Phonetic and phonological vowel reduction in Central Catalan

Abstract

In Central Catalan, phonological vowel reduction causes the stressed seven-vowel system to reduce in number in unstressed position, where only the three reduced vowels [iəu] can occur. Exceptionally, full vowels (typically expected in a stressed syllable only) can appear in unstressed syllables in certain contexts. This study explores the acoustic characteristics of phonologically unreduced vowels found exceptionally in unstressed position in Central Catalan and compares them to stressed full vowels and corresponding unstressed (phonologically reduced) vowels. Results show that, contrary to traditional descriptions, presumably phonologically unreduced vowels in verb + noun compounds sporadically undergo phonological vowel reduction. When they do not, they are shorter than stressed vowels and more centralized in the F1*F2 vowel space. In addition, stressed full vowels do not differ in accented vs. unaccented contexts in duration or vowel quality, indicating that vowels are hyperarticulated under lexical stress, but not when they receive intonational pitch accent. The findings contribute to a body of cross-linguistic research dealing with the influence of prosody at the segmental level.

Stressed postverbal pronominals in Catalan

“…Thus, at least for Central Catalan, while the stressed vowel in the verb is stressed and accented in all conditions in our study, it only receives a nuclear pitch accent in condition 1. The longer duration of the stressed vowel in condition 1 is consistent with findings indicating that accent, not only lexical stress, is cued by duration (Ortega-Llebaria and Prieto 2010) and contrasts with the absence of durational effects in Nadeu (2016). In Nadeu (2016), the target words appeared in prenuclear position, unlike in the present study and in OrtegaLlebaria and Prieto (2010).…”

“…The longer duration of the stressed vowel in condition 1 is consistent with findings indicating that accent, not only lexical stress, is cued by duration (Ortega-Llebaria and Prieto 2010) and contrasts with the absence of durational effects in Nadeu (2016). In Nadeu (2016), the target words appeared in prenuclear position, unlike in the present study and in OrtegaLlebaria and Prieto (2010). Taken together, these findings (as well as those on vowel quality in Central Catalan) indicate that, in Catalan, the correlates of prosodic prominence are specific to different types of prominence -stress vs. accent, but also prenuclear vs. nuclear accent -and not necessarily cumulative (see also Cho and Keating 2009, on English), which may be related to their diverse functions (see the discussion in Nadeu 2016).…”

“…Our finding that /a/ and /ɔ/ are lower in condition 1, where they receive a nuclear pitch accent, is consistent with the enhancement of linguistically encoded information (the low feature of these vowels) due to higher prominence. Note that, for Catalan, Nadeu (2016) observed no localized hyperarticulation due to the presence of prenuclear accent. However, a different type of accent, emphatic stress, did result in an expansion of the vowel space (i. e. more peripheral vowel qualities, Nadeu and Hualde [2012]).…”

“…Similarly, in exceptional cases, in which phonological vowel reduction fails to apply despite absence of lexical stress (i. e. V + N compounds like rentaplats 'dishwasher'), vowel centralization is also observed (Nadeu 2016). Despite vowel quality being such a consistent and reliable cue to the Catalan stress distinction, other correlates of stress (controlling for accent) include duration and spectral balance, the former being the most robust (Astruc and Prieto 2006;Nadeu 2016).…”

“…A study on emphatic stress -i. e. when a pitch accent is exceptionally associated with a lexically unstressed vowel -reveals that despite exhibiting phonological vowel reduction, vowels with emphatic stress are less centralized and longer than their purely unstressed, unaccented counterparts (Nadeu and Hualde 2012). 7 On the other hand, a comparison of stressed vowels in (prenuclear) accented and unaccented contexts shows no difference in either size of the vowel space or vowel duration (Nadeu 2016). 8 This suggests that the correlates of accent may depend on the type of accent (nuclear vs. prenuclear) or on information structure or degree of contrastiveness (broad vs. narrow vs. contrastive focus), although this has not been explored for this language.…”