Systematic durational patterns in languages
The acoustic duration of intervals associated with consonants and vowel has been shown to vary systematically as a function of a number of influencing factors.
These durational modulations are necessary to make speech sound
- natural
- native
Gestural causes of durational adjustments
The changes in measured duration of intervals are caused by
- the dynamics of individual gestures (e.g., how long it takes for their constrictions to form)
- the relative timing of gestures.
Specific gestural mechanism that causes durational modulation may vary for different influencing factors.
Most of our knowledge of durational patterns is, however, based on acoustic measurements. We do not know the underlying gestural mechanisms.
Control of duration
Some influence factors are universal.
(presumably these have some physical, physiological cause, not controlled by a speaker).
Others must be actively controlled by speakers as part of the their knowledge of the phonetic or phonological structure of their language.
Influencing Factors
Gestural modulations responsible may differ as a function of factors.
Universal
Lower vowels are longer than higher vowels (Lehiste, 1970):
Labial stops are longer than coronals or dorsals.
Language-specific
Fricatives are usually longer than stops.
Consonant "gemination" (doubling)
Geminate consonants are roughly twice the duration of single consonants
- Italian
- Japanese
- Finnish
- Arabic
Vowel length
Long vowels usually differ from short ones in quality as well
Czech (Lehiste,1970)
Why do we consider the the length contrast as primary when there are also quality differences?
Quality differences are regular and predictable
Serbo-Croatian (Lehiste, 1970)
Speakers could be producing the same gestures for short vowels as for long vowels, but the short vowels are truncated.
Length contrast vs. tense-lax contrast
English can be described as having a length contrast:
Tense vowels are long.
Lax vowels occur only in closed syllables.
Tense-lax contrast is neutralized before /r/ and /N/.
Long (tense) vowels do not have twice the duration of short (lax) vowels.
Variation due to vowel quality much greater than in language with clear length contrast (e.g., Finnish).
Vowels shortest before labials (universal).
Vowels shorter before voiceless consonants than voiced (almost universal).
Vowels shorter before stops that before fricatives.
Word effects
The greater the number of syllables in a word, the shorter each vowel is.
The last vowel in a word is longest.
Interaction of multiple factors (Port, 1981):
- word
- length (tense-lax)
- local context
Stress effects
Vowels in stressed syllables are longer than in unstressed syllables in many (but not all) languages.
English (Lea, 1977)
Not true in other languages, e.g., those with length contrasts
Czech
First syllables are accented (Lehiste, 1970):
Falling vs. rising accent in Serbo-Croatian (Lehiste, 1970):
Phrase Effects
Vowels (and consonants) are longer at the end of phrase than within a phrase.
Gestures are slowed down near phrase boundaries.
Types of Phrase boundaries
Do different kinds of boundaries exert different amounts on slowing down on the boundary gestures?
Byrd & Saltzman (1998)
Example data Results:
PI-gestures slow down speech production clock