CHILDREN DO ASK, BUT DO NOT KNOW HOW TO DO ASKING: EPI-PRAGMATIC VS. META-PRAGMATIC SKILL1

Old fi ndings on children’s comprehension of ask and tell were subject to different interpretations refl ecting progress in the fi eld of language acquisition. We want to show that acquiring a particular skill does not necessarily include competence of its intentional control and use. Development of linguistic skills takes place at different levels starting from early spontaneous, implicit abilities to the level of meta-pragmatic refl exive knowledge that enables deliberate monitoring, planning, and practice. The present study was aimed at exploring two extreme points in development: early epi-pragmatic and late refl exive metapragmatic competence. The fi rst part aims at fi nding the earliest instances of children spontaneous ability to pass ask-instructions, and the evidence is provided for the ages as early as 22 to 40 months (much earlier than recorded in the previous studies). The second part is experimental and focuses on children’s ability to respond to askand tell-instructions in the context of a cancelled conversational rule (Gricean Maxim of Quantity) which requires deliberate monitoring and use. The results show that this meta-pragmatic refl exive ability becomes stable only at the age of 6 years.

Old fi ndings on children's comprehension of ask and tell were subject to different interpretations refl ecting progress in the fi eld of language acquisition.We want to show that acquiring a particular skill does not necessarily include competence of its intentional control and use.Development of linguistic skills takes place at different levels starting from early spontaneous, implicit abilities to the level of meta-pragmatic refl exive knowledge that enables deliberate monitoring, planning, and practice.The present study was aimed at exploring two extreme points in development: early epi-pragmatic and late refl exive metapragmatic competence.The fi rst part aims at fi nding the earliest instances of children spontaneous ability to pass ask-instructions, and the evidence is provided for the ages as early as 22 to 40 months (much earlier than recorded in the previous studies).The second part is experimental and focuses on children's ability to respond to ask-and tell-instructions in the context of a cancelled conversational rule (Gricean Maxim of Quantity) which requires deliberate monitoring and use.The results show that this meta-pragmatic refl exive ability becomes stable only at the age of 6 years.
After a long period of strong interest in development of syntactic ability to pose questions, the fi eld of language acquisition as well as developmental psychology seem to provide more ground to explore posing questions as a speech act.Main question of this study of pragmatic development is when children become capable of posing questions on someone's request.We propose that there is a gradual development of this communicational skill, and we want to clarify the difference between early pragmatic ability to pass questions after ask-instruction in spontaneous interaction, and meta-pragmatic ability to perform questioning willingly, an ability that seems to emerge later.We try to demonstrate a difference between the spontaneous skill of responding to ask-instructions that is early practiced in spontaneous interaction, and the meta-pragmatic, refl exive ability of intentional monitoring and use.
Since Caroll Chomsky published her study on acquisition of complex linguistic structures children's understanding of the ask-and tell-instructions attracted the attention of researchers which resulted in different accounts of their comprehension (Chomsky, 1969;Clark, 1971).Further contributions to the issue (Tanz, 1983;Warden, 1981;Bock & Hornsby, 1981) as well as recent disputes on conversational awareness and its effect on performance in experimental cognitive tasks (Siegal, 1996;Siegal, 1999;Siegal & Peterson, 1994;Smith, 1999;Astington, 1999;Deleau, 1999;Lillard, 1999;Laurenco & Machado, 1999) have yielded opportunities for reviewing the same fi ndings in the light of new developments in the fi eld.

Syntactic account
Among many of the intriguing fi ndings of the Chomsky's (1969) study of acquisition of complex linguistic structures was that 5-years-olds and 6-years-olds faced with ask-and tell-instructions tended to impose a general tell-interpretation on ask-constructions.When expected to differentiate between the structures: (1) Ask Laura what to feed the doll, (2) Tell Laura what to feed the doll, children provided responses as if supposed to tell in both situations ('Cucumber!').The fi nding was explained by the twofold meaning of the English verb ask -posing question and making request: (3) Ask q Laura what to feed the doll, (4) Ask r / Tell Laura to feed the doll.
Since the English verb ask can occur with different complements, proper interpretation of the twofold meaning is possible only by relying on syntactic structure.Thus, the children's misunderstanding was attributed to the lack of syntactic knowledge.

