object when the adult held it out. If after approximately five seconds the child had not produced a request, the experimenter gave him/her an item so that the game could continue. For assertions, the decision was based on whether the child's behavior corresponded to a conventional way of making an assertion. If after approximately five seconds the child had not produced an assertion, the experimenter went on to the next item so as not to interrupt the interaction and allow the trial to corttinue. Of course, the number of conventional nonverbal expressions for requesting is potentially greater than for asserting. Three items were presented per situation (satisfaction, clarification, and refusal). For requests, the order of the items and the order of the situations were varied randomly across subjects. For assertions, the order of the books and the order of the situations were varied randomly across subjects. The three items in each book (theme) 3 The types of speech acts studied and the need for a procedure that would be suited to young children necessitated two spatial relations (face-to-face for requests and side-by-side for assertions). Prior experiments and data have shown that in these two cases, children call upon the entire range of nonverbal behaviors with the same ease.
788
H. Marcos, J. Bernicot / Journal of Pragmatics 27 (1997) 781-798
were presented in succession for each speech act and each child. The entire sequence lasted between five and seven minutes and was recorded on videotape.
3. Data coding The gestural and linguistic behaviors were coded from the videotapes. For at least 25% of the corpus for each speech act, the inter-coder agreement rate 4 was above 0.85. The first coder was the experimenter. The second was a student who was familiar with the coding system but was unaware of the objective of the experiment. Inter-coder agreement was achieved whenever the same category was attributed to a behavior. In cases of coder disagreement, a discussion was held to determine which category to choose. Two types of coding were performed. 3.1. F i r s t t y p e o f c o d i n g
The first type of coding concerned the gestural, linguistic, and visual behaviors manifested during the initial production in the satisfaction situation, and during the reformulation in the clarification and refusal situations. The behaviors were classified using the following categories. N o n - r e s p o n s e s . In the satisfaction situation, for both requests and assertions, the child did not produce a message (whether gestural or linguistic) during the five seconds that followed the item change. In the clarification and refusal situations, again for both speech acts, the child did not produce a message (whether gestural or linguistic) during the five seconds that followed the experimenter's response to his or her initial message. G e s t u r a l p r o d u c t i o n s . For requests, we noted arm and hand movements (forward and backward arm movements, other arm movements, hand movements, pointing, handling of the puzzle board) and body movements (forward and backward). For assertions, pointing was the most common behavior (pointing to left side of picture, pointing to right side of picture). L i n g u i s t i c p r o d u c t i o n s . Vocalizations like 'Eu', 'Ta', etc., which were extremely rare, were not analyzed. In the request task, the requests were classified into the four conventional categories: direct orders ('Give me that'), 5 truncated imperatives ('The cow'), expressions of a personal desire or need ('I want the cow'), and allusions ('Oh, the cow!'). We added the 'insisting' category ('But yes!'), observed in the 4 The inter-coder agreement rate was the ratio of the number of identical ratings to the total number of ratings. 5 The initial French forms of the examples were as follows: 'Give me that': Donne-moi qa - 'The cow': La rathe - 'I want the cow': Je veux la rathe - 'Oh, the cow! ': Oh, la rathe, t. For 'But yes', the French word used by children was Si?, a form of 'yes' used to oppose or refute a negative statement, achieved in English via intonation (e.g. "He's not sleeping', 'Yes he is'). 'Painting': p e i n t u r e - 'He's drinking his milk': II boit son l a i t - 'They're going on vacation': Ils vont en vacances 'Look. he packed his suitcase': Regarde, il a pr~pard sa valise.
