Matthew K. Gordon
Dept. of Linguistics, UCLA, 405 Hilgard Ave., Los Angeles, CA 90024
Languages differ in the way they signal stress. In Estonian, the primary stressed syllable is marked by increased onset duration [Lehiste, Consonant Quantity and Phonological Units in Estonian (Indiana Univ., Bloomington, 1966)] and by a pitch rise [Liiv, Sovetskoe Finno-Ugrovedenie 21(1), 1--13 (1985)]. The acoustic properties of secondary stress are subtle, resulting in controversy over its location [Hint, Eesti Keele Sonafonoloogia (Eesti NSV Teaduste Akadeemia, Tallinn, 1973); Eek, EPP, 20--59 (1982)]. To determine further the acoustic properties of Estonian stress, amplitude, F0, and consonant and vowel durations were measured for di-, tri-, and tetrasyllabic words. Results for primary stress conformed to those found by Lehiste and Liiv. Secondary stress was characterized by an interruption in the F0 decline (Eek, 1982) and by lengthening of the onset, but not the rhyme. Estonian is thus unusual in signaling stress by pitch and by lengthening of the onset, but not the rhyme. Unstressed open syllables were significantly longer after primary stressed syllables of both the CV [Lehiste, Language 41(3), 447--456 (1965)] and the CVV type, than after CVC, arguing against foot isochrony [Eek and Remmel, Speech Communication Seminar, 179--185 (1974)]. Implications of these phonetic measurements for proposed algorithms of stress assignment in Estonian will be considered.