伝 康晴

1. 研究目的 本研究では，アンケート調査により音声認識システムに対するニーズの高い診療領域・診療場面を明らかにし，それらの領域・場面での辞書構築を図ることを目的とした．

国立国語研究所 音声言語研究領域国立国語研究所 音声言語研究領域 非常勤研究員国立国語研究所 音声言語研究領域 非常勤研究員国立国語研究所 音声言語研究領域 非常勤研究員国立国語研究所 音声言語研究領域国立国語研究所 音声言語研究領域 非常勤研究員国立国語研究所 音声言語研究領域 非常勤研究員千葉大学国立国語研究所 コーパス開発センター 非常勤研究員国立国語研究所共同研究プロジェクト「大規模日常会話コーパスに基づく話し言葉の多角的研究」では，『日本語日常会話コーパス』（CEJC）の構築を進めている。CEJCは，日常会話の多様性を捉え自然な会話行動が観察できるよう，様々な種類の会話をバランスよく収めることを目標に掲げている。2021年度末に予定している本公開に先立ち，コーパスの利用可能性や問題などを把握するために，目標とする200時間のうち50時間の会話データについて，2018年12月にモニター公開を開始した。本稿ではまず，コーパスの設計について，会話の収録法，データの公開方針，調査協力者の内訳，コーパスの規模や構成などの観点から概観する。次に，収録されているデータが設計通りバランスがとれているかを，話者と会話の両面から検証する。最後に，コーパスを用いた予備的分析を通して，CEJCモニター版を活用した研究の可能性を示す。

本WSの目的は、祭りの支度を共に行うことで、以下に示す共同体固有の〈心体<br /> 知〉を後継者世代が仲間内や現役世代と多対多の相互作用から習得していく過<br /> 程を紐解くことである。<br /> <br /> 〈心〉 成員たちがもつ価値観や見識、信頼感といったエートス<br /> （e.g. 他者への気配り、自己犠牲の精神）<br /> 〈体〉 成員間で力や身体位置の配分が必要な協働活動技法<br /> （e.g. 環境や道具に応じた他者との協調的動作）<br /> 〈知〉 成員たちがもつ祭りに固有の共有知識<br /> （e.g. 縄の結び方、唄、舞など）<br /> 本WSでは、共同体〈心体知〉を学習する成員たち自身のやり方を相互行為分析か<br /> ら炙り出す。

本稿では，国立国語研究所共同研究プロジェクト「均衡性を考慮した大規模日本語会話コーパス構築に向けた基盤整備」（リーダー：小磯，2014年7月～2015年8月）の活動について報告する。本プロジェクトの目標は，21世紀初頭の日本語母語話者の多様な会話行動を納めた大規模な日本語日常会話コーパスの構築を目指し，その基盤整備として，（1）均衡性を考慮した会話コーパスの設計，（2）種々の日常場面での会話を収録するための方法論，（3）日常会話を適切・効率的に転記するための方法論の策定を進めることである。本稿ではこのうち（1）に焦点を当て，均衡性を考慮したコーパス設計案を策定するために実施した，一日の会話行動の種類と従事時間に関する調査について報告する。調査では，首都圏在住の成人約250人を対象に，起床から就寝までの間に行ったそれぞれの会話について，いつ，どこで，誰と，何をしながら，どのような種類の会話を，どのくらいの長さ行ったか，などを問う調査項目に回答してもらった。その結果，日常会話には以下の傾向が見られることが分かった。（1）雑談や用談・相談が多く，会議・会合や授業・レッスン・講演は少ない。（2）少人数・短時間の会話が多い。（3）自宅や職場・学校など，私的あるいは公的に主たる空間での会話が多い。（4）いくつかの調査項目の間には対応関係が見られる。これらの分析結果にもとづき，均衡性を考慮した日常会話コーパスの設計方針について議論する。In this paper, we report on the activity of a preparatory project to build a large-scale corpus of conversational Japanese (NINJAL collaborative research project, 2014/7/1-2015/8/31). The overall aims of this project are: i) to establish a corpus design for collecting various kinds of everyday conversations in a balanced manner, ii) to develop a methodology of recording naturally occurring conversations, and iii) to create a transcription system suitable for effectively transcribing natural conversations. This report focuses on the first issue of establishing a corpus design. We first describe our survey of everyday conversational behavior, conducted with about 250 Japanese adults last year, in order to reveal how diverse our everyday conversational behavior is, and to build an empirical foundation for corpus design. The questionnaire included when, where, how long, with whom, and in what kind of activity informants were engaged in conversations. We found that ordinary conversations show the following tendencies: i) they mainly consist of chats, business talks, and consultations; ii) in general, the number of participants is small and the duration is short; iii) many conversations are conducted in private places such as homes, as well as in public places such as offices and schools; and iv) some questionnaire items are related to each other. Based on these results, we discuss how to design a balanced corpus of conversational Japanese.

