Hanae Koiso - Academia.edu (original) (raw)

Papers by Hanae Koiso

国立国語研究所千葉大学国立国語研究所National Institute for Japanese Language and LinguisticsChiba UniversityNationa... more 国立国語研究所千葉大学国立国語研究所National Institute for Japanese Language and LinguisticsChiba UniversityNational Institute for Japanese Language and LinguisticsCSJ-RDB Version 2.0 利用の手引きおよびサンプルファイル（参考資料として CSJ-RDB Version 1.0 の講習会資料を同梱

Proceedings of LPSS 2019, Oct 30, 2020

Journal of the Phonetic Society of Japan, 2000

Technical report of IEICE. HIP, 2004

In this paper, we focus on weak clause boundaries in Japanese monologs in order to investigate th... more In this paper, we focus on weak clause boundaries in Japanese monologs in order to investigate the relationship of the length of constituents following weak boundaries to three acoustic and linguistic features: 1) occurrence rate of fillers, 2) occurrence rate of boundary pitch movements, and 3) degree of lengthening of clause-final morae. We found that all these features were significantly correlated with the length of following constituents. Most importantly, boundary pitch movements had an additional effect that can be distinct from the effect of clause-final lengthening. These results suggest that Japanese speakers have earlier-occurring items that help them deal with cognitive load in speech planning, in addition to fillers and other clause-initial disfluencies.

In this paper, we try to lay the foundation for an informational model of human conversations tha... more In this paper, we try to lay the foundation for an informational model of human conversations that formally specifies, for each stage of a conversation, what information is or is not made available to conversants through various forms of “cuings” that occur in the conversation. Squarely facing the fact that multiple lines of cuings often co-occur and interact with each other in the course of an actual conversation, we classify, illustrate, and mathematically characterize their interactions on the basis of Barwise and Seligman’s general theory of information flow (1997). A conversation is what conversants construct. Thus, to explain the construction of a conversation is to explain the conversants’ behaviors. We may try to do the latter in various ways. With “conversation analysts” (e.g. Sacks et al. 1974), we may appeal to some social conventions that the participants actually attend to and comply with. Or with “discourse analysts” (e.g. Labov and Fanshel 1977), we may appeal to general rules specifying possible sequences of speech acts. Or we might combine two approaches (Traum 1994) or take still another approach. Whatever path we may take, such an endeavor must involve or presuppose some explanation of what information is or is not available to the conversants at a given stage of the conversation. For example, the application of a particular item of the turn-exchange rules would crucially depend on the information available to conversants about the turn-occupancy state at the point; likewise, depending on what information is assumed to be available to a conversant concerning the prior sequence of speech acts, the sequencing rules on speech acts predict different behaviors of the participant. In most cases, theorists manage to correctly guess the available information to a participant (by “putting themselves in his or her position”) to make specific predictions about the conversant’s behaviors. This practice, however, runs the risk of trivializing whatever theory one may have ∗Also with Nara Advanced Institute of Science and Technology.

le: and s c . Historically, dynamic semanticists (Kamp 1981, Heim 1982, Groenendijk and Stokhof 1... more le: and s c . Historically, dynamic semanticists (Kamp 1981, Heim 1982, Groenendijk and Stokhof 1991) focused on utterances of sentences in c, and studied how they update scoreboards about the topic of c. Thus, their interests were in the tertiary relation s t , where updating events e are confined to utterances of sentences and s t and s t are scoreboards about c's main topic. Some authors, including Lewis himself and Stalnaker (1978), emphasized that the update potentials of utterances may depend on conversational parameters such as speaker, addressee, referential salience, and point of reference. Thus, they were interested in a slightly different relation, hs c ; s t i . Even in their cases, however, e was confined to utterances of sentences, and the focus was on the shift from s t to s t . These confinements are only natural, given that the project's main concern was the interpretation of a sentence, namely, the information carried by a sentential utterance co

In this paper, we present the outline of a new model of turn-taking that is applicable not only t... more In this paper, we present the outline of a new model of turn-taking that is applicable not only to smooth transitions but also to transitions involving overlapping speech. We identify acous-tic, prosodic, and syntactic cues in overlapped utterances that elicit early initiation of a next turn, based on a quantitative anal-ysis of Japanese three-party conversations, proposing a model for predicting a turn's completion in an incremental fashion us-ing sources from units at multiple levels.

