To ish or not to ish: The influence of author characteristics on the use of suffix -ish

In English, the suffix -ish is used when one wants to express the quality of being close to a more exact quality (greenish, 10:30ish), or similar to an entity (foolish, Disneyish). 'Ishing' is not a new phenomenon: it was already done by Shakespeare (Neuhaus and Spevack 1975). One might imagine ishing to be more popular for, say, some authors or some genres. In this talk, we therefore investigate whether the presence of the suffix -ish in contemporary English texts is significantly influenced by various extralinguistic features.

We extracted 5,639 instances of -ish attached to an adjective or adverb (greenish), a numeral (10:30ish), a common noun (foolish) or a name (Disneyish) from the 100Mw British National Corpus. For individual values of features derived from the sample metadata, like gender, age, medium, genre and target audience, and additionally of the Flesch reading ease score, we determined whether the average frequency of -ish with that feature value deviate significantly from that for other values for the same feature. Furthermore, we used logistic regression to determine which of the features are taken to have significant influence when all features are available simultaneously as predictors. Finally, in order to examine whether the choice of corpus has any effect on our findings, we replicated our experiment for the features gender, age and readability using 12,207 instances extracted from the 140Mw Blog Authorship Corpus (Schler et al. 2006). We present our findings at the conference.

Neuhaus, H. and M. Spevack (1975). A Shakespeare Dictionary, Some Preliminaries for a Semantic Description. Computers and the Humanities 9, pp. 263-70.
Schler, J., M. Koppel, S. Argamon and J. Pennebaker (2006). Effects of Age and Gender on Blogging. Proceedings of 2006 AAAI Spring Symposium on Computational Appraoches for Analsying Weblogs.

Presented at: The 20th meeting of Computational Linguistics In the Netherlands (CLIN-20), 5 February 2010, Utrecht University, Utrecht, the Netherlands.
Poster (pdf; 949kB)

back to presentations and posters