In this study, we investigated segment lengthening in spontaneous Japanese based on a quantitative analysis of a large-scale corpus, focusing on the following three locations at which lengthening frequently occurs: the final segments of (i) clause-initial preface tokens (fillers and conjunctions), (ii) clause-initial wa-marked topic phrases, and (iii) clause-final particles. Two cognitive factors, namely clause complexity and boundary depth, were precisely analyzed using statistical models that also accounted for several phonological and syntactic factors. The results showed that in addition to the reliably strong effects of some phonological factors such as the presence of a following pause and the presence of boundary pitch movement, the effects of two cognitive factors were also evident. The way in which lengthening is related to the cognitive factors, however, varies significantly by location and token type. Lengthening of clause-final particles was affected by boundary depth, while lengthening of the topic marker wa of clause-initial topic phrases was influenced by clause complexity. Lengthening of the filler e was affected by both factors. A significant interaction between the two factors was also observed for the filler ano. We discuss the implications of these results as well as agendas for improving the current analysis.

Conversational interactions contribute not only to the sharing of information and <br>establishment of consensus but also to the construction and sustenance of mutual trust<br>among conversational participants in our daily lives. The interrelationship between<br>trust and conversational interactions has not been studied extensively in cognitive sci-<br>ence. One reason for this lack of research is the fact that a study of social emotions such<br>as trust requires real fields, since social emotions in their natural, non-artificial forms<br>are not readily observable in laboratory settings. We introduce a notion of concern<br>alignment to describe the surface conversational processes toward mutual trust forma-<br>tion. Focusing on medical communications as our research field, we collected health<br>guidance conversations between nurses and patients who were diagnosed as having<br>metabolic syndrome, and we provide a qualitative analysis of the structure of conver-<br>sations in terms of a set of dialogue acts we propose for the description of concern<br>alignment processes. We demonstrate that the idea of concern alignment enables us to<br>capture and elucidate both the local and the global structures of mutual trust formation<br>in conversational consensus-building processes. We also discuss underlying mechanisms<br>connecting concern alignment and mutual trust.

The balanced corpus of contemporary written Japanese (BCCWJ) is Japan's first 100 million words balanced corpus. It consists of three subcorpora (publication subcorpus, library subcorpus, and special-purpose subcorpus) and covers a wide range of text registers including books in general, magazines, newspapers, governmental white papers, best-selling books, an internet bulletin-board, a blog, school textbooks, minutes of the national diet, publicity newsletters of local governments, laws, and poetry verses. A random sampling technique is utilized whenever possible in order to maximize the representativeness of the corpus. The corpus is annotated in terms of dual POS analysis, document structure, and bibliographical information. The BCCWJ is currently accessible in three different ways including Chunagon a web-based interface to the dual POS analysis data. Lastly, results of some pilot evaluation of the corpus with respect to the textual diversity are reported. The analyses include POS distribution, word-class distribution, entropy of orthography, sentence length, and variation of the adjective predicate. High textual diversity is observed in all these analyses.

