UALL 2004

PSYCHOLINGUISTICS

LECTURE 8:

Word Recognition

1
 Learning Outcomes
At the end of this lecture, students will be able to:-
• Understand context effect in word recognition.
• Identify stages involved in spoken word
recognition process.
• Identify and explain models of lexical access
(Models of Speech Recognition – Interactive
Models)
• Discuss a special problem for the mental lexicon.
• Determine the time course of activation for
lexically ambiguous words.
2
 Context Effect in Word Recognition
• Does context affect spoken word recognition?

• The nature of the context being discussed also depends

on the level of analysis.

• For example, we might have word-level context

operating on phoneme identification, and sentence level
context operating on word identification.

3
 Context Effect in Word Recognition
• To show that context affects recognition, we need to
demonstrate top-down influences on the bottom-up
processing of the acoustic signal.

• There are 2 opposing positions on the role of context in

recognition, which can be called the autonomous and
interactionist positions.

• The autonomous position says that context CANNOT

have an effect prior to word recognition. It can only
contribute to the evaluation and integration of the output
of lexical processing.
4
 Context Effect in Word Recognition
• For example, information flow is allowed between words
within the lexicon, not from the lexicon to lower level
processes such as phoneme identification.

• The interactionist position allows different types of

information to interact with one another.

• For example, information about the meaning of the

sentence or the pragmatic context might affect
perception.

5
 Context Effect in Word Recognition
• In an interactionist view, context has an effect before
or during the access and selection phases.

• In an autonomous view, context can only have an

influence after a word has emerged as the best fit to the
sensory input.

6
Context Effect in Word Recognition

7
 Spoken Word Recognition Process
• (1)Initial contact, (2)lexical decision (selection) and
(3)word recognition are the three stages involved in
spoken word recognition process.

– Initial contact – the listener takes the speech as input

and generates the representations which contact the
internally stored form-based representations
associated with each lexical entry.

8
Spoken Word Recognition Process
– Lexical decision (selection) – accumulating sensory input
continuesly to map onto this subset until the intended
lexical entry is eventually selected.

– Word recognition – the end point of the selection phase

when a listener has determined which lexical entry was
actually heard.

9
 Models of Lexical Access
(Models of Speech Recognition – Interactive Models)

• Cohort Model and TRACE Model are concerned with

auditory word recognition.

• For these models, the end result is a meaningful

utterance rather than a meaningless syllable.

• These models aim to describe the interaction between

the processes of phoneme recognition and word
recognition.

10
 Cohort Model
(Marslen-Wilson, 1987; Marslen-Wilson & Welsh, 1978; Marslen-
Wilson & Tyler, 1980, 1981)
STAGE 1: The acoustic-phonetic
information at the beginning of a target
word activated all words that resembles
it, making up a “word-initial cohort”
• John was trying to get some bottles
down from the top shelf. To reach
them he headed to sta… stab, stack,
stagger, stagnate…
STAGE 2: All other sources of
information (frequency, context) may
influence the selection of the target
word from the cohort
This model was designed to STAGE 3: Final word chosen when
account only for auditory word
only a single candidate remains in the
recognition.
cohort.
11
 TRACE Model
(Elman and McClelland, 1984, 1986)
• Features, phonemes, and words
constitute nodes that represent
different levels of processing.
• Each node has a resting level, a
threshold, and an activation level –
signifies the degree to which the
target word is consistent with the
unit in the node.
• When input is appropriate to the
nodes, the activation level of a
node rises towards its threshold.
• E.g. excitation of nodes
representing the phonemes [b]
This model states that processing occurs and [o] would lead to excitation of
through excitatory and inhibitory word nodes containing candidates
connections among numerous processing
whose first sounds are [bo], such
units called nodes.
as boat and bone
12
Cohort and TRACE Models

13
 A Special Problem for the Mental
Lexicon: Lexical Ambiguity
• No strict one-to-one mapping between words and
meaning.

