| 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 | 2018-07-29 10:30 Explaining the Morpheme Acquisition Order with Contingency and Type Frequency ○Zoe Pei-sui Luk（Edu U of HK）  TL2018-34 |
|
| 抄録 | （和） | Although Brown (1973) and others identified an acquisition order of English grammatical morphemes, there is little consensus as to why the order resulted. Brown (1973) rejected parental frequencies as a determinant because of the insignificant correlation between the adult and child data, but his analysis using token frequency is questionable. The present study reanalyzes the role of frequency in the forms of contingency and type frequency, and tests whether they can predict the acquisition order. The contingency (ΔP) model postulates that judgements about the reliability of a cue that leads to an outcome depend on how often the cue and the outcome co-occur, how often only the cue or only the outcome occurs, and how often neither is present. Following Ellis (2006a), I apply this to the learning of grammatical morphemes. I treated a phonological form that is homophonous to a morpheme as a cue, and the meaning of the morpheme as an outcome. For instance, the phonological form [d] is used to signify pastness (e.g., called). However, there are cases where [d] occurs at the word-final position but has no past meaning (e.g., bend, which may be analyzed as benned), and cases where the phonological form is absent but the past meaning is present (e.g., ran). The contingency of a morpheme is the differential of the proportions of the events for which an outcome is present with and without the cue (Ellis, 2006a, 2006b). I hypothesized that the larger the contingency of a morpheme is, the earlier the morpheme will be acquired, because the morpheme?s meaning is less ambiguous for a large contingency. To test this, I examined the distributions of five morphemes, (progressive -ing, plural -s, third person singular -s, possessive -?s, and regular past -ed) in the data from the caregivers of the three children in Brown?s (1973) study. The data were annotated for whether the phonological form (cue) of a morpheme and the meaning of it (outcome) is present. I also examined the number of word types to which each morpheme is attached. According to Bybee (2010), the higher the type frequency of a morpheme is, the easier it is for learners to abstract patterns from the input, and thus the earlier the morpheme will be acquired. Results from an ordinal logistic regression analysis show that there is a main effect of contingency (t = 2.179, p < 0.05). The main effect of type frequency is insignificant (t = 1.185, p = 0.236), but the interaction between contingency and type frequency is significant (t = -2.279, p < 0.05). These findings suggest that both contingency and type frequency are important predictors of the acquisition order, in that morphemes with higher contingencies will be acquired earlier, but the order can be altered by the number of word types a morpheme is associated with. The results support both Ellis?s (2006a, 2006b) and Bybee?s (2010) claims that language acquisition can be explained by the statistical distribution of the input without resorting to a domain-specific mechanism in neither L1 nor L2 as claimed in the morphemes studies. |
| （英） | Although Brown (1973) and others identified an acquisition order of English grammatical morphemes, there is little consensus as to why the order resulted. Brown (1973) rejected parental frequencies as a determinant because of the insignificant correlation between the adult and child data, but his analysis using token frequency is questionable. The present study reanalyzes the role of frequency in the forms of contingency and type frequency, and tests whether they can predict the acquisition order. The contingency (ΔP) model postulates that judgements about the reliability of a cue that leads to an outcome depend on how often the cue and the outcome co-occur, how often only the cue or only the outcome occurs, and how often neither is present. Following Ellis (2006a), I apply this to the learning of grammatical morphemes. I treated a phonological form that is homophonous to a morpheme as a cue, and the meaning of the morpheme as an outcome. For instance, the phonological form [d] is used to signify pastness (e.g., called). However, there are cases where [d] occurs at the word-final position but has no past meaning (e.g., bend, which may be analyzed as benned), and cases where the phonological form is absent but the past meaning is present (e.g., ran). The contingency of a morpheme is the differential of the proportions of the events for which an outcome is present with and without the cue (Ellis, 2006a, 2006b). I hypothesized that the larger the contingency of a morpheme is, the earlier the morpheme will be acquired, because the morpheme?s meaning is less ambiguous for a large contingency. To test this, I examined the distributions of five morphemes, (progressive -ing, plural -s, third person singular -s, possessive -?s, and regular past -ed) in the data from the caregivers of the three children in Brown?s (1973) study. The data were annotated for whether the phonological form (cue) of a morpheme and the meaning of it (outcome) is present. I also examined the number of word types to which each morpheme is attached. According to Bybee (2010), the higher the type frequency of a morpheme is, the easier it is for learners to abstract patterns from the input, and thus the earlier the morpheme will be acquired. Results from an ordinal logistic regression analysis show that there is a main effect of contingency (t = 2.179, p < 0.05). The main effect of type frequency is insignificant (t = 1.185, p = 0.236), but the interaction between contingency and type frequency is significant (t = -2.279, p < 0.05). These findings suggest that both contingency and type frequency are important predictors of the acquisition order, in that morphemes with higher contingencies will be acquired earlier, but the order can be altered by the number of word types a morpheme is associated with. The results support both Ellis?s (2006a, 2006b) and Bybee?s (2010) claims that language acquisition can be explained by the statistical distribution of the input without resorting to a domain-specific mechanism in neither L1 nor L2 as claimed in the morphemes studies. |
|
| キーワード | （和） | Contingency / morpheme acquisition order / type frequency / / / / / |
| （英） | Contingency / morpheme acquisition order / type frequency / / / / / | |
| 文献情報 | 信学技報, vol. 118, no. 163, TL2018-34, pp. 107-107, 2018年7月. | |
| 資料番号 | TL2018-34 | |
| 発行日 | 2018-07-21 (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 | TL2018-34 |
|
| 研究会情報 | |
| 研究会 | TL |
| 開催期間 | 2018-07-28 - 2018-07-29 |
| 開催地（和） | 慶応大学 |
| 開催地（英） | Keio University |
| 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 | 2018-07-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） | Explaining the Morpheme Acquisition Order with Contingency and Type Frequency |
| サブタイトル（英） | |
| キーワード(1)（和/英） | Contingency / Contingency |
| キーワード(2)（和/英） | morpheme acquisition order / morpheme acquisition order |
| キーワード(3)（和/英） | type frequency / type frequency |
| キーワード(4)（和/英） | / |
| キーワード(5)（和/英） | / |
| キーワード(6)（和/英） | / |
| キーワード(7)（和/英） | / |
| キーワード(8)（和/英） | / |
| 第1著者 氏名（和/英/ヨミ） | Zoe Pei-sui Luk / Zoe Pei-sui Luk / |
| 第1著者 所属（和/英） | The Education University of Hong Kong (略称： Edu U of HK) The Education University of Hong Kong (略称： Edu U of HK) |
| 講演者 | 第1著者 |
| 発表日時 | 2018-07-29 10:30:00 |
| 発表時間 | 30分 |
| 申込先研究会 | TL |
| 資料番号 | TL2018-34 |
| 巻番号（vol） | vol.118 |
| 号番号（no） | no.163 |
| ページ範囲 | p.107 |
| ページ数 | 1 |
| 発行日 | 2018-07-21 (TL) |
[研究会発表申込システムのトップページに戻る]