Cfrtjryk

In one of the Nature articles related to Google n-grams site [1], as well as in the book [2], the authors describe (and quantify the rate of) the process of regularization of English irregular verbs.

The word "built" seems to have moved the other way, from regular to irregular, since I found the variant "builded" listed in a dictionary as archaic. Is there an explanation for this exceptional occurrence?

[1] https://www.nature.com/articles/nature06137
[2] https://www.amazon.com/Uncharted-Data-Lens-Human-Culture/dp/1594487456

Regular and Irregular, Weak and Strong

Past tenses and participles such as built or the British past tense spelt, learnt, spilt or burnt are orthographically irregular, but that does not make them irregular verbs in the sense the Nature article is using that term. There, an irregular verb is a Germanic strong verb which forms its past and participle through vowel change: sink, sank, sunk; drive, drove driven.

The article gives statistical underpinnings to the long-observed phenomenon that infrequently used Germanic strong verbs tend to become weak, though even common verbs such as help — OE helpan, healp, holpen, ModGer helfen, half, geholfen — could become weak as well.

“Irregular” in this sense also includes English verbs that became strong through analogy, such as strive, strove, striven, like drive, which is becoming regular again for many speakers: strive, strived, strived, or dove as the past of dive, though no one ever said *diven. And no one knows where dig, dug, dug came from — there is no obvious cognate verb in other Germanic languages and an anology formation to stick, stuck, stuck seems a bit far-fetched.

A weak (regular) verb is one which forms its past and participle by means of a dental suffix (d or t) without vowel change. A hybrid is one that does both: think, thought, thought; bring, brought, brought. In this sense, build is still a regular verb: it has no vowel change and it ends in a dental consonant. It is still irregular in the sense that its past and participle cannot be predicted. The same goes for spilt, learnt, etc., despite the obvious pattern of the verb stem ending in l or rn: burn > burnt, but not turn > *turnt.

Noah Webster

The reason why Americans write spilled, learned and burned and, unlike in Commonwealth nations, only use spilt and burnt as adjectives is because of the spelling reforms of Noah Webster, not because of some deep tendency in the language itself toward more regular forms. Curiously enough, he didn’t make a case for builded; the only problem he had with built was the u. He proposed instead to spell it bild and bilt, and did so — inconsistently — in his 1832 History of the United States:

On a small isle, in a fresh water pond, within the large isle, he bilt a hut, and remained in it about six weeks. — Noah Webster, History of the United States, New Haven, 1832, 90.

360. First Shipping bilt in New England. The first attempt to bild water craft in New England, was in 1626. Ibid., 173.

In other passages, whether out of habit or through the intervention of a zealous typesetter, build and built launch a comeback:

As the first colony which Columbus left was cut off, he sought a more convenient and healthful situation, marked out the plan of a town, erected a rampart, and built houses. Ibid., 75.

Bild and bilt — along with other proposed reforms like masheen, tung, and dawter — never caught on.

Middle and Early Modern English

Though the Old English verb was quite rare, built and builded, in their varying form and orthography, were used concurrently in both Middle and Early Modern English. The prolific writer and monk John Lydgate of Bury seems to prefer bilt:

Þis myȝti stronge cite..bilt of lym and stoon. — John Lydgate, Troy Book, 1412–1420 (ms., Aug A.4, 1425), 5.545.

In the manuscript tradition of his Siege of Thebes, however, there is a variety of forms: bilt, beld, bilded, bylde, and even ybylt with the old Germanic perfective prefix (cf. Ger geholfen above, in English ge- > y(e) > Ø).

A castel…Strong and myghty belt [vrr. bilt, bildid, bylde, ybylt] vpon a roche. — John Lydgate, Siege of Thebes, ca. 1420–22 2269.

The two forms coexisted in Early Modern English:

but he was dryuen thennes by brunmylda the quene and come in to almayn / and builded many abbayes… — William Caxton, trans., Prolicionycion [sic], 1482. EEBO

for vpon thys mounte of sion was the temple of the Lorde builded. — Nicholas Udall, trans., Desiderius Erasmus, Paraphrase of the New Testament v. 1, 1548. EEBO

who haue builded our schooles and vniuersities in england? — Robert Parsons, A Temperate VVard-VVord, 1599. EEBO

cities: those which are seated neere rivers, for the most part are builded of stone; … — Robert Johnson, trans., Giovanni Botero, Relation of the Most Famous Kingdomes and Common-wealths, 1630. EEBO

there was a certayne houszholder which planted a vynyarde, and hedged it roude aboute, and dygged a wyne presse in it, and built a tower, and let it out vnto huszbandmen, and wente in to a straunge countre: — Miles Coverdale, trans., Biblia the Byble, that is, the holy Scrypture of the Olde and New Testament, faithfully translated in to Englyshe, 1535. EEBO

if one see but a house strongly and magnificently built, wisely and usefully contrived, gloriously and sumptuously adorn'd and furnish'd, though he as yet see no man appear in it, or about it, will he be such a fool as to say, it had no builder? for every house is builded by some man; but he that built all things is god, heb: iii: — Clement Ellis, The Folly of Atheism, 1692. EEBO

While the KJV follows Coverdale in using built in Mk 12.1, Clement Ellis in following Heb 3.4 of the KJV in using builded and built in the same verse. The first use belongs to the author.

In the entire translation, the KJV uses builded 70 times, but built 198. This was likely enough to help preserve builded as a high register, archaic word suitable for poetic use far longer than any secular usage would warrant. Plus, one never knows when an extra syllable would add resonance or fill out the meter.

I have only seen the abstract of the Nature article, but note that it says the authors proceeded by “tracking inflectional changes to 177 Old-English irregular verbs.” So this article simply doesn’t seem to be about verbs that started out regular and later became irregular. Knowing that irregular verbs tend to become regular over time doesn’t tell you that the number of irregular verbs in a language will go down over time.

As mentioned in the comments, there are various other examples of verbs that have become irregular. In some cases, there may be a phonological/phonetic motivation; in other cases, analogy is probably the cause.