| 電子情報通信学会 研究会発表申込システム 講演論文 詳細 |
技報閲覧サービス [ログイン] 技報アーカイブ |
| トップに戻る | 前のページに戻る | [Japanese] / [English] |
| 講演抄録／キーワード | ||
| 講演名 | 2017-07-22 11:10 Context and semantic composition of multiple dimension representations in real-time comprehension ○Yao-Ying Lai（Yale Univ.）・David Braze（Haskins Lab.）・Maria Pinango（Yale Univ.）  TL2017-14 |
|
| 抄録 | （和） | We investigate the role of context in sentential meaning composition of aspectual verbal predicates (AspV). This semantic class involves morpho-syntactically underspecified composition: Sentences with aspectual verbs, e.g., "Dave began/continued/finished this CD of classic Jazz hits" are semantically ambiguous between (i) an agentive reading along the eventive dimension (e.g., Dave began/continued/finished recording this CD...), in which the complement is a participant of an event, and (ii) a constitutive reading along the spatial/informational dimension (e.g., Dave's song was the first/medial/final track on this CD...), in which the complement refers to a physical entity or a body of informational content. We hypothesize that AspVs (e.g., begin/continue/finish) require their complement to denote a structured individual, which is construed as an axis to be mapped along various dimensions (spatial, informational, temporal, eventive). This mapping is done via a set of lexical functions encoded in AspVs (e.g., f-space, f-info, f-time). The composition of an AspV + complement yields multiple dimension representations, and the ultimate sentence interpretation is determined by the specific dimension chosen in context. Results of an eye-tracking experiment show a main effect of verb type, such that AspV sentences engendered longer reading times than the control condition (p<0.001). We did not observe an effect of context facilitation. This suggests that the lexical-function retrieval and dimension-ambiguity resolution process are initially encapsulated from the larger contextual circumstances of AspV utterances, and are therefore required as part of the meaning implementation of AspVs. Conclusion: This study reveals that in real-time comprehension, (i) sentences involving dimension ambiguity engender more cost; (ii) exhaustive lexical retrieval and semantic composition of multiple dimension representations take place regardless of context. The role of context is likely to constrain, rather than determine, the appropriate sentence reading and only after multiple dimension representations are construed. |
| （英） | We investigate the role of context in sentential meaning composition of aspectual verbal predicates (AspV). This semantic class involves morpho-syntactically underspecified composition: Sentences with aspectual verbs, e.g., "Dave began/continued/finished this CD of classic Jazz hits" are semantically ambiguous between (i) an agentive reading along the eventive dimension (e.g., Dave began/continued/finished recording this CD...), in which the complement is a participant of an event, and (ii) a constitutive reading along the spatial/informational dimension (e.g., Dave's song was the first/medial/final track on this CD...), in which the complement refers to a physical entity or a body of informational content. We hypothesize that AspVs (e.g., begin/continue/finish) require their complement to denote a structured individual, which is construed as an axis to be mapped along various dimensions (spatial, informational, temporal, eventive). This mapping is done via a set of lexical functions encoded in AspVs (e.g., f-space, f-info, f-time). The composition of an AspV + complement yields multiple dimension representations, and the ultimate sentence interpretation is determined by the specific dimension chosen in context. Results of an eye-tracking experiment show a main effect of verb type, such that AspV sentences engendered longer reading times than the control condition (p<0.001). We did not observe an effect of context facilitation. This suggests that the lexical-function retrieval and dimension-ambiguity resolution process are initially encapsulated from the larger contextual circumstances of AspV utterances, and are therefore required as part of the meaning implementation of AspVs. Conclusion: This study reveals that in real-time comprehension, (i) sentences involving dimension ambiguity engender more cost; (ii) exhaustive lexical retrieval and semantic composition of multiple dimension representations take place regardless of context. The role of context is likely to constrain, rather than determine, the appropriate sentence reading and only after multiple dimension representations are construed. |
|
| キーワード | （和） | Semantic composition / Context / Underspecification / Semantic ambiguity / Aspectual verbs / Lexical retrieval / Structured individuals / dimension representations |
| （英） | Semantic composition / Context / Underspecification / Semantic ambiguity / Aspectual verbs / Lexical retrieval / Structured individuals / dimension representations | |
| 文献情報 | 信学技報, vol. 117, no. 149, TL2017-14, pp. 7-11, 2017年7月. | |
| 資料番号 | TL2017-14 | |
| 発行日 | 2017-07-15 (TL) | |
| ISSN | Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 | |
| 著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します．(許諾番号：10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) | |
| PDFダウンロード | TL2017-14 |
|
| 研究会情報 | |
| 研究会 | TL |
| 開催期間 | 2017-07-22 - 2017-07-23 |
| 開催地（和） | 国立国語研究所 |
| 開催地（英） | NINJAL |
| テーマ（和） | 人間の言語処理と学習 |
| テーマ（英） | Human Language Processing and Learning |
| 講演論文情報の詳細 | |
| 申込み研究会 | TL |
| 会議コード | 2017-07-TL |
| 本文の言語 | 英語 |
| タイトル（和） | |
| サブタイトル（和） | |
| タイトル（英） | Context and semantic composition of multiple dimension representations in real-time comprehension |
| サブタイトル（英） | |
| キーワード(1)（和/英） | Semantic composition / Semantic composition |
| キーワード(2)（和/英） | Context / Context |
| キーワード(3)（和/英） | Underspecification / Underspecification |
| キーワード(4)（和/英） | Semantic ambiguity / Semantic ambiguity |
| キーワード(5)（和/英） | Aspectual verbs / Aspectual verbs |
| キーワード(6)（和/英） | Lexical retrieval / Lexical retrieval |
| キーワード(7)（和/英） | Structured individuals / Structured individuals |
| キーワード(8)（和/英） | dimension representations / dimension representations |
| 第1著者 氏名（和/英/ヨミ） | Yao-Ying Lai / Yao-Ying Lai / |
| 第1著者 所属（和/英） | Yale University (略称： イェール大) Yale University (略称： Yale Univ.) |
| 第2著者 氏名（和/英/ヨミ） | David Braze / David Braze / |
| 第2著者 所属（和/英） | Haskins Laboratories (略称： Haskins Lab.) Haskins Laboratories (略称： Haskins Lab.) |
| 第3著者 氏名（和/英/ヨミ） | Maria Pinango / Maria Pinango / |
| 第3著者 所属（和/英） | Yale University (略称： イェール大) Yale University (略称： Yale Univ.) |
| 講演者 | 第1著者 |
| 発表日時 | 2017-07-22 11:10:00 |
| 発表時間 | 30分 |
| 申込先研究会 | TL |
| 資料番号 | TL2017-14 |
| 巻番号（vol） | vol.117 |
| 号番号（no） | no.149 |
| ページ範囲 | pp.7-11 |
| ページ数 | 5 |
| 発行日 | 2017-07-15 (TL) |
[研究会発表申込システムのトップページに戻る]