Auditory Word Recognition

Question 1

What is auditory word recognition?

Answer 1

Understanding spoken vs written words

Question 2

What is a theoretical problem about speech signals?

Answer 2

The same phonemes can sound differently in different circumstances, but people appear to be able to cope with the variability relatively well.

Question 3

What is assimilation (theoretical problem of speech signals)?

Answer 3

Phonemes can be influenced by some acoustic properties of their neighbor.

Question 4

What is the speaker variability in the theoretical problem in speech signals?

Answer 4

No two speakers produce the same phonemes in the same way due to different accents and different size of vocal tract. Also Socio economic accents (posh) or regional accents/dialects

Question 5

What is the segmentation problem in speech signals?

Answer 5

Words are often uttered without a clear break between them. Words are articulated.

Question 6

What are the two segmentation strategies?

Answer 6

Possible word constraint and metrical segmentation strategy.

Question 7

What is the possible word constraint strategy?

Answer 7

Speech is segmented in such a way that there is no syllable left attached to words.

Question 8

What is the metrical segmentation strategy?

Answer 8

Subjects inserted word boundaries are clear, so they are used for segmentation (syllable based segmentation)

Question 9

What is phoneme perception?

Answer 9

(Pre-lexical processing).

- some phenomena in phoneme perception, phoneme perception and phoneme mediation in word recognition.

Question 10

What is categorical perception?

Answer 10

People perceive a speech sound as one phoneme or another - nothing in between.

Question 11

Which study did Liberman and co (1957) do?

Answer 11

The ba, da and ga study. They manipulated the voice onset time. 0ms was ba and 60ms was pa but 20-40 was both.

Question 12

What is the McGurk effect?

Answer 12

Phoneme perception can be influenced by input from another modal (interaction between visual and auditory modals).
Senses work together. We relay on information from more than 1 modality.

Hear ba, see ga and actually produce pa.

Question 13

Who has done research on phoneme perception?

Answer 13

Floss and blank (1980) but also Foss and Gernbacher (1983)

Question 14

What does Foss and Blank (1980) investigate with their phoneme monitoring task?

Answer 14

Word recognition in phoneme perception. So is information related to the word accessed before a phoneme within the word identified.
They found it was affected by word predictability (return book/bill).

Question 15

What are the phoneme perception model?

Answer 15

Dual code model which is based on pre-lexical (phonetic) code and post lexical (phonetic) code.

Question 16

What is pre-lexical (phonetic) code in the dual code model?

Answer 16

Phonetic is computed directly from the perceptual analysis of the input acoustic information about the word.
- people use the pre lexical information when the task is easy and also when processing non-words.

Question 17

What is post lexical (phonetic) code in the dual code model?

Answer 17

Computed from information derived from higher level units (words). People use the post lexical code when the task is difficult (the target word unpredicted by context).
Both bottom up and top down

Question 18

What did Foss and Gernbacher (1983) find in the phoneme monitoring task?

Answer 18

Post lexical code may be used but only in very exceptional circumstances.
Phoneme identification is primarily based on the pre lexical code. To process phonemes, we don’t have to use information in which the phonemes are embedded in.
Both bottom up and top down

Question 19

Do phonemes have to be identified before the spoken word is recognized?

Answer 19

No, you don’t need to understand phonemes to understand words. Only bottom up processes.

Question 20

What is the lexical decision task by Marslen-Wilson and warren (1983)?

Answer 20

Three different non-word of smob.
3 different types of smob where 1 smo is taken from a real word and b from non word. The other 2 conditions from non word smo and non word b.
The found a co articulation (phonetic not phonemic) effect. There are no phonological mediation).

Question 21

What is phoneme restoration?

Answer 21

Missing phonemes in words can be filled by preceding sentential context

Question 22

Where do we often see phoneme restoration?

Answer 22

In media when they use swear words

Question 23

What did warren (1970) find in a lexical identification task?

Answer 23

People can report the missing phonemes in words and even if multiple phonemes are missing.

Question 24

What does phoneme restoration prove?

Answer 24

That we use semantic and syntactic information beyond individual phonemes.

Question 25

What is “the *eel was on the orange” an example of?

Answer 25

Phoneme restoration. In this example we can restore the phoneme based on context.

Question 26

Which 3 models explain auditory word recognition?

Answer 26

Cohort model (original), Zwitserlood’s experiment and cohort model (revised).

Question 27

What is the cohort model?

Answer 27

As we hear a speech steam, we set up a cohort of possible items that the word could be.
Items are eliminated until only one is left in the cohort.
The mental representations of words are activated (in parallel) on the basis of bottom up input (sounds) and top down input (contexts).

Question 28

According to the cohort model when does word recognition happen?

Answer 28

After pre lexical but before post lexical

Question 29

What is the uniqueness point in cohort model?

Answer 29

The point at which a word becomes uniquely identifiable (in theory).

Question 30

What is the recognition point in the cohort model?

Answer 30

The point at which a word is empirically identified. It could: coincide with uniqueness point, be earlier if there is strong supporting information (sentential context) and be later if the stimulus is degraded (poor sound quality).

Question 31

Which test did Zwitserlood (1989) used to test the cohort model?

Answer 31

The cross model priming (with (visual) lexical decision).

Question 32

What is cross modal priming (with (visual) lexical decision) test of cohort model?

Answer 32

Visually seeing a word which could be associated with two things.
For example, captain (ship or money).
The test looks at probe position (early or late) and context (conditions: biasing, neutral, carrier, control)

Question 33

What is the findings of the cross modal priming (with (visual) lexical decision) test of cohort model?

Answer 33

Frequency effects: words with HF had bigger effects than words with LF.
No context in early stage of processing: biasing context didn’t have a larger priming effect.
Context effects at later stage of processing: context effects were found when the visual words were presented later in the words.

Question 34

How is the original cohort model different from the revised cohort model?

Answer 34

The revised model goes against the original model.

Question 35

What are the changes in the revised cohort model?

Answer 35

Bottom up priority, frequency effects, no “all or none” elimination and re-activation of decayed candidates.

Question 36

What are the common effects in auditory word recognition?

Answer 36

Frequency effect, neighbourhood effects and context effects.

Question 37

What is neighborhood effects?

Answer 37

It can be neighborhood size or frequency.

Question 38

What is the neighborhood size effects in auditory word recognition?

Answer 38

Words with a smaller neighbor are recognized faster than those with a bigger neighborhood

Question 39

What is neighborhood frequency effect in auditory word recognition?

Answer 39

Words with low frequency neighbors are recognized faster than words with high frequency neighbors

Question 40

What is context effects in auditory word recognition?

Answer 40

Words are recognized faster when they are embedded in a coherent context. However, the effects takes place at a later stage of processing!

Question 41

What are the evaluations of the revised cohort model?

Answer 41

It can explain frequency effects, neighborhood size effects and context effects.
However, it cannot explain neighborhood frequency effects!