Chapter 3 Short-term memory coding by deaf people
3.8. What subject characteristics relate to the use of a speech-based code?
Some deaf people can use a speech-based STM code. As a speech-based code is often regarded as fundamental to cognitive development of language abilities, much emphasis has been placed on determining the factors that relate to the efficient use of such a code. Conrad (1979, section 3.3.iii) identified the following factors as related to the use of a speech-based code:
Age at onset o f deafness - The percentage of congenitally, prelingually and postlingually deaf children classed by Conrad as using a speech-based code was 47%, 46% and 93% respectively. Although, early auditory speech exposure is important to the development of a speech-based code, these data clearly demonstrate that even people who are bom deaf or become deaf at an early age often have access to such a code.
Hearing loss - There was a negative relationship between the use of a speech-based code and hearing loss, such that children with higher levels of hearing loss were less likely to use a speech-based code. Furthermore, the average inner speech ratio (IS-R) of those classed as using a speech-based code fell as deafness increased, indicating that the use of a speech-based code is not ‘all or nothing’. Rather, at higher levels of hearing loss a speech-based code may be less well specified or strategic use of this code may be less efficient than when there is some residual hearing.
Non-verbal IQ - This was measured using Raven’s Progressive Matrices. Use of a speech-based code increased as NVIQ increased and was independent o f hearing loss.
Speech intelligibility - Hearing loss and NVIQ are both related to speech intelligibility (Musselman, Lindsay & Wilson, 1988). Conrad showed that even when these factors had been controlled for there was a significant relationship between speech intelligibility and the use of a speech-based code. However, some researchers have not replicated Conrad’s findings (e.g., Dodd, Hobson, Brasher & Campbell, 1983; Hanson & Fowler, 1987). A possible reason for this discrepancy is that there is no agreement on how speech intelligibility should be measured.
Reading ability - Conrad found that the use of a speech-based code was related to reading ability. Many other studies have also found this relationship (e.g., Hanson, Liberman & Shankweiler, 1984). However, establishing the direction of causality in the relationship between these two variables is problematic. This will be discussed in more detail later in this chapter.
Educational background - As the majority of Conrad’s subjects attended oral schools, he was unable to compare their performance with that of deaf children attending TC schools. Although Conrad showed that over half of congenitally or prelingually deaf oral deaf children did not use a speech-based code, it is tempting to hypothesise that
‘oral’ children are more likely to establish a speech-based code than children educated in TC environments. In one of the earliest reported studies of STM of deaf people, Pintner and Patterson (1917) found that digit span of oral deaf subjects was greater than that of ‘manual’ subjects. Although educational practices today differ vastly», to those of Pintner and Patterson’s time, Wallace and Corballis (1973) also found differences between oral and TC deaf subjects. They showed that both groups of children relied heavily on a visual code for the retention of letters. However, when list length was increased to exceed span, oral subjects resorted to a speech-based code to enhance performance whereas TC subjects resorted to a fmgerspelling code.
In contrast, Lichtenstein (1985; see also Lichtenstein, 1998) found no difference in the use of a speech-based code by deaf college students who had attended oral or TC
schools. However, this college-educated sample is not representative of the deaf population at large. Their higher levels of literacy may have enhanced their use of a speech-based STM code. Therefore, the evidence regarding the influence of linguistic background on the use of a speech-based code is inconclusive. This is investigated further in Experiment 2 by comparing STM performance by deaf pupils attending oral and TC schools who were of normal attainment within the deaf schooling system.
In summary, there is a range of subject characteristics that relate to the use of a speech-based code by deaf people such as age at onset of deafness, degree of hearing loss, NVIQ, speech intelligibility and perhaps educational background. Despite awareness of these factors, why some deaf children are more able to use such a code than others is still not clear. Dodd and Murphy (1992) highlight this in the case studies of two very similar deaf girls:
“ They are.... a closely matched pair: same sex, age, cause, and degree of hearing impairment, non-verbal intellect, socioeconomic group, family support (both mothers became fluent in sign English), educational experience and provision of speech therapy” (1992, pg. 48).
Performance by these girls on a rhyme task showed that one used a speech-based code to a far greater extent than the other. Dodd and Murphy propose that ‘individual choice’ may be the factor that distinguishes the two girls. Unfortunately, individual choice is not something that can be controlled for experimentally.
3.9. The relationship between the use of a speech-based STM code and higher