John Ingram & Thu Nguyen
University of Queensland
Compounding is a highly productive process in English and most languages. Compounding is a word formation process that provides an interface between the compositional semantics of the phrase and the direct or non-compositional form-meaning mappings that characterize lexical items. Phonologically speaking, at least in English, compounding involves a phrase taking on the prosodic attributes of a word i.e., a culminative stress pattern, in which there is not only a contrasting (left headed) pattern of stress-prominence (compared with the right-headed pattern of the phrase), but where the internal timing of the compound is fitted to word-prosody requirements, such as compensatory shortening in polysyllabic constructions and changes of foot structure. The phonological features which distinguish compounds from their phrasal counterparts, are also closely related to the prosodic cues that (optionally) permit speakers to resolve 'attachment ambiguities' of the kind shown in (1) and (2), which are traditionally termed 'juncture':
(1) He helped [the lady [over the road]].
Restrictive: within NP
(2) He helped [the lady] [over the road].
Both juncture (timing) and prominence (accent placement) are involved in the compound-phrasal contrasts illustrated in (3) - (6). Compare:
(3) a red [book-cover]
(4) a [red-book] cover
(5) a light [house-keeper]
(6) a [light-house] keeper
Because compound formation takes place at the interface of the lexicon and the syntax it has attributes of both domains. In recent years there has been considerable interest in the role that typological prosodic differences may play in perceptual parsing of the speech signal for word boundary detection and 'prosodic bootstrapping'. In view of the universality of compounding, it might be expected that there will be strong similarities in the prosodic features that distinguish compounds from phrases and which resolve attachment ambiguities in phrasal constructions across languages.
In this paper we undertake an acoustic analysis of prosodic features that distinguish compounds and phrases and which resolve structural ambiguities in NPs in three typologically different languages: English (a stress-accent language, with stress timing), Japanese (a pitch-accent language, with moraic timing) and Vietnamese (A tonal language with syllable timing). We investigate minimal prosodic pairs such as:
| JAPANESE | ENGLISH | |
| (1) | [nurí gasa] ire | [lacquered umbrella] case |
| case for lacquered umbrellas | ||
| (2) | nuri [kasá ire] | lacquered [umbrella case] |
| lacquered umbrella case | ||
| (3) | óokina [nooen-no óonaa] | a big [farm owner] |
| farm owner who is big | ||
| (4) | [óokina nooen-no] óonaa | a [big farm] owner |
| owner of big farm | ||
Examples (1-2) above are compound expressions and (3-4) are phrasal constructions in Japanese. The application of the voicing assimilation rule rendaku which only applies word-internally clearly signals the compound status of (1-2) and the inflectional suffix (genitive case) on nooen 'farm' signals the phrasal status of (3-4). Vietnamese also makes extensive use of compounding and has the ability to resolve left versus right branching ambiguities of the type illustrated in (1-4) above.
The aims of this study are:
1. To compare the prosodic features which disambiguate left and right branching readings of compound and phrasal constructions in Japanese, Vietnamese, and English as representative cases of three prosodic linguistic types.
2. To determine whether the prosodic features used to resolve left and right attachment ambiguities in phrasal compounds are the same or different from those in phrasal constructions of comparable phonological complexity in each of the languages under discussion.
Method: A common methodology was used in the three experiments reported here. Native speakers who had considerable experience in recording items for language teaching or phonetic experiments produced pairs of phrases spoken in 'the careful but natural style of a news reader' which contrasted left versus right branching readings in pairs of noun phrases, or phrasal compounds. Multiple tokens of the contrasts were recorded, digitized, phonetically annotated, and analysed for the principal temporal, voice pitch, and intensity parameters that have been established as acoustic cues to accent prominence, and phrase boundary marking (juncture). Analysis of variance, with Phrase type and Direction of branching as independent variables, provided the main analytical tool in this study. The dependent variables were, by turn: juncture duration, syllable duration (separately analysed for position in the phrase), fundamental frequency (f0) and vowel intensity, also separately for each position in the phrase.