Semantic account
However Eve Clark (1971a) argued that the syntactic structures associated with particular words are inherent to their meaning.Thus, Chomsky's fi ndings might refl ect subtle semantic changes in the acquisition of the verbs ask r , ask q and tell, as proposed in the hierarchical model of semantic features (Donaldson & Wales, 1970;H. Clark, 1970;E. Clark, 1971bE. Clark, , 1973)).Two additional semantic features occur in the verb ask q (meaning question) comparing to the verb tell: (5) tell: 'I order you -you say to X -complement S (6) ask q : I order you -you say to X -you request to X -X say to you -complement S.
This may cause young children to impose a general tell-interpretation on both ask r -and ask q -constructions until all the features are learned (E.Clark, 1971a).

Pragmatic account
Further developments in the fi eld have yielded yet another interpretation of the Chomsky's fi ndings.Warden (1981) and Tanz (1983) have emphasized an important role of pragmatic constraints in experimental settings.In his efforts to overcome certain methodological defi ciencies of Chomsky's study, Warden (1981) argued that 5-yearolds are sensitive to pragmatic constraints of particular contexts.In order to respond properly to an ask q -instruction (instruction to ask a listener something), the child must assume either that she has to fi nd something for the experimenter or for herself, or that she has to test the listener's knowledge on the experimenter's behalf.Similarly, in responding properly to tell-instructions, the child must assume that the listener missed particular information, or must assume that she is supposed to demonstrate her own knowledge either to the listener or to the experimenter.In an experimental setting that replicated natural interpersonal context of spontaneous interaction Warden obtained signifi cantly higher frequency of correct responses to the verb ask and the verb tell.He concluded that the pragmatic properties of experimental setting in Chomsky's study were biased towards the tell-interpretation of both tell-instructions and askinstructions.Tanz (1983) provided evidence that the children's knowledge of the answers to the questions signifi cantly determined what speech act they will perform in particular task.When requested to ask someone a question, children aged 5;5 to 9;9 simply apply the rules from ordinary discourse: if they do not know the answer to the question, they will pass the question, but if they do know it, they will supply it straightaway.
These studies revealed that comprehension of the ask q -and tell-instructions cannot be reduced to syntactic and semantic issues.It requires development of pragmatic skills: how seeking and giving information are to be performed, what are the rules under which these speech acts are usually applied, and what are other participant's expectations in a particular situation.5-year-olds can respond appropriately to instructions to ask and tell when these instructions are given in an appropriate context (Warden, 1981;Tanz 1983).
More recently, perspectives from discourse analysis have again brought to the researchers' attention the fact that the experimental setting in developmental studies is an extremely complex source of conversational rules.Children do not consider the aims of interaction in experimental setting in the same way the experimenter does and they obviously may do it at different levels of conversational awareness (Siegal & Peterson, 1994;Siegal, 1999).Siegal (1996) argues that by the age of 3 children expect adults to hold on Gricean maxims -quantity, quality, relevance and manner (Grice, 1975).Cancellation of one of these that happens frequently in the experimental settings of developmental studies requires additional efforts from a child in interpreting the experimenter's intentions.

EPI-LINGUISTIC VS META-LINGUISTIC DEVELOPMENT
A gradual development of communicational skills is proposed in this study and we want to differentiate at least two levels of it.
The fi rst level of language development includes early epi-linguistic competence.It is an ability of use and comprehension of language which enables children to participate smoothly in spontaneous verbal exchange.In using language at early stages, children are not aware of its complex rule-governed nature.It is implicit, internal and un-refl ected knowledge and consequently cannot be deliberately guided and performed.However, competence in language requires not only spontaneous production and comprehension but also the ability to refl ect upon language and the deliberate use of it.Meta-linguistic competence refers to refl exive ability of intentional monitoring, manipulation and modulation of language activities (Vygotsky, 1985;Gombert, 1992;Ely, 2005).Furthermore, it includes an awareness of the relationship between language and the social context in which it is being used (Ninio & Snow, 1996).
In this study we try to demonstrate that the emergence of a skill does not inevitably imply the ability for its intentional activation and use.We explore the comprehension of ask q -and tell-instructions at two different developmental levels.We demonstrate that the competence of posing questions to a listener on someone's request emerges quite early in spontaneous three-parties exchange, but gaining deliberate control of the same pragmatic skill develops only several years later.
Maxim of Quantity phrased in a form of prescriptive command says: 'Make your contribution as informative as is required for the current purposes of the exchange.Do not make your contribution more informative than is required.'In the study that we present here, ability of cancellation of the Maxim of Quantity is proposed as a sign of high level of communicational skills, in other words, as a sign of meta-pragmatic competence which include ability of deliberate monitoring, use and cancellation.

