Détail de la contribution
Auteur: Aaron BRAVER
Co-Auteur(s): Shigeto KAWAHARA, Rutgers, USA
Titre:
Japanese Monomoraic Vowel Lengthening as Incomplete Neutralization
Abstract/Résumé: INTRODUCTION: Incomplete neutralization is when underlyingly distinct segments become nearly identical on the surface. This experiment shows that in Japanese, vowel lengthening due to bimoraic minimality in PWds creates an incompletely neutralized vowel length contrast. Thus, the typology of incomplete neutralization should include not just segment- and feature-level phenomena like final devoicing (e.g., Dinnsen and Charles-Luce 1984, Port and O'Dell 1985), but also processes that directly affect suprasegmental structure like moras. Japanese prefers minimally bimoraic PWds (Poser 1990, Itô 1990). Monomoraic nouns, e.g., 'fu' (gluten), combine with particles, e.g., 'mo' (also), to meet this requirement. When no particles are present, monomoraic nouns lengthen to meet the requirement. Mori (2002) found that monomoraic nouns without particles, like 'fu' in (2), are 40–50% longer than monomoraic nouns followed by particles, like 'fu mo' in (1). Our experiment shows that this lengthening is incompletely neutralizing by (a) replicating Mori's finding, and (b) showing that vowels in lengthened nouns are shorter than underlyingly long vowels in nouns with identical segments (not examined by Mori). METHOD: 15 sets of minimal triplet sentences were constructed, each containing (1) a monomoraic noun with a particle (short/prt), (2) a monomoraic noun without a particle (short/Ø), and (3) an underlyingly long noun (long). 12 speakers read each sentence 7 times, in random order. (1) fu mo nokoshita yo gluten also remained EMPH (2) fu Ø nokoshita yo gluten Ø remained EMPH (3) fuu Ø nokoshita yo seal Ø remained EMPH RESULTS: Short/Ø vowels were significantly longer than short/prt vowels (dif.=69.96ms, t=15.69, p<0.001). Vowels in underlyingly bimoraic long nouns were significantly longer than in the (lengthened) short/Ø nouns (dif.=32.47ms, t=4.05, p<0.001). This three-way distinction (short/prt > short/Ø > long) holds for all speakers and items. DISCUSSION: These results suggest that the short/long vowel length contrast in Japanese is incompletely neutralized in monomoraic noun lengthening: these nouns must be bimoraic on the surface, yet their vowel durations lie between those of underlyingly short and underlyingly long vowels. We know of no previously reported cases where a vowel length contrast is incompletely neutralized (modulo, perhaps, iambic lengthening, e.g., Hayes 1995). Further, this lengthening cannot be treated as phonetic, since its trigger is a clearly phonological constraint (as opposed to, e.g., intrusive stop in English, Ohala 1974). These results expand the typology of processes leading to incomplete neutralization to include not just processes at the segment- and feature-level, but also processes motivated by suprasegmental structure.
Co-Auteur(s): Shigeto KAWAHARA, Rutgers, USA
Titre:
Japanese Monomoraic Vowel Lengthening as Incomplete Neutralization
Abstract/Résumé: INTRODUCTION: Incomplete neutralization is when underlyingly distinct segments become nearly identical on the surface. This experiment shows that in Japanese, vowel lengthening due to bimoraic minimality in PWds creates an incompletely neutralized vowel length contrast. Thus, the typology of incomplete neutralization should include not just segment- and feature-level phenomena like final devoicing (e.g., Dinnsen and Charles-Luce 1984, Port and O'Dell 1985), but also processes that directly affect suprasegmental structure like moras. Japanese prefers minimally bimoraic PWds (Poser 1990, Itô 1990). Monomoraic nouns, e.g., 'fu' (gluten), combine with particles, e.g., 'mo' (also), to meet this requirement. When no particles are present, monomoraic nouns lengthen to meet the requirement. Mori (2002) found that monomoraic nouns without particles, like 'fu' in (2), are 40–50% longer than monomoraic nouns followed by particles, like 'fu mo' in (1). Our experiment shows that this lengthening is incompletely neutralizing by (a) replicating Mori's finding, and (b) showing that vowels in lengthened nouns are shorter than underlyingly long vowels in nouns with identical segments (not examined by Mori). METHOD: 15 sets of minimal triplet sentences were constructed, each containing (1) a monomoraic noun with a particle (short/prt), (2) a monomoraic noun without a particle (short/Ø), and (3) an underlyingly long noun (long). 12 speakers read each sentence 7 times, in random order. (1) fu mo nokoshita yo gluten also remained EMPH (2) fu Ø nokoshita yo gluten Ø remained EMPH (3) fuu Ø nokoshita yo seal Ø remained EMPH RESULTS: Short/Ø vowels were significantly longer than short/prt vowels (dif.=69.96ms, t=15.69, p<0.001). Vowels in underlyingly bimoraic long nouns were significantly longer than in the (lengthened) short/Ø nouns (dif.=32.47ms, t=4.05, p<0.001). This three-way distinction (short/prt > short/Ø > long) holds for all speakers and items. DISCUSSION: These results suggest that the short/long vowel length contrast in Japanese is incompletely neutralized in monomoraic noun lengthening: these nouns must be bimoraic on the surface, yet their vowel durations lie between those of underlyingly short and underlyingly long vowels. We know of no previously reported cases where a vowel length contrast is incompletely neutralized (modulo, perhaps, iambic lengthening, e.g., Hayes 1995). Further, this lengthening cannot be treated as phonetic, since its trigger is a clearly phonological constraint (as opposed to, e.g., intrusive stop in English, Ohala 1974). These results expand the typology of processes leading to incomplete neutralization to include not just processes at the segment- and feature-level, but also processes motivated by suprasegmental structure.