Morgan Sonderegger (University of Chicago)

“Longitudinal phonetic and phonological dynamics on reality television”

There has been much recent interest in to what extent an individual’s
phonetics and phonology changes over time, from two perspectives.  In
short-term laboratory settings, aspects of one’s speech shift in
response to the speech of others; such ‘convergence’ effects are
mediated by social and linguistic factors, and are well-attested
(Goldinger, 1997; Pardo, 2006; Nielsen, 2008; Babel, 2009).  It has
been hypothesized that an accumulation of such shifts over time is an
important source of accent change in individuals and sound change in
communities (Pardo, 2006; Delvaux & Soquet, 2007).  However, studies
where phonetic or phonological variables are remeasured for
individuals at times years apart have found huge variability: there is
often no evidence for any change for a majority of individuals, while
a minority change significantly (Harrington, 2006; Evans & Iverson,
2007; Sankoff & Blondeau, 2007; Siegel, 2010).  What is the link
between the different patterns seen in short-term convergence and
long-term dynamics?

We address this question by investigating ‘medium term’ phonetic and
phonological variation in a British reality television show where
speakers live in an isolated house for three months, subject to
constant recording.  The house is a socially and linguistically closed
system, making it possible to trace the dynamics of phonetic and
phonological variables in contestants’ speech, and to test hypotheses
about their sources.  We consider three variables — VOT, vowel
formants, and t/d deletion — in 8 hours of speech from 12 speakers.
We analyze each variable’s dynamics over the course of the season
after controlling for linguistic factors.  Different speakers show
extremely different dynamics for particular variables: some do not
change at all over time, some show significant short-term fluctuations
without long-term trends, and some show long-term trends. The most
common pattern is for a speaker’s use of a variable to fluctuate
between recording sessions on different days, in part due to shifts in
the topic of conversation.  We also tentatively find effects of social
interaction on observed dynamics, and individual differences in
plasticity across all variables.  Our findings suggest that short-term
shifts in individuals’ speech (days) are common, but only accumulate
into longer-term change for some speakers.