ASK-AND TELL-INSTRUCTIONS IN SERBIAN LANGUAGE
Before we move to the research details, we present some semantic and grammatical properties of the requests in consideration.Serbian is a South Slavic language with highly developed infl ectional and derivational morphology.In regards to ask r vs. ask q distinction (as explained by C. Chomsky, 1969), the Serbian verb pitati has the same semantic properties as the English verb ask.It can be used in both meanings -question (ask q ) and request (ask r ): (7) Pitaj q Mariju koliko ima sati.'Ask q Maria what time it is.' (8) Pitaj r Mariju da nahrani lutku.'Ask r Maria to feed the doll.' (9) Reci Mariji koliko ima sati.
'Tell Maria what time it is.(10) Reci Mariji da nahrani lutku.'Tell Maria to feed the doll.However, morpho-syntactic properties of the Serbian language are different from those of English in some aspects that may be relevant for the comprehension of ask-and tell-instructions.In the following examples, the complement verb feed is infi nitival in English, but fi nite in the Serbian sentences (nahrani, nahraniš), and this makes the subject of the wh-clause omitted from the surface structure in English, but transparent in Serbian: (11) Pitaj q Laur-u čime da nahrani-š lutku.ask q .IMP.2SG Laura.ACC what to feed.PRES.2SGdoll.ACC 'Ask q Laura what to feed the doll.'Although it was necessary to present the morphological and syntactic properties of Serbian ask q -and tell-requests, it should be noted that these properties are not of crucial importance for our discussion, since we are primarily interested in the pragmatic aspect of these requests.According to these aims, only sentences of the type ( 7) and ( 9) were used in the second, experimental part of the present study.

PART I: ASK Q -INSTRUCTIONS AND ACCOMPANIED RESPONSES IN SPONTANEOUS INTERACTION Aims
In the fi rst part of the study we were primarily interested in the spontaneous development of a particular pragmatic skill.We focused on fi nding the earliest instances of passing ask q -instructions to a listener in natural setting.We expected to answer two main questions: When do adults start to deliver ask q -instructions to children (eg

Sample
The search was made on the Serbian Corpus of Early Child Language (Anđelković, Ševa & Moskovljević, 2001) compiled according to the CHILDES system (MacWhinney, 1989;MacWhinney & Snow, 1985).It contains longitudinal recordings of spontaneous interaction within the families of eight children (4 boys and 4 girls).The recordings lasted approximately 90 minutes, and they covered 16 longitudinal samples of interaction at the ages between 18 and 48 months (2-months intervals).

Corpus retrieval
The retrieval was focused on the situations in which the target children were given the ask q -instructions, that is requested to pose a question to a third person in interaction.The search was aimed at all incidence of the request pitaj (imperative form of Serbian verb ask) addressed to the target children.

Data analysis
The analysis focused primarily on the comprehension of the pragmatic aspect of ask q -requests.Therefore, the pragmatic appropriateness of children's responses to ask q -instruction was evaluated, and the responses were classifi ed in regards to pragmatic suitability and fulfi lment.The children's linguistic performance was not of prime interest.Thus, if a child's response had no indications of an imposed tellinterpretation (e.g.answering question), but rather provided indications of ask qinterpretation instead (i.e.passing the question to the listener), it was classifi ed as fulfi lment, no matter whether the child properly dealt with all morpho-syntactic rules or not.

Earliest records of ask q -instructions and accompanied responses
The Table 1 presents the earliest ages at which ask q -instructions directed to children were recorded, as well as the ages of the earliest fulfi lments of ask q -instructions.The earliest fulfi lment of ask q in the overall corpus was found in a child at the age of 22 months.By the age of 40 months, proper fulfi lment was found in all 8 children, which is much earlier then it was suggested by the previous fi ndings (Chomsky, 1969;Warden, 1981;Tanz, 1983).Although those studies were not aimed at fi nding the earliest indications of this ability, they placed it about the age of 5 years.
The earliest records of the adults' ask q -request were found when the children were 18 months old.The age of the earliest ask q -request recorded for each particular child differed, which may be attributed to several reasons: a) different developmental maturity of children at the same chronological age; b) different collocutors may have different expectations from the children; c) situational and contextual variety in interaction sequences.It should be noted that the syntactic structure of the requests that children received was very different, varying from the simple to very complex ones.It should be also noted that requests were sometimes shortened and elliptic.The recorded sequences in which the children were not able to fulfi l the request due to developmental immaturity were particularly interesting, since they provided evidence of how adults guide children through the course of communicational exchange.After not receiving any reply from the child, the adult would sometimes try to adjust the expectations according to the child's developmental level either by giving up on his/her request, or by playing the entire sequence of interaction for the child: fi rst by giving a request, and then, changing voice, fulfi lling the request in the role of the child by passing the question to a third party.

