The terminology used to describe speech problems is rooted in classificatory systems derived from different academic disciplines. In order to understand the rationale behind the psycholinguistic approach, it is helpful to examine other approaches and compare how speech problems have been classified from different perspectives. Three perspectives that have been particularly influential are the medical, linguistic, and psycholinguistic perspectives.

In a medical perspective, speech and language problems are classified according to clinical entity. Commonly used labels include dyspraxia, dysarthria, and stuttering. Causes of speech difficulties can be identified (e.g., cleft palate, hearing loss, neurological impairment) or an associated medical condition is known (e.g., autism, learning difficulties, Down syndrome).

Viewing speech and language disorders from a medical perspective can be helpful in various ways. First, through the medical exercise of constructing a differential diagnosis, a condition may be defined when symptoms commonly associated with that condition are identified; two examples are dyspraxia and dysarthria. Second, for some conditions, medical management can contribute significantly to the prevention or remediation of the speech or language difficulty, such as by insertion of a cochlear implant to remediate hearing loss or by surgical repair of a cleft palate. Third, the medical perspective may be helpful when considering the prognosis for a child's speech and language development, such as when a progressive neurological condition is present.

However, the medical approach has major limitations as a basis for the principled remediation of speech problems in individual children. A medical diagnosis cannot always be made. More often the term “specific speech and/or language impairment” is used once all other possible medical labels have been ruled out. Moreover, even if a neuroanatomical correlate or genetic basis for a speech and language impairment can be identified, the medical diagnosis does not predict with any precision the speech and language difficulties that an individual child will experience, so the diagnosis will not significantly affect the details of a day-to-day intervention program. To plan appropriate therapy, the medical model needs to be supplemented by a linguistic approach.

The linguistic perspective is primarily concerned with the description of language behavior at different levels of analysis. If a child is said to have a phonetic or articulatory difficulty, the implication is that the child has problems with the production of speech sounds. A phonological difficulty refers to inability to use sounds contrastively to convey meaning. For example, a child may use [t] for [s] at the beginning of words, even though the child can produce a [s] sound in isolation perfectly well. Thus, the child fails to distinguish between target words (e.g., “sea” versus “tea”) and is likely to be misunderstood by the listener. The cause of this difficulty may not be obvious.

The linguistic sciences have provided an indispensable foundation for the assessment of speech and language difficulties (Ingram, 1976; Grunwell, 1987). However, this assessment is still a description and not an explanation of the disorder. Specifically, a linguistic analysis focuses on the child's speech output but does not take account of underlying cognitive processes. For this, a psycholinguistic approach is needed.

The psycholinguistic approach attempts to make good some of the shortcomings of the other approaches by viewing children's speech problems as being derived from a breakdown in an underlying speech processing system. This assumes that the child receives information of different kinds (auditory, visual) about an utterance, remembers it, and stores it in a variety of lexical representations (a means for keeping information about words, which may be semantic, grammatical, phonological, motor, or orthographic) within the lexicon (a store of words), then selects and produces spoken and written words. Figure 1 illustrates the basic essentials of a psycholinguistic model of speech processing. On the left there is a channel for the input of information via the ear and on the right a channel for the output of information through the mouth. The lexical representations at the top of the model store previously processed information. In psycholinguistic terms, top-down processing refers to an activity whereby previously stored information (i.e., in the lexical representations) is helpful and used, for example, in naming objects in pictures. A bottom-up processing activity requires no such prior knowledge and can be completed without accessing stored linguistic knowledge from the lexical representations; an example is repeating sounds.

Figure 1..  

The essentials of a psycholinguistic model of speech processing. (From Stackhouse, J., and Wells, B. [1997]. Children's speech and literacy difficulties 1: A psycholinguistic framework. London: Whurr. Reproduced with permission.)

A number of models have been developed from this basic structure (e.g., Dodd, 1995; Stackhouse and Wells, 1997; Hewlett, Gibbon, and Cohen-McKenzie, 1998; Chiat, 2000). Although these models differ in their presentation, they share the premise that children's speech difficulties arise from one or more points in a faulty speech processing system. The aim of the psycholinguistic approach is to find out exactly where a child's speech processing skills are breaking down and how these deficits might be compensated for by coexisting strengths. The investigative procedure to do so entails generating hypotheses, normally from linguistic data, about the levels of breakdown that give rise to disordered speech output. These hypotheses are then tested systematically through carefully constructed tasks that provide sufficient data to assemble a child's profile of speech processing strengths and weaknesses (Stackhouse and Wells, 1997; Chiat, 2000).

