Ü Spoken Duration in STM Performance

Findings o f a significant effect o f word length on recall gave rise to the suggestion that the capacity o f the phonological loop is limited (Baddeley et al., 1975). Word length effects have been found in STM recall for spoken words in young children (Hitch & Halliday, 1983, Ford & Silber,

1994, Gathercole & Adams, 1994, Hulme et al., 1984).

Word length effects have been linked with spoken duration o f material and speech rate. The view is that longer words take longer to say, take longer to rehearse, and therefore, fewer o f them can be maintained within the phonological loop. Baddeley et al. (1975) argue that the word length effect is due to the articulatory duration o f words, and not just to the number o f syllables within the words. They carried out a study in which recall o f two syllable words, with short durations, e.g. ‘wicket’ and ‘bishop’, was compared to recall o f two syllable words, matched for number o f phonemes, but with longer durations, e.g. ‘harpoon’ and ‘Friday’.

Significantly more o f the shorter words were recalled suggesting that the phonological loop is sensitive to the articulatory length o f the stimuli, rather than the number o f syllables. This finding was confirmed by Cowan et al. (1992) who also compared recall o f words with shorter or longer articulatory durations, matched for number o f phonemes and number o f syllables and found that more o f the shorter words were recalled.

However, Caplan, Rochon and Waters (1992) and Caplan and Waters (1994) did not find the expected word length effects when comparing span for lists o f two syllable words that were matched for number o f phonemes and syllables, but differed for articulatory duration. The spoken duration of the words was manipulated by choosing words with short vowels, e.g.

‘picnic’ or words with at least one long vowel in them, e.g. ‘balloon’. In fact there was significantly better recall for the words o f longer duration, than for the shorter words. A further experiment (Caplan et al., 1992) matched sets o f stimuli for the number o f phonemes and syllables, but differed in articulatory complexity, using CCV words like ‘crow’ versus CVC words like ‘hat’. There was a significant difference in articulatory duration, with the CCV words taking longer to articulate than the CVC words. But there was no significant difference in recall between the difficult-to-say (i.e. CCV) and the easy-to-say (CVC) words. These two findings led Caplan et al. (1992) to conclude that word length effects were determined by phonological structure and not by features o f articulatory length, and that they arise at the level o f speech output processing with the increased complexity o f planning motor speech gestures.

This issue o f articulatory complexity is another hypothesis to consider in interpreting word length effects longer words will require more complex motor planning, because they will contain more phonemes and syllables, and will be more likely to contain consonant clusters. Effects o f articulatory complexity have been found in NW R in children, with non­ words with clusters being harder to repeat accurately than non-words without clusters (Gathercole & Baddeley, 1989, 1990b, Bishop et al.,

A further interpretation o f word length effects on recall is that it is the amount o f phonological material to be recalled that is significant, rather than the time taken to say it, or its articulatory complexity. On this view the phonological loop might be considered to have a capacity for number o f units (i.e. phonemes to be recalled) rather than a time- based capacity.

Lovatt et al. (2000) matched two sets o f disyllabic words for number o f phonemes, but different spoken durations. This was measured by recording and timing a number o f different subjects reading the words aloud. The words were recalled verbally or by picture pointing. A number o f different experiments, with adults, resulted in the finding that significantly more o f the words with a longer spoken duration were recalled than words with shorter durations. This is the opposite to what would be predicted by the traditional interpretation o f word length effects reflecting effects o f spoken durations. Although why the words with shorter durations were more difficult to recall is not clear, given that they contained the same number o f phonemes.

Another approach to this issue was taken by Service (1998) who contrasted recall o f three sets o f stimuli. Non-words based on Finnish phonology were created. A non-word (e.g. /tepa/) was matched with: i) a non-word

that contained the same phonemes, but with some sounds lengthened e.g. / te:p;a/; ii) a non-word that contained more phonemes, e.g. /tiem pa/. There

was no significant difference in recalling a list o f non-words with lengthened phonemes, i.e. o f longer spoken duration, rather than non- words with shorter phonemes. There was a significant effect on span in recalling the words with more phonemes, i.e. the more phonologically complex words.

A number o f studies have now challenged the traditional view that word length effects are the result o f spoken durations. It may be that at least some o f the word length effect is related to the amount o f phonological material to be recalled, rather than how long it takes to say the material. As discussed in Chapter 2, Cowan et al. (1992) and Henry (1991b) suggest that word length effects may arise at spoken recall. It would be interesting to compare spoken and picture-pointing recall o f stimuli containing different amounts o f phonological material to see if there is more effect o f phonological complexity on spoken recall than on non-spoken recall.

5.3 RATIONALE FOR THIS EXPERIMENT

The experiment presented here was designed to compare performance on STM tasks using different combinations o f input and output mechanisms. This would allow the different speech processing requirements o f these STM tasks to be considered. Sets o f stimuli were used that were matched for spoken duration but varied for number o f phonemes in the word. This was to allow a different interpretation o f the word length effect to be evaluated, i.e. that it arises from the amount o f phonological material to be recalled, rather than from spoken duration.

5.3.! Predictions

The 3 STM tasks have different speech processing requirements. It is hypothesised that speech processing requirements will affect STM recall, so performance will differ on the STM tasks that involve different input and response mechanisms.

STM capacity may be affected by the amount o f phonological information to be recalled, rather than by the spoken duration o f the stimuli. It is hypothesised that phonologically simple word lists will be easier to recall than phonologically complex word lists that are matched for spoken duration.

It has been suggested that word length effects might arise at the level o f spoken recall rather than within the sub-vocal rehearsal process. If the phonological complexity o f the stimuli is responsible for the word length effect, and the word length effect arises at spoken output, then there will be a greater effect o f complexity on spoken recall than on picture pointing recall.

5.4 METHOD

5.4.: Experimental Design

This experiment investigated within-subject differences in performance on 3 different STM tasks, with 2 conditions that differed in degree o f phonological complexity.

5.4.Ü Participants

Twenty-two children aged between 4;06 and 5;00 years at the start o f the study were recruited from 3 main-stream primary schools. Parents signed a consent form to allow their child to participate. Criteria for selection were: i) no history o f speech and language difficulties or o f significant hearing loss

ii) considered by their teacher to be within the average range, with no evidence o f any general or specific learning difficulty

iv) no significant medical or neurological condition

The use o f a range o f speech and language measures in the whole test battery used (see Chapter 6) allowed for children to be screened for speech and language difficulties. This resulted in two children being rejected from this participant group.

The participants were first tested when aged between 4;06 and 5;00 years (T l). They were tested again on a similar test battery approximately one year later, when aged between 5;06 and 6;00 years (T2).

5.4.iii Task Materials and Procedures