Word-level Dependency-structure Annotation to Corpus of Spontaneous Japanese and Its Application Kiyotaka Uchimoto* Yasuharu Den † *National Institute of Information and Communications Technology (NICT) † Chiba University
Outline Background Dependency Structure in the CSJ Dependency-structure Annotation Word-level Dependency-structure Analysis Towards Construction of Middle Words Summary and future work
Background (1) Corpus of Spontaneous Japanese (CSJ) [Maekawa et al., 2000] l The largest spontaneous-speech corpus in the world l Include transcriptions of speeches as well as audio recordings l One tenth of the CSJ has been manually annotated with Morphemes, sentence boundaries, syntactic structures, discourse structures, prosodic information, etc
Background (2) Syntactic structure of a sentence l Represented by dependency relationships between bunsetus l As represented in the Kyoto University text corpus Syntactic structure of a bunsetsu is not considered nihon gata kokusai kouken ga (Japanese style international contribution) motome rare te iru (is required)
Dependency Structure in the CSJ (1) Dependency relationships between bunsetsus l Annotated within “sentences” in the CSJ Dependency relationships between words l Annotated within bunsetsus l Word segments in the word-level dependency structure: short words Short word approximates a term found in an ordinary dictionary Long word represents various compounds nihon gata kokusai kouken ga (Japanese style international contribution) motome rare te iru (is required)
Dependency Structure in the CSJ (2) Disfluencies characteristic to spontaneous speech l Self-correction Represented as dependency between bunsetsus, and label D is assigned to them Yamada (Yamada) Yamada san wa (Mr. Yamada) kyoujin na (strong) nikutai no (body) mochinushi da to (possessor) it te mashi ta ne (said) D (Yamada, Mr. Yamada said that he had a strong body.)
Dependency Structure in the CSJ (3) Disfluencies characteristic to spontaneous speech l Self-correction Represented as dependency between words, and label D is assigned to them kokuritsu (national) Nihon (Japanese) go (word) kokugo (Japanese language) kenkyuu (research) jo (institure) de case marker D (At National Japanese word, Japanese language research institute)
Dependency-structure Annotation Manual annotation l 199 speeches for dependency relationships between bunsetsus l 50 speeches for dependency relationships between words Human annotation by using a tool l Initial: every bunsetsu depends on the next l Step 1: two annotators examined each dependency and modified it if it was inappropriate l Step 2: a checker examined all dependencies Referred to audio recordings as well as transcriptions
Each line represents a bunsetsu Modified by mouse drag- and-drop Self-corrections, coordination, and appositives can be annotated with labels D, P, and A by right-clicking the mouse
Each line represents a word Modified by mouse drag- and-drop
Word-level Dependency-structure Analysis (1) Finding a modifiee for each word in a bunsetsu l Each dependency goes from left to right l The rightmost word is assumed to have no modifiee Existing methods were applied l Ex. shift-reduce method [Nivre and Scholz, 2004] nihon/noun gata/Suffix kokusai/noun kouken/noun ga/ppp gatanihon … kokusaikoukenga Input words stack
Word-level Dependency-structure Analysis (2) Experiments l 50 speeches in the CSJ Word-level dependencies (total: 33,429) –Every rightmost dependency in a bunsetsu was not counted l 10-fold cross validation l Features: words and their POS categories MethodDependency accuracy Baseline Shift-reduce (Nivre & Scholz, 2004) MST parser (McDonald et al., 2005) CaboCha (Kudo and Matsumoto, 2000) 98.6% 99.1%
Application of Word-level Dependency-structure In text-to-speech synthesis l Basic unit is required to indicate appropriate pronunciation and accent Long word dandanba^takegairaigokanahyoukimanyogana Short word da^ndan (layered) hatake (fields) gairai (foreign) go (word) kana (kana) hyouki (orthography) manyo (myriad) kana (kana) Long word dandanba^takegairaigokanahyoukimanyogana Middle word dandanbatakegairaigokanahyoukimanyogana Short word da^ndan (layered) hatake (fields) gairai (foreign) go (word) kana (kana) hyouki (orthography) manyo (myriad) kana (kana) “rendaku” (Weijer et al., 2005)
Long word dandanba^takegairaigokanahyoukimanyogana Middle word dandanbatakegairaigokanahyoukimanyogana Short word da^ndan (layered) hatake (fields) gairai (foreign) go (word) kana (kana) hyouki (orthography) manyo (myriad) kana (kana) Application of Word-level Dependency-structure A sound change or an accent change are blocked by right branched tree structures (Kubozono, 1995)
Construction of Middle Words Construction rule l Combining adjacent short words that have dependency relationships under the condition that a middle word is not longer than a long word Morphological information l If a middle word corresponds to a long word Extracted from the long word. l Otherwise Extracted from the rightmost short word in the middle word. Example kihon / shuuha / suu / pataan Noun Noun Suffix Noun (basic frequency pattern) kihon | shuuha suu pataan Noun
Middle Words and Accent Phrases Relationships between middle words and accent phrases (BI=2, 2+p, 2+b, 2+bp, 3) in the CSJ Long words (LW) (97,167) No accent phrase boundary (APB) in LW Accent phrase boundary (APB) in LW 94,0383,129 LW = MWLW > MWAPB in MW No APB in MW MW boundary corresponds to LW boundary or APB MW boundary corresponds neither to LW boundary nor to APB 93, , ,07554 nihonjin/gakushuusha rittai/chuushajou kaku|zokusei gen|jiten zen|shikiichi emuten|chuuouchi/heikatsuka yuudo/saidaika|kijun should be reduced
Summary and Future Work Dependency structure of a large, spontaneous, Japanese-speech corpus, Corpus of Spontaneous Japanese (CSJ) Application of a word-level dependency-structure l Constructing new basic units, middle words l Middle words: useful as constituents of accent phrases Annotation to the Balanced Corpus of Contemporary Written Japanese (BCCWJ) l Supported by the priority area program ‘Japanese Corpus’, a five-year ( ) project