Published online by Cambridge University Press: 28 October 2014
Rivalry between the two English nominalising suffixes -ity and -ness has long been an issue in the literature on English word-formation (see esp. Marchand 1969; Aronoff 1976; Anshen & Aronoff 1981; Romaine 1983; Riddle 1985; Giegerich 1999; Plag 2003; Säily 2011; Baeskow 2012; Lindsay 2012; Bauer et al. 2013: ch. 12). Both regularly attach to adjectival bases, producing nouns with (mostly) synonymous meanings. Most standard accounts assume that stronger restrictiveness of -ity is an effect of -ity being less productive than -ness, and that the observed preferences are an effect of selectional restrictions imposed on bases and/or suffixes. The focus of the present study is on the productivity of the two suffixes in synchronic English and on the diachronic development of that productivity in the recent history of the language. The article presents a statistical analysis and a computational simulation with an analogical model (using the AM algorithm, Skousen & Stanford 2007) of the distribution of -ity and -ness in a corpus comprising some 2,700 neologisms from the Oxford English Dictionary from three different centuries (the eighteenth, nineteenth and twentieth). Statistical analysis of the OED data reveals that -ity preference for pertinent bases is far more widespread than hitherto thought. Furthermore, the earlier data show a consistent development of these preference patterns over time. Computational modelling shows that AM is highly successful in predicting the variation in synchronic English as well as in the diachronic data solely on the basis of the formal properties of the bases of nominalisation. On the basis of a detailed analysis of the AM model it is shown that, unlike many previous approaches, an analogical theory of word-formation provides a convincing account of the observed differences between the productivity profiles of the two nominalising suffixes and their emergence over time.