H. Marcos. J. B e r n i c o t / J o u r n a l o f P r a g m a t i c s 2 7 (1997) 7 8 1 - 7 9 8
789
refusal situation. In the assertion task, the children produced linguistic forms which varied in degree of explicitness and elaboration. Some examples are 'Painting', ' H e ' s drinking his milk', ' T h e y ' r e going on vacation' and 'Look, he packed his suitcase'. Direction o f g a z e . For requests, the child looked at the item, the adult, the board, or elsewhere. For assertions, the child directed his/her gaze towards the left-hand page or the right-hand page, visually explored the picture, or looked at the adult or elsewhere. Looking elsewhere was very rare. Gazes were noted in all cases, even when no gestural or linguistic response was produced. The productions considered were the ones that occurred between the beginning of the child's message and the beginning of the adult's reaction. A request or assertion lasted at most four seconds in all situations. For each category (gestural productions, linguistic productions, and direction of gaze), the occurrence of a production was noted, irrespective of its duration and regardless of whether the productions took place simultaneously or in succession. 3.2. Second
type of coding
The second type of coding was designed to enable a finer analysis of the linguistic productions for each type of request in the clarification and refusal situations. It was designed to reflect the change made by the child between the initial request and its reformulation. Only the linguistic components of the children's productions were considered. Five categories of reformulations were distinguished. Non-linguistic reformulation. For requests and assertions alike, there was no speech accompanying the child's reformulation. Repetition. The request (or assertion) was simply repeated. For example, for requests, the child repeated 'that '6 or 'the horse'. For assertions, the child repeated ' T h e y ' r e painting' or ' T h e y ' r e going on vacation'. C h a n g e o f r e f e r e n t . For requests, the child requested a different object. For example 'the cow' became 'the goat'. For assertions, the child changed the character, the action, or the object to which the assertion referred. For example 'painting' became 'Mrs. Mouse' or 'going to bed '7 became ' S h e ' s bringing his pillow'. Further specification o f r e f e r e n t . For requests, the child provided further specifications about the referent of the request. For example, 'There' became 'There, there,
6 The initial French forms of the examples were as follows: "That': ~'a - 'The horse': L e c h e v a l 'They're painting': lls f o n t d'la p e i n t u r c - 'They're going on vacation': Ils vont en vacances. 'The cow': La r a t h e - "The goat': Le rpouton - . 'Painting': P e i n t u r e - 'Mrs. Mouse': M a m a n Souris 'Going to bed': F a i s dodo - 'She's bringing his pillow': E/le a m d n e son coussin. 'There': LtJ - 'There, there, there': L~i./d. hJ - 'Another': Un autre - 'That one' - celui Igt - "That one there': Un autre l~ - 'G-night': D o d o -- 'Night, night, night': Dodo, dodo. dodo - "They're doing the paper': lls f o n t les p a p i e r s - "It's green': C ' e s t vert. 'But yes': Si - "Want more': V e u x enco~e - "The horse': Le c h e v a l - 'Look': R e g a r d e "They're sleeping' : lls f o n t d o d o - 'They're sleeping, look' : I/s f o n t dodo, r'garde. 7 D o d o is a French form used by the children to designate the action of sleeping, but also elements of the action as bed, bedroom, etc.
790
H. Marcos, J. Bernicot / Journal of Pragmatics 27 (1997) 781-798
there' or 'Another' became 'That one' or 'That one there'. For assertions, the child further specified the character, action, or object to which his/her assertion referred. For example 'G-night' became 'Night, night, night' or 'They're doing the paper' became 'It's green'. Reinforcement of intention. For requests, the child maintained and reinforced his/her intention by opposing the partner's behavior with the expression 'But yes'. This category only appeared in the refusal situation. For example 'want more' became 'But yes' or 'the horse' became 'But yes'. For assertions, the child maintained and reinforced his/her intention to assert by attracting the partner's attention with the term 'look'. For example 'They're sleeping' became 'They're sleeping, look' or 'Titi' became 'Look'.
4. Analysis of results As above for the coding, the data were analyzed twice. The first analysis concerned the gestural, linguistic, and visual behaviors manifested during the initial production in the satisfaction situation, and during reformulation in the clarification and refusal situations. Remember that the behaviors obtained in the satisfaction situation were taken to represent the children's initial productions, for requests as well as for assertions. The second analysis dealt with the changes made by the child between the initial request or assertion and its reformulation. Only the linguistic components of the children's productions were considered. Obviously, this second analysis only concerned the clarification and refusal situations. For each child and each act (request or assertion) in each situation (satisfaction, clarification, or refusal), the total number of possible responses was three. 4.1. Gestural, linguistic, and visual behaviors
For each speech act and situation, Table 2 gives the number of non-responses 8 and responses, and within the responses, the number of gestural and linguistic behaviors. Three types of productions could be distinguished: those consisting solely of gestural elements (gestural productions: G), those combining gestural and linguistic elements (gestural-and-linguistic productions: G+L), and those consisting solely of linguistic elements (linguistic productions: L). The total number of non-responses was very low (approximately 0.50 out of a possible 9). For both speech acts, responses combining both gestural and linguistic behaviors largely outnumbered the others (approximately 5 out of a possible 9). For each of the four types of productions (non-responses, gestural, gestural-andlinguistic, and linguistic), the variation in the mean number of responses per subject was tested using an analysis of variance with two factors: Type-of-act (2) × Situation (3). Non-responses were cases where a gestural or linguisticresponse was lacking: the children made no gestures and said nothing.
