‘… normal silent reading takes place via direct visual access with phonological representations extracted from the lexicon for use in working memory. The indirect phonological route is seen as a back-up route which only comes into play when access via the direct visual route proves difficult or impossible. [or we could] … suppose that the lexicon represents words in a hierarchy of different sized sublexical units. These units are both orthographically and phonologically defined. The phonological units may or may not be given much weight during lexical processing, depending upon the task requirements. If the task requires overt pronunciation, homophone decision, or a heavy load on verbally coded memory, then the phonological representation is given more importance than if a simple lexical decision response is required.’ (Taft 1991 pp. 91-92).

‘The common ground for all positions [in this debate] is that direct visual access is important and that sound encoding plays some part.’ (Rayner and Pollatsek 1989 p. 109.)

‘… when a skilled reader fixates a very familiar word, access to its meaning occurs directly from print, with the sound of the word having only a minor role. When a skilled reader fixates a somewhat less familiar word, but one that has nevertheless been encountered before, then access to meaning directly from print and indirectly via the sound of the word may occur more or less simultaneously and in parallel.’ (Ellis 1993 p. 36.) [Though logically these two recognitions cannot be exactly simultaneous, because text is a visual signal which must be visually appreciated.]

Reading is primarily a visual activity. Text is a purely visual signal (it is silent) and the most direct, simplest, fastest and cheapest way to deal with it is purely as a visual item. ‘Phonological recoding’ (‘hearing’ the text) may be the result of automatic spreading activation. On the other hand, when the fluent reader is reading very simple text very fast such phonological recoding may not always occur at all. Possibly (the evidence is not voluminous) such recoding into sounds is a means of helping to hold what has been read in memory for long enough to enable comprehension of large and unruly pieces of text like, for example, this overweight sentence (Adams 1990). Ellis (1984 p. 59) says that ‘… the inner speech which accompanies much of ordinary reading does not provide the initial access to meaning but may afford a useful second pass through the language comprehension process’.

And to take all this into practice: Frank Smith says ‘...rather than phonics making reading possible, it is reading that makes phonics seem to work.’ (Smith 1985 p. 50). He also makes the points - blindingly obvious once made - that ‘… in order to apply phonic rules, words must be read from right to left.’ or at least read completely and then considered before their meaning or pronunciation can be accessed - as if they were read backwards (Smith [Frank] 2004 p. 143) (how will you know otherwise what to do with the ‘ph’ in graphic and shepherd, or how to pronounce stable or stability?) and ‘…phonics rules look simple if you already know what a word is.’(ibid. p. 143).

Smith’s examples include the letters ‘ho-’ at the beginning of eleven common words and pronounced in eleven different ways. (Hot, hope, hook, hoot, house, hoist, horse, horizon, honey, hour and honest.) I, for the same purposes, will give you the ‘magic E rule’ or vowel/single consonant/vowel, VCV rule. This is a widely taught ‘phonic rule’ which states that a vowel followed by a single consonant and then another vowel is sounded long - it ‘says its own name’. Examples would be ‘cut’ becoming ‘cute’, ‘bit’ becoming ‘biting’, ‘lad’ becoming ‘lady’, ‘not’ becoming ‘notorious’, ‘a venerable bed’ becoming ‘the Venerable Bede’, and so on.