lexical and morphosyntactic minimal pairs. evidence for different processing luca cilibrasi, vesna...
TRANSCRIPT
Lexical and morphosyntactic minimal pairs. Evidence for different processing
Luca Cilibrasi, Vesna Stojanovik, Patricia Riddell,
School of Psychology, University of Reading
Minimal pairs
Minimal pairs are defined as pairs of words in a particular language which differ in only one phonological element and have a different meaning (Roach, 2000)
Pen – Pan Hat – Had Teach – Reach
Minimal pairs in language remediation
Minimal pairs are widely used in the treatment of phonological disorders, such as Dyslexia (Brancalioni et al, 2012, Earl, 2011, Pulvermuller et al, 2001, Crosbie et al, 2005, Blache et al, 1981, Barlow & Gierut, 2002).
However, several studies report problems, i.e. no generalisation to untreated items (Sabem & Ingham, 1991, Williams, 2000).
Representations involved
We believe there is a gap in our comprehension of how minimal pairs discrimination takes place
Lexical representations
Mono VS polymorphemic real words discrimination
How are polymorphemic words stored in the lexicon?
Plays :
[Play + s] [Plays]
Competing approaches
Inflected forms are stored in the lexicon as units (Stemberger & MacWhinney; 1986, Bertram et al, 2000)
During language acquisition, the first strategy is to store inflected forms in the lexicon as units (Tomasello, 2006; Diessel, 2012)
Irregular forms are stored in the lexicon, regular forms are derived using a rule (Pinker & Ullman, 2001)
How is this linked to minimal pairs?
In many languages, bound morphemes, which are used to mark inflection, generate minimal pairs. These sets are referred to as “morphosyntactic minimal pairs” (Law & Strange, 2010)
Mangio – mangia – mangi
Plays – played
Null hypothesis
If inflected forms are stored as units in the lexicon, discriminating lexical minimal pairs and morphosyntactic minimal pairs should not be different processes
Badge – Back
Asks – Asked
Two monosyllabic forms differing in the final phoneme
Participants
20 monolingual native speakers of English were recruited trough wall advertising in the department of Clinical Language Science, University of Reading
Graduate students, 9 males, 11 females, mean age 25.5, standard deviation, 2.03.
Stimuli
30 monosyllabic lexical minimal pairs 30 monosyllabic morphosyntactic minimal
pairs 60 pairs of identical words (30 from the first
condition, 30 from the second condition)
Task: Two words appear on the screen Participants are instructed to press white if the
two words are identical, black if they are different
Method
Task programmed using E-prime
Measures of accuracy (number of items coded correctly)
Measures of Reaction Times (msec)
Results
Accuracy is at ceiling for all subjects in all conditions and therefore will not be considered further.
RTs mean are compared in the Lexical vs. morphosyntacic condition.
Only correct responses are taken. Trimmed means (95% of the range of reaction
times recorded for each individual) are used.
Results
t (19) = -4.486, p < .001 The difference is highly significant. Morphosyntactic
minimal pairs take more time than lexical minimal pairs to be distinguished
Null hypothesis
If inflected forms are stored as units in the lexicon, discriminating lexical minimal pairs and morphosyntactic minimal pairs should take the same time
The null hypothesis is rejected
Discussion
This suggests that lexical and morphosyntactic minimal pairs might require two different forms of processing
This could be because morphosyntactic minimal pairs require decomposition in stem + affix in order to be analysed (Pinker & Ullman, 2001)
Our result is not consistent with the hypothesis that inflected forms are stored as units (Bertram et al, 2000)
Alternative explanations
Elements within morphosyntactic minimal pairs are semantically linked, elements within lexical minimal pairs are not.
The distinction between two elements in morphosyntactic minimal pairs is semantically subtle, even if the elements belonging to the pair are stored in the lexicon as units with the bound morpheme
Alternative explanations
Morphosyntactic minimal pairs differ always on the same phonemes /s/ - /z/ VS /d/ - /t/ while the possible distinctions in the lexical condition are more varied.
Thus, the morphosyntactic condition is more predictable so it should be easier but it is NOT
Alternative explanations
Verbs are slower than nouns
There is evidence of dissociation but is there evidence that verbs are slower?
Conclusion
We report evidence that discriminating elements in morphosyntactic minimal pairs takes longer than discriminating elements in lexical minimal pairs.
Our tentative conclusion is that morphosyntactic minimal pairs are decomposed in order to be processed syntactically, while monomorphemic words do not require this.
Future work (in progress) Can we operate minimal pairs discrimination
without appealing to the lexicon?
Odd ball paradigm RTs + ERPs
We expect RTs to correlate with phonological short term memory
MMN not to vary in latency in the four conditions
In progress
Presented aurally:
Side VS size Bud VS buzz Cared VS cares Chewed VS chews
Past work
In a previous work we showed that we can predict reading performance using accuracy and RTs in non-words minimal pairs discrimination
The general picture
Minimal pairs discrimination can take place at the sub-lexical level and this is the level we have to focus on in order to improve reading performance. However, the lexical level tends to be involved when we use real words, as is demonstrated by the fact that polymorphemic words require more time than monomorphemic words.
Acknowledgements
My supervisors and co-authors of the presentation, Vesna Stojanovik and Patricia Riddell
My monitors, Doug Saddy and Theo Marinis The faculty of social sciences of the University
of Reading for funding the project