Pragmatic adequacy of early responses to ask q -instructions
The most important issue in this analysis is the pragmatic adequacy of children's accompanied responses.Classifi cation of the responses regarding pragmatic suitability and specifi c properties of children's (between 18 and 48 month of age) behaviour is presented in the Table 2. Appropriate fulfi lment of the ask q -instructions in the overall sample was found in 31% percent of cases.In an additional 20% of cases, children exhibited indications of correct request interpretation, although they had not performed the requested speech act (categories 2 and 3).The tell-interpretation was imposed on the ask q -instruction only in 5% of cases.In overall sample the proportion of misinterpretation and obvious inability for suitable response was relatively low (Category 4).
Objective obstacles barred the children's fulfi lment of ask q -instructions in a relatively high percentage of cases (Category 6).Sometimes children were distracted by another event, person, or object, which is not unusual in spontaneous communication.In a smaller number of situations the fulfi lment of the request was objectively impossible when the collocutor directed the child to someone who was absent at the moment, either because he/she was referring to a later occasion, or because he/she was unaware of that person's absence.In addition to that, there was about 18% of cases in which interpretation was not possible simply because the child did not perform any speech act (Category 5).Both are quite normal for the analysis of spontaneous interaction sequences.Category 5 included different cases in which children's hesitation may refl ect different states of mind (pragmatic immaturity, shyness).

PART II: META-PRAGMATIC MANAGEMENT OF RESPONSES TO ASK Q -INSTRUCTIONS
It was already argued that the ability to understand the collocutor's intention is an important aspect of pragmatic development (Tanz, 1983).In a situation of request like Ask Laura what to feed the doll, it is plausible to believe that the addressee does not know the answer.From a pragmatic point of view it is completely appropriate to answer the question when knowing the answer, since it fulfi ls the collocutor's expectations.Instead of posing the question to someone else, the children in Tanz's study who knew the answer gave it straightaway, while those who did not know it passed the question to another person.Important to note here is that due to context and situational varieties, some children in this kind of experiment participate in an 'atypical' way.Tanz (1983) mentions that only one child from the fi rst 'informed' group passed the question to another person even though he knew the answer to that question.The 'unusual' behaviour was interpreted as an indication of an even higher pragmatic maturity, since the child was able to take the experimental situation into consideration and play as if he did not know the answer, or as if the everyday pragmatic rules were cancelled for the experiment.Sharing intentions and conversational rules in experimental settings attracted lot of attention in the more recent debates (Siegal, 1999;Siegal & Peterson, 1994, Astington, 1999;Deleau, 1999;Lillard, 1999;Lourenco & Machado, 1999;Smith, 1999;Lee & Eskritt, 1999;Bialystock, 1986).Children do not share the purposes of the experimental setting with the researchers (Siegal, 1999;Siegal & Peterson, 1994), and they are often confused about the violation of Gricean maxims (Grice, 1975).In other words, when a child fi nds reasons to believe that a conversational rule or a Gricean maxim does not apply in a particular experimental setting, he is faced with a new task at a high level of conversational awareness.

Aims
The aim of this part of the study was to differentiate low level skills of understanding ask q -and tell-instructions from meta-pragmatic grasp of conversational rules and adjustment of one's own response in accordance with them.For this reason we expect children to fi nd it diffi cult to fulfi l the ask q -and tell-instructions in the situations in which Gricean Maxim of Quantity is cancelled (maxim that specifi es that one should speak no more or less than required).
In order to achieve these aims, it was necessary to conduct an experiment in which the answers to the questions were known to all the parties engaged in the interaction, but nevertheless one of the participants asks a question about the matter observable at the context.
In conditions of cancellation of this conversational rule we expect children to have more diffi culties with ask q -instructions then with tell-instructions, even though semantic, syntactic and pragmatic prerequisites were presumably fulfi lled at a much younger age.

