The Frequency and Degree of “intrusive /r/” in New Zealand English 45the full paradigms. Participants were warned that some of the sentences they
would be asked to read would be a bit weird, and they were asked just to
humor us! Note that the ¿ ller sentences, with which the test sentences were
randomized, were relatively normal.
Each utterance was given a binary analysis, indicating whether an /r/ was
produced at the boundary or not. Six examples were excluded from the overall
analysis due to misreading by the participant. Of the 48 environments in which
each speaker could potentially produce an intrusive /r/, the range of /r/s actually
produced was between 0 and 36, with the mean being 13 and the median 10.5.
Overall 206 tokens were analyzed as containing an intrusive /r/.
For those tokens which were analyzed as containing an intrusive /r/, we
also took an F3 measurement at the lowest point of F3 during the /r/, the
most salient acoustic correlate of /r/. Acoustic analysis was carried out using
Praat (http://www.fon.hum.uva.nl/praat/) using the standard settings (25 ms
analysis frame, Gaussian window, 10 pole LPC ¿ lter) with formants manually
corrected when it was clear that Praat’s reading was incorrect. Correction was
most often needed when the female speakers’ fundamental frequency was
so high that accurate reading of the ¿ rst formant was dif¿ cult. Figures 2.1
and 2.2 give examples of where the measurement was taken. Both are male
speakers producing the word clawify, and in both cases an /r/ was produced.
Note, however, that the F3 descends relatively lower for the ¿ rst speaker
(Speaker 14) than for the second speaker during this word. We will return to
the potential relevance of this difference.
Figure 2.1 Utterance of the word clawify produced by Speaker 14, with an indication
of where the F3 reading was taken.