The classic example of this is all those ‘-sign’ words of similar derivation with their silent ‘g’ - sign, resign, design, assign and so on. Much is sometimes made of their irregularity. However, they all have forms in which the ‘g’ is sounded – signal, resignation, designated, assignation and so on. There are many other examples: nation but native; stationary but static; optician but optical, clinician but clinical; permission but permissive; division but divisive. There is medicine, but medical and medicate, publicity but public. You will easily find many other examples.
Units (e.g. morphemes) which look the same probably mean the same, those which don’t probably don’t.
Units with related meanings or derivations share common spellings, and at least one of these will probably be phonically regular. Sign and signal look the same and mean similar things whereas rode and road do not and do not, for example. This gives the reader information of such value that it is surely worth the sacrifice of a little phonic regularity.
‘Though this be madness, yet there is method in’t.’ (Hamlet, of course.)
Spelling ‘reform’ is still, spasmodically, being suggested. (Look on the internet for several passionate societies and some frankly mad ones.) ‘Reform’ invariably means phonic regularisation, turning English spelling into a more transparent phonological system where letter-sound correspondences are far more regular, less diverse and more predictable. This would make life easier for the absolute beginner learning to write, but a tremendous amount of value to everyone else would be lost in the process. We fluent readers, of course, are almost completely unaware just how much the spelling system helps us take meaning from text so easily and at such speed and just how many clues to meaning we would lose if our spelling were to be reorganised on purely phonetic principles.
What would we do about regional differences? If the spelling system is to be democratically regular, to be phonetically true everywhere and for everyone, then a Glaswegian will have to use a radically different writing system to the Prince of Wales, and both will have to use a different system to a Liverpudlian, or me - never mind the vast numbers of English speakers in other countries; what about the dog breeder from Dallas, the solicitor from Sydney, the judge from Jamaica, the barman from Bermuda, the physiotherapist from Fiji, the doctor from Delhi, the cricketer from Christchurch, the urologist from Utah, the typist from Toronto, or the mechanic from Mombasa? If it is to be one single, phonically regular system, should it be yours, or theirs, or mine?
More importantly, large swathes of semantic and derivational/relational clues will disappear. Wee wud hav to rite notist, hawdid and plaset; ride hawsez on rodez and sale botes; rite sine but signl; medik and medikl but medikate with medisin; nativ but nashunal; wud and wud, cud, shud, doo and dun, way, way, wade and wade and soe on and soe on and soe on, mor or less indeffinnitly. Will this help fluent readers? If not, will spelling be rendered sufficiently more immediately accessible to the very early learner to compensate?
Many linguists, Chomsky among them, consider that English spelling allows the fluent reader to home in on the meaning-bearing parts of text which really matter, to skip along sampling a minimal amount of the actual text. If this facility were to disappear we would be obliged to examine all of every word we read, something we don’t do at present. When we read we probably use a cascading, proactive and predictive, minimal cue, flexible, visual feature analysis system which certainly tends to consider units larger than letters regardless of their ‘sound’ - common patterns, morphemes or even whole words (common words, remember, are often the most phonically irregular of words). If these flexible and minimalist features were replaced with a one-dimensional phonically regular system, reading would be slower, and comprehension more difficult, for all but the very earliest readers.
The most compelling argument against reforming English spelling until it is phonically regular is that we don’t read or spell primarily phonically in the first place. We don’t access written language mainly through its sound. We read, and we spell, visually. Phonics is a secondary system used, when it is, as back-up to the visual system. Very early readers use a degree of phonic attack, but even very phonological readers, as they become fluent, wean themselves onto predominantly visual attack. (Goswami & Bryant 1990, Frank Smith 2004) Fluent literacy is visually mediated and English spelling is a visual system, signalling to a visual reader. Wholesale reform might benefit the very early learner at the very beginning of the process, to an unknown degree, but it would seriously disadvantage the more fluent user. ‘It is worth noting a healthy trend towards trying to understand our orthographic system instead of trying to reform it.’ (Brazil 1982, cited in Peters 1985 p. 35).