1.3. MODELS OF READING ACQUISITION.
1.3.1. Stage and phase models
A family of models that gained particularly wide currency presents reading acquisition as a succession of stages or phases, with each phase defined by a qualitatively different processing strategy (e.g. Marsh, Friedman, Welch & Desberg, 1981; Frith, 1985; Ehri, 1995). Models of this type vary considerably in terms of the number of phases proposed, their labelling and description (see table 1-1). However, in a majority of cases, this variability can be reduced into the same three broad developmental periods. The initial one is characterised by a predominantly global, instance-based processing of words. Words are identified mainly via unsystematic use of visual cues (such as salient letter parts), with decoding skills playing a secondary (if any) role. Some authors (e.g. Ehri, 1995) maintain that skills developed during this initial period still do not constitute ‘proper’ reading insofar as no appreciation of the alphabetic principle is evident. Insight into the nature of alphabetic orthography is the main achievement o f the second period, where children acquire basic decoding skills, focusing on basic, ‘small unit’ letter-to- sound correspondences. The following third period involves the mastery of orthography-specific knowledge, characterised by complex, conditional rules and corresponding to the morphological level of language. This is usually described as the end point of development, as far as word recognition (and spelling) are concerned. A few authors (e.g. Chall, 1996) proposed further stages characterising the development of reading comprehension.
Marsh et. al. (1981)
Frith (1985)
Kirby (1990)
Ehri (1995) Bakker (1990) Chall (1996)
prereading g la n ce and ^uess logograp h ic global pre-alphabetic elem entary reading
(right hem isphere, perceptually-based) sop h isticated g u e ssin g partial alphabetic a cq u isition o f sim p le graphem e -p h on em e corresp on d en ces
alph abetic analytic fu lly alphabetic initial read ing or d eco d in g stage
the sk illed reader orthographic synthetic con solid ated alphabetic
advanced reading (left hem isph ere, lin g u istica lly based)
c o n firm ation , flu en cy , u n g lu in g from print
m eta- com p re hension
reading for learning the n ew
m u ltip le v ie w p o in ts con stru ction and reconstruction - a word v iew
Table 1-1. Schematic comparison o f some influential stage and phase m odels o f literacy acquisition (based on Krasowicz-Kupis, 1999, p. 81).
Stage and phase models are attractive through their descriptive parsimony. A simple set of underlying notions (stages, strategies) encompasses both quantitative and qualitative developmental changes. Stage models are particularly valuable in the applied context of education, as they take into consideration overt reading behaviour (such as sounding out) and allow for easy classification of learners as ‘falling within a certain stage’. However, they were also met with serious criticisms, especially during the last decade. Some of these have challenged the very assumption of reading as a stage process. Following Piaget, the notion of a stage process bears a tacit assumption o f a universal (i.e. culture independent) sequence of discrete cognitive gains. This aspect of universality, in particular, may be inappropriate in the context of reading, which (unlike such ‘natural’ skills as speaking or object recognition) is not acquired spontaneously and relates to a cultural artefact. At least some purported universal stages of reading acquisition may simply be a reification of the type of reading instruction children happen to be exposed to. This criticism is particularly directed at the notion of an initial, purely visual (‘logographic’: Frith, 1985) phase, which probably occurs only in the context of the ‘look and say’ teaching method that was developed as an attempt to deal with the specific demands of inconsistent English orthography (Wimmer & Hummer, 1990; Wimmer, 1990; Ellis, 1993).
Most phase models avoid this ‘false universality’ criticism, however, as they explicitly reject the strong stage assumption (stage A is a precondition o f B) and talk about reading as a sequence of phases (phase A usually precedes B, but B does not require A to develop). However, many authors also question this weaker assumption. Literature reviews (e.g. Stuart & Coltheart, 1988; Share, 1995) suggest that processing strategies applied to written stimuli at any particular point in time vary considerably depending on stimuli properties (such as their familiarity and orthographic complexity: see the previous section). Thus, the choice of processing strategy is predominantly stimulus-driven, and children may have a number of strategies at their disposal at any given time (Share, 1995). Phase models, then, result in an oversimplified view of reading acquisition if they treat the strategy that is used most frequently at any given phase of development (for those written stimuli that lie within a typical range of experience) as the only one that is available at the time. Some phase models (e.g. Kirby, 1990) partially accommodate this criticism as they describe the transition into higher phases as the broadening of a strategic repertoire, rather than the mere superseding of one strategy by another.
Even if we do assume the predominance of one strategy at any given point in development, we must take individual differences into account. The same absolute level of performance accuracy may be achieved with the help of different processing strategies (Stuart & Coltheart, 1988; Beminger, 1994). Such dissociation between quantitative (attainment) and qualitative (processing strategies) aspects of performance is, again, difficult to accommodate within a standard phase framework.
Finally, phase and stage models tend to be more descriptive than causal, although some of them (e.g. Frith, 1985 with her notion of developmental steps) do propose some mechanisms that bring about the transition between the stages. Crucially, these models may confound description and explanation, due to implicit assumption that the output of a cognitive system (reading behaviour) directly reflects the internal organisation of this system.
1.3.2. “Processes and resources” models.
