TP187: Usage-based and rule-based approaches to phonological variation
Individual words may differ in their phonological behavior depending on usage frequency. Finnish has two variable lenition processes that show robust frequency effects: Assibilation (t --> s /_ i) and Apocope (i --> 0) (Paunonen 1974, Laalo 1988, Anttila 2006). For example, in South-Eastern Finnish the verb /lentä-i/ ?fly-PAST? has four possible past tense forms: lenti ~ lensi ~ lent ~ lens. (Note that the rules interact in a counterbleeding fashion.) The generalization is familiar: lenition is more common in high-frequency words than in low-frequency words (see e.g. Bybee 2001).
What is surprising, though, is that frequency is relevant only under very narrow structural circumstances. In verbs, the frequency effect only emerges if the initial syllable has exactly two moras. In all other environments, frequency is irrelevant. For example, monomoraic verbs never assibilate despite their high usage frequency and trimoraic verbs virtually always assibilate despite their low usage frequency. This is exactly the opposite of what one would expect based on frequency. As a general statement about language, the generalization that high frequency leads to lenition is thus clearly false.
The Finnish evidence shows that frequency effects are embedded in categorical phonological structure that overrides frequency. In this case, segmental reduction is triggered by incomplete metrical parsing (extrametricality): high-frequency words tend to be incompletely parsed, low-frequency words tend to be exhaustively parsed, but the choice is available only in environments where two different metrical parses are possible. The relationship between frequency and lenition is thus indirect and mediated by categorical metrical constraints. We conclude that frequency effects are real, but can only be understood against the backdrop of categorical phonology.
Session: Themed Panel (part 1)
Usage-based and rule based approaches to phonological variation
Friday, April 4, 2008, 13:45-15:15