Why English words are not pronounced as they're spelt
Edward Finegan and Niko Besnier, from Language: Its Structure and Use, pp. 34-35.
There are five principal reason for the discrepancy between the written representation of many English words and their actual pronunciation:
English orthography had several diverse origins with different spelling conventions:
The system that had evolved in Wessex before the Norman Invasion of 1066 gave us such spellings as ee for the sound in words like deed and seen.
The system that was overlaid on the Old English system by the Normans, with their French orthographic customs, gaves us such spellings as queen (for the earlier cween) and thief (for earlier theef).
A Dutch influence from Caxton, the first English printer, who was born in England but lived in Holland for thirty years, gave us such spellings as ghost (which replaced gost) and ghastly (which replaced gastlic).
During the Renaissance, an attempt to reform spelling along etymological (that is, historically earlier) lines gaves us debt for earlier det or dette and salmon for earlier samon.
A spelling system established several hundred years ago is still used for a language that continues to change and develop its spoken form. Thus the initial k in knock, know, knee, and certain other words was once pronounced, as was the gh in knight and thought, among others. As to vowels, change in progress when the system was developing and continuing change in pronunciation have led to such matched spelling for mismatched pronunciations as beat/great and food/foot.
English is spoken differently in different countries throughout the world (and in different regions within a single country), despite a relatively uniform standard for written orthography. Though this orthographic uniformity certainly facilitates international communication, it also increases the disparity between the way English is written and spoken in any given place.
Words (and their meaningful subparts) alter their pronunciation depending on the adjacent sounds and stress patterns. For example, in electric the second c represents the sound [k] as in kiss, but in electricity it represents the sound [s] as in silly. Compare also the pronunciation of i in senile (pronounced like the i of I’ll) with its pronunciation in senility (in which it has the i of ill).
Spoken forms differ from one set of circumstances to another—for example, in formal and informal situations. While some degree of such variation is incorporated into the written system (do not/don’t; was/’twas), there is relatively little tolerance for such spelling variation as gonna (‘going to’), wanna (‘want to’), gotcha (‘got you’), and jeat yet? (‘did you eat yet?’). Such variable spelling of variable speech would force readers to determine the pronunciation of the represented speech before arriving at meaning, instead of directly for meaning, as adult readers normally do, with the necessity of silent pronunciation.