Chapter 2 Flashcards

You may prefer our related Brainscape-certified flashcards:
1
Q

The task of speech percetion is an extraordinarily complex one, for two reasons:

A
  1. Environmental context often interferes with the speech signal
  2. The variability of the speech signal
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Phonetics

A

The study of speech sounds

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Articulatory phonetics

A

The study of the pronunciation of speech sounds

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Bilabial consonants

A

Articulated at the lips (b and p)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Alveolar consonants

A

Formed by placing the tongue against the alveolar ridge (d and t)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Velar consonants

A

Are produced in the back of the mouth (g and k)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Stop consonants

A

Obstruct the airflow completely

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Fricatives

A

Obstructing without completely stopping the airflow

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Affricate

A

A stoplike closure followed by the slow release characteristic of fricatives

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Voiced speech sound

A

Vibrating vocal cords

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Voiceless

A

If the voical cords are seperated and the air is not obstructed at all

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Spectogram

A

One of the most common ways of describing the acoustical energy of speech sounds. It is produced by presenting a sample of speech to a device.

Vertical = the frequency of the speech

Horizontal = the time

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Intensity in spectogram

A

Darkness of the spectogram at various locations

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Formants

A

Dark bands in spectogram, at various frequency levels.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Formant transitions

A

Large rises/drops in formant frequency that occur over short durations of time. I between is the formant’s steady state.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Parallel transmission

A

different phonemes of the same syllable are encoded into the speech signal simultaneouslty

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

Context-conditioned variation (phenomenon)

A

The exact spectographic appearance of a given phone is related to (conditioned) the speech context

We do not process speech sounds one at a time

The information for each phonetic segment is spread throughout the syllable

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

Coarticulation (phenomenon)

A

Producing more than one speech sound at a given time

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

A cognitive system is modular if:

A
  1. It is domain specific
  2. It operated on a mandatory basis
  3. It is fast
  4. It is unaffected by feedback (higher levels of processing)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

The lack of invariance

A

There is no one-to-one correspondence between acoustic cues and perceptial events

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

Categorical perception

A

To comprehend speech, listeners must impose an absolute/categorical identification on the incoming speech signal. It is our job to identify whether a sound is a p or a d.

A failure to discriminate speech sounds any better than you can identify them.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

The motor theory of speech perception Liberman et al. (1957):

A

Alternative explanation to categorical perception

Listeners use implicit articulatory knowledge (how sounds are pronounced) as an aid in perception

It deals effectively with the lack of invariance (the link between perception and articulation is more direct than the link between acoustics and perception)

Experimental evidences for the theory:

Exp: Students taking foreign language classes are often encouraged to practice articulating new sounds as a means of hearing them better

Exp: The lack of visual information about mouth movements while talking by telephone in a foreign language make comprehension more difficult

Exp: McGurk effect: when visual and auditory information are in conflict, perceivers use both sources of information to arrive at a stable perception

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

Exp: Pollack and Pickett (1964) recorded conversations and spliced out individual words of the tape-recorded conversation

A

Acoustic information may be insufficient to permit identification of speech sounds = higher order syntactic and semantic factors will contribute to comprehension

context > isolation

word recognition in the presence of noise is better in the context of a sentence than in isolation

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
24
Q

The “Trace” model of speech perception: McClelland & Elman (1986,1988)

A

Changes the assumption from the modularity view that phonemic processing is unaffected by higher levels of processing.

Several levels of processing (distinctive features, phonemes and words) are simultaneously active during speech perception and interact in parallel with each other

How it works:
1. Given a particular input (host) different units from different levels will reach a given level of activation

  1. The activation of a given unit inhibit units from the same level and activate units from different levels
25
Q

Orthography

A

A method of mapping the sounds of a laguage into a set of written symbols

26
Q

Languages differ in their orthographies (three main types):

A
  1. Logography
  2. Syllabary
  3. Alphabet
27
Q

Logography

A

paires the unit (word/morpheme) with some pictorial symbol

more transparent in meaning

VB: Chinese characters

28
Q

Syllabary

A

Takes the syllable and associates it with some visual representation

Ma-ca-ro-ni –> 4 symbols

29
Q

Alphabet

A

System in which each letter is supposed to represent a phoneme

involve much less memorization than the other systems

30
Q

Saccades

A

Movements of the eyes during reading (10-20 ms; 10 letters)

31
Q

Regressions

A

saccades moving backward

32
Q

Fixations

A

The time spend at a give location between eye movements (last about 225 ms)

33
Q

Experiment (Neisser 1964)

A

Letter detection speed depends upon the accompanying letters

Letter detection (Z) is faster when flanker letters are constituted by different features (round vs. angle features)

34
Q

Double Route model of reading (Coltheart, 1982; Humphrey, Evetts, 1985; Gough, 1972):

A

Account for normal reading and our ability to read irregular or new words and nonwords. We have two different systems that enable us to read individual words: a rule system and a memory system.

