My main concern is the socialization process of the child through interaction with adults and I am investigating it with particular regard to conversation.
This report attempts to investigate how we observe `children' in conversation. How is it that a participant can be defined as a `child' in conversational sequences? The answer to this question needs to be addressed as a prerequisite for any research into socialization. Before investigating some conversation as a child-adult interaction, we must first confirm that the entity that we have described as a `child' is actually observable in conversational data.
To do so, we can follow the approach proposed by Harvey Sacks, which can be characterized as self-descriptive reconstruction of our practical reasoning, using conversational data as a trigger (Sacks 1972, 1974). We, as members of conversational community, may be able to identify some utterances in conversational data as turns by a child even in the absence of clear contextual clues. It is argued here that this can be achieved by members undertaking a `conversation analysis' on the basis of `asymmetrical rules', which are referred to interpretive-constructive rules adopted in a conversation between asymmetrical participants. By the word `conversation analysis' I mean a series of practices by members to observe-describe-formulate conversations. I focus here on the asymmetrical aspects of conversational structure because they are believed to be fundamental factors in conversational community. I attempt to propose experimental procedures by which we can confirm the observability of `children' in conversation and also the asymmetric nature of conversation.
The catcher in the conversation
Unexpectedly, it is not easy for us to observe `children' in conversational data. If you felt it easy, you provably could do it not by considering the way of their interaction but by categorization with audio/visual clues, ex. appearance, tones of voice of participants. And once some categorization was made, it will be used as a framework for following observation/interpretation of conversation.
How can we avoid such a trap? One possible tip for that is so-called `Turing test.' A.M.Turing, a founder of computer science, produced a procedure called the 'imitation game' in order to make it testable the question, i.e. "Can machines think?" (Turing 1950).
According to Turing, "...It is played with three people, a man(A), a woman(B), and an interrogator(C)... The interrogator stays in a room apart from the other two. The object of the game for the interrogator is to determine which of the two is the man and which is the woman..." A's object in the game is "...to try and cause C to make the wrong identification..." And B's object is "...to help the interrogator..." The interrogator is allowed to talk with the other two, but it must be done without any direct clues, such as tones of voice by tele-communicating between two rooms. Now let a machine take the part of A in this game. If "...the interrogator decides wrongly as often when the game is played between a man and woman...", we should admit the machine can think.
Turing test proposes a theme that the objects should be identified without contextual clues other than interaction itself and also proposes how to do that. Not a few sociological subjects, ex. socialization, gender, class and power, assume certain asymmetricity between objects (child-adult, man-woman etc.) I believe researchers investigating those subjects using conversational data should take the part of the interrogator not the experimenter who knows the answer in advance in some part of his/her research program.
`Turing test' in conversation analysis
How can we revive Turing's suggestion in conversation analysis? Transcript of conversational data may be useful for that.
As you know, there are fundamental limits at many different levels on the extent to which the original conversation can be reproduced from transcript. However, if we are to attempt to reconstruct how participants and researchers carry out conversation-conversation analysis, it is the transcript with a wide range of moments from the original conversation abstracted which may enables us to observe the problems involved. In other words, this is because factors such as voice quality and appearance are removed, which accompany the conversational situation, constitute part of it, and are essential constituent moments in leashing out the structure of conversation and which contribute to the conversation analysis of the investigator and observer. These moments impose categories(e.g. sex or age) for us and provide a pre-existing framework in the form of ground for the configuration which constitutes the meaning of utterances and recede into the background without their effect being noticed.
However, before using transcript for `Turing test', we must investigate categorization of speakers in transcript, for example, indicating each speaker by heading such as "Mother:"-"Child:" or "Man:"-"Woman:". Identifying speakers as such and the way we write that is a choice made from many different alternatives. For example, we could use the symbols "A:"or "B:" or even nothing at all. Categorizing speakers as "Man"-"Woman" and the way we write that is not something which can be taken for granted. The investigator has chosen one of a variety of different possible categories. This does not means that in the actual conversation the participants see themselves and their counterparts only in terms of the same categories and have made their utterances to each other using these categories. In spite of this, the categorization and labeling used is added as a construct of the transcripter/investigator. This means that the way in which third party understands the conversational data exercises an important influence.
Now the problem is identifying/categorization of speakers in transcript. We are to try manipulating the labels given to the speakers for the `Turing test' in conversation analysis. I would like to call them `Turing procedures.'
Turing procedure (1) - Anonymizing of participants
This is very simple manipulating, which deletes the labels/categories as headers in transcript to help identifying speakers statuses and gives the mere symbols "A:" or "B:." After that, we, as investigators, observe conversational sequences in transcript without direct clues about categories of each speaker. If we could suppose categories of each speaker or feel some asymmetricity between participants even after that, we would do self-descriptive reconstruction of our understanding. Why and how could I do that? It would be important data with regard to such as socialization and gender and so on.