H. Marcos, J. Bernicot / Journal of Pragmatics 27 (1997) 781-798
791
Non-responses. No significant effects were obtained. As a whole, the number of non-responses was very low. Gestural productions and linguistic productions. For gestural productions, the overall analysis yielded a significant effect of the Type-of-act factor (F(1,22)=4.45, p<0.05). The number of gestural productions was greater for requests than for assertions. The significant interaction between the Type-of-act and the Situation (F(2,44)=4.96, p<0.01) and subsequent partial comparisons indicated a significant difference between requests and assertions in satisfaction (F(1,22)=7.33, p<0.01) and in refusal (F(1,22)=4.76, p
Respo~ases
Gestural behaviors
Gestural-andlinguistic behaviors
Linguistic behaviors
0.000 0.083 0.250 0.333
3.000 2.917 2.749 8.667
1.000 0.417 1.083 2.5
2.000 2.500 1.5 6.000
0.000 0.000 0.167 0.167
0.167 0.083 0.333 0.583
2.833 2.917 2.666 8.416
0.000 0.333 0.333 0.666
1.583 1.917 1.333 4.82
1.250 0.667 1.000 2.916
Requests Satisfaction Clarification Refusal Total
Assertions Satisfaction Clarification Refusal Total
Direction of gaze. Two gaze directions were analyzed, fixed gazes and changing gazes. Fixed gazes were directed solely towards the requested object for requests, and solely towards a specific point in the book for assertions. Changing gazes alter-
792
H. Marcos, J. Bernicot / Journal of Pragmatics 27 (1997) 781-798
nated between the object and the partner for requests, and between the book and the partner or between two locations in the book for assertions. For each speech act and situation, Table 3 gives the number of fixed and changing gazes. The sum of the two types of gazes was close to 9 in all cases (maximum number of possible responses). This finding confirms the fact that child gazes not directed at the adult or the focal object were extremely rare, for both requests and assertions. Table 3 Per-subject mean of the number of responses exhibiting each type of gaze (fixed and changing), by speech act and situation. Maximum number of responses = 3. Fixed gaze
Changing gaze
Total
1.833 1.583 1.083 4.499
I. 167 1.333 1.667 4.167
3.000 2.916 2.750 8.666
1.333 2.167 1.583 5.083
1.667 0.583 1.417 3.667
3.000 2.750 3.000 8.750
Requests Satisfaction Clarification Refusal Total
Assertions Satisfaction Clarification Refusal Total
For each type of gaze, the variation in the mean number of responses per subject was tested using an analysis of variance with two factors: Type-of-act (2) x Situation (3). For fixed gazes, the overall analysis indicated a significant effect of the Situation factor (F(2,44)=3.51, p<0.04 and an interaction between the Type-of-act and the Situation (F(2,44)=4.33, p<0.02). The number of fixed gazes decreased as the situation went from clarification to satisfaction to refusal. For both requests and assertions, the number of fixed gazes was greater in clarification than in refusal. It was the number of fixed gazes in satisfaction that differentiated the two speech acts: in requests, this number was higher than in the other two situations, and in assertions, it was lower. For changing gazes, the overall analysis yielded a significant effect of the Situation factor (F(2,44)=3.92, p<0.03) and an interaction effect between the Type-of-act and the Situation (F(2,44)=4.11, p<0.02). The number of changing gazes decreased as the situation went from refusal to satisfaction to clarification. For both requests and assertions, the number of changes in direction of gaze was greater in refusal than in clarification. The number of changing gazes in satisfaction is what differentiated the two speech acts: for requests, this number was lower than in the other two situations, while for assertions it was higher.
