![]() ![]() ![]() The two frames were created by cross-splicing and PSOLA method resynthesis. The starting point for the creation of these frames was (1) above. Any differences observed across conditions in the experiments that follow can only be attributed to context. ![]() These stimuli accordingly present a fairly conservative manipulation, changing only context to ensure that properties of the target sound itself did not shift listeners' perception. The two prominence conditions in the present study, which will be referred to as the nuclear pitch accent (NPA) and post-focus conditions were created to manipulate only the context surrounding the target (with the target identical across conditions), in such a way that listeners' perception of target prominence was roughly equivalent to the (phonological) prominence distinction exemplified in (1) and (2). In contrast, in (2) the target follows narrow focus marking, realized with a rising L + H* accent on the word “say” the target is therefore unaccented and non-prominent. In (1), the target is prominent, bearing the nuclear pitch accent in the phrase which contains a standard declarative tune. One outstanding question is thus how listeners relate contextual prominence information to how a prominent high front vowel should be realized, and if listeners, like speakers, are variable in this regard. Taking this literature as a whole, we can conclude that speakers appear to variably prioritize these two prominence strengthening patterns in high front vowels. Also, apparent is within-study inter-speaker variation ( Cho, 2005). Some suggest sonority expansion ( Houde, 1967), others hyperarticulation ( Kent and Netsell, 1971), and others some intermediate patterns ( Cho, 2005). 1 Empirically, various speech production (cinefluorographic, EMA, acoustic) studies of American English /i/ and /ɪ/ (primarily /i/) show variation. In comparison, sonority expansion would entail a more open vowel articulation, not as closely approximating a high front target (acoustically with raised F1, and possibly lowered F2). A hyperarticulated variant of these vowels should be produced with a more closed articulation, that is, an articulation more closely approaching a high/front target (acoustically with lowered F1, and raised F2). In the particular test case adopted here, sonority expansion and hyperarticulation in vowel articulations and their consequences on formant structure can conflict: this is for high front vowels such as American English /i/ and /ɪ/. ![]()
0 Comments
Leave a Reply. |