Information Conveyed by Vowels (Ladefoged and Broadbent) Three kinds of information conveyed Linguistic Meaning of utterance
Information Conveyed by Vowels (Ladefoged and Broadbent) Three kinds of information conveyed Linguistic Meaning of utterance Sociolinguistic Social class Region of origin Ethnicity
Information Conveyed by Vowels (Ladefoged and Broadbent) Three kinds of information conveyed Linguistic Meaning of utterance Sociolinguistic Social class Region of origin Ethnicity Idiosyncratic Gender Personal vocal characteristics
Information Conveyed by Vowels (Ladefoged and Broadbent) Most information about vowels is in formants
Two theories Vowel identification is made by absolute values of formants (with a little wiggle room) It's [I] because F1 is about 270 and F2 is about 2290
Two theories Vowel identification is made by absolute values of formants (with a little wiggle room) It's [I] because F1 is about 270 and F2 is about 2290 Vowel identification depends on relative formant frequencies for individual speaker
Two theories Bob's [e] in red Bill's [e] in blue Bob's vowels are all lower than Bill's If you heard Bob use Bill's [e] you'd hear [i]
Two theories Bill's and Bob's [i, e, æ] are identified in respect to each other not to a particular formant frequency [i] [e] [æ]
Two theories “The effect that the phonetic quality of a vowel depends on the relationship between the formant frequencies for that vowel and the format frequencies of other vowels pronounced by that speaker.” It's the relationships between vowels not the absolute F1 and F2 frequencies
The experiment Please say what this word is produced by synthesizer Six different versions Sounded like different people saying same thing
The experiment Four test words synthesized Vowel ranged between bit, bet, bat, but
The experiment Four test words synthesized Vowel ranged between bit, bet, bat, but People heard a test sentence then a test word They chose which word (bit, bet, bat, but) they heard
The experiment If vowel identification depends on absolute frequencies the different versions of Please say what this word is shouldn't influence test word choice
The experiment If vowel identification depends on absolute frequencies the different versions of Please say what this word is shouldn't influence test word choice If vowel identification depends on relative difference between an individual's vowels, the different version of the test sentence will influence test word choice
The experiment
Connections with formal approaches Formal approaches require static, unchanging symbolic units to be manipulated
Connections with formal approaches Formal approaches require static, unchanging symbolic units to be manipulated Vowels aren't static with absolute frequencies
Connections with formal approaches Formal approaches require static, unchanging symbolic units to be manipulated Vowels aren't static with absolute frequencies Vowel identification depends on relative frequencies
Connections with formal approaches Formal approaches require static, unchanging symbolic units to be manipulated Vowels aren't static with absolute frequencies Vowel identification depends on relative frequencies How would formal approach account for the results of this experiment (done in 1957)?