The Δ columns specify the difference in identification of the response vowel between the short and the long vowel durations. The %-columns express this difference as a percentage relative to the total number of responses given by the group of listeners, i.e., 1720 by the native listeners and 3526 by the PA EFL learners. It is then easily seen that the effect of vowel duration is much larger, also relatively, for the nonnatives than for the native listeners. In the table, the rows containing lax vowels are marked in yellow. The ratio of all lax vs. all tense vowel responses by native listeners is 40:60 for short vowels, against a ratio of 31:69 for long vowels.
For the EFL learners, these ratios are 58:42 and 18:82, respectively. For both language groups, there is a significant association between tenseness and length of the vowels (long = tense, short
= lax), but the association (expressed by the Phi coefficient) is much stronger for the EFL group (φ = .418) than for the native group (φ = .095). The difference between the two strengths of association is significant, χ2(1) = 183.3 (p < .001). This shows, once more, that duration is a (much) stronger cue in the tense-lax contrast in English for EFL learners than for native listeners.
AE norm exceeded 50% only for two response categories, namely, /i/ (51.4%) and /u/ (51.6%).
Agreement with the native modal response was much lower for the other nine vowel types and never reached 35%. This is strong evidence that the PA EFL learner’s perceptual mapping of the AE vowel space is significantly different from that of the AE natives.
4.4.2. Overall accuracy of perceiving vowel duration
The results of the vowel identification experiment reveal that PA EFL learners have a fairly good perceptual representation of the duration of the vowels of American English. Figure 4.7 shows that the EFL learners assume short durations for the four lax vowels of AE and choose longer durations for the seven tense vowels. The correlation of the EFL and native durations preferred for each of the eleven vowel types is high but is compromised by two vowel types, viz. /æ/ and /ɑ/. These vowels are long in the perceptual representation of the native AE listeners but substantially shorter, although not as short as any of the lax vowels, in the perceptual representation entertained by the EFL learners. A tentative explanation for this phenomenon might be that some of the PA EFL learners consider the vowels /æ/ and /ɑ/ to be phonetically short, possibly due to exposure to the pronunciation of these vowels in British English, where /æ/ is always short (as it should be given that it is a lax vowel on distributional grounds) and where AE /ɑ/ is pronounced as short /ɔ/ in doll. At the same time, Figure 4.7 shows that duration plays a much more prominent role in the perceptual representation of the AE vowels for the EFL learners than for the native listeners. I observed that the PA learners assumed (much) larger duration differences between the AE vowels than were seen in the responses of the native AE listeners. This exaggeration of the duration differences has been reported earlier in the active production of the AE vowels by Arabic EFL speakers from a wide range of regional Arabic varieties (Munro, 2003; see Chapter 5 for details).
4.4.3. PA EFL learners’ confusions in depth
In relation to the third question, PA learners’ confusions at the vowel level were also accounted for by vowel quality. The most important deviations of the AE vowel space from the natives’
norm in nonnative PA mental conceptions are visualized in Figure 4.9. These confusions are argued to require serious attention at the pedagogical level, as they will negatively impact the learner’s perception and word recognition skills in EFL and most likely lead to pronunciation errors (see Chapter 5). Among these confusions in perceptual representation, seven confusions (indicated with thick red arrows in Figure 4.9) occurred in more than 20% of the cases (or more). These prominent confusions, however, never involved the tense-lax contrast; rather, the
most serious confusions are related to differences in quality within either lax or tense pairs of spectrally adjacent vowels. Learners seem not to discriminate between them correctly and apparently consider them to be very similar to each other. Serious systematic confusions in tense vowel pairs were observed only occasionally (in only 3 out of 24 arrows): /æ/→/e/, /o/→/ɑ/, /ɑ/→/u/, and among mid and low-back vowels only, and even fewer within lax pairs (8.3%).
Curiously, the perceptual representation as entertained by the PA EFL learners is superior even to that held by the native listeners for vowel types in the low back part of the AE vowel space. The results show that the contrast between the vowels /ɑ/ and /ɔ/ is upheld better by the EFL learners, as indicated by the negative numbers (and green arrows) in Figure 4.9. I will ignore this phenomenon in light of the consideration that the poor separation between /ɑ/ and /ɔ/ (and hence the confusion with adjacent vowel types) is caused by the low back vowel merger in Californian English, which is the regional variety of the native AE listeners.
4.4.4. Comparing assimilation to L1 with mapping L2
PA learners show good discrimination of AE vowels in the PAM test when they assimilated them to their L1 vowel inventory and only with difficulty in discriminating adjacent vowels that are not part of the tense-lax contrast (i.e., pairs of spectrally adjacent tense-tense or lax-lax pairs) and do not exist in learners L1. Based on the problematic contrasts in the perceptual assimilation test in Chapter 3, members of these contrasts should yield poor vowel mapping (i.e., larger confusions) in the results of the present chapter. Comparing the results in Figure 3.2 with the results in Figure 4.9 reveals a similar trend. Members of the contrasting pairs in PAM that do not have a (near)counterpart in L1 yielded serious mapping misconceptions but with different AE vowels (than those of the contrasting pairs in PAM), mainly with vowels that contrast in the tense/lax feature. In other words, none of the confusing pairs in the PAM test were among the serious confusions in mapping AE vowels according to the results of the present chapter, and misconceptions only were in adjacent vowels that contrast in the tense/lax feature with vowels lacking in L1. This calls attention to two things. First, there is the role of duration in organizing the perceptual mapping in the mind of the nonnative learners in comparison to the native norm. The PA results show a significant correlation between the length of the stimuli and the tenseness feature of the chosen vowels. Second, there is the role of L1 interference in warping the perceptual mapping of EFL. Here, the results show that most of the confusions involve (at least) one L2 vowel that has no clear counterpart in L1 or confusions between two vowels that do not exist in L1, which in return suggest a clear L1 interference in
filtering L2 features. These L2 vowels with no counterpart are considered the most marked vowels (MDH term) in L1 and were automatically the most difficult to discriminate, yielding serious confusions. Less marked or unmarked vowels either yielded less serious confusions or no confusion at all.
The good mapping of the high tense and lax vowels for the nonnative learners in the present study conforms with other studies that reported Arabic vowel inventory to have such features, and that differences in Arabic vowel inventory are not exclusive to duration but also involve differences in vowel quality for the short/long counterparts. A detailed comparison of the native results with the nonnative results shows that native listeners attach more equal weight to quality and duration in the mental representation of the AE vowels, whereas nonnatives depend more on duration as a reliable cue for mapping AE vowels, while only partially taking quality into consideration. Arguably, even if incorrect representations based on duration differences comprise most of the serious confusions (in the tense-lax contrast), they can still be corrected more easily than confusions in quality differences if targeted and explicit learning tasks in the teaching process are employed.
Despite the fact that AE low back vowels are supposed to be NEW sounds for PA learners by definition, they were more correctly mapped by nonnative learners than by the natives.
However, this good mapping will not help L2 learners much because many American native speakers have merged these vowels into a single category. On the other hand, other NEW sounds were the most problematic to discriminate and the most difficult to be correctly mapped, which support L2LP postulations in this context for NEW sounds.