Felicitas Kleber

This study investigates the realization of narrow and broad focus in Albanian, an understudied language. Production and perception data were analyzed for narrow and broad focus. Speakers were found to produce broad focus with a low f0-trough (L*) and narrow focus with a high f0-peak (L + H*) in both final and initial positions. Listeners preferred L + H* stimuli for narrow focus in initial position regardless of whether they heard a narrow or a broad focus question. These findings show that focus marking in Albanian is influenced more by position than prosodic cues.

This study is concerned with the Bavarian German dialect feature of complementary length in vowel plus consonant sequences according to which tense (long) vowels always precede lenis stops (short closure) and lax (short) vowels always precede fortis stops (long closure). The study investigates whether a vowel length contrast is developing before fortis stops due to dialect leveling. We measured vowel and consonant duration in trochaic words differing only in vowel length which were read by 40 older and younger Bavarian and Saxon (control group) German speakers. Older Bavarians distinguished tense and lax vowels by means of vowel to vowel plus consonant ratios that indicate an inverse timing pattern. Saxons and younger Bavarians signal the tense–lax distinction independently of the following stop length. A perception test showed that this sound change in progress also affects perception. We argue for a contact-induced change which is triggered by external as opposed to internal factors.

The study tests a model of sound change based on how prosodic weakening affects shortening in polysyllabic words. Twenty-nine L1-German speakers produced minimal pairs differing in vowel tensity in both monosyllables /zakt, zaːkt/ and disyllables /zaktə, zaːktə/. The target words were produced in accented and deaccented contexts. The duration ratio between the vowel and the following /kt/ cluster was less for lax than tense vowels and less for disyllables than monosyllables. Under deaccentuation, there was an approximation of tense and lax vowels towards each other but no influence due to the mono- vs. disyllabic difference. On the other hand, Gaussian /a/ vs. /aː/ classifications of these data showed a lesser influence due to the syllable count in deaccented words. Compatibly, when the same speakers as listeners classified synthetic sackt-sagt and sackte-sagte continua, they were shown to compensate for the syllable count differences, but to a lesser extent in a deaccented context. Deaccentuation may therefore provide the conditions for sound change to take place by which /aː/ shortens in polysyllabic words; it may do so because the association between coarticulation and the source that gives rise to it is hidden to a greater extent than in accented contexts.