| The phonetics∼phonology interface has long been debated; some linguists argue for a modular approach (Keating 1984, Pierrehumbert 1990, Zsiga 1997, Cohn 1998), while others argue that there is no interface, and that phonetics and phonology are one and the same (Browman & Goldstein 1989-1992, Ohala 1990). Recent proposals by Gafos (2002), and Petrova et al. (2006) have proposed that language-specific phonetic differences, such as the timing of articulatory gestures and the implementation of a two-way voicing contrast in stops, are grammatically controlled through constraints on gestural alignment and laryngeal features. These proposals blur the line between the two components by including language-specific phonetic facts within the phonological grammar.;This dissertation investigates the phonology∼phonetics interface by documenting and analyzing data from Turkish and Urban East Norwegian, both cited by Petrova et al. (2006) and Bradley (2007). The results of a production study in each language and one perception study in Turkish demonstrate that the data do not pattern as previously described, or as predicted to occur by these proposals.;Results demonstrate that Norwegian coalescence of r+coronal sequences (producing a retroflex consonant), and voicing assimilation within words in Turkish pattern as categorical processes. However, devoicing of word-initial stops in Turkish is non-neutralizing, and vowel intrusion/deletion for r+non-coronal sequences in Norwegian is not a categorical process (Kristoffersen 2000, Bradley 2007), but rather gradient and dependent on morphological, consonantal and prosodic context. The gradience across contexts is best captured through a coupled-oscillator model of gestural timing (Nam & Saltzman 2003), while the task-dynamic phonetics implements this coordination of gestures.;The data from both languages are best captured if the phonology accounts for complete neutralizations by making qualitative featural changes, while a task-dynamic phonetics produces a gradient quantitative acoustic signal. This dissertation, then, argues for a modular approach to the phonetics∼phonology interface, following Keating (1984), Pierrehumbert (1990), Zsiga (1997), Cohn (1998), in that the phonology manipulates features and segments, and a separate language-specific phonetic component interprets features and produces gestural timing patterns. The universal task-dynamic phonetics, then, operates on this output, causing gradient and quantitative changes to articulatory trajectories and the acoustic output. |
