Presentation is loading. Please wait.

Presentation is loading. Please wait.

Inferring conversational functions in Japanese discourse with Discourse Marker Complex Eiji Tomida & Shunichi Maruno Faculty of Human-Environmental Studies,

Similar presentations


Presentation on theme: "Inferring conversational functions in Japanese discourse with Discourse Marker Complex Eiji Tomida & Shunichi Maruno Faculty of Human-Environmental Studies,"— Presentation transcript:

1 Inferring conversational functions in Japanese discourse with Discourse Marker Complex Eiji Tomida & Shunichi Maruno Faculty of Human-Environmental Studies, Kyushu University

2 Growing interest in Discourse Process nMany social and cognitive scientists have been interested in discourse processes. nObserving discourse, we can find important interactions for socially-shared reasoning. lWhat kind of utterance facilitates reasoning? lHow a new idea emerge in conversation? nIn most studies in psychology and other social sciences, they analyze discourse data manually. For example…

3 Manual coding procedure 1.Construct a coding scheme. 2.Using the scheme, 2 independent coders assign one of these categories to a conversational turn. 3.The total numbers of each assigned category are calculated with respect to each participant. 4.The frequencies of these categories are compared with other variables.

4 Some categories in a coding scheme (Tomida & Maruno, 2005) FunctionsDescriptions Counter- argument Providing one's own ideas which are in opposition to another members ideas. DoubtDoubting certainty what another member said. InterpretationInterpreting what another member means in his/her previous utterance. ConfirmationMaking sure whether ones own understanding of another members utterance is correct or not. ExplanationAdding more detailed explanation for ones previous utterance.

5 However… nMethodological Problems lRelatively low reliability lTime consuming nAutomatic coding system is needed. nWhat index can be utilized for the automatic coding system?

6 Possible index for functional inference nDiscourse Marker (Schiffrin, 1987) nSingle words or lexicalized phrases that are supposed to have a function of organizing discourse structure. nExample: and, but, because, now, then, and I mean etc. nBy the way signals the start of a digression. nAnyway signals the return from a digression.

7 Limitation of the discourse maker nDiscourse marker is inductively assumed as index to signal a specific function. nHowever, a typical marker is not always accompanied with all utterances that surely have such a function. Not enough for accurate detection nIf many different markers are combined, more accurate and more robust inference system will be possible

8 Concept of Discourse Marker Complex Original discourse marker Discourse Marker Complex Number of wordOne or a fewMore than 10 or so Form of markerWord or phraseConditional expression with word or phrase By what markers function is determined Theoretical assumption Empirical examination

9 Corpus construction nParticipants: Japanese College students n43 participants, divided into 10 groups. nDiscussion: 30 min. nTask: To jointly construct a naïve model which explains causal mechanisms of Japanese teenagers aggression. nDiscussions were transcribed and tagged. lSpeakers name lUtterance function

10 DMC construction process Explore candidate words for DMC, referring to manually coded utterances. Calculate coverage rates of the candidate words. Construct a DMC, combining all the candidate words. The constructed DMC is repeatedly examined and modified. Analysis tool: HK-Coder (Higuchi, 2001) (internally, also MySQL and Chasen are used)

11 Exploring candidate words for DMC Grou p Utteranc e No. Ss No. Functio n Utterance Content J24528 Counter- argumen t But G1669 Counter- argumen t My opinion is a little bit different from yours C316114 Counter- argumen t No isnt it

12 Results nWe have constructed DMCs for: lCounter-argument lConfirmation nAlso found some categories cannot be distinguished from each other. lCounter-argument & Explanation lWhen people make a counter-argument, they usually add detailed reasons for being against.

13 An abbreviated sample of DMC for counter-argument or ( or ) or ( ' ' ) or ( or ) or ( and ' ' and ( or ) ) or ( ' ' and ( or ) ) or ( ' ' or ' ' ) or

14 Coverage & correlation of DMCs DMC for Counter- argument DMC for Confirmation FunctionCoverage rp Coverage rp Counter- argument 89/ 125 (71.2%).74.00 6 / 125 (4.8%).03.83 Doubt 18 / 60 (30.0%).18.26 5 / 60 (8.33%).44.00 Interpretation 27 / 119 (22.69%).16.31 22 / 119 (18.49%).27.08 Confirmation 14 / 178 (7.9%).32.04 91 / 178 (51.1%).53.00 Explanation 108 / 225 (48.0%).84.00 24 / 225 (10.7%).02.91

15 Preliminary validation of DMCs nCorrelations with self-rated conversational behaviors during discussion ( 7-point scale ). Counter-argumentConfirmation DMCManualDMCManual How often did you argue against other members?.52 (p >.001).52 (p >.001).22 (.17).23 (.14) How often were you challenged by other members?.30 (.05).32 (.03).001 (.99).09 (.57)

16 Conclusions nAccuracy of DMC is not perfect, but enough. lNot enough for one-to-one precise matching. lEnough for discovering individual differences among people: Who is more likely to generate targeted utterance? nDMC is useful for discourse analysis.

17 Further Task nConstruct DMCs for other conversational functions. nValidate with other similar corpus. nUtilize contextual information. nClassify some meaning words. nUtilize other techniques (hopefully). l Interactive Evolutionary Computation (IEC) for automated exploration of words and phrases.

18 Thank you netedu@ybb.ne.jp


Download ppt "Inferring conversational functions in Japanese discourse with Discourse Marker Complex Eiji Tomida & Shunichi Maruno Faculty of Human-Environmental Studies,"

Similar presentations


Ads by Google