Procedure
The experiment was conducted in a kinder-garden.The participants were engaged in a three-party exchange in which the experimenter was giving tell-and ask q -instructions to the child (e.g.Reci Jovanu koje je ovo boje, 'Tell Jovan what color this is'; Pitaj Jovana čemu ovo služi, 'Ask Jovan what this is for').The child was supposed to perform the speech acts by telling something to the third party in the fi rst situation, or passing him/her a question in the second.
The instructions always referred to an object perceivable in the context (e.g.'Tell/Ask Jovan what color this is' referred to a red ball).In this setting, the child knew the answer, and knew that other participants knew it also, so the setting was biased against asking questions, as well as against passing them to a third party.Conversely, it was not biased against making assertions, since commenting and talking about what is known may have a function of maintaining joint attention in everyday communication.Therefore, a conversational rule was cancelled -the Grice's Maxim of Quantity -one wouldn't ask if all the parties knew the answer, and if all the parties knew that all others knew.Thus, the three parties played a game of performing the speech acts themselves without intrinsic motives for asking and telling.We believe that profi cient participation under these circumstances requests for highly developed conversational awareness at meta-pragmatic level.
Moreover, in order to impel children to engage in an intentional alternation of speech acts, randomized addressee change in instructions was introduced.Children were supposed to address either the listener, the experimenter, or a toy-listener.We presumed that addressee switch additionally burdens the task and makes it more complex for deliberate performance.The addressees differed in properties important from pragmatic point of view: dyad/triad exchange and alive/non-alive addressee.

Design
Three factors were systematically varied in the experiment.The fi rst was age (4, 5, and 6 years old children), the second was request type (ask q -and tell-instructions), and the third factor was addressee (experimenter, listener, and toy listener).

Material
After 6 exercise items the child was exposed to a randomized list of ask q -and tell-instructions.There were 18 sentences in total, 3 sentences in each of the following 6 situations: Sample 31 children of three age levels were included (4, 5, and 6 years) -evenly distributed in age samples.Age span of particular groups was 8 months -between 3;8 and 4;4 for the youngest, between 4;8 and 5;4 for the 5-years-olds and between 5;8 and 6;4 for the 6-years-old group.

Data analysis
The number of correct responses was counted for every child, and average scores calculated for the age groups.Appropriate interpretations of ask-and tell-instruction were considered as correct answers (ask-and tell-interpretation respectively), which in case of ask-instruction meant passing question to another party, and not providing answer to the question.Three-way ANOVA by subjects was applied for the 3x3x3 design in which the factors of request-type and addressee were repeated (see the section Design).
However, analysis revealed signifi cant interaction between the factors of age and request type: F(2,28)=9.24,p<0.01. Figure 1 shows that the ask q -instructions were more diffi cult than the tell-instructions for the young children, which was not the case for the 6-years-olds.However, the tell-instructions were performed equally well by the children at all age levels.No signifi cant main effect for addressee was obtained, as well as for interaction between addressee and age (Figure 2).The factor of addressee did not affect performance of speech act -the basic effect of addressee and interaction with age were not signifi cant.
To summarize, even though the corpus analysis revealed that the proper interpretation of ask-instruction (ability to pass a question to another person) emerges at the early ages (from 22 months to 40), the experimental part of the study has shown that children before the age of 6 years are not capable of performing this competence under all circumstances.Very young children can do it smoothly in a natural spontaneous interaction, but the violation of a Gricean maxim seems to burden their understanding of collocutors intentions.This can provoke even older children to perform at a developmentally lower level, i.e. to attribute tell-interpretation to the ask-instructions.Moreover, request from an experimenter to a child to pose a question to him/her (the experimenter) is proved even more diffi cult than request of posing a question to another participant (listener or toy-listener).The situation is rare and probably seems unnatural to children, and may require even more efforts for reading the collocutor's intentions.
The experimenter reported on developmental differences among the children that became prominent throughout the testing.Every child was able to fulfi ll the tell task, but not all of them responded properly to ask q -requests (Table 3).

Table : Number of children at particular levels of performance
Age Levels of performance 4y 5y 6y N 1. performs tell, and never performs ask q 8 3 0 1 1 2. performs tell, and sometimes performs ask q 1 4 2 7 3. performs both tell and ask 1 4 8 13 Obviously the fulfi lling of tell-request is developmentally less demanding then a passing of request ask q .It is important to note that the number of children who succeeded in passing questions increases with age, indicating developmental transition towards higher conversational awareness and more fl exible management of speech acts.

