Tid: Torsdag 22 maj 2014, kl 15:00–17:00
Plats: C 307, Södra huset, Frescati

Postseminarium följer direkt efter seminariet i institutionens pentry.


In this talk I will present (an outline of a) model for factors that predict the speed at which a word will get replaced in the lexicon.

Words change and are replaced. But not all words are replaced at the same pace, or for the same reasons. For example, many people would agree with the observation that words for bodily fluids (like words for 'vomit', 'urine') seem to be replaced more often over time than do words for inanimate objects in nature ('stone', 'tree'). From this simple observation arise a host of questions – What are the other factors that affect whether a word will change or not? How much faster are some words replaced than others? How does the replacement actually happen?

Pagel et al (2007) have suggested that frequency plays a major role in which type of words get replaced fast and which get replaced more slowly. The more frequent a word is, the less likely that it will be replaced. They calculated the speed of lexical replacement for the 200 item Swadesh list, and found that this correlated separately for four languages with the frequency of the translation equivalents in the languages (r around 0.35, p<0.0001 for all languages). If information on Word class was added, the correlation was even higher (r around 0.7 for all languages, p<0,0001).

In this talk, I will address some issues with the Pagel et al study. I will present a more diversified model of the factors that co-occur with lexical replacement. Looking only at open word class items, frequency alone has an effect size of .27 (p<.001) in determining the speed of lexical replacemen. Other factors are the number of synonyms typically available for the word (r=.23, p<.01); the level of imageability and concreteness (r=.27,
p<.001) and the collocational strength the word has with other words in corpora (r=.3, p<.001).

I will also report on some interesting negative results: potentially likely factors that I have not been able to demonstrate affect the speed of lexical replacement. Some examples are the number of word senses, the level of emotional (taboo) charge, and the number of readily elicited association words.


Bernhard Wälchli & Ljuba Veselinova