4.2. Type of linguistic" reformulation For the clarification and refusal situations, five reformulation categories were considered: non-reformulation, repetition, referent changes, object specification, and
H. Marcos, J. Bernicot / Journal of Pragmatics 27 (1997) 781-798
793
intention reinforcement. For each category, the variation in the mean number of responses per subject was tested using an analysis of variance with two factors: Type-of-act (2) x Situation (2). Table 4 presents the results for non-linguistic reformulations, repetitions, referent changes, referent specifications, and intention reinforcements. Tables 3 and 4 together clearly show that most of the children's reformulations contained linguistic elements; compared with reformulations containing only gestural elements, the ratio was 1 to 3 for requests and 1 to 7 for assertions. For both speech acts, the referent specification category was the most frequent. For requests, there were many repetitions, although the number of referent changes and intention reinforcements was low. For assertions, the repetition, referent-change, and intention-reinforcement categories occurred in equal proportions, representing together half of the occurrences of the predominant category, referent specifications. Non-linguistic reformulations. The overall analysis indicated a nonsignificant effect for the Type-of-act factor, a significant Situation effect (F(1,22)=14.47, p<0.001), and a significant interaction effect between the two (F(1,22)=5.21, p<0.03). The number of non-reformulations was greater in refusal than in clarification. This situation-linked difference was more pronounced in requests than in assertions, where it is very small. Moreover in clarification, there was virtually no difference between the non-reformulations of requests and assertions, whereas this difference did exist in the refusal situation (and was much larger for requests). Repetitions. The overall analysis yielded a significant effect of the Situation factor (F(1,22)=11.08, p<0.003) and an interaction between the Type-of-act and the Situation (F(1,22)=4.33, p<0.05). The number of repetitions in the clarification situation outnumbered those in refltsal, and this difference was more pronounced for requests than for assertions. In addition, there was virtually no difference between the number of repetitions of requests and assertions in refusal, while this difference did exist in clarification (and was much larger for requests). Referent changes. The overall analysis yielded a significant effect of the Type-ofact factor (F( 1,22)=24.20, p<0.0001 ), the Situation factor (F( 1,22)= 14.01, p<0.001 ), and the interaction between the two (F(1,22)=5.67, p<0.03). The number of reformulations of the change-of-referent type was greater for assertions than for requests, and greater in refusal than in clarification. Moreover, since this type of response was very rare for requests, the difference between refusal and clarification was more marked in assertion. Referent specifications. Only the Situation effect was significant (F(1,22)=9.32, p<0.006). Referent specifications were more numerous in clarification than in refusal. There was no difference between requests and assertions. Intention reinforcements. The only significant effect was the interaction between the Type-of-act and Situation factors (F(1,22)=8.60, p<0.008). For requests, intention reinforcements in refusal outnumbered those in clarification (where they did not occur), whereas this difference did not exist for assertions (an opposing, but nonsignificant tendency was obser~ ed).
H. Marcos, J. Bernicot / Journal of Pragmatics 27 (1997) 781-798
794
Table 4 Per-subject mean of the number of responses involving non-linguistic reformulation or linguistic reformulation via repetition, referent change, referent specification, or intention reinforcement by speech act and situation (clarification and refusal). Maximum number of responses = 3. The number of linguistic reformulations is equal to the sum of the number of repetitions, referent changes, referent specifications, and intention reinforcements.