Under this label I describe models of reading acquisition, which borrow some crucial elements from the stage framework (e.g. the concept of reading strategies), yet dispose of the assumption of a single sequence of phases. Instead, they propose a parallel development of two or more basic modes of processing, the rudiments of which are present from the very beginning of learning to read. The interaction of those distinct modes, processes or strategies eventually brings about proficient reading skills.
A model of reading acquisition proposed by Goswami and Bryant (Goswami & Bryant, 1990; Bryant, MacLean, Bradley & Crossland, 1990; Goswami, 1999) is an example of this category. Following Frith (1985), they suggest an initial dissociation between reading and spelling. Phonemic awareness and the alphabetic principle are first learned and applied for spelling. Initial reading, on the other hand, proceeds with the help of the analogy-making mechanism that is underpinned by the awareness of larger, intrasyllabic phonological units (onsets and rimes), not phonemes. Eventually, the two different strategies merge, and the alphabetic recoding ability developed for spelling becomes incorporated into reading. The model stresses the importance of early sensitivity to intrasyllabic phonological units. This sensitivity contributes to reading both directly (enabling analogy making) and indirectly (facilitating the development of phonemic awareness that is cmcial for spelling and, later on, for reading as well).
The model proposed by Goswami and Bryant retains the phase framework insofar as it proposes age-related, qualitative change in reading strategies. Also, the
hypothesis of a universal, sequential development of phonological awareness (from large to small units) reflects the stage assumption. However, the emphasis is put on different skills and strategies that operate at the same time in different contexts (reading vs. spelling) and their interaction as the condition of learning success.
Another model proposing the availability of different developmental resources at the same time was proposed by Seymour (1997, 1998) and referred to as the dual foundation model. It postulates two parallel processes that account for the acquisition of orthographic knowledge. One - logographic - deals with words as units and allows one to acquire word-specific information (sight vocabulary). The learning mechanisms proposed here are, however, markedly different from those entailed in Frith’s (1985) concept of the logographic stage: global shape recognition or visual cue identification play but a marginal role; word representations encoded by the logographic process normally consist of at least partial information about letter identities. The other, alphabetic process involves sequential letter sounding and assembly. It is triggered by unfamiliar words and is closely tied to phonemic awareness. Interaction o f the alphabetic and logographic processes leads to the development of the Orthographic Framework. It contains knowledge o f correspondences between spelling and pronunciation that extends beyond individual letter-phoneme mappings but links letter sequences with the phonological structure of a syllable. The orthographic framework is a generalised mode of representation: it entails word-specific knowledge yet also allows unfamiliar items to be read without requiring the alphabetic process (that is, sequential letter sounding). The capacity of the orthographic framework is essentially limited to single morpheme mono- and bi-syllables. Longer words of complex morphemic structure (e.g. containing prefixes or suffixes) are processed within the Morphographic Framework. This component emerges from the Orthographic Framework (which contains representations of individual syllables building longer words) and linguistic - especially morphological - awareness.
Seymour’s model is consistent with the stage assumption insofar as the highest, morphographic framework requires earlier formation of the orthographic framework, which, in turn, cannot develop unless the two foundation processes have already been established at least partially. However, no sequential order in the development of those foundations is proposed.
An interactive-compensatory model of reading proposed by Stanovich (1980) should also be mentioned. It was set up to solve the problem of the relationship between the different levels of processing involved in word recognition: low-level (sublexical
analysis o f individual words) and high-level (syntactic and semantic, which allow for context-based inferences). Stanovich argues against the serial model of processing (whereby information is relayed sequentially through the processing levels) and for the interactive one. Different levels of analysis (e.g. feature extraction, orthographic, lexical, syntactic and semantic) operate simultaneously, each seeking to synthesise the stimulus on its own, yet using and being constrained by the information coming from other levels. This way, a process at any level can compensate for deficiencies at any other level. It implies that, as one level is deficient, the relative importance of other levels increases. For example, a reader with poor word decoding skills may rely more on the contextual factors as a compensatory source of information.
The model was developed primarily to account for individual differences in processing strategies used by good and poor readers of the same age. However, it also has potential for explaining developmental change in reading strategies (e.g. a decrease in reliance on context-based inferences accompanying an improvement in decoding ability) and has been used in this capacity (cf. Share, 1995).
In comparison with the phase framework, the processes and resources models seem more capable of accounting for individual differences in reading at the same level of ability. In fact, explaining individual variance may be their explicit aim (cf. Stanovich, 1980). Individual preferences for certain strategies are explained by the relative strength or weakness of different component (foundation) processes. An account is given for the existence of different subtypes of developmental dyslexia (Stanovich, 1980; Seymour, 1999) - instead of explaining them in terms of developmental arrest at different phases of reading development, as phase models do (Frith, 1985). Also, process and resources models are generally more specific with respect to the mechanism of development, especially the role of phonological skills (although they may differ significantly in terms of how this role is formulated - the issue I shall return to in the next chapter). However, just as for the stage models, the process models seem much more concerned with the early period of development than with the more advanced one. Most of the empirical evidence collected or interpreted in the context of those models relates to the basic (foundation: Seymour, 1999) processes, and attempts to map those developmental models onto the adult architecture of skilled reading are sketchy.