Wednesday, November 22, 2006
turn units and emoticons.
I appreciate the discussion about emoticons as turn units, by the way. When I teach speech acts in my computer-mediated discourse analysis course, there is always much discussion as to whether an emoticon can function as a proposition (i.e., a speech-act bearing unit). I adopt the view that if the emoticon constitutes the entire message (in chat) or appears on a separate line (in, say, email), it is a proposition, but if it appears on the same line as text, it (usually) inflects a textual proposition. This is theoretically not a very coherent treatment of emoticons, but it reflects my intuitions. Any ideas as to why it should work this way?
comments to follow.
re: the distinction kris brought up about 'turns' and 'posts' - i also don't think you can conflate the two by any means, though i'm not sure i agree with the idea of multiple turns occurring in one post (or transmission, to use baron's terminology). i'm going on the conversation analytic definition that takes into account turn-relevance points, and since the idea of a conversational floor is realized differently in CMD due to the message composition process and the message tranmission process being two separate things, i can't conceptualize multiple turns occurring within a single post (since it seems like the TRP for an utterance can only occur at the end of the transmission, all syntactic and other content aside). i know what you're talking about when you mention multiple units used within a single post, but i would consider this a multi-unit turn rather than multiple turns.
in regards to multiple posts consisting of a single turn at talk, well, i was thinking the same thing as far as a single-post emoticon goes (and was also delayed from responding by a conference and a mess of grading). i've seen emoticons get used as second post 'afterthoughts' in chat, bulletin boards, and a few other mediums that could be considered part of the same turn as the first post, and that can be oriented to by other speakers as functioning the same as would a sentence-final emoticon occurring within the first post. so i agree with kris that the sequential organization of the single-turn emoticon has a lot to do with how it's interpreted or oriented to, rather than whether it just occurs on a separate line. i also think that there's something to susan's idea that the emoticon can function as a proposition in the right contexts of the talk, and that looking at these would probably help get rid of the idea that emoticons are *just* graphical representations of or replacements for extralinguistic cues :)
(hope NCA went well, by the way - AAA was great.)
NCA went well, although it seems there are still some people out there (at least in communication) who need to be convinced that chat interaction is conversation (as opposed to "writing"). But anyway....
I was scratchign my head for a while, trying to figure out why, in my diss, I hadn't talked about transition relevant places (TRPs for you non-CA types). So here I've come up with an answer I'm going to throw out into cyberspace. I think that, for some kinds* of chat environments anyway, TRPs are not really a relevant concept. What I mean is that, in chat, the turn taking possibility is inherent, i.e. it is always already anyone's turn to talk. I argue that in chat, we need to reconceptualize the turn to see it not as a unit of speaker exchange, but as a unit of thread coherence. A turn in chat is a sensemaking tool, not a device for deciding who is next. That said, I certainly have evidence that chat speakers, through the construction of their turns, can make specific others, or any others, relevant future speakers (for example, by using a 1st pair part of an adjacency sequence), and thus in that limited sense, the end of a turn might be thought of as a TRP. However, in systems where the chat conversation is persistent, I would still move away from the use of the TRP concept, b/c even when using a 1st pair part (like a question) to make another speaker relevant, you are not *guaranteeing* that the next slot in the chat conversation will be turned over to that other person. Rather, you are creating (or continuing) a conversational thread in such a way that makes future turns in that thread more easily aligned to your turn. I'm not sure if that last sentence made much sense, but basically I'm referring hewre to another of my positions, which is that conversations in chat* are organized primarily by conversational thread, as opposed to the turn taking (exchange) system in spoken interaction.
Thus, I separate post (one unit of text) from turn (one idea unit), and stated earlier that in theory a post could contain more than one turn (though I have no data showing this, and I htink it probably works against thread continutity to do so). Additionally, I do have data that shows that participants are oriented to the possibility of spreading one turn (idea unit) over multiple posts. I don't think this affects TRPs b/c I really don't think the TRPs exist as a functional device in the turn organization system.
*Major caveat to all of the above; my data (and thus observations) come from small group interactions (no more than 6 people) in a chat room that allowed for persistence of conversation. I know that the persistence did sometimes affect how the participants created/constructed turns, and I am fairly certain that group size has an effect on turn organization and turn construction in chat. Thus is it quite possible that, in larger groups (though I don't know at what exact point a group becomes 'large') the rules for turn organization will vary. I think, for example, that threads might become less important as conversations evolve into multiple dyadic exchanges, and then, perhaps, TRPs become more relevant? But I don't have that data, so 'tis all speculation.
Turn-taking management (or the absence thereof) could be different in other situations, such as the small groups you studied. Perhaps the fewer the participants and the more focused the topic, the more lengthy threads become possible. And perhaps in established threads, participant roles emerge and it is less necessary to assign next turns because topic and/or turn order already indicate who the next speaker should be?
I think we should keep turn allocation separate from TRP in the narrower sense of 'appropriate place to start the next turn so as not to cause gap or overlap with the ongoing turn', though, right? Those temporal issues seem to be less relevant in CMD (aside, perhaps, from Hillary Bays' work on rhythm in chat).
By the way, Condon and Cech have written about single messages containing more than one turn -- according to your definition of a turn as an 'idea unit', that makes sense.
since the chat discourse you've written on is fairly similar in makeup to the ones i've studied, i wondered why your transcripts would make it appear as if the speakers are orienting only to threads instead of TRPs. i think one answer may be that in all of the examples of talk you provided in your paper, i couldn't find any that lacked any type of clear syntactic or even pragmatic completion, and thus didn't make use of any type of turn-maintenance device that relied on other speakers' orientation to TRPs (which would illustrate this orientation). this is possibly explained by the fact that your speakers were engaging in a somewhat formal, educational slash organizational conversation where that type of 'break' might not be seen as acceptable, whereas the speakers i looked at were informal groups of friends where it might be (an analogue to spoken discourse might be looking at a business meeting versus a group of teenagers hanging out). i think this is a point where tannen's idea of conversational style actually becomes very relevant to a discussion of turn-taking.
i think doctor herring is right that TRPs aren't being oriented to as 'places where one could begin to speak to avoid gaps or overlap', though i think speakers can and do orient to them as 'places where the other person is finished with the floor, and where my own composition of responses to their utterance will have maximum relevance'. not that anyone actually thinks so long-windedly, or anything. while they may be less relevant in some environments (as the work discussed here has illustrated), i don't think we can discount temporality or the sequential organization of utterances in all formats of cmd just yet (nor do i think we have to).