Requests Clarification Refusal Total
Non-linguistic
Linguistic
Repetition
Referent change
Referent specification
Intention reinforcement
0.500 1.333 1.833
2.500 1.667 4.167
1.250 0.167 1.417
0.000 0.167 0.167
1.250 0.5000 1.750
0.000 0.833 0.833
1.250 0.667 1.917
0.583 0.417 1.000
Maximum number of reformulations: 6 Total number of reformulations: 5.666 Total number of purely-gestural reformulations: 1.500 Total number of reformulations containing linguistic elements: 4.167
Assertions Clarification Refusal Total
0.417 0.667 1.084
2.583 2.334 4.917
0.583 0.333 0.916
0.167 0.917 1.084
Maximum number of reformulations: 6 Total number of reformulations: 5.583 Total number of purely-gestural reformulations :0.666 Total number of reformulations containing linguistic elements:4.917
5. Discussion The results obtained here shed some light on the emergence of communication skills in young children. The subtle adaptation to the partner's behavior observed in these 2;6-year-olds during request reformulation raises the question of what social knowledge underlies reformulation. We hypothesized that social knowledge of this type is defined by flexible conversational routines. A comparison between request reformulation and assertion reformulation, which differ in adjustment direction, communication channel, and the age at which experience as an interaction partner begins, supported our hypotheses: (1) these 2; 6-year-old children did not reformulate their assertions any less than their requests, (2) the form of the reformulations varied with the partner's behavior for both speech acts, and (3) the variations observed for the assertions were partially different from those observed for requests. For these 2;6-year-old children as a whole, the number of non-responses was very low for assertions as well as for requests. There was no significant difference between the two speech acts, whether both gestural and linguistic behaviors or only linguistic behaviors were considered. For both acts, non-linguistic reformulations were more numerous in refusal situations than in situations of clarification; this difference was more pronounced for requests than for assertions. Note that for refor-
H. Marcos, J. Bernicot / Journal of Pragmatics 27 (1997) 781-798
795
mulation via repetition - an unelaborated form of reformulation - there was still no significant difference between assertion and request. For both acts, repetitions were more numerous in clarification than in refusal, and this difference was greater for requests than for assertions. How can one account for the discrepancy between these results and those obtained by Shatz and Watson O'Reilly (1990), who found that in situations of clarification, children linguistically reformulate significantly more requests than assertions? One possible explanation might be related to the processing of the data. As mentioned above, the Shatz and Watson O'Reilly (1990) study pooled children's reformulations following all types of adult requests for clarification. In our study, the requests for clarification were always unspecific requests for repetition ('What do you want' or 'What did you say?'). Another possible explanation is that Shatz and Watson O'Reilly's (1990) data were taken from a natural setting, while ours were collected in an experimental situation. The more constraining experimental conditions may have guided the children's behavior and promoted assertion reformulation. This argument is difficult to defend, however, insofar as it implies that the experimental situation would have also favored the reformulation of requests. The most important point here appears to be that in the Shatz and Watson O'Reilly (1990) study and in our own, the reformulation rate of 2;6-year-old children is very high (more than 75%) for both assertions and requests. Children's reformulations of assertions and requests are adapted to the partner's behavior. For some of our measures, these variations differed, while for others they did not. In accordance with the fact that asserting, unlike requesting, implies the use of the linguistic channel, the children produced more purely-gestural behaviors for requests than for assertions. Likewise, they produced more solely-linguistic behaviors for assertions than for requests. For requests, gestural elements were not as prevalent in clarification as they were in the: other two situations, and for assertions, linguistic elements were the less prevalent ones in these same situations. The two speech acts had one point in common: productions combining gestural and linguistic elements were more numerous in clarification. The behavior which distinguished assertions from requests was the children's