• A single word can have multiple meanings. (bank, right,

drill).

• One meaning of an ambiguous word can be more

common than the others.

• Ambiguous words may foil comprehension because of

their multiple meanings.
14
 A Special Problem for the Mental
Lexicon: Lexical Ambiguity
• However, we seldom encounter words in isolation; they
tend to appear in sentences, and these sentences often
provide a CONTEXT for determining which of several
meanings of a lexically ambiguous word is intended.

• Sentence context usually constrains which interpretation

an ambiguous word receives.

1. The canoeist rowed up to the bank

2. The sergeant ordered the men to drill before they
had recovered from morning exercises.
15
 A Special Problem for the Mental
Lexicon: Lexical Ambiguity
• “The men started to drill before they were ordered to do
so”
– The sentence context does not specify which
interpretation of the word is intended.
– It thus provides subjects with an ambiguity (Simpson
& Burgess, 1988).

• There are 2 theoretical camps about the role of context

in influencing which of several meanings of lexically
ambiguous words are activated.

16
A Special Problem for the Mental
Lexicon: Lexical Ambiguity
– Selective access view
• Context biases the interpretation of an ambiguous
word, so that only intended meaning is accessed
(Glucksberg, Kreuz, & Rho, 1986; Schvaneveldt,
Meyer, & Becker, 1976; Simpson, 1981).

– Exhaustive access view

• Even with context provided in a sentence, multiple
meanings of a lexical ambiguous term are activated
(Onifer & Swinney, 1981; Seidenberg, Tanenhaus,
Leiman, & Bienkowski, 1982; Swinney, 1979;
Tanehaus, Leiman, & Seidenberg, 1979).

17
 Time Course of Activation
• Simpson and Burgess (1985) attempted to test pure
lexical access with a lexical decision task using prime-
target pairs (for example, bank-money, or bank-river)
without any sentence context.

• At 16 millisecond interstimulus interval (ISI) - target

appeared immediately after the prime went off – only
dominant meanings were facilitated.

18
 Time Course of Activation
• The word “money” would be recognized faster when it
appeared after bank because money is related to the
most common meaning of bank.

• However, with a 300 millisecond delay before the target

appeared on the screen, both meanings were equally
activated, and even the subordinate meaning (river) was
primed.

• At longer ISIs, the dominant meaning appears to remain

active while the less-frequent meaning subsides.
Moreover, results showed an inhibition of meanings
associated with the less-frequent meaning (Simpson &
19
Burgess, 1988).
 Time Course of Activation
• It appears that multiple meanings of a word may be
activated in parallel, with the dominant meaning
“popping up” first. Evidence supports that this is an
automatic process not influence by subject intentions.

• Context sometimes may speed access to one of

several meanings, but does not restrict access to all
interpretations of an ambiguous word (Prather &
Swinney, 1998).

20
 The Time Course of Sentence
Contexts vs. Word Pairs
• Tanenhaus, Carlson, and Seidenberg (1984) and
Swinney (1979) found immediate activation of multiple
word meanings after reading a sentence; but only the
contextually specified meaning of an ambiguous word
remained by 200 milliseconds.

• Ambiguity may begin activation of a word unit prior even

to its interpretation, so that access to multiple meanings
is available sooner than in word-pair studies.

21
 Summary
• Overall, the evidence suggests that activation of all
meanings of ambiguous words takes place even in the
face of biasing contexts. (Supports the exhaustive view)

• It thus appears that lexical ambiguity is a dynamic

process, with the various interpretations of a word racing
against each other based on (1) frequency of meaning
and (2) the degree to which context biases one
interpretation over the other.

22
 Summary
• Studies on lexical ambiguity are an effective way to
determine the link between words and meanings. And
findings will clearly have implications for the models of
lexical access.

• Lexical ambiguity provides :

– a window into the complex processes that the human
mind uses to navigate between our conceptual and
linguistic systems.
– A vivid illustration of how language serves both
communication and symbolic functions.

23