Language Flashcards
Two main models of word production
Spreading activation model: an interactionist model
Computational model – spreading activation consists of four levels
Semantic level – the meaning of words or sentences
Syntactic level - The grammatical structure of words in the planned sentence

Morphological level – the morphemes in the planned sentence
Phonological level – The phones or sounds in words in a sentence
A representation is formed at each level
Processing at each level is in parallel and interactive however normally the higher levels e.g. semantic level is activated earlier than the lower levels e.g. phonological level
The spreading activation model
The spreading activation model can account for the following speech errors – semantic word substitutions – the word that replaces the intended word is in 99% of the cases semantically related.
Phonological word substitutions (or malapropism)
Makes error – both semantic and phonological relation
These errors exhibit both a somatic and phonological relationship. Mixed errors suggest an interactive influence of semantics phonology in lexical selection
Strengths for the spreading activation model
Predict many speech errors that occur in every day language
Mixed error affect – processing can be highly interactive
Notion of spreading activation provides links to other cognitive processes
E.g. word recognitions generative Nature of spoken language may be due to widespread activation between processing levels.
Limitations for the spreading activation model
Single word production
De-emphasises processes related to the semantic level
Not designed to predict the time taken to produce spoken words
Interactive process is more apparent in errors than error – free speech
Generally predicts too many errors in speech.
Weaver ++ a top down model
Proposed by Levelt and his colleagues and based on levels (1989)
The model focuses on the production of individual spoken words and not in the production of sentences
Main characteristics of the model
– It’s a computational model
- It’s a feedforward activation spreading network
There are three main levels
- Nodes representing lexical concepts
- Nodes representing Lammas or abstract words from the mental lexicon
- Nodes representing morphemes and phonemes
Processing is cereal, i.e. one stage proceeds the following one
self monitoring avoids speech errors
Some neurological evidence, sequence of processes for picture naming from Indefrey (2011).
Strengths of the Weaver a top down model
Indefrey and Levelt’ (2004) meta analysis supports notion that word production moves from.
Lexical selection
Morphological encoding
Phonological encoding
Shift the focus away from errors and towards precise timing of word production processes
Simple and elegant model that makes testable predictions.
Limitations of the weaver and a top down model
Emphasis on single word production
There is likely much more interaction between processing stages than assumed
Fails to explain speech errors that seem to suggest parallel processing
Components of language production
We have mentally represented
The phonological or sound - part of words
Morphology (morpheme’s).
The semantic or meaning of words
The syntactic category of the word and syntactic rules.
Facts about Language
The estimate vocabulary (lexicon) of a typical educated person is between 50-100 thousand words.
By the age of 21 a typical person has spoken an average of 50 million tokens (tokens is every time you produce a new word but it can be another token for the plural meanings)
We are able to produce an average of 120-150 words per minute. (In pressure we can double it)
The error rate of language production is about 1 error per 1000 words produced.
Aging and language production
Applied issues
Burke & Shafto (2004) the ability to produce familiar words declines with aging.
Semantic processing is as good for older adults than younger adults
However, older adults experience more tip of tongue states compared with younger adults
Burke & Shafto (2004) propose that the representational system for work phonology is vulnerable to transmission deficits, impairing retrieval.
Anomia (without name)
Patients find it difficult finding the right word for an object or action., but are good at recognition.
Damage to the temporal pole causes anomia for common nouns
Damage to the frontal cortex, in and around Broca’s area causes anomia for verbs
Speech errors
1 error per 1000 words
1 speech error every 7 minutes
1 speech error every 3 minutes for fast speech
Discourse markers
( ok, actually, yeah, I mean)
These markers do not contribute to the context of speech
Different functions such as change of topic, or to show politeness
We use different discourse markers in writing then speaking
Written language - although, moreover
Spoken language - anyway, well, oh, I’m
Prosody
The particular rhythm, tempo, melody, pattern used when speaking a language
When we speak we also use parsodic cues which include rhythm and stress
