Baddeley’s phonological loop paradigm has led to an utter appreciation, as well as to scepticism. The simplicity of his model is indeed very tempting, given the complexity of cognitive science. However, in Baddeley’s case, major controversies have arisen in relation to his model of the phonological loop. One of them is described by Lovatt, Avons and Masterson (2000). The evaluation will therefore base itself on this paper. For this, it is necessary to first give a description of Baddeley’s phonological loop and the importance of the word-length effect.
The second part will focus on the actual analysis of the theory, the experimentation, and will offer a brief discussion. It is generally accepted, within the psychological field and elsewhere, that there is a cognitive function called the working memory (WM). Baddeley (1986) has divided the WM into three components: the central executive and its two ‘slave systems’, the visual-spatial sketchpad (VSS) and the phonological loop (PL). According to Baddeley, the PL consists of a passive phonological store and of an articulatory control process (ACP).
The phonological store holds auditorily presented verbal information and is subjected to time decay. In principle, the ACP’s rehearsal mechanism could maintain the information indefinitely in the phonological store if the time taken to rehearse is kept within the decay time (about 1. 6 to 1. 8s). Another function of the ACP is the transcoding of visual input, after which this latter accesses the phonological store. Consequently, Baddeley claims there is a correlation between speech rate and memory span, the recall for short words would be then greater than for the long words, as short words would be rehearsed more quickly.
This theory is called the word-length effect and represents a cornerstone for the evidence of Baddeley’s phonological loop. Yet the recent findings of Lovatt et al. (2000) demonstrate that there is not any difference in recall between short and long words, thus they question the phonological loop itself. It has to be observed that Baddeley’s tripartite model of working-memory is recurrent in cognitive literature. He is extensively quoted and referenced to, and his studies of the working-memory are innumerable. In addition, the structure of the working memory model he proposes is parsimonious.
This would be very advantageous, taken into account the complexity and abstraction of cognition. However, this over-simplicity could be questioned, especially when the PL is claimed to include such a variety of functions and processes: “the simple model fails to reflect the complexity of ‘real’ cognition (… ). A broader application of the model would require the addition of extra subcomponents or processes, removing the original virtue of simplicity and potentially creating a less coherent model than one that began life as a complex model”( Andrade, 2001).
Baddeley’s model of the working memory has been subjected to constant changes, thus bringing numerous criticisms; or the other way around. The criticisms brought forward by other researchers could be seen as having greatly contributed in the refinement of the model. For instance, what is called today the phonological loop (1990) was at first the phonemic buffer in 1974, then the articulatory loop in 1986. Although one of his colleagues, Andrade (2001), claims “the essence of the model has been retained” (p. 11), it is difficult to follow up with the modifications.
Also, the operationality of Baddeley’s model is not always clear: rehearsal is somewhat obscure as it is only “assumed to be a process resembling covert speech” (Lovatt et al. , 2000. pp. 2). This is an important part of the PL and it is not clear whether it is covert speech (that one could assume to be articulatory) or an unconscious or not fully conscious process that occurs in around 500 milliseconds. Baddeley’s experiment on the word-length effect is relatively simple, and a wide number of researchers have retested his hypothesis.
For instance, some have tested four years old children and supported, to an extend, his theory1. Others have examined cross-linguistics effect2. Nevertheless, if the model enables ‘generativity’, it does not account for a general effect. Indeed, finding adequate words in their length significance is dubious. It is a laborious process that bases itself on many variables such a context, familiarity, phonological similarity etc. There have been various evidences accounted for the phonological loop that irrevocably support it, although one has to be cautious when examining them.
For instance, Cowan, Day, Saults, Keller, Johnson and Flores (1992) have replicated Baddeley’s experiment, using the same word set, which led them to support the word-length effect. Referring back to Lovatt et al. ‘s (2000) paper, it was actually demonstrated that the word-length effect, using disyllabic words, has failed as either long words were better recalled (experiments 1a and 1b) or there were not any significant difference between short and long words (experiments 2a and 2b). Nevertheless, in experiment 3, the word-length effect succeeded when using Baddeley’s word set.
Baddeley accepted he committed “language-as-fixed-effect fallacy” (Andrade, 2001. pp. xvi). This shows that the words used for the serial recall were inadequately related to the context. For instance, where Lovatt et al. have carefully measured the words on the bases of: word duration, standard frequency, number of phonemes, phonological similarity and familiarity. This latter variable does not appear to have been taken into consideration. Words such as ‘zygote3’ could appear awkward.
On the other hand, it could be argued that different people are familiar with different types of words based on various aspects, such as education, careers, regional dialects etc. Thus, the familiarity with the words should have been tested as it has been in Lovatt’s empirical report; thereby reducing this bias. This suggests that the actual meaning of the word is more important than its length. Of course, Baddeley refutes this argument by stating the word duration, rather than its complexity, is significant. It can then be said that the word-length is not a general effect, thus it destabilises Baddeley’s empirical evidence.
Lovatt’s findings tremble the very foundation of the phonological loop by questioning time-based rehearsal and time-based decay. Baddeley’s theory indeed bases itself on only two parameters, a phonological store that decays and an articulatory control process that rehearses information. The word-length effect demonstrates that there is decay and limited speech capacity in the phonological loop, as short words are better recalled than long words in serial recall experiments. Yet, based on Lovatt et al. ‘s experiments, the word-length effect does not exist.
If this is the case, then there is not such a component as the phonological loop. In the foreword of Working Memory in Perspective (Andrade, 2001), Baddeley acknowledged a problem with his model but defended it by writing “the model is not greatly affected by changing from a simple decay to a rather general stimulus-response associative interference theory4. ” This answer demonstrates either that the description given by Baddeley of the phonological loop is too vague; or a ‘simple’ denial, deprived from a constructed argument, of Lovatt’s findings.
In either case, this reply cannot possibly be satisfactory as for an empirical evidence for the PL. Furthermore, it has already been mentioned earlier that cognitive structures and processes can be very complex to understand and to analyse. Any empirical research of such magnitude requires a minimal leap of faith, as it deals with abstract notions and despite being based in controlled conditions, it is still subjected to numerous variables that could have been discarded or believed insignificant5.
The research is indeed subjected to one’s interpretation when ‘translating’ internal processes to external ones. This subjectivity can be extended to the question of controlled conditions, which bases itself on stimuli-response. Harri?? and Secord (1972) have pointed out that “we do not always react mechanistically or deterministically to a given stimulus. Rather, we might think, interpret, attribute meaning or seek further information” (pp. 390). They claim that experiments deny the participants as being characteristically active.
In addition, despite controlling a range of variables, it remains difficult to accept a ‘number’ for the decay time (1. 6-1. 8s. ), as in ‘real’ life conditions, we are exposed to an extensive array of other causal variables that bombard us. Another question could be posed regarding the actual length of the words. Is 530 msec really significant from 693 msec6 to allow such a major effect as a fundamental proof of the phonological loop? 7 As history has over demonstrated it, we all are subjected to the prejudices and materials of our time.
It seems essential not to fall into a dogmatic approach but keeping in mind the wonderful subjectivity of which we consist. It is not the end that imports, but the journey, which is so exciting. Baddeley has proven to be a perfect example of this idea. Furthermore, he has offered us a new dimension of understanding of the working memory. However, Lovatt’s evaluation contributed to the circle we are engaged in by opening another page on Baddeley’s model.