Turing procedure (2) - Exchange of participants' categories
This is also simple manipulating. Using the transcript of conversation you can exchange participants' categories, for example exchange "Child:" and "Mother:." (Or we may add some manipulations such as making even category "Adult:" or "Child:" in originally child-adult interaction.) After that, we observe conversational sequences in transcript. If we could feel discomfort in category-reversed transcript, we should do self-descriptive reconstruction of our feeling. It would be important clues for us to understand the way of our practical reasoning about asymmetrical members.
Asymmetrical data (1)
Let us have a look at the following transcription altered according to procedure(1);
[Case 1] #1 A: Hi. 2 B: Hi. 3 A: Hi, Annie. (x C: .... ) 4 D: Annie, don't you hear someone say hello to you? 5 A: Oh, that's okay, she smiled hello. 6 D: You know you're supposed to greet someone, don't you? 7 C: [Hangs head] Hello. (Sacks 1974 p.227.)
This is a fragment of an ordinary conversation which involves some important aspects of conversation analysis by members and illustrated the socialization process.
Firstly, as Sacks suggested, this is an example of a case in which members notice a break of turn-taking (i.e.lack of greeting) and refer it. That is to say, the nature of the conversation itself is topicalized. But why can members do that? Of course they are not only uttering and hearing each other's utterance but also monitoring the whole process of conversation (ex. sequencing). By this we can respond to someone's utterance. In this sense "listening" is a very important factor in conversation, but strangely it is not transcribed at all. This is applicable to most conversational data. On #1 member B does not do nothing. He is " listening" and observing the conversation process (so he could immediately respond to A).
Secondly, although member C failed to exchange greetings, A accepted C's incomplete turn (i.e.smiling) as a proper one. We can find a lot of examples of this in conversations involving infants or babies. We often take other's incomplete turns (it may be an utterance or a gesture) as complete ones and extract various meanings from them intentionally/unintentionally. When they are incomplete, communication doesn't necessarily stop. This fact helps and encourages candidate members without sufficient interactive competence to attend to ordinary conversation.
And I would like to add another point suggested in former section. When we consider #1-7, it does not seem to involve complete turn-taking, i.e. it seems to represent a lack of greeting. However, we also feel this fragment is enough to be thought of as conversation. Because we can assume that member C, Annie, seems to be a young child. And if so, we can admit this conversation is complete. However why are we likely to think C is a child? There is no direct clue about age in the example. And it is often the case that some adults do not respond to others, isn't it? Yes, we can think this is a conversation between adults in #1-4 , but after hearing sequence #5-7 we are much less likely to think so. With no direct clue we can imagine participants' categories to some extent, i.e. A and D are adults and C is a child.
I would like to say here that we can assume there are some special rules to interpret conversations involving candidate members, which allow us to categorize participants without direct clues. How can you find them? There is a simple technique. Using the transcript of conversation you can exchange participants' categories, for example exchange "Child:" and "Mother:". Or you may try to substitute Adult or Child for A, B, C and D in Case 1. If you cannot take proper conversation, there are some rules of interpretation above-mentioned. As I have said, I would like to call them `asymmetrical rules.' Of course there are various asymmetrical rules corresponding to various aspects of conversation.
Asymmetrical data (2) : A-Q-A sequence
Let us consider another example. Look at the following sequence from my field notes;
[Case 2]
(Participants are watching TV in a room.)
#1 Adult: This is Ponkikki! ...I've seen it!
(10)
2 Child: What is this?
3 Adult: Ponkikki!
4 Child: Have you ever seen it?
5 Adult: Yes. I always watch Ponkikki.
*Translated from Japanese ** " Ponkikki" is a Japanese TV program for children.
This is a transcript exchanged participants' categories by Turing procedure(2). Can you find the above-mentioned fact? Probably you can feel discomfort in this sequence. Actually, A is a child (3years) and B is an adult (his teacher). If reversed we won't be able to accept it as a possible conversation. Why and how do you think so? It is thought that certain sequence of question and answer belongs only to a certain combination of participants.
This part of conversation consists of some sets of questions and answers. Interestingly, question in #2 and 4 are situated in A-Q-A sequences such as "#1-2-3" and "#1-4-5", which cannot be seen in ordinary conversation between adults. Why can we suppose asymmetrical participants in it, when we observe A-Q-A sequences? What sort of background knowledge is used for identifying questioners and answerers? To confirm about that, we need to consider the way of questions.
Now please see following passages;
[Case 3]
Then the man and his wife heard the sound of the Lord God as he was walking in the garden in the cool of the day, and they hid from the Lord God among the trees of the garden. But the Lord God called to the man, " Where are you?"
He answered, " I heard you in the garden, and I was afraid because I was naked; so I hid."
And he said, " Who told you that you were naked? Have you eaten from the tree I commanded you not to eat from?" (Genesis Chap.3 8-11.)