direction of gaze in situations of satisfaction. Compared to the other two situations, requests in the satisfaction situation were characterized by fixed gazes (at the object), while assertions were characterized by changing gazes (alternating between the partner and the object in 60.:23% of the cases and visual exploration of different parts of the object in 39.76% of the cases). Gazes which alternated between the partner and the object certainly reflect verification of the visual locus of the partner's attention, whereas gazes which explored the object no doubt reflect information intake from the materials. One of our results was found for both speech acts: the number of fixed gazes was grewEer in clarification than in refusal, and inversely, the number of changing gazes was greater in refusal than in clarification. As far as the different types of linguistic reformulations are concerned, we found both differences and similarities between assertions and requests. Reformulations of the changing-referent type were characteristic of assertions in refusal, but occurred
796
H. Marcos, J. Bernicot / Journal of Pragmatics 27 (1997) 781-798
rarely in requests. For assertions, the children seem to have been seeking to equate their own referent with the assumed referent of the adult. Reformulations of the intention-reinforcement type, which consisted for the child of contesting the adult's behavior, occurred in refusal only for requests, while they appeared in clarification and refusal for assertions. For both acts, there were many reformulations of the referent-specification type, and such reformulations were always more numerous in clarification than in refusal. Our results provide clear evidence that children at the age of two-and-a-half reformulate as many assertions as requests, and that adaptation to the partner's behavior is just as subtle for assertions as it is for requests. They also pointed out that the various means used to adapt when asserting, differ in some respects from those used when requesting. Thus, children appear to differentiate between the two speech acts and to act accordingly. These data are consistent with our point of view, which lies halfway between the positions of Shatz and Watson O'Reilly (1990) and Golinkoff (1986, 1993). In our minds, the reformulations of 2;6-year-old children are underpinned by conversational routines which are neither rigid, nor simple, nor automatic. On the contrary, they are characterized by their richness, their flexibility, and their complexity: the fine variations observed no doubt testify to the non-global, multielement nature of these routines. However, at the age of 2;6, children do not have social knowledge of a 'metacognitive' nature which would allow them to render explicit the mental states of their conversational partner. This knowledge can be viewed as 'implicit' knowledge (Clements and Perner, 1994) or 'practical' knowledge (Dunn, 1988); it can only be implemented in a real setting. When the need to communicate arises, the situation guides the child's behavior, and that behavior is tied to flexible and complex conversational routines. We can conclude that for children at the age of two-and-a-half, the 'meeting of minds' occurs through action.
References Anselmi, Dina, Michael Tomasello and Mary Acunzo, 1986. Young children's responses to neutral and specific contingent queries. Journal of Child Language 13: 135-144. Astington, Janet W. and Alison Gopnik, 1991. Developing understanding of desire and intention. In: A. Whithen, ed., Natural theories of mind, 39-50. Oxford: Blackwell. Baroni, Maria R. and Giovanna Axia, 1989. Children's meta-pragmatic abilities and the identification of polite and impolite request. First Language 9: 285-297. Bates, Elizabeth, Laura Benigni, Inge Bretherton, Luigia Camaioni and Virginia Volterra, 1979. The emergence of symbols. New York: Academic Press. Bemicot, Josie, 1991. French children's conception of requesting: the development of metapragmatic knowledge. International Journal of Behavioral Development 14: 285-304. Bernicot, Josie, 1994. Speech acts in young children: Vygotsky's contribution. European Journal of Psychology of Education 9:311-320. Bernicot, Josie and Virginie Laval, 1996. Promises in French children: Comprehension and metapragmatic knowledge. Journal of Pragmatics 25: 101-122. Bernicot, Josie and Suzanne Legros, 1987. Direct and indirect directives: What do young children understand? Journal of Experimental Child Psychology 43: 346-358. Bloom, Lois, Lorraine Rocissano and Lois Hood, 1977. Adult-child discourse: Developmental interaction between information processing and linguistic knowledge. Cognitive Psychology 8: 521-552.