Collation of these profiles shows that some children with speech difficulties have problems only on the output side of the model. However, many children with persisting speech problems have pervasive speech processing difficulties (in input, output, and lexical representations) that impede progress. For example, when rehearsing new words for speech or spelling, it is usual to repeat them verbally. An inconsistent or distorted output, normally the result of more than one level of breakdown, may in turn affect auditory processing skills, memory, and the developing lexicon. It is therefore not surprising that children with dyspraxic speech difficulties often have associated input (Bridgeman and Snowling, 1988) and spelling difficulties (Clarke-Klein and Hodson, 1995; McCormick, 1995).

The case study research of children with developmental speech disorders, typical of the psycholinguistic approach, has shown that not only are children unintelligible for different reasons but also that different facets of unintelligibility in an individual child can be related to different underlying processing deficits (Chiat, 1983, 1989; Stackhouse and Wells, 1993). Extending the psycholinguistic approach to word finding difficulties, Constable (2001) has discovered that such difficulties are long-term consequences of underlying speech processing problems that affect how the lexical representations are stored, and in particular how the phonological, semantic, and motor representations are interconnected.

The psycholinguistic approach has also been used to investigate the relationship between spoken and written language and to predict which children may have long-term difficulties (Dodd, 1995; Stackhouse, 2001). Those children who fail to progress to a level of consistent speech output, age-appropriate phonological awareness, and letter knowledge skills are at risk for literacy problems, particularly when spelling. Psycholinguistic analysis of popular phonological awareness tasks (e.g., rhyme, syllable/sound segmentation and completion, blending, spoonerisms) has shown that the development of phonological awareness skills depends on an intact speech processing system (Stackhouse and Wells, 1997). Thus, children with speech difficulties are disadvantaged in school, since developing phonological awareness is a necessary stage in dealing with alphabetic scripts such as English. Further, these phonological awareness skills are needed not just for an isolated activity, such as a rhyme game, but also to participate in the interactions typical of phonological intervention sessions delivered by a teacher or clinician (Stackhouse et al., 2002).

An individual child's psycholinguistic profile of speech processing skills provides an important basis for planning a targeted remediation program (Stackhouse and Wells, 2001). There is no prescription for delivering this program, nor is there a bag of special activities. All intervention materials have the potential to be used in a psycholinguistic way if analyzed appropriately. Principled intervention is based on setting clear aims (Rees, 2001a). Tasks are chosen or designed for their psycholinguistic properties and manipulated to ensure appropriate targeting and monitoring of intervention. To this end, each task is analyzed into its components, as follows:

Rees (2001b) presents seven questions for examining these four components. She demonstrates how altering any one of them can change the nature of the task and thus the psycholinguistic demands made on the child.

In summary, a psycholinguistic approach to intervention puts the emphasis first on the rationale behind the design and selection of tasks for a particular child, and then on the order in which the tasks are to be presented to a child so that strengths are exploited and weaknesses supported (Vance, 1997; Corrin, 2001a, 2001b; Waters, 2001). The approach tackles the issues of what to do, with whom, why, when, and how.

In a review of psycholinguistic models of speech development, Baker et al. (2001) present both box-and-arrow and connectionist models as new ways of conceptualizing speech impairment in children. This discussion has focused on the former, since to date, box-and-arrow models have arguably had the most impact on clinical practice by adding to our repertoire of assessment and treatment approaches, and also by promoting communication and collaboration between teachers and clinicians (Popple and Wellington, 2001). The success of the psycholinguistic approach may lie in the fact that it targets the underlying sources of difficulties rather than the symptoms alone (Holm and Dodd, 1999). Although it is true that the outcome of intervention depends on more than a child's speech processing profile (Goldstein and Geirut, 1998), the development of targeted therapy through the setting of realistic aims and quantifiable objectives should make a contribution to the measurement of the efficacy of intervention.