Lexical route: activated the entry for the corresponding word in our internal lexicon

Nonlexical route: this system allows us to correctly pronounce onwords as well as irregular words

35
Q

Phonologic dyslexia

A

difficulties reading nonwords

36
Q

Deep dyslexia

A

read bed instead of chair

37
Q

Word-Superiority effect

A

Cattel: letter perception was faster in a word context than in nonword letter strings or isolated letters

38
Q

Interactive activation model (McClelland y Rumelhart, 1980):

A

Similar to the TRACE model for speech comprehension

    • Levels: features, letters and words
    • Processing occurs simultaneously in the different levels
    • Interactions between levels may be excitatory or inhibitory

The model account for the effects described previously (word-superiority effect, context effects in isolated letter detection…)

Layers:

  1. Orthographic layer: spelling
  2. Phonological layer: pronunciation
  3. Semantic layer: meaning

Single route, does not require the assumption of a mental lexicon

Through experience the learner gradually comes to develop weights between letters and sounds that approximate those of a mature learner.

39
Q

STG

A

superior temporal gyrus

40
Q

Neural bases of speech comprehension

Two different brain networks enable two different tasks:

A

Ventral network “Conceptual system”: connect linguistic sounds and meanings (lexical path)

Dorsal network “Articulatory-motor system”: connect linguistic sounds and articulatory based representations (phonologic output; sublexical path)

41
Q

Left STG

A

Activation associated with the processing of short-time scale patterns (i.e., phonemes)

42
Q

Right STG

A

Activation associated with the integration of phonemes into temporally complex patterns (i.e., words)

43
Q

Internal lexicon

A

Words in permanent memory

44
Q

Lexical acces

A

the proces by which we activate the properties of the words (meaning, spelling, pronunciation, grammatical class, relation to other words)

45
Q

agramatism in Broca aphasia

A

lack of closed-class or functional words

46
Q

Free morphemes

Bound morphemes

A

Free morphemes: words

Bound morphemes: attache to free morphemes

free morphemes + bound morpheme that generate new words
(for example: good (adj) + -ness = goodness (noun)

47
Q

Reference

A

the relationships between a word and an object or event in the world.

48
Q

Sense

A

the place of the word in a system of relationship which contracts with other words in the vocubulary

49
Q

Denotation

A

objective or dictionary meaning of a word

50
Q

Connotation

A

aspects of meaning beyond that which it explicitly names or describes

51
Q

Hierarchical Network Model by Collins & Quillian (1969, 1970, 1972):

A

A network is hierarchical if some of these elements stand above or below other memebers of the network

Cognitive economy principle: The space available for the storage of semantic information is limited, so it would be beneficial to store information only in one place (at the higher possible node). –> Typically effect –> not a good candidate for a model for the internal lexicon

52
Q

Semantic verification tasks

A

A person is presented with a statement of the form An Animal is a Bird, and asked to determine as quickly as possible whether the sentence is true or false

Reaction times taken to answer are thought to reflect the organization of information in the internal lexicon => time = distance between different words in the internal lexicon

53
Q

Category-size effect:

A

in a statement of the form an Animal is a Bird, the higher the location of Bird in the hierarchy in relation to Animal, the longer RTs

Bird –> Animal = Higher RT

Bird –> Salmon = Lower RT

54
Q

Problem for Hierarchical Network Model by Collins & Quillian (1969, 1970, 1972):

Typicality effect:

A

Items that are more typical of a given subordinate take less time to verify than atypical items in true statements

55
Q

Spreading activation model by Collins & Loftus (1975):

A

Words are represented in the internal lexicon in a network (not strictly hierarchical)

with the distance between nodes determined by both structural characteristics such as taxonomic relations and considerations such as typicality and degree of association between related concepts

Limitations: little attention to phonological, syntactic, and morphological aspects of words (it is a model for concepts rather than words)

56
Q

Spreading activation model Bock y Levelt (1994)

A

Knowledge of words exists at three different levels

a. - Conceptual level: nodes that represent concepts and their relation
b. - Lemma level: Refers to syntactic aspects of word knowledge
c. - Lexeme level: word’s phonological properties, or how word sounds

Individuals in the TOT state can fail to recall the correct word but still retrieve similar-sounding words (speakers knew the word’s meaning-concept- and syntactic category-lemma- but not its phonologica features).

The model appears to be useful in understanding lexical access in both comprehension and production

57
Q

Autonomous search model of Forster (1976, 1979)

A

Activation of words from the lexicon is not directly influenced by syntactic or semantic factors.

In the model the word recognition system is divided into several different components:

a. - Orthographic analysis component (spelling)
b. - Phonetical analysis component

58
Q

Logogen Model (Morton, 1969): activation model

A

Each word (morpheme) in the lexicon is represented as a logogen

The logogen is activated in either two ways:

a. - Sensory input: when orthographic or phonological features of a word are detected
b. - Contextual information: semantic and syntactic information may temporally reduce a logogen activation threshold.

These two routes are assumed to work in parallel. Sensory and contextual matches increase the same counter

59
Q

Cohort model (Marslen-Wilson et al., 1987, 1990)

A

Developed to account for auditory word recognition: word recognition is fast, listeners are sensitive to the recognition point of a word (point at which the word diverges from other possible candidates)

Three stages for spoken word recognition::

a.- Activation of the initial cohort of candidates: based on the acoustic-phonetic analysis of the input

Bottom-up process; ; Ej: Ángel ha perdido su bo… (bolso, bolígrafo, bombilla, bocadillo…)

The activation level of the elements of the cohort depends on their similarity to the input

b.- Selection of the members for analysis (Is sensitive to multiple sources of information)

Elimination by: (a) sentence context and (b) similarity with the incoming signal

c.- Integration: the selected word is fit into the connected discourse