I would like to suggest there are common features between Case 2 and 3. It is concerned about type of question and answer. That is, in both cases each questioner;
a) knows the answer;
b) knows the other party knows the answer;
c) knows what the other party will answer;
Let me call these questions type G. This type of question suggests asymmetrical relations between questioners and answerers. When it is used in ordinary conversation, it is probably closely connected with the socialization process. Can you think the questions known the answer by questioner as meaningless? No. Because it imposes some task to the possible answerer. For infants taking their turns properly is very difficult. They can say something, but cannot necessarily say it at the proper time or as proper turn. Please remember the phenomenon called collective monologue. It seems to indicate a developmental stage which infants can utter but they cannot do enough `conversation analysis.' Question type G helps them to take their turns and communicate with others.
By the way, strictly saying, the observation as "A-Q-A" sequences is a reflexive recognition by observer. Of course "A" in #1 is not responded to "Q" in #2. But we, as members of conversational community, can take it as an answer after observing following sequence. So this type of sequence seldom occurs between standard members of conversation.
And those facts mentioned above are well-known by members of daily conversation. They are embedded in members and in the structure of conversation as asymmetrical rules. By them we can immediately identify participants while observing A-Q-A sequences.
Asymmetrical data (3)
In this section I introduce further data connected with asymmetrical rules. However, they are different from former ones in that asymmetrical aspect in them may be thought as culture-bound one. Any way, please follow these transcripts (upper part of each lines is original Japanese data and lower part bracketed by ' ' is translated in English);
[Case 4] #1 A: Buubuu notte ittoide. 'Ride in the car and go.' 2 B: Een, mama mo, mama mo. 'Waah, mama too, mama too.' 3 A: Mama mo? Mama ii no. Mama ii no. 'Me too? I'm all right(without going). I'm all right.' (Clancy 1986, p.238) [Case 5] #1 A: Boku wa ima nani site asonderu nokana? 'What are you playing?' 2 B: Ojisan dare? 'Who are you?' 3 A: Ojisan wa ne ... 'I'm ...'
Native speakers of Japanese can immediately suppose participants' categories from Japanese data (Member A in each case is an adult and B is a child). On the other hand it is not easy for native speakers of language other than Japanese to identify participants' categories from transcript translated into English.
Though I do not have enough space to explain in detail, I would like to suggest that asymmetricity in each data is based on the Japanese way of terminology of person, including way of exchange of personal pronouns, ex. "I" and "you", following turn-taking of speakers. As you probably know, Japanese has so many personal pronouns, or has no personal pronoun as an independent grammar unit (Suzuki 1978-1984). We just use properly so-called personal pronouns and other items (ex. kinship names, status names and so on) according to social relationship.
And we Japanese do not necessarily request to the counter party nor impose on ourselves exchange of personal pronouns in case of conversation with infants or young children. Because we suppose it is rather difficult for them. Case 4 and 5 (Japanese version) include such sequences that same items are used both as terms of self-reference and as address terms. So we immediately identify one of the participants as a young child,or someone who has a asymmetrical status in some sense. This fact is connected with another asymmetrical rule. But I am afraid it is difficult to translate meaning of interaction with a nuance of asymmetricity. I do not know whether English has such an aspect or not. However, using same item both as terms of self-reference and as address terms may suggest asymmetrical relationship between participants in English conversation. Please consider the words in a movie, "My name is Forrest Gump. People call me Forrest Gump."
Asymmetrical rules as fundamental factors
It is thought conversations involving candidate members (ex. babies, infants) have their specific rules, which since then are an extension of ordinary rules makes it easy for them to do conversation with other members. I call them `asymmetrical rules.' Though I am not sure whether the same type of asymmetrical rule will be seen in various cultures, I believe that some form of asymmetrical rules must be found in each culture. Because without these types of rules, it is thought that the community of conversation cannot get new members at all and so cannot survive.
References
Abe,K. 1997, "Kaiwa niokeru `kodomo' no kansatsu kanousei nituite", Japanese Sociological Review 47(4): 445-460.
Clancy, P.M. 1986, The acquisition of communicative style in Japanese, in Language socialization across cultures, ed. by B.B. Schieffelin & E. Ochs, Cambridge University Press, pp.213-250.
Schegloff,E.A.& Sacks,H. 1973 Opening up closing, Semiotica 7: 296-327.
Sacks,H. 1972 An initial investigation of the usability of conversational data for doing sociology, in Studies in Social Interaction, ed. by D.Sudnow, The Free Press, pp.31-74.
── 1974 On the analyzability of stories by children, in Ethnomethodology, ed. by R.Tuner, Penguin, chap.17.
── 1992 Lectures on Conversation, 2 Vols., Oxford:Blackwell.
── Jefferson,G.&Schegloff,E.A 1987 A simplest systematics for the organization of turn-taking for conversation, Language 50(4): 696-735.
Seiya,Y. 1994 "Ningen Keisei no Ethnomethodology", Toyokan shuppan.
Suzuki, T. 1978-1984, Words in Context : Japanese Perspective on Language and Culture, Kodansha International.
Turing,A.M. 1950 Computing machinery and intelligence, Mind 59: 433-460.