The Frequency and Degree of “intrusive /r/” in New Zealand English 43/r/ are virtually non-existent, and all work on this topic has analyzed intrusive
/r/ as a discrete variable.
The lack of empirical data on intrusive /r/ likely relates to the scarcity
of appropriate environments for its occurrence. In the studies reported here,
then, we decided to sacri¿ ce the advantages of natural speech in favor of
reading and production studies, which ensured collection of suf¿ cient data,
enabled us to systematically manipulate potentially relevant linguistic fac-
tors, and facilitated recording of comparable environments across different
speakers. Some preliminary analysis of the data-set used here was reported
in Hay and Warren (2002). That paper presented a simple binary analysis
of rates of /r/ production and concentrated on the role that different af¿ xes
played in facilitating /r/-production.
This paper analyzes these recordings with four research questions in
mind. First, does the use of intrusive /r/ act as a sociolinguistic variable in
New Zealand English? Second, can the variability be predicted by any lin-
guistic conditioners? Third, can the likelihood of /r/ insertion after /au/ be
predicted by the phonetic quality of a speaker’s /au/ vowel, i.e., is the docu-
mented change in production of /au/ in NZE (Gordon et al. 2004, Maclagan,
Gordon and Lewis 1999, Woods 2000) affecting the rate of /r/ insertion? And
¿ nally, is there any evidence for socially or linguistically conditioned varia-
tion in the degree of constriction of the /r/? That is, we hy pothesized that there
may be interesting variation not only in the likelihood of /r/ being produced,
but also in how /r/-like the /r/ is when it is produced.
Introductory sociolinguistics textbooks drive home the important differ-
ence between “discrete” (or categorical) sociolinguistic variables and contin-
uous ones (see e.g., Milroy and Gordon 2003: 138–139), acknowledging that
perhaps continuous ones are sometimes analyzed as categorical for statistical
convenience. However our analysis reveals that NZE intrusive /r/ acts both as
a categorical and as a continuous sociolinguistic variable. A straightforward
analysis of the presence vs absence of /r/ reveals socially conditioned varia-
tion in the likelihood /r/ will be produced. However when an /r/ is produced,
the degree of the constriction is also socially conditioned. This difference in
degree echoes the more straightforward difference in frequency.
- Methodology
Because relevant environments for /r/-intrusion are relatively rare in sponta-
neous speech, we set out to answer these questions through a series of sen-
tences containing relevant environments, which participants were asked to