Abstract
We investigated Zipf’s law in fluent and non-fluent aphasics’ spontaneous speech in English, Hungarian, and Greek. A previous study showed that the word frequency distribution in Dutch non-fluent aphasic speech conforms to Zipf’s law, although with a different slope. In this project we investigated to what extent these results can be generalized to other languages and to fluent aphasic speech. The results suggest that both the fluent and the non-fluent aphasic speech of English, Hungarian and Greek conform to Zipf’s law, and that differences in slope can be related to a language’s morphological properties and a group’s particular language impairments.
Acknowledgements
This research was conducted at the Utrecht Institute of Linguistics, as part of the dissertation of Marjolein van Egmond.
Notes
1. The frequency distribution of aphasic speech was first investigated by Howes and Geschwind (Howes, Citation1964; Howes & Geschwind, Citation1964). Unfortunately, they did not use Zipf’s law, but a cumulative version concerning the percentage of words that occur with frequencies up to and including each frequency value. This formulation is not very sensitive: disruptions in the higher frequency classes are easily concealed if the lower frequency classes do follow a Zipfian distribution. van Egmond et al. (Citation2015) were the first to study Zipf’s law in aphasic speech as such.