Page 38 - Fall 2011
P. 38

  Fig. 3. Ultrasound imaging of the tongue. a: Schematic midsagittal vocal tract for an \[š\]-like articulation. The red line shows the approximate location of the plane of the ultrasound image. b: Schematic of the tongue shape in the coronal plane shown by the line in (a). The arrows mark the peaks and trough of the tongue groove. c-e: Three ultrasound images of the tongue during c) \[θ\] d) \[s\] e) \[š\]. The speaker's left is on the left and the speaker’s right on the right. Selection of the three points was guided by repeated viewing of the video image, where the coherent structure of the tongue is more obvious than it is in any sin- gle frame. depend on. Explaining this long-distance effect has been challenging. The operation of this harmony can be seen in the follow- ing forms, where the underlined letter corresponds to the word segment “-s-”, the marker for the first person singular subject: ɛsk'ɑ: ɛšdžini ɛθdu:θ ‘I’m gutting fish’ ‘I’m singing’ ‘I whipped him’ As can be seen from this last form, this agreement can work across intervening segments (here, /d/ and /u/) that neither promote nor block harmony. It looks like “action at a distance.” What if there is something that makes this process local after all? Gafos (1996) proposed just such a solution. He claimed that the differences in the amount of tongue grooving seen in these fricatives could extend through the intervening vowels and non-participating consonants, thus making the agreement local. Our efforts, supported by the National Science Foundation’s Office of Polar Programs, have begun to test this hypothesis using ultrasound images of the tongue. Ultrasound allows us to look at the tongue during speech (e.g., Stone and Lundberg, 1996) in a minimally invasive way that is appropriate for a broad range of speakers, including the elderly, and in non-laboratory conditions as well (Gick et al., 2005). We generally take a “sagittal” view of the tongue, showing a two-dimensional image from near the tip to the back of the tongue near the uvula. However, we can just as easily take “coronal” sections that go across the tongue. This is ideal for measuring tongue grooves, and it is what we did in our study. Figure 2 shows one of our speakers, Margery Inkster, with the ultrasound probe under her chin. Figure 3 shows three of the cross-sections of her tongue during the fricatives \[θ s ʃ\] The groove depth differs among the three, and this same groove persists through the vowel. This is in accord with Gafos’s prediction. (See Figs. 2 and 3) To understand where this pattern might have come from, however, we need to know whether the same kind of persist- ence appears in languages without the consonant harmony. Phonetics of Endangered Languages 37 

   36   37   38   39   40