👀
5. Word Recognition
Created @August 21, 2025 1:49 AM
Class PS2021
Spoken Word Recognition
Why understanding speech is often difficult is due to factors that decrease
speech intelligibility
Mattys et al. (2012):
Energetic masking (bottom-up processing): Distracting sounds degrade
intelligibility of target words (e.g., multiple conversations, traffic noise)
Informational masking (top-down processing): Cognitive load (e.g.,
performing a second task while listening to speech) hinders speech
perception
Characteristics of Speech that Make Word
Recognition difficult
Segmentation
Distinguishing phonemes (units of sound) and words from patterns of
speech sounds
Most speech has few periods of silence, making it hard to identify when
one word ends and the next begins
Noticeable in unfamiliar foreign languages
Covariance
Pronunciation of a phoneme is influenced by surrounding phonemes
5. Word Recognition 1
, Example (Harley, 2010): The /b/ in ‘bill’, ‘ball’, ‘able’, and ‘rub’ sounds
different due to coarticulation
Problem: Coarticulation increases variability in speech signals
Advantage: Helps listeners anticipate the next phoneme
What helps to cope with segmentation and covariance
Dividing the speech signal into its constituent word
Mattys et al. (2012) this involves using cues
Acoustic Phonetic: coarticulation
Dependence on listener’s knowledge
The immediate context
Point Method Results
Identified three main
categories of cues in his
hierarchal approach
Auditory stimuli: Tier 1: Lexical
Sequences of spoken (syntax; word knowledge)
words varying in
background noise or Tier 2: Segmental
pseudowords to identify Use segmental (sound)
how well participants could cues (coarticulation) and
Mattys et al. (2005)
identify words how they change in
different contexts
Asked to identify words or (one phoneme may be
segments to measure for associated with two or more
effectivess of different similar sounds like /p/ in ‘pit’
segmentation cues and ‘spit’)
Tier 3: Metrical prosody
Focusing on which syllables
are stressed
Miller et al. (2023) Same methodology Coarticulation was more
significant than other
More coarticulation within Had to distinguish between acoustic cues, such as
than between words words based on lexical and metrical cues,
coarticulation sounds and especially in conditions
5. Word Recognition 2
5. Word Recognition
Created @August 21, 2025 1:49 AM
Class PS2021
Spoken Word Recognition
Why understanding speech is often difficult is due to factors that decrease
speech intelligibility
Mattys et al. (2012):
Energetic masking (bottom-up processing): Distracting sounds degrade
intelligibility of target words (e.g., multiple conversations, traffic noise)
Informational masking (top-down processing): Cognitive load (e.g.,
performing a second task while listening to speech) hinders speech
perception
Characteristics of Speech that Make Word
Recognition difficult
Segmentation
Distinguishing phonemes (units of sound) and words from patterns of
speech sounds
Most speech has few periods of silence, making it hard to identify when
one word ends and the next begins
Noticeable in unfamiliar foreign languages
Covariance
Pronunciation of a phoneme is influenced by surrounding phonemes
5. Word Recognition 1
, Example (Harley, 2010): The /b/ in ‘bill’, ‘ball’, ‘able’, and ‘rub’ sounds
different due to coarticulation
Problem: Coarticulation increases variability in speech signals
Advantage: Helps listeners anticipate the next phoneme
What helps to cope with segmentation and covariance
Dividing the speech signal into its constituent word
Mattys et al. (2012) this involves using cues
Acoustic Phonetic: coarticulation
Dependence on listener’s knowledge
The immediate context
Point Method Results
Identified three main
categories of cues in his
hierarchal approach
Auditory stimuli: Tier 1: Lexical
Sequences of spoken (syntax; word knowledge)
words varying in
background noise or Tier 2: Segmental
pseudowords to identify Use segmental (sound)
how well participants could cues (coarticulation) and
Mattys et al. (2005)
identify words how they change in
different contexts
Asked to identify words or (one phoneme may be
segments to measure for associated with two or more
effectivess of different similar sounds like /p/ in ‘pit’
segmentation cues and ‘spit’)
Tier 3: Metrical prosody
Focusing on which syllables
are stressed
Miller et al. (2023) Same methodology Coarticulation was more
significant than other
More coarticulation within Had to distinguish between acoustic cues, such as
than between words words based on lexical and metrical cues,
coarticulation sounds and especially in conditions
5. Word Recognition 2
