Skip to main content

Abigail Cohn

Cornell University

Thursday, October 10th at 2:00 PM in AP&M 4301.

Sources of variation in phonology

Abstract: Increased incorporation of experimental and naturalistic data in phonological analysis in recent years has provided an important complement to introspection and elicited data. These data sources highlight that there is greater gradience (temporal/spatial) and variability (variation in frequency of occurrence) than often thought.

In this talk I discuss two examples from Indonesian, highlighting the interplay between phonological patterning and both inter- and intra- speaker variation, based on joint research with Okki Kurniawan. To address limitations of prior work focused on Standard Indonesian (SI), we investigate observed phonological and morphophonological patterning in Jakarta Indonesian (JI), a rapidly developing colloquial variety spoken in and around Indonesia's capital. We draw on data from a phonetically transcribed naturalistic spoken corpus of Jakarta Indonesian (Gil and Tadmor 2015) along with a production study and acoustic analysis.

First, we investigate the realization of the active prefix N- in JI (cognate with the meN- prefix in SI). We investigate the nature of observed inter- and intra-speaker variation, drawing on the available corpus data and a production study. Second, we investigate the status of schwa. Based on the corpus data and acoustic analysis of corresponding audio files, we show that multiple factors together condition the optional deletion of schwa, including orthography, historical source, morphological structure, and phonological structure.

These data from a naturalistic corpus highlight the extent and complexity of variation seen in phonological and morphophonological patterns. Much can be learned from such corpus work, but careful analysis is needed. Multiple factors including phonological environment, morphological structure, speaker identity, historical source, encoded in part in the orthography, may all play a role. Each of these factors needs to be encoded in the corpus and included in the analysis to fully understand these complex interactions.