由分類看認知: 以英語量詞為例

Abstract

人類日常生活中最常做的認知活動即是將物體或事件做分類，而分類的概念也是其他認知活動的基礎。人類對於物體或事件的記憶也是來自於將新的事物做分類之後與先前的舊知識做比對，再將其歸納到不同的記憶區塊中，因此將週遭的物體或事件做分類是人類認識並探索這個世界的第一步。了解分類的機制是得知人類認知思考與運作方式的核心議題，因此也是了解人何以為人的基礎 (Lakoff 1987)，故人們到底是如何在認知系統中將事物做分類，是許多科學家從上個世紀起便極力研究的領域之ㄧ。分類在語言中最直接的呈現即是藉由使用分類詞與度量詞，因此本研究觀察28個英語度量詞及其搭配詞探討三個主要議題: 分類詞與度量詞之分別、英語度量詞功能區辨、與英語度量詞分類。 首先藉助English Sketch Engine與Suggested Upper Merged Ontology (SUMO)探討觀察blade, sheet, loaf, slice, piece, leap, brood, litter, flight, crowd, pride, colony, pack, group, herd, flock, cluster, clump, swarm, wad, line, coil, string, row, stack, heap, pile,與bunch等28個英語量詞，發現即使英語被歸類為沒有分類詞的語言(Allan 1977b)，有些英語量詞仍具備類似分類詞的功能，而如此的觀察引發我們重新思考分類詞與度量詞之間的分別。過去文獻中時常將兩種詞類做簡易二分法(Allan 1977b, Tai 1990, Ahrens1994)，但本研究認為兩之間應形成ㄧ連續線段，其中包含兩者間過渡的模糊地帶，但兩端仍為典型的分類詞與度量詞。我們將此線段稱之為分類詞與度量詞連續線段。 因為此連續線段上過渡模糊地帶的存在，我們再次證實伴隨分類而來的模糊性對於人類認知這個世界實為必要。做分類時不同的切入角度也會影響分類的結果，在將具有分類特性的度量詞做分類時也是ㄧ樣的過程。觀察28個英語量詞與其搭配詞後發現，英語量詞可因其功能上的不同加以分類，過去Allan (1977a, b)雖將英語度量詞做過分類，但未表現出不同英語量詞間功能上的差異，所以在此研究中，我們依據英語量詞的功能將其重新做分類。英語量詞可分成客觀與主動使用兩大類，客觀使用的英語量詞多與科學上專有單位詞相同，而主觀使用的量詞常經由英語其他詞類相借而來，並形成分類詞與度量詞連續線段。 經此研究發現，英語使用者即使在文法上沒有規定必須使用量詞的情形下，仍然因為希望能傳達給聽者最精確的訊息而選用量詞。在選用的過程中，對於這個世界的認知與將事物的分類便會影響其所選定的量詞。研究過程中所使用的輔助工具English Sketch Engine與SUMO，對於大量語料的初步分析有化繁為簡的功效，但仍有其不完備之處，在本文最後也對於語料庫語言學中所使用的工具做了反思，希望能為本領域貢獻ㄧ己之力。

Categorization is one of the basic cognitive activities conducted by human beings in everyday life. This study is an investigation of the relation between human categorization and English measure words. Thoroughly analyzing 28 English measure words, we have touched upon the issues on distinguishing between the measure word and the classifier, functions of English measure words, and the taxonomy of them. Analyzing English measure words blade, sheet, loaf, slice, piece, leap, brood, litter, flight, crowd, pride, colony, pack, group, herd, flock, cluster, clump, swarm, wad, line, coil, string, row, stack, heap, pile, and bunch with aid of the English Sketch Engine and the Suggested Upper Merged Ontology, we found that some English measure words are functionally similar to classifiers. As a result, it is claimed that English has classifier-like measure words regardless of English being considered as a non-classifier language in previous literature (Allan 1977b). The observation of classifier-like measure words inspired us to reconsider the distinction between measure words and classifiers. While many studies (Allan 1977b, Tai 1990, Ahrens1994) attempted to make a binary distinction between measure words and classifiers, the present study proposes that the relation between classifiers and measure words should be seen as a continuum with fuzzy boundaries. There are true measure words and classifier-like measure words in English. The classifier-to-measure word continuum proposed in this study demonstrates that categorization may not always result in clear-cut boundaries between categories. Fuzziness is necessary for human cognition to understand and conceptualize the world. Categorization is also a matter of perspective, in which different perspectives would result in various forms of taxonomies of entities. The English measure words chosen for analyses are all used subjectively according to the speaker’s perspective. However, there is another group of measure words that are objectively used, including standard measures of quantity and quality. Though all are known as “measure words,” their functional statuses vary. We, therefore, suggest there is a need to differentiate between these two groups of measure words, namely objective measure words and subjective measure words. Allan’s (1977a, b) taxonomy failed to point out this difference, so this study re-categorized English measure words, and proposed a new taxonomy.

English measure words, unlike Chinese or Thai, do not require classifiers or measure words before noun phrases. However, measure words are frequently used by English speakers. Considering language economy and efficiency, we wonder why speakers would bother to choose and make use of English measure words if not obligated. This study suggests that the additional information given eases the process of communication, and categorization plays an important role during the cognitive activity of choosing the appropriate measure word. Finally, through the process of investigating these 28 English measure words, it is found that the reliability of the search results is still questionable, because it is not guaranteed that all the data provided by the corpora are correct or relevant, which would then affect the credibility of the follow-up analyses. In addition to the usage of corpus, applying the Suggested Upper Merged Ontology to the present study gave forth to the finding that SUMO and WordNet may not cover all aspects of how human perceive and conceptualize the world. Hence, though computational tools aid the process of data collection, manual check is still suggested.