CARACTERÍSTICAS DE LOS PACIENTES Y DE LAS ESTRATEGIAS

It was suggested in Section 8.2 that an inefficacious memory code would explain the Auslan, SE, and Oral subjects' decreased reliance upon communication-based rehearsal. There are three reasons for postulating reduced effectiveness in the codes used by these deaf subjects.

First, when compared to hearing subjects, all the deaf groups rehearsed relatively slowly, despite attempts to overcome this by abbreviation. This indicates less efficacious coding, because fewer items could be rehearsed (hence recalled) before the onset o f decay. In this respect, the signers' code was less efficacious than the Oral group's code, because signed items usually took longer to rehearse. The features o f Auslan which contribute to its slower rehearsal rate are discussed in Section 8.3.1.

Secondly, the Oral group deployed a less discriminable trace which lacked the richness provided for hearing subjects by the auditory component. Therefore, as proposed by Conrad (1972a) and others, the elements o f this code may have been insufficiently distinctive for deaf phonological coders to recover the original response, or may have decayed more quickly to the point where recognition was not feasible. Section 8.3.2 discusses this issue.

Finally, signers may have been disadvantaged by a code which was more difficult to sequence, thus leaving fewer resources available for storage purposes. This idea is developed in Section 8.3.3 and supported by a reanalysis o f the data in Section 8.4.

8.3.1 Sequential rehearsal in sign and speech

The demands o f serial recall may prevent sign from being used in its most efficacious way. It will be argued that various features o f sign language (but not speech) are impossible to incorporate during rehearsal o f sequences. This in turn increases sign length, thereby reducing the effectiveness o f sign coding relative to speech coding in short-term memory tasks.

Naveh-Benjamin and Ayres (1986) propose: "there is some universal (biologically determined) optimum rate o f information flow towards which all languages naturally evolve in order to suit the human capacity for message transmission and reception" (p. 223). These researchers reported data which registered a trade-off between syllable number and syllable speed: the greater the number o f syllables per word in a particular language, the smaller the average time taken to articulate that syllable. Similarly, Mayberry and W aters (1991) discerned no difference in the fingerspelling rate o f 3-, 4-, and 5-letter words. Additional temporal parities between sign and speech have been noted in investigations o f signing rate, o f pauses, and o f information conveyed per unit o f time (Bellugi, 1980; Bellugi & Fischer, 1972; Grosjean, 1979; Klima & Bellugi, 1979).

How are these comparable production rates achieved, given that articulation o f single items with the hand is slower than with the mouth?

First, at the individual sign level, the meaning o f a sign can be modulated by simultaneously imposing changes in its parameters (Klima & Bellugi, 1976), rather than by the addition-o f affixes or modifiers, the strictly sequential strategies which occur in English. For example, in Auslan the difference in emotional intensity between feeling merely "angered" as opposed to "very angered" is conveyed in the latter case by using two hands in place o f one.

Secondly, sign language can switch to a time-saving mode, utilising various grammatical forms which rely on the simultaneous interrelationships between signs to reduce the number o f signs required (Newport & Meier, 1986). Usually, ideas are expressed by using the minimal number o f words necessary. Some English prepositions and articles are not used in sign language, or they may be incorporated into "classifiers" o f one kind or another and/or expressed via movement and/or justaposition o f signs. The upshot is that signed and spoken propositions are expressed at roughly comparable rates (Bellugi & Fischer, 1972).

The reliance on simultaneous elements in sign, and on sequential elements in speech, may hold the key to signers' poor serial recall. In normal discourse, the simultaneous elements o f Auslan reduce the duration o f signed propositions to the point where they are comparable to spoken ones, thereby assisting serial processing o f Auslan (we know this because signers' serial recall is sensitive to sign length). In contrast, the imposition o f serial recall - a task which is typically constructed to minimise semantic coding - reduces cherological efficacy by divesting sign o f its customary mode o f simultaneously processing sign clusters, thereby putting an artificial barrier to the achievement o f a comparable processing rate to speech. This transpires even though attempts may be made at the single-item level to shorten response time wherever possible, which effectively translates into an abbreviation o f signs at long lengths only. Since speech is organised serially anyway, its length is unaffected by the requirement to

remember the order o f items. The conclusion follows that individual signs are less suited than individual words to serial recall because they take longer to execute - which is totally compatible with the rehearsal speed hypothesis.