PsycEXTRA Dataset

The study reported in this paper focuses on different functions of echoing in Japanese dialogues.... more The study reported in this paper focuses on different functions of echoing in Japanese dialogues. Echoing is defined as a speaker's lexical repeat of (parts of) an utterance spoken by a conversation partner in a previous turn. The phenomenon was investigated in three task-oriented, informal dialogues. Repeats in this corpus were labelled in terms of whether or not the speaker had integrated the other person's utterance into hislher own body of knowledge. The investigation brought to light that the level of integration is reflected in a number of lexical and prosodic correlates. These features are discussed regarding their information potential, i.e., their accuracy and comprehensiveness as signals.

… and discourse processing, …, 1994

Joho Shori Gakkai Kenkyu Hokoku, 2001

Abstract; A large-scale corpus of spontaneous Japanese speech is being compiled as a joint work o... more Abstract; A large-scale corpus of spontaneous Japanese speech is being compiled as a joint work of the National Institute for Japanese Language, the Communications Research Laboratory, and Tokyo Institute of Technology. This corpus is designed to contain about 700 hours of speech (about 7 million morphemes), its transcription, and various tagging information such as POS information. In this paper, the transcription criteria designed specifically for CSJ is described after a brief overview of the general architecture of the ...

SIG SLUD, 2005

Abstract; The effects of language-internal and language-external factors upon the moraic-nasaliza... more Abstract; The effects of language-internal and language-external factors upon the moraic-nasalization of particle/no/was analyzed using the Corpus of Spontaneous Japanese. Three obvious factors and one seemingly factor were found out. First, the rate of moraic-nasalization differed considerably depending on the type of particle. The rate was much more higher when the particle was a nominalization-particle than it was a case-particle. Second, moraic-nasalization was favored when the onset consonants of the following ...

In Search of Basic Units of Spoken Language

5th International Conference on Spoken Language Processing (ICSLP 1998)

The study reported in this paper focuses on different functions of echoing in Japanese dialogues.... more The study reported in this paper focuses on different functions of echoing in Japanese dialogues. Echoing is defined as a speaker’s lexical repeat of (parts of) an utterance spoken by a conversation partner in a previous turn. The phenomenon was investigated in three task-oriented, informal dialogues. Repeats in this corpus were labeled in terms of a 5-point scale which expressed the level to which a speaker had integrated the other person’s utterance into his/her own body of knowledge. Kappa statistics showed that the labels could reliably be reproduced by three independent subjects. The investigation brought to light that the level of integration is reflected in a number of lexical and prosodic correlates. These features are discussed regarding their information potential, i.e., their accuracy and comprehensiveness as signals.

Proceedings of LPSS 2019, Oct 30, 2020

Journal of the Phonetic Society of Japan, 2000

Technical report of IEICE. HIP, 2004

PsycEXTRA Dataset

… and discourse processing, …, 1994

Joho Shori Gakkai Kenkyu Hokoku, 2001

SIG SLUD, 2005

In Search of Basic Units of Spoken Language

5th International Conference on Spoken Language Processing (ICSLP 1998)

The study reported in this paper focuses on different functions of echoing in Japanese dialogues.... more The study reported in this paper focuses on different functions of echoing in Japanese dialogues. Echoing is defined as a speaker’s lexical repeat of (parts of) an utterance spoken by a conversation partner in a previous turn. The phenomenon was investigated in three task-oriented, informal dialogues. Repeats in this corpus were labeled in terms of a 5-point scale which expressed the level to which a speaker had integrated the other person’s utterance into his/her own body of knowledge. Kappa statistics showed that the labels could reliably be reproduced by three independent subjects. The investigation brought to light that the level of integration is reflected in a number of lexical and prosodic correlates. These features are discussed regarding their information potential, i.e., their accuracy and comprehensiveness as signals.