4.2 Perception of [aw], [o] and [u]

4.2.1 [aw] as an "odd" variant

Around half of the Sunyerites see the [aw] variant as an odd pronunciation (52%), so apparently according little prestige to this variant. Having said that, the degree to which the sound is thought odd differs from word to word: [aw]ració (oració - prayer) is seen as the oddest instance, and [aw]perat (operat - operated) and [aw]lives (olives - olives) the most normal instances (see figure 1). These general replies have to be put alongside results from the closed question questionnaire, since the villagers of Sunyer scarcely tolerate pronunciations which they do not usually use themselves.

Figure 1. Degree of oddness and probability of maintaining [aw]

Turning to the social factors, it can be seen that, in general terms, the degree to which the pronunciation [aw] is viewed as odd is in direct (inverse) proportion to age of interviewee: the oldest group scarcely find it odd or marked, among the youngest its oddness or markedness is very great, and here again this result connects directly with the results we obtain for production. (See figure 2).

Figure 2. Degree of perceived "oddness" of [aw] and probability
of maintaining

Schooling also emerged as a relevant factor related to subjective evaluation of [aw]: the interviewees with little or no knowledge of written Catalan, and either had no education or only elementary education, are the only ones to consider the [aw] variant as normal or unmarked, and this falls off proportionally as educational level rises (see figure 3). And here too, production follow in step with perception, such that interviewees with secondary education maintain the [aw] variant at lower level than other speakers.

Figure 3. Degree of perceived oddness of [aw] of maintaining

Lastly, social status of the speakers is also a modifying factor: speakers with a higher sociocultural level have a greater tendency than others to view the [aw] pronunciation as odd: consultants (informants) with greater status view this as odd in some 64% of instances, while those with lower sociocultural status, only do so in 8% of cases.

4.2.2 The contrast between [aw] / [o] and [u] and evaluations of [aw]

4.2.2.1 The contrast between [aw] / [o] and [u]

In general terms, the phonic contrast between [aw] and [o] or [u] well perceived in Sunyer, as one would expect. The speakers who see this clearest of all are speakers aged between 19 and 39 and are the most innovative in analysis of production. Apart from that, the oldest interviewees, aged from 66 to 85, are the least equipped to perceive the contrast in vowels in pretonic position and are also the most conservative. This state of affairs leads us to relate what we have just described with what in psychology is called the restoration effect, (9) whereby the auditory system is capable of restoring correctly a sound either not uttered at all, or different from the utterance that one expects (Fowler, 1986). In the case we are concerned with here, after hearing the variants [o] and [u] in  pretonic position it seems that the elderly consultants automatically replace them with the [aw] form, which they use in words of a similar nature. And, consequently, they do not note substantial qualitative differences in two words supposedly the same. This line of argument comes from the field of psychology, there are also factors of a biological nature that could be taken into consideration, such as the auditory capacity of those questioned (Walley 1988), (10) when explaining these auditory mismatches.

4.2.2.2 Evaluations of [aw], [o] and [u]

What we find overall is that 67% of evaluations made by Sunyerites take [o] as the "good" variant in pretonic position in olor (smell) ovella (sheep) orella (ear) and [u] as the "good" one in ufanosa (luxuriant); nonetheless, in more than 80% of instances the interviewers state that [aw] is the variant that they themselves use and also the one generally used in the village (see figure 4):

Figure 4. Evaluation of pre-tonic realisations [aw] and [o]

Looking at how evaluations made by the inteviewees in relation to lexical items, we see that ovella is the word where [o] is considered most normal; on the other hand, in the case of ufanosa, [aw] is seen as normal. Lastly, in the case of orella speakers are more insecure: 75% of the interviewees deem [o] the "good" variant 100% claim [aw] is the variant they use and consider this is the form used in Sunyer. These results tie in, once again, with the data obtained in the analysis of production (see figure 5) and make it possible to track the process of decline of the [aw] realisation from the subjective point of view of the speakers.

Figure 5. Evaluations and probability of maintaining [aw]

The consultants' impressions regarding the pretonic [aw] variant vary significantly in relation to age: 1) [aw] is considered own variant (the variant used by the speaker) in 97% of the elderly group; among the youngest speakers this falls by some 30% but still occurs at a comparatively high rate (66%). 2) [aw] is considered the form used in Sunyer and occurs at a high level among all speakers, above all the elderly. 3) Lastly, only the elderly (mostly) view the [aw] diphthong as the "good" form, (at 77%) while in the case of the other speakers the percentage who think so falls to a mere 10% (see figure 6). From all this, we can deduce that, on a subjective level, the sound change has reached an intermediate stage in the case of the youngest speakers and practically speaking has not even begun among the oldest age group, the only ones who have a high level of linguistic security. (11)

Figure 6. Evaluation and probability of maintaining [aw]
by age group of interviewee

As regards schooling, the only evaluations that present interesting differences among the interviewees are those that have to do with which vowel is considered the good one. In effect, the interviewees who have no knowledge of written Catalan and those who did not receive any schooling consider [aw] to be good; all the others view [aw] as the good variant in a progressively decreasing curve in inverse relation to education (see figure 7). This, combined with the results obtained in the analysis of production enable us to detect the beginning of the sound change among the more educated, who are in turn show greater discrepancy between their evaluations and production.