The Phonology and Phonetics of Nasal Obstruent Sequences

This dissertation explores the relationship between the phonological
patterning of nasal-obstruent sequences (NC sequences) and their phonetic
realizations. I argue that there are distinct NC patterns in the phonology,
specifically unary segments (such as prenasalized stops) vs. clusters, and that
these patterns are reflected in the phonetics. The data for these investigations
come primarily from phonological and phonetic studies (both acoustic and
aerodynamic) collected through fieldwork on four Austronesian
languages—Tamambo and Erromangan of Vanuatu, and Pamona and
Manado Malay of Indonesia.

Clear cases of prenasalized stops, in languages like Fijian, and clusters,
in languages like English, provide ample evidence for different phonological
NC entities, despite the fact that some languages are more difficult to classify.
I propose a methodology for determining the status of an NC, which separates
the often-conflated issues of tautosyllabicity and unary segmenthood.
Grouping NC-types along two principle divisions—unary vs. cluster, and
voiced vs. voiceless obstruent—I argue that only six of sixteen possible
patterns (combining 0-4 NC types) are attested. The unattested cases are
attributed to two factors: the lack of prenasalized voiceless stops, and the lack
of contrasting unary vs. cluster NCs of the same voicing specification. An
investigation of phonetic properties of NC sequences, including total NC
duration, duration of a preceding vowel, and degree of nasalization in a
preceding vowel, reveals that total duration does correlate with phonological
NC structure while the other two factors do not. Data on the relative nasaloral
timing of NC sequences reveals interesting similarities across the types:
voiced NC sequences, NC sequences with a voiceless stop, and NC affricates
each have distinct realizations, regardless of phonological status.

Not only are phonological NC patterns reflected in the phonetics, but
the phonetic realizations also have important consequences for the phonology.
The phonetic characteristics of unary vs. cluster NCs—both their differences
and similarities—are argued to explain gaps in the phonological patterns.
Phonological representations of NCs are proposed that crucially include both
prosodic and segmental structure, as well as being consistent with the
phonetic facts. These representations have implications for other reported
types of partially nasal segments.