Distributional Hypothesis: Words for 'Human Being' and their Estonian Collocates

The article was inspired by the Distributional Hypothesis by Zellig Harris, which states that words occurring in similar contexts tend to have a similar meaning. The hypothesis was tested by a comparison of the 10 most frequent collocates of the Estonian words for 'human being'. In the present study, the word collocate is used in a neo-Firthian sense, covering all the words that co-occur with the node word the most often. The collocates of the words inimene 'human', mees 'man', naine 'woman', laps 'child', tüdruk 'girl', poiss 'boy', tütar 'daughter', poeg 'son', ema 'mother' and isa 'father' were drawn from the context 'three words to the left' of the node word as occurring in the Newspaper subcorpus of the Balanced Corpus of Estonian. The comparison involved the 30 most frequent collocates for each node word. Assuming that a bigger number of shared collocates means a greater semantic closeness, intersections of collocates of the Estonian words for ‘human being’ were computed. It turned out that antonymous words had the highest number of collocates in common, which indicates that syntagmatic relations of words may also reflect some of their paradigmatic relations. In addition, what may be decisive for the part of speech of collocates, is analysed.

並列關鍵字

collocations ； distributional hypothesis ； human being ； syntagmatic relations ； antonymy ； Estonian

國際替代計量

全文下載

主題瀏覽