8.3.2 O ral subjects: An indistinct speech code?

An indistinct memory trace which reduces the efficiency o f phonological coding might explain the smaller measured span o f the Oral deaf groups. This receives support from two observations:

1. The deleterious effects on memory o f indistinct stimuli, irrespective o f hearing status;

2. The reduced magnitude o f phonological processing effects in the Oral deaf population.

The phonological code o f orally educated deaf people may be relatively inefficient because it is based on speechreading, which only permits coarser distinctions o f the items in memory. There are fewer "visemes", or visually-distinct speech units, than there are phonemes, their auditory counterpart. It follows that the elements o f the phonological code o f deaf people, lacking the extra discriminability imparted by an auditory component, are less able to be differentiated than are the units o f the speech-based code o f hearing people. Being a less discriminable code, it is less efficient, thus leads to poorer recall.

This interpretation is consistent with the results o f Luce et al. (1983), who reported relatively poor recall o f synthetic speech in hearing adults, presumably because the extra resources required for identification left fewer available for storage. These results bear on the performance o f Oral deaf subjects with auditory (effectively, lipread) presentation, where the difficulties o f speechreading would be expected to disadvantage recall because o f a processing-storage trade-off.

It is also relevant that memory spans for printed rhyming stimuli are reduced in hearing subjects. In these instances, since perceptual (orthographic) similarity can be controlled, rhyming effects can be attributed to a less discriminable memory representation (Baddeley, 1966, 1986; Conrad, 1971a; Schweickert et al., 1990).

Such manipulations o f phonological similarity have been extended by various researchers to include deaf subjects. Typically, smaller differences are found in this group between the memory spans for rhyming and non-rhyming stimuli (Campbell & Wright, 1989; Hanson, 1982a; Hanson, Liberman, & Shankweiler, 1984). This trend is also apparent in this study, as discussed in Section 8.1. There are two ways in which this could be interpreted as evidence for an indistinct code. Either the deaf subjects relied less than hearing subjects on phonological coding throughout the experiment (because it was too indistinct to be useful). Alternatively, deaf and hearing subjects may rely to the same extent on a phonological code, but the deaf subjects were less influenced by the rhyme manipulation because their indistinct code contributed less effectively to memory span.

8.3.3 Sign: A code which is difficult to sequence?

As discussed in the literature review, deaf speech coders are characterised by better ordered recall than cherological coders (Krakow & Hanson, 1985). Additionally, the performance o f deaf and hearing bilinguals reveals that although sign may be used widely in other tasks, speech is the preferred medium for serial recall (Guttentag & Schaefer, 1987; Hanson, 1982a, 1990; Lichtenstein, 1985). These findings suggest that sign may be less suited than speech to serial recall. One reason for this, as discussed in Section 8.3.1, may be that signs take longer to execute. A second possibility is that sign is harder to sequence than speech.

Mayberry and Waters (1991) provide data suggesting that sequencing in sign is especially difficult for young children, and that, unlike speech, a major change concerns the rate at which they can repeat sign sequences rather than single signs. That a similar developmental progression is not seen in spoken sequences (Hulme et al., 1984) implies that ordering speech is not particularly difficult, even for children.

Each sign must be simple enough to be used easily, but sufficiently complex to be distinctive. It may be the complexity and length o f signs which renders them difficult to sequence. Historical changes show a reduction o f embellishments over time (Frishberg, 1979; Kyle & Woll, 1985) with ease o f articulation and a reduction in duration suggested as simplifying forces (Rimor, Kegl, Lane, & Schermer, 1984). For example, the Auslan sign RETURN is a blend o f GO and COME - where the movement o f GO has been constricted and its handshape has actually vanished.

In this context it is worth noting that Auslan, ASL, BSL and other visuospatial languages all incorporate a body o f signs that are specialised for serial recall. These are the letters o f the alphabet, fingerspelled in rapid succession to convey English words when single signs are unavailable or not the first preference. In this case, location no longer serves as a distinguishing feature because all letters are signed in neutral space, and few incorporate movement as an integral part. Thus apart from hand configuration, fewer parameters are required to distinguish letters from each other than is the case for signs-as-words. This is true o f the two handed alphabet used in Auslan and BSL, and applies equally to the ASL alphabet, which uses just one hand. In this sense, the relative simplicity o f fingerspelled letters, plus that fact that all have short durations, may act to facilitate their serial processing.