Asking children to memorise scores of high frequency words (e.g. Dolch words, or the HFWs listed in the DCSF's Letters and Sounds programme) as random strings of letters without phonic decoding is a dangerous practice, and why no genuine synthetic phonics programme expects the learning of more than a handful of words as logographs. Firstly, because words viewed as whole units form abstract visual patterns which humans find difficult to memorise; examination of different writing systems reveals that the usual memory limit for whole words is around 2,000-2,500, since no true writing system, past or present, has expected users to memorise more than this number of abstract symbols. When children reach their visual memory limits they will struggle to read texts containing more unusual words if they haven't, in the meantime, been taught, or deduced the alphabet code for themselves. Secondly, for most children memorising words seems easy at first and, if its use is encouraged, it will become their main strategy, subverting their phonological abilities and setting up a habit or reflex in the brain which can be hard to shift.
'If children 'receive contradictory or conflicting instruction, most children prefer to adopt a 'sight word' (whole word) strategy. This seems 'natural', it is easy to do initially, and has some immediate success, that is until visual memory starts to overload...becoming a whole-word (sight-word) reader is not due to low verbal skills, but is a high risk factor in the general population, and something that teachers should curtail at all costs.'
(D. McGuinness. RRF newsletter 51 p19)
Homeschooling dad, Timothy Power, learnt the dangers of teaching global sight word memorisation the hard way. He says, of his small daughter, ''The skill of sounding out simple words, that she had been able to do shortly after she turned three, had been completely lost. If she didn't know a word by sight, she was stuck. Now, with that memory of hers that was able to memorize the 50 states by age two, she could get around this problem without too much trouble: she could just get someone else to read it for her a time or two, and then she would remember the word thereafter, and could even recognize it in new sentences. But this was still a work-around (although an effective one); even if a word was in her spoken vocabulary, she couldn't recognize it on the page if she hadn't seen it before in print, even if it was totally phonetically regular, with all short-vowel sounds. And when she came to these words she didn't recognize, she would try to guess, coming up either with nonsense words or with words that were similar-looking (same starting and ending letter, totally different middle), or with a synonym that bore no visual resemblance to the correct word on the page.'' tdpower.blogspot.com/2007/09/phonics-vs-sight-recognition-reading.html