H. Marcos, J. Bernic6t / Journal of Pragmatics 27 (1997) 781-798
797
Blum-Kuika, Shoshana, 1990. You don't touch lettuce with your fingers: Parental politeness in family discourse. Journal of Pragmatics 14: :259-288. Blum-Kulka, Shoshana, Juliane House and Gabriele Kasper, 1989. Cross-cultural pragmatics: Requests and apologies. Norwood, NJ: Ablex. Bruner, Jerome S., Carolyn Roy and Nancy Ratner, 1982. The beginnings of request. In: K.E. Nelson, ed., Children's language, Vol. 3, 91-138. New York: Garden Press. Dore, John, 1975. Holophrases, speech acts, and language universals. Journal of Child Language 2: 21-40. Clements, Wendy A. and Josef Perner, 1994. Implicit understanding of belief. Cognitive Psychology 9: 377-395. Dunn, Judy, 1988. The beginnings of sc,cial understanding. Cambridge, MA: Harvard University Press. Ervin-Tripp, Susan, Jiansheng Guo and Martin Lampert, 1990. Politeness and persuasion in children's control acts. Journal of Pragmatics 14.: 307-331. Ervin-Tripp, Susan, Mary C. O'Connor and Jarret Rosenberg, 1982. Language and power in family. In: C. Kramerae and M. Schultz, eds., Language and power, 117-135. Urbana-Champaign, IL: University of Illinois Press. Garcia, Carmen, 1993. Making a request and responding to it: A case study of Peruvian Spanish speakers. Journal of Pragmatics 19: 127-1:52. Garton, Alison F. and Chris Pratt, 1990. Children's pragmatic judgements of direct and indirect requests. First Language 10: 51-59. Garvey, Catherine, 1977. The contingent query: A dependent act in conversation. In: M. Lewis and L. Rosenblum, eds., Interaction, conversation and the development of language, 63-94. New York: Wiley. Golinkoff, Roberta M., 1986. The prew~rbal negotiation of failed messages. Journal of Child Language 13: 455-467. Golinkoff, Roberta M., 1993. When is communication a 'meeting of minds'? Journal of Child Language 3, 199-209. Halle, Tamara and Marilyn Shatz, 1994. Mother's social regulatory language. First Language 14: 83-104. Hogrefe, G. Juergen, Heinz Wimmer arid Josef Perner, 1986. Ignorance versus false belief: A developmental lag in attribution of epistemic states. Child Development 57: 367-382. Houtkoop-Steenstra, Hanneke, 1990. Accounting for proposals. Journal of Pragmatics 14:111-124. Johnson, C.E., 1980. Contingent query: The first chapter. In: H. Giles, W.P. Robinson and P.M. Smith, eds., Language: Social psychological perspectives, 11-19. Oxford: Pergamon Press. Koike, Dale A., 1994. Negation in Spanish and English suggestions and requests: Mitigating effects? Journal of Pragmatics 21 : 513-526. Littman, David C. and Jacob L. Mey, 1991. The nature of irony: Toward a computational model of irony. Journal of Pragmatics 15: 131-151. Marcos, Hayd6e, 1991. Reformulation requests at 18 months: Gestures, vocalizations and words. First Language 11 : 361-375. Marcos, Hayd6e and Josie Bernicot, 1~'94. Addressee co-operation and request reformulation in young children. Journal of Child Language 21 : 677~92. Marcos, Hayd6e and Mila Komhaber-ke Chanu, 1992. Learning how to insist and how to clarify in the second year. International Journal of Behavioral Development 3: 359-377. Place, Karen S. and Judith A. Becker, 1991. The influence of pragmatic competence on the likeability of grade-school children. Discourse Processes 14: 227-241. Searle, John R. and Daniel Vanderveken, 1985. Foundations of illocutionary logic. Cambridge: Cambridge University Press. Shatz, Marilyn, 1983. Communication. In: J.F. Flavell and E. Markman, eds., Cognitive development. In: P. Mussen, ed., Carmichael's m~mual of child psychology, 841-889. Fourth Edition, New York: Wiley. Shatz, Marilyn and Anne Watson O'Reilly, 1990. Conversational or communicative skill? A reassessment of two-year-olds' behaviour in miscommunication episodes. Journal of Child Language 17: 131-146.
798
H. Marcos, J. Bernicot / Journal of Pragmatics 27 (1997) 781-798
Sinclair, Anne and Ruth Van Gessel, 1990. The form and function of questions in children's conversation. Journal of Pragmatics 14: 923-944. Vanderveken, Daniel, 1990a. Meaning and speech acts. Vol. 1: Principles of language use. Cambridge: Cambridge University Press. Vanderveken, Daniel, 1990b. Meaning and speech acts. Vol. 2: Formal semantics of success and satisfaction. Cambridge: Cambridge University Press. Wellman, Henry M., 1990. The child's theory of mind. Cambridge, MA: Bradford Books and MIT Press. Wellman, Henry M. and Karen Bartsch, 1988. Young children's reasoning about belief. Cognition 30: 239-277. Wellman, Henry M. and Jacqueline D. Wotley, 1990. From simple desires to ordinary belief: The early development of every day psychology. Cognition 35: 245-275. Wimmer, Heinz and Josef Perner, 1983. Beliefs about beliefs: Representation and constraining function of wrong beliefs in young children's understanding of deception. Cognition 13:103-128.