Phonology schools

Experimental phonology
Nonsegmental phonology
Second language phonology

Prosody models, and Spontaneous Speech

Prosody models, and Spontaneous Speech
By Hiroya Fujisaki from Computing Prosody: computational models for processing spontaneous speech 1996 edited by Yoshinori Sagisaka, Nick Campell, Norio Higuchi
Views on prosodoy
1. Concrete—taking measurements
2. Abstract—building models
3. The author’s view
Three types of input information
1. Linguistic
2. Para-linguistic
3. Non-linguistic
Process of spontaneous speech production
1. Message planning
2. Utterance planning
3. Motor command generation
4. Speech sound production
Role of generative models
1. Finding a model
2. Use of a model in speech synthesis
3. Use of a model in speech recognition

Ch 16

Speech perception
1. Often viewed as a branch of experimental phonetics or of psycholinguistics
2. Should be incorporated in phonology
a. The research reported in this chapter:
Applying explicit statistical pattern-recognition concepts to the perception of phonological categories
b. The results
Illustrates how a careful examination of existing empirical data together with an analytic examination of alternate statistical models of production patterns can give insight into perceptual processes
The relationship between f0 vowel quality
1. Two theoretical positions on the role of f0 in the specification of vowel quality
a. Incorporate f0 directly to the initial feature representations of vowels
b. Separation of the source (f0) and filter (formants, spectrum envelope) properties in perception
2. The position adopted by the authors
a. Statistical relations between and formant patterns are at play in perceptual results

Ch 10

Ohala has gave substance to Baudouin’s insight
a. Misperception as a significant source of sound change
b. Investigation of the nature of such misperceptions by experimental methods
Two fundamental implications of Ohala’s research
a. The innocent misperception can lead directly to attested recurrent sound patterns
b. Sound change is non-teleological
The sources of the resistance to non-teleological models
a. Experimental results are simply ignored
b. Interpretations of perception experiments are not empirically motivated, and fail to recognize lexical effects
c. Simplification of the model

Ch 8

Substance-based approach to phonology
Using phonetics to determine and possibly, model how the emergence of formal systems could be shaped by the perceptuo-motor substance of speech communication.
Both vowel and consonant systems seem to combine dispersion and MUAF principles
Dispersion-focalization theory (DFT)
Two costs determine a sound system
Structural dispersion cost
Local cost
Energy function
MUAF
Based on available controls
Learnability
Speech perception
Auditory theories
Motor theories
The author's view

Ch 7

The use of a large phonological database to test hypotheses about cross-language pattern
The development of the hypothesis
1. The humanistic principle that languages are equal in serving communicative demands
2. Principle of equal complexity
3. Languages will undergo adjustments to equalize their overall complexity across different subsystems
(Rejected by the comparisons conducted in this research)
a. Historical processes
b. Processing considerations
The properties to be examined
a. The complexity of the maximal syllable structure the language permits
b. The complexity of the tone system
c. The consonant inventory size
d. The vowel-quality inventory size
e. The total vowel-inventory size

Ch 5

A reductionist view
Is necessary to bridge the broad gap from VT deformation to acoustic patterns and then to percept
a. VT representation by tubes
b. Articulatory modeling (AM)
F-pattern and VT
Maeda’s articulatory model
a. Guided PCA=> profile of VT
b. Midsaggital distances=> VT area function
c. Transfer function
d. Sound
Illustrations of the method

Ch 4

Three scientific revolutions in the study of language and the emergence of dualism in the study of the sounds of language
1. The positivist codification of “historical linguistics”
a. Descriptive phonetics
b. Experimental phonetics
2. The behaviorist codification of “structural linguistics”
a. Phonology
b. Phonetics
3. The mentalist codification of “generative linguistics”
Phonology was incorporated into the grammar
Two attempts to reconcile phonetics and phonology
1. Laboratory phonology
Phonology-going-into-the-lab
Bridges the phonetics-phonology schism
2. Experimental phonology
Phonology–coming-out-of-the-lab
a. Has the elucidation of speech communication as its goal
b. Subordinates all phonological categories and phonetic measurements to the relevance for communicative functions

Ch 3

Two perspectives to approach language
a. Structural perspective
b. Ecological perspective
The exemplar-based orientation to sensory memory
a. Has a long history in cognitive psychology
b. Can be used to account for recognition and categorization
c. Can be applied to phonology
Two general approaches to language sound systems
a. Generalization
b. Exemplar-based
Two decisions in exemplar-based phonology
a. to choose a unit of representation
b. to represent the dimensions of exemplars
Two mechanisms:
Methods for calculating activation of exemplars in response to input and the spread of that activation in a network of exemplars
a. Similarity matching
b. Exemplar resonance
Permitting activation to spread through the set of exemplars via non-phonetic properties