Neat theories, messy realities: How to apply absolute definitions to gradient phenomena

Allomorphy is a term that applies to a situation in which a single meaning is expressed by two or more forms in a language. For example, in English plural can be expressed differently depending upon the final sound in a noun: -[s] as in cats, -[z] as in dogs, or -[?z] as in foxes. In Russian one expresses ?apart? with the prefix raz- before vowels and voiced consonants as in razbit? ?break apart? (from bit? ?break?), but with ras- before voiceless consonants as in raspilit? ?saw apart? (from pilit? ?saw?). Traditionally this definition is stated in absolute terms: the meaning must be identical and the forms must be in complementary distribution. But this traditional definition is too narrow since there are situations where the meaning is not identical (cf. leaves vs. [Toronto] Maple Leafs -- a hockey team), and also situations where the distribution is not perfectly complementary (for example, the English plural can also be expressed by no change in all the same environments where we find the three endings, as in sheep, deer, and fish). We also see situations where both criteria of the traditional definition are violated, yet the forms show an important relationship that should not be ignored. An example is Russian prefix s- and suffix -nu, both of which express a single action, but overlap in their distribution and tend to have slightly different meaning, with s- preferred for behaviors (sglupit? ?do one stupid thing?), but -nu preferred for repetitive physical motions (prygnut? ?jump once?). The traditional definition fails to recognize many important relationships like these. Our project explored many examples of allomorphy in Russian. Specifically, we investigated 15 Russian prefixes and analyzed both corpus and experimental data. We found a range of relationships, among them prototypical allomorphy, standard allomorphy, non-standard allomorphy, and non-allomorphy. We present comprehensive data supporting the hypothesis that Russian verbal prefixes constitute a verb classifier system akin to those found in some Australian and East Asian languages. These conclusions give us a considerably improved description and typological perspective on Russian. We also looked at the extension of the category of diminutives to verbs in Russian via prefixes. We made some cross-linguistic comparisons, examining the behavior of prepositional vs. postpositional allomorphs in North Saami. Finally, we explored what may be thought of as the converse of allomorphy, the situation where two forms with a similar meaning compete in the same or overlapping environments, an understudied phenomenon we have dubbed ?rival forms?. In the course of our investigations we have experimented with various types of statistical modeling methods for linguistics. As results of the Neat Theories, Messy Realities project, we published the following: 2 anthologies of articles, both strategically placed as special issues of prestigious journals 1 book on Russian prefixes that serves both scholarly and pedagogical purposes 1 edited anthology on quantitative approaches to linguistics plus over 20 articles in addition. Our two PhD students both completed their databases, which are publicly available on the TROLLing site (opendata.uit.no), and completed their dissertations (one defended in August 2014 and the other one will defend January 16, 2015). We hosted five conferences, including: two NORKOG conferences (one in Oslo in 2012, and one in Tromsø in 2014), a conference entitled "Time and Variation in Cognitive Linguistics" in Tromsø in 2011, a conference entitled "Time and Space in Petersburg" in Petersburg, Russia in 2012, and a conference entitled "Cognitive Linguistics in the Triangle: Slavic and Beyond" at the University of North Carolina at Chapel Hill (USA), February 2013. We also made 80 presentations of our works at scholarly conferences in over a dozen countries across the US, Europe, and Asia. We created a website with comprehensive materials for teaching of Russian prefixes (http://emptyprefixes.uit.no/methodology_eng.htm). Additionally, we collaborated with the library at UiT to create the Tromsø Repository of Language and Linguistics, an international digital resource for linguists worldwide (opendata.uit.no).

"Neat Theories, Messy Realities" is a project that explores the structures of languages. Traditional definitions of language phenomena such as allomorphy (a type of relation between the forms of a language and their meanings) are stated in terms of absolu te criteria. However, such definitions are inadequate for describing the real complexity found in language data. As a result, our descriptions of languages are often inaccurate and incomplete, hindering both scholars and language learners. "Neat Theories, Messy Realities" examines data from Russian and other languages that have defied analysis. A national (UiT & UiO) and international (US, Belgium, and Russia) network of researchers will tackle a series of case studies. Two PhD students will receive world -class mentoring while collaborating with the team. The team will establish best practices for interpreting a range of form-meaning relationships found in language. The team's work entails analysis of large amounts of electronically-available language dat a, such as the Russian National Corpus, with 150 million words. The goal is not to discard traditional definitions, but to determine standards for applying them even when absolute criteria are not met. Statistical methods are key to evaluating the data an d establishing such standards. The team will debunk some myths about form-meaning relationships and unravel some long-standing mysteries. The results will be valuable to linguists as a model for allomorphy analysis, and the focus on complex data that depa rt from traditional definitions has implications for science in general, since scholars across many disciplines face similar problems. We plan not only theoretical achievements, but practical applications as well. Our case studies will yield online pedago gical materials. Russian is both an Arctic language and a world language, and is designated for academic emphasis in Tromsø, so it is appropriate that this project be carried out at the University of Tromsø.