Layman's Summary

There is a common sound pattern where voiced stops (such as b) are pronounced as voiced spirants (similar to v) between vowels. (Spanish has one version of this pattern.) Usually, this pattern is explained as an articulatory one: pronouncing a stop requires the speaker to completely close off his or her mouth (using the lips in the case of b), while a speaker making a spirant leaves a narrow opening. This means that speakers don't have to move their lips as far when saying a spirant as when saying a stop; therefore, it's believed that spirants are easier to pronounce between vowels than stops are. In other words, this sound pattern takes less effort than simply pronouncing stops as stops all the time. An alternative hypothesis has to do with the way these sounds are perceived by listeners: that this pattern happens because people are likely to mishear a voiced stop as a voiced spirant between vowels.

My dissertation involves a series of experiments that test both of these hypotheses. In one set of experiments, I look at how these sounds are perceived by listeners; in the other, I look at how these sounds are pronounced by people who have consumed alcohol - on the assumption that drinking alcohol will cause people to produce more "easy" sounds and fewer "hard" sounds. So far, the results suggest that both of these hypotheses explain part of what's going on in these sound patterns, but neither hypothesis can be the whole story. The two UCSC Graduate Research Symposium posters below give preliminary results from both types of experiment for a non-linguist audience.

Summary

Lenition has commonly been described as an articulatorily motivated pattern: as a form of articulatory effort reduction in weak positions (such as between vowels). However, articulatory effort is very difficult to measure; however carefully designed, studies that measure or model articulatory effort in leniting contexts are forced to assume that less articulator movement means less articulatory effort: a reasonable assumption, but an assumption nonetheless.

Lenition of intervocalic voiced stops to spirants can be seen as an instance of the "too-many-solutions" problem: why do these voiced stops spirantize as opposed to becoming, for example, voiceless stops (a less marked class of segments)? One approach to too-many-solutions, known as the P-map, locates the directionality of many sound patterns in perception: the more likely sound changes are those that involve a perceptually small change.

In one set of experiments for my dissertation, I test whether the P-map could explain why intervocalic voiced stops spirantize but do not devoice. Experimental results suggest that perception does have something to say about the direction of lenition: voiced stops are perceptually more similar to voiced spirants than they are to voiceless stops. These results do not preclude a role for articulation in our understanding of lenition; however, they do show that we need more direct ways of measuring articulatory effort than are currently available.

Currently, I am analyzing the results from a production experiment designed to test the articulatory hypothesis more directly. This experiment compares the production of intervocalic stops for subjects in two conditions: sober and intoxicated. Since alcohol is known to impair cognitive and motor function, I hypothesize that intoxication will encourage articulatory "easy" sounds. Preliminary results suggest that effort reduction does indeed result in neutralization of certain contrasts intervocalically, but not necessarily by causing lenition-like changes.

Downloadable Stuff

(See CV for citation information. Listed from newest to oldest.)

NELS Workshop Handout
2009 Articulatory and perceptual similarity in intervocalic lenition
Results for six subjects from the production experiment, plus a condensed version of the perceptual results.

UCSC Graduate Research Symposium Poster
2009 Effort reduction in intoxicated speech
Some early results from the production experiment with alcohol, intended for non-linguists. Note that these preliminary conclusions have since been revised! More up-to-date material coming soon.

UCLA Phonology Seminar Handout
2009 Perceptual and systemic pressures on lenition
Similar to the Phorum handout, but includes more data from non-English-speaking subjects.

LSA Handout
2009 Perceptual pressures on lenition
Includes only the perceptual results from the QE.

Phorum Handout
2008 Perceptual, articulatory, and systemic influences on lenition
Includes many of the results from the QE in condensed form. Also includes preliminary results from two subjects with native languages other than English.

Qualifying Exam
2008 Perceptual, articulatory, and systemic influences on lenition
The most thorough write-up to date. Includes data for all English-speaking subjects and discussion of related typological facts.

UCSC Graduate Research Symposium Poster
2008 What factors affect sound patterns in language?
A condensed version of the TREND presentation, intended for non-linguists. See also this article in the Santa Cruz Sentinel for a brief summary.

TREND Handout
2008 Perceptual and articulatory influences on phonological alternations
A write-up of the experimental results from the first ten subjects.

LASC Handout
2008 Perceptual and articulatory influences on phonological alternations
Presentation of the basic research question and analysis of some previously published data. (Note that this data, problems with which are acknowledged in the handout, suggested the opposite of what the experiment shows.)