In this paper, we describe the design and development of a new version of the Corpus of Spontaneous Japanese (CSJ), which is a large-scale spoken corpus released in 2004. CSJ contains various annotations that are represented in XML format (CSJ-XML). CSJ-XML, however, is very complicated and suffers from some problems. To overcome this problem, we have developed and released, in 2013, a relational database version of CSJ (CSJ-RDB). CSJ-RDB is based on an extension of the segment and link-based annotation scheme, which we adapted to handle multi-channel and multi-modal streams. Because this scheme adopts a stand-off framework, CSJ-RDB can represent three hierarchical structures at the same time: inter-pausal-unit-top, clause-top, and intonational-phrase-top. CSJ-RDB consists of five different types of tables: segment, unaligned-segment, link, relation, and meta-information tables. The database was automatically constructed from annotation files extracted from CSJ-XML by using general-purpose corpus construction tools. CSJ-RDB enables us to easily and efficiently conduct complex searches required for corpus-based studies of spoken language.

In this paper, we propose an experimental method for building a specialized corpus for training and evaluating backchannel prediction models of spoken dialogue. To develop a backchannel prediction model using a machine learning technique, it is necessary to discriminate between the timings of the interlocutor's speech when more listeners commonly respond with backchannels and the timings when fewer listeners do so. The proposed corpus indicates the normative timings for backchannels in each speech with millisecond accuracy. In the proposed method, we first extracted each speech comprising a single turn from recorded conversation. Second, we presented these speeches as stimuli to 89 participants and asked them to respond by key hitting whenever they thought it appropriate to respond with a backchannel. In this way, we collected 28983 responses. Third, we applied the Gaussian mixture model to the temporal distribution of the responses and estimated the center of Gaussian distribution, that is, the backchannel relevance place (BRP), in each case. Finally, we synthesized 10 pairs of stereo speech stimuli and asked 19 participants to rate each on a 7-point scale of naturalness. The results show that backchannels inserted at BRPs were significantly higher than those in the original condition.

Because of the tremendous effort required for recording and transcription, large-scale spoken language corpora have been hardly developed in Japanese, with a notable exception of the Corpus of Spontaneous Japanese (CSJ). Various research groups have individually developed conversation corpora in Japanese, but these corpora are transcribed by different conventions and have few annotations in common, and some of them lack fundamental annotations, which are prerequisites for conversation research. To solve this situation by sharing existing conversation corpora that cover diverse styles and settings, we have tried to automatically transform a transcription made by one convention into that made by another convention. Using a conversation corpus transcribed in both the Conversation-Analysis-style (CA-style) and CSJ-style, we analyzed the correspondence between CA's 'intonation markers' and CSJ's 'tone labels,' and constructed a statistical model that converts tone labels into intonation markers with reference to linguistic and acoustic features of the speech. The result showed that there is considerable variance in intonation marking even between trained transcribers. The model predicted with 85% accuracy the presence of the intonation markers, and classified the types of the markers with 72% accuracy.

Conversational interactions contribute not only to the sharing of information and establishment of consensus but also to the formation and maintenance of mutual trust among conversational participants. We introduce a notion of concern alignment to describe the process of mutual trust formation in consensus-building dialogues. We propose a conceptual picture of dialogue processes taking place toward concern alignment and trust formation, and provide a qualitative analysis of health guidance dialogues between nurses and patients who were diagnosed as having metabolic syndrome to support the picture. We argue, based on our dialogue analysis result and other research in psychological and anthropological fields, that sharing of knowledge on values and preferences of others negotiated through concern alignment makes it possible for people to choose their actions based on trust on their partners. (C) 2013 The Authors. Published by Elsevier Ltd.

Conversational interactions contribute not only to the sharing of information and establishment of consensus but also to the formation and maintenance of mutual trust among conversational participants. We introduce a notion of concern alignment to describe the process of mutual trust formation in consensus building dialogues. We propose a conceptual picture of dialogue processes taking place toward concern alignment and trust formation, and provide a qualitative analysis of health guidance dialogues between nurses and patients who were diagnosed as having metabolic syndrome to support the picture. We argue, based on our dialogue analysis result and other research in psychological and anthropological fields, that sharing of knowledge on values and preferences of others negotiated through concern alignment makes it possible for people to choose their actions based on trust on their partners.