Antecedentes

6.2.1 Perceptual distance

Charts that compare the values of the first and second formant values for vowels are conventionally used to visualize vowel positions within the vowel space. These charts typically place the first formant on the vertical axis, and the second formant on the horizontal axis, both of which are inversely labeled because of the inverse relationship to traditional articulatory

parameters (Ladefoged & Johnson, 2011). Because the vowel positions are plotted in respect to vertical and horizontal axes, the distance between any two points on the chart can be determined by using the Euclidean distance formula (Equation 3). This formula determines the distance between any two points by treating this distance as the hypotenuse of a triangle, where the differences in X and Y values serve as the other two sides. D represents the perceptual distance. X2 and Y2 represent the SAE standard’s first and second formant values, respectively. X1 and Y1 represent the participants’ first and second formant values, respectively.

(3)

The Euclidean distance formula has previously been used to compare distances between vowels. Some applications of this formula include comparing vowel position differences for male and female speakers (Fant, 1975); analyzing cross-linguistic perceptual distances of the same vowels to test the predictions of the adaptive dispersion theory (Yang, 1996), which posits that the distinctive sounds of a language are positioned within the phonetic space to maximize the perceptual distance (Flege, 1989; Johnson, 2006); determining what factors are distinctive for vowel identification by assessing mean distance among vowels (Neel, 2008); comparing how

different language accents pronounce the same vowel (Vieru-Dimulescu & Boula de Mareuil, 2006); and analyzing the proximity of dialectal vowel pronunciations to an accepted standard (Wright & Souza, 2012). Due to this widespread application of the Euclidean distance formula, it is generally accepted as a measure for comparing vowel positions.

In this study, the Euclidean distance formula was employed to compare the participants’ vowel pronunciations with standard SAE vowel pronunciations by quantifying the perceptual distance between them. This goal most closely resembled that of Wright & Souza (2012). Before comparison, participants’ average formants for each of the SAE target vowels were converted into their Bark scale equivalents. Bark scale equivalents were also created for the standard SAE vowel formant measurements, which were compiled from previous research on SAE and are available in Appendix K. The Bark scale is a perceptual scale that takes into account the fact that some formant differences are more salient than others (Steels & de Boer, 2008). The Bark scale has been used frequently to compare vowels (Neel, 2008; Shrosbree, Narita, Grenon & Kubota, 2011; Wright & Souza, 2012). The formant values were converted into the Bark scale using the formula proposed by Traunmüller (1990) via the University of Stockholm website (Traunmüller, 2005). The average formant values and Bark scale equivalents for the participants and the SAE standards are available in Appendix M. The Euclidean distance formula was used from the Bark scale F1 and F2 measurements to calculate the perceptual distances of the participants’ pronunciations and the SAE standards.

6.2.2 Vowel unmerging of /ie/

6.2.2.1 Description and measurements

This study contained 24 tokens chosen to examine the unmerging of /ie/ into /e,eɪ/, /eǝ/ and /ɪǝ/. In SAE, nine of these words contain the /e,eɪ/ vowel, six contain the /eǝ/ vowel, and

nine contain the /ɪǝ/ vowel. The non-phonemic SAE diphthong /eɪ/ was collapsed with /e/. In Patwa, the initial vowel in all the words is /i/, whereas the initial vowel segments present in the target SAE pronunciations can be /ɪ/ or /e/. Similar to the method used by Labov (1963), the values of the first and second formants were recorded at the beginning of the vowel segment for all tokens. For tokens beginning with bursts, fricatives, or affricates, the formants were

measured as soon as the vowel began. The initial vowel sound in words beginning with nasals, liquids, and glides was measured at the earliest point in the vowel at which the consonant was no longer heard.

Perceptual distance was used to compare the participants’ initial formant values with the target SAE formant values (Appendix K). The target words were grouped according to the initial vowel sound present in the target SAE pronunciation, for example /e/ or /ɪ/. Then, an average was taken for each of the specific SAE vowels to represent the participants’ formants for that vowel. The perceptual distance was calculated from the participants’ average formant values to the SAE average formant value for each vowel (Equation 3).

After comparing the initial vowel segments of all the tokens, the data was subdivided further to allow for a more detailed analysis. Specifically, the target words that begin with /e/ in SAE were split into two categories: /e,eɪ/ and /eǝ/. This split was completed in order to look at differences in diphthong movement. To illustrate the movement of these diphthongs, the final vowel sound of each of the target words was also measured. The formants of the final vowel sound were measured at the latest point in the vowel at which the following consonant was not heard. The initial and final vowel formants for each of the three conditions, /e,eɪ/, /eǝ/, and /ɪǝ/ were then compared.

6.2.2.2 Spectrogram examples

As described previously, the first and second formant values were measured for all tokens examining the unmerging of /ie/. Following Labov (1963), these formants were measured at the beginning of the vowel. The formants at the end of the vowel were also measured. Figure 12 and Figure 13 provide visual examples of where these measurements were taken.

Figure 13 JCP4 pronunciation of “face”

6.2.3 Vowel unmerging of /a/

6.2.3.1 Description and measurements

This study contained 36 tokens chosen to examine the unmerging of the low central vowel. In the SAE pronunciation of these words, 12 contain /æ/, 3 contain /ɔ/, 12 contain /ɑ/, and 9 contain /ɒ/. Because /ɑ/ and /ɒ/ are distinguished by rounding (Ladefoged & Johnson, 2011), they were collapsed in all following analyses as /ɒ,ɑ/. For all tokens, the first and second formant values were recorded from the midpoint of the vowel segment.

Perceptual distance was used to compare the participants’ formant values with the target SAE formant values (Appendix K). The target words were grouped according to the vowel present in the target SAE pronunciation, for example /æ/ or /ɒ,ɑ/. Then, an average was taken for each of the specific SAE vowels to represent the participants’ formants for that vowel. The perceptual distance was then calculated from the participants’ average formant values to the SAE average formant value for each vowel (Equation 3).

6.2.3.2 Spectrogram examples

The first and second formants were measured for all tokens examining the unmerging of /a/. These formants were measured from the midpoint of the vowel segment. Figure 14

demonstrates the selection of the entire vowel segment.

Figure 14 JCP2 pronunciation of “socks”