DISCUSSION
The results obtained in the corpus-based part of this study indicate that adults start early with ask q -instruction directed to children.Early requests are rewarded with early responses, and fi rst instances of satisfactory fulfi lment were found in children already between 22 and 40 moth of age.This is several years earlier than it was found in the classic experimental studies (Chomsky, 1969;Warden, 1981;Tanz, 1981).Moreover, the analysis of pragmatic suitability reveals that approximately one third of requests are suitably fulfi lled already before the age of 4, and only in 5% of cases (in overall sample) children impose tell-interpretation on ask q -instructions.This is an evidence of an early epi-pragmatic competence which grows in spontaneous interaction, and originates from all clues available in the context.At this level children do ask, but do not know how to do asking.The child is performing speech acts, but is not consciously monitoring and manipulating them.It is a functional, but implicit and un-refl ected knowledge, and cannot be intentionally changed.
The experimental part of the study has shown that ask-instruction is more diffi cult then tell-instruction up to the age of 6. Responding to ask q -and tell-instructions carry attendant diffi culties derived from all layers of language function (semantics, syntax, pragmatics).The social context may require cancellation of particular maxim, where refl ection on language and ability of deliberate control are needed.At this point a child should have already compiled a relatively rich conversational experience, and developed a competence of intentional monitoring and planning.It provides the subject with prerequisites to read conversational rules, enabling him/her to manipulate speech acts accordingly to social context.Thus, profound understanding of wide variety of social exchange, and refi ned adjustment according to interpersonal context requires a long way to run.It starts from spontaneous participation in exchange of communicational acts, and aims at meta-pragmatic refl ection on applying and cancellation of maxims and principles that underlie human communication.Iako je u navedenoj studiji ovaj nalaz samo usputan, on već decenijama privlači pažnju istraživača koji ga tumače na različite načine.E. Klark je ovaj rezultat pripisivala semantičkom razvoju.Značenje svake reči, prema teoriji semantičkih crta, može se razložiti na kombinaciju jedinica značenja.Razvoj semantičkog znanja sastoji se od dodavanja novih crta značenja, sve dok se kombinacija crta ne podudari sa kombinacijom kod odraslih.Promene nastale u oblasti u tokom poslednjih decenija omogućile su da se nalaz postavi u sasvim drugačiji, pragmatski kontekst.Vorden tvrdi da su 5-godišnjaci veoma osetljivi na pragmatske karakteristike eksperimentalne situacije i da je glavni izvor "zabune" kod dece interpersonalni kontekst.Da bi odgovorilo adekvatno na zahtev ask 'pitaj nekoga nešto' (odnosno da bi postavio pitanje slušaocu) dete mora imati nameru da sazna ono što interesuje eksperimentatora ili njega samog, ili nameru da proveri da li slušalac ima to znanje.Slično, da bi odgovorio adekvatno na zahtev tell 'reci nekome nešto' ono mora pretpostaviti da slušaocu nedostaje ta informacija, ili pretpostaviti da se od njega očekuje da pokaže svoje znanje.
CHILDREN DO ASK, BUT DO NOT KNOW HOW TO DO ASKING: EPI-PRAGMATIC VS.META-PRAGMATIC SKILL 1 Nevena Buđevac, Darinka Anđelković 2 and Maja Savić Laboratory of Experimental Psychology, University of Belgrade, Serbia

Figure 1 :
Figure 1: Ask/Tell and Age Interaction

Figure 2 :Figure 3 :
Figure 2: Interaction of Addressee and Age (not signifi cant) When do children start to respond to those requests properly, that is at what age children can pass a question to a listener after the collocutor's request to do so.
. Ask John what time it is);

Table 2 : Children's responses to ask q -instructions in spontaneous interaction
Jovana koje je ovo boje 'Ask Jovan what colour this is' Reci Jovanu koliko ovde ima bombona 'Tell Jovan how many candies are there' Pitaj mene čemu ovo služi 'Ask me what this is for' Reci meni koja je ovo životinja 'Tell me what animal this is' Pitaj medu kako se ovo zove 'Ask teddy-bear how this is called' Reci medi šta radi beba 'Tell teddy-bear what the baby is doing' Only well-known materials and objects were used for testing.