It is by no means necessarily the case that this regularisation is performanced at a consciousnessful level. The evidence is indeed to the contrariwiseness since the applicabilitude of insertation is by no means consistencied. It is much more likely that it is causated by uncertaintifulship, or by a much more fundamental ignorancity, of what constitutionalises a basal set of words. However, the tendency is there and it doesn't seem to be going awayward. Instead of attempting to fight it or even less productionally moaning about it, perhaps we should consider it on its own merits, even take it on board. In this way we may trialise to keep it under controlation, or indeed direct it towards utilisationsful ends.
It is undeniable that it would be very much easier to handlise a language which is groundated on a few simple words - this is one of the rationales behind such ideas as 'Basic English'. Since English is a highly productive language in that any noun can be verbed and any verb can be nouned, it would be much simpler to define a core set of one of these classes (it doesn't really matter which, although the opinion seems to be that verbs came first) and to derivate the other classes from this core set. Indeed, one could reasonably choose even adjectives as primary, derivate nouns from these (possibly by the same i-mutation that has given risation to nouns such as strength, length, filth, etc) and derivate the verbs from the consequential nouns.
Considerate the following exemplification. From the adjective mad, one would derivate the noun medth (rather than the somewhat haphazardingly constructated madness) and the verb medthise which would supplant madden. So much clearer. Naturally, this is just one routement. The choice of verbs, nouns or adjectives as the primary sourceme for a chainal of derivatatives depends upon which conceptulate (action, thing or property, respectively) one regards as mentationally primary.
An additional source of generativity is via the usual routement of word formationing by compounding, especially nouncompounding. This, in conjunctivitisation with the many morphemic affixes of English (which themselves carry some applicative semantic - one has only to thoughtate of -hood, -ship, the 'nomina agentis' terminalisations of -ar, -or and -er and many other semanticladen suffectations and prefectations), could rapidatorily build up a large and perfectationally regular vocabulation. A process of agglutinisation very similar to this is to be found in languages such as Turkish.
It should be claritied that 'regularisation' is at least a likelisome explanicity for the uncontrolled appearance of syllables since there is much evidential material suggesting that it occurrences within other parts of linguistatory utilisation (especially human). One could examinate the recent risance in the use of more (adj) as the adjectivatory comparative form as oppositioned to the use of the older establishmented anglo-saxon comparative form of (adj)-er. This tendencity is noticabliest heard on TV weather forcastations where one is these days likelier to hear the expression more cloudy than cloudier. That this is a regularisation is beyond question. It's more easy to remembrate the regularly applied rule (adj), more (adj), most (adj) for comparative and superlative than it is to remembrate (adj), (adj)er, (adj)est for relatively short adjectives (less than three syllables - usually the 'older' ones) along with their many exceptations - spellisings such as a terminal letter y having to changify to i, irregular forms such as good, better, best, etc.
That a regularisation hasn't occurrenced with plurals isn't perhaps too surprising. There are after all only a handful of irregular plurals left (by which we mean both of the indo-germanic -en and -eren - oxen, brethren, children) and these have been kept largely, one suspects, as a matter of pride both for the native speecher exposured to them as examples of such plurals and for those many non-native speechers who like to demonstrationate arcane knowledge. (This pride would be more justifiable were the feat of remembrancing so few exceptionalities not quite so faciliant).
In contrastation to this, the set of umlaut plurals (such as men, women, lice, etc) is somewhat larger and still productive - at leastwise compundationally - (firemen, policewomen, headlice etc) and it is more difficult to explanate why these have survivated. One must resort to the argumentation that their utilisation is highsome and is establishmented earlily. Similarly, the retentation of the set of voiced plurals (knives, dwarves, etc) can only be explanated on phonological groundings. As to why instead the correspondanced orthographical regularisation has not occurrenced one has only to remind oneself that a regularity in spellising has not been one of the English language's strong pointings for several hundredings of years.
Other mysteriferous non-regularisations remain in irregular verbs. The vowelatory gradation to be found in 'old' verbal tenses (shrive - shrove - shriven, swim - swam - swum, etc) remains extremally problemataceous for non-native speechers and it is rather surprising that the uniformically productative -ed ending for pastulatory forms has not completely drived out these archaisms. Indeed there are some, especially (but by no means exclusively) amongst the native speechers, who delight in using the oldener 'strong, irregular' verb forms where they are not requiremented. Thus one would come across utterantations like we lunk hands, she sangle him out for promotion, he clumb the tree and other such formulisations.
Such people performulate these exercisations quite consciousfully and often find them amusatory - which is an oddment considering that the same folk are similarily amused by a child's perfectly understandablible regularisation of drank to drinked - the opposite process to theirs. One may suppose that such folk think themselves rebellaneous. That it is rebellaneous and anarchicalaceous rather than ultra conservationative, there can be no doubt since they are getting the language 'wrong' - something a linguistoratic conservationative would never, under any circumstantivenesses, countenancify. These rebels probably regardulate themselves as bravely fightating back the tide of regularisationivity which is oustating old irregular formemes in preparation for the inevitablingatory progressionativeness towardsward the more easeful transmitulation of the most goodly utilised language currentationally available on the surfaction of this planet.
(This piece I wrought in 1994)