Chapter 12: Models and Theories of Speech Production and Perception

Model

a simplification of a system or any of its parts into a real world system but only its essential variables

How are models constructed?

Models are constructed to represent the system in some way that can then be manipulated in a controlled manner

Physiological models

We use specimens taken from animal or human cadavers to determine how a particular structure responds under different conditions

Theory

a theory is a statement about a particular phenomenon, incorporating the underlying principles and assumptions

What do theories help to explain?

Theories help to explain observed data and information and can be used to make predictions about events related to the phenomenon in question

How do theories and research interact?

Research and theory interact with each other to generate new research, theories, and practices

The serial order issue

a sequence of sound elements from sound, phonemes, syllables to a larger unit

Degrees of Freedom

a pre-structured set central commands capable of carrying out a movement aka the many ways in which articulators can move

Context-Sensitivity Problem

Sounds vary within the context in which they are produced and are influences by speaking rate, stress, clarity of articulation, and other factors

What do Dynamic Systems Models address?

The degrees of freedom problem.

How do Dynamic Systems Models limit degrees of freedom?

By reducing them from infinite to a few useful degrees of freedom.

What is synergy?

Muscles working together in groups

What are Connectionist Models?

Connectionist models are defined as a non-linear, non-hierarchical set of components where various elements process information simultaneously

Example of the connectionist model

The node for phoneme /k/ might be connected in one direction to all morphemes containing that phoneme (e.g., cat, truck, tick, tock, cap, etc.) and in another direction to the distinct features of that phoneme (e.g., voiceless, velar, stop)

Target Models

describe speech production as a physical specification of a planned phonological unit such as a syllable or word

What are the two types of target models?

spatial (articulatory) or acoustic (auditory)

Spatial Models

There is an internalized map of the vocal tract in the brain that allows the speaker to move his or her articulators to specific regions within the vocal tract

Feedback Models

Feedback is the transfer of part of the output of a system back to the input to regulate and correct any errors in the output

Sensory feedback

transfer of a portion of the system's output to the input for regulation and error correction

How do feedback models work for speech?

Feedback models compare speech with intended action

Lack of variance

There is no one-to-one relationship between acoustic feature and perceived sound

Lack of segmentation

the acoustic waveform is not consistent with how you hear it

Duplex perception

Simultaneous perception of nonspeech and speech stimuli extracted from segments of the speech signal

McGurk Effect

differences between the visual and auditory stimuli resulted in a unique speech sound perceived

Two types of spatial-temporal organization

spatial and acoustic

Spatial organization

articulators aim for the targets from various initial positions to produce the desired acoustic result

Acoustic organizaton

Articulatory movements are based on an acoustic outcome rather than a particular placement

Unit of analysis

Moving on after finishing or achieving a goal, e.g., sound, syllable, word, gesture

Feedforward Models

proposes that the articulators can execute a desired message due to their experience performing similar tasks. Sensory information is "accessed" before it is made available so that errors are minimized, and the task is completed accurately.

Example of how feedback and feedforward models work

Feedback models will tell the person that they said a word wrong. After getting the word correct 9/10 times with the corrective help of feedback, the feedforward model allows them to say the word correctly moving forward.

Acoustic (auditory) targets

The goal to be achieved is the acoustic output; the articulatory movements used to achieve the acoustic output may vary. Thus, a speaker may use different articulatory movements to achieve a particular speech sound, depending on factors such as the adjacent sounds, rate of speech, and different patterns of stress.

What can impact acoustic targets?

Limitation in acoustic feedback, such as hearing impairment may negatively impact speech production

Aerodynamic pressures

managing airflow from the lungs to create sound

Example of aerodynamic pressure

To produce "boy" you must put the lips together to build pressure for sudden release

Articulatory gesture

The speaker has an internal map composed of spatial targets of the vocal tract that guides the articulators.

What is the problem with articulatory gestures?

A very large number of gestures would have to be stored in the brain and readily accessed.

How is novel sound production learned?

Novel sound production is learned through motor programs, which are enhanced through repetition.

Do children under 12 have the same speech motor control as adults?

No, children under 12 years demonstrate spatiotemporal patterns similar to but not as stable as adults.

coarticulation

The adjustment of articulator movements to target more than one speech sound simultaneously

How do biomechanical properties affect coarticulation?

Biomechanical properties constrain movement velocities, influencing the production of speech sounds sharing the same border.

Directions Into Velocities of Articulation (DIVA)

a neural network framework explaining how the brain learns and produces speech through integrated feedforward and feedback control systems. It maps speech sounds to motor commands via auditory and somatosensory feedback, simulating vocal tract movements, particularly during early acquisition and for correcting speech errors.

Motor Theory

Assumes comprehension of speech is based on ability to produce speech

What does the motor theory say about articulatory gestures?

Articulatory gestures are thought to be innately mapped by allowing an individual to picture the movements of all of the articulators despite hidden nature of majority of them.

What is the motor theory based upon?

Mirror neurons

Native Language Magnet Theory

Phonetic categories of one's native language are organized as prototypes (perceptual magnets). This is done by facilitating processing of speech information distinguishing irrelevant distinctions close to prototype from distinction near category boundaries.

Perception of speaker identity

Information arises from sociolinguistic factors such as: region of origin, social status, gender, ethnicity, etc.

How does speech science connect to model/theory development?

It allows researchers and clinicians to quantify the complex, typically subjective act of speech by verifying claims a theory proposes. Also, it allows them to accurately assess and treat disordered speech.

Acoustic Landmarks and Distinctive Features

Words are represented in memory as a sequence of segments, each of which consists of a bundle of binary distinctive features

What are the three sets of acoustic landmarks?

1) Vocalic - Frequency

2) Glide - Amplitude

3) Consonantal - Closure and distribution of energy