Polysemy and word frequency: A replication. (24th July 2018)

• Record Type: Journal Article
• Authors: Kuiper, Koenraad; Fromont, Robert; Gerhard, Daniel
• Is Part Of: Journal of research design and statistics in linguistics and communication science. Volume 4:Number 2(2017)
• Page Start: 144
• Page End: 155
• Publication Date: 2018-07-24
• Subjects: bnc -- wordnet -- polysemy -- zipf -- lexical frequency; Linguistics -- Research -- Periodicals; Linguistics -- Statistical methods -- Periodicals; Mathematical linguistics -- Periodicals; 410.72
• DOI: 10.1558/jrds.33751
• Languages: English
• ISSNs: 2052-417X

Abstract

Abstract : One piece of evidence adduced by George Kingsley Zipf for his eponymous law (Zipf, 1935) and its explanation of the principle of least effort (Zipf, 1949) is the hypothesis that a word's polysemy is proportional to the square root of its frequency (Levelt, 2013). Pawley (2006) following Zipf, also proposes that 'there is a strong general correlation between frequency and the extent of polysemy'. This paper replicates Zipf 's approach but with data drawn from different sources to those available to Zipf, namely, for word frequency, the Kilgarriff most frequent word list drawn from the BNC (Kilgarriff, 1995) and, as a measure of polysemy, the WordNet data for the polysemy of the words in Kilgarriff's list. It also takes note of the syntactic category of lexemes. More advanced statistical modelling is used. Zipf 's observations are confirmed with some provisos. Their utility is examined. Explanations for this relationship remain to be established.