March 01, 2004

A shibboleth of gentility: [h] from William Shakespeare to Henry Higgins

I wrote earlier that "it seems that there was a period centered around 1800 when 'an hero' was common, as suggested by this histogram of the death dates of the 60-odd authors that finds for the search string 'an hero'. By 1900, "a hero" is all that is found; and the pre-1700 citations also seem to be mostly of that form, though there are not many of them."

Thanks to Bill Labov, I can now give the story behind this little graph, based on information from Henry Sweet, Otto Jespersen and others. The summary: h-loss was a feature of some London speakers since at least the mid-16th century, and probably affected Shakespeare; during the 18th century h-loss spread rapidly to all classes and regions of England except the far north, without being stigmatized at first; in the 19th century, it was beaten back, in part by explicit prescriptivist pressure, so that [h] became a strong "shibboleth of gentility"; this pressure leaked out to cause the introduction of [h] into words where it had always been spelled but never pronounced, like hospital and humour. Meanwhile, Americans missed the action almost entirely, though the English fad for h-loss did apparently affect a few colonials such as John Adams.

Henry Sweet wrote in his New English Grammar (published in two parts, 1892 and 1895):

894. Initial (h), which was preserved through First and Second MnE, began to be dropped at the end of the last century, but has now been restored in Standard E. by the combined influence of the spelling and of the speakers of Scotch and Irish E., where it has always been preserved. It is also preserved in American E., while it has been almost completely lost in the dialects of England -- including Cockney E. -- as also in vulgar Australian.

Otto Jespersen, in his Modern English Grammar (1949), gives many details (section 2.943) about the history of the "mute h" (originally in words taken from French), observing that "in many ... words where h is marked as mute by early orthoepists, the tendency to pronounce according to the spelling has become increasingly powerful." Among his examples are humble, inherit, heretic, homely, hypocrisy, hospital, heritage, humour, for which he documents considerable variation in the opinions of various authorities. He says that "humour and hotel are now pronounced with [h] by some educated speakers, without [h] by others." Finally, he says that "[i]n such words as are taken directly from Latin or Greek or as suggest a learned origin, though they may originally have come from French, h is pronounced: heredity, hero, heroism, hemisphere." [emphasis added -- it seems that hero was a good probe].

In section 13.6, Jespersen discusses the history of "loss of /h/", breaking the topic down into "several different phenomena, some of which are universal, while others belong to vulgar or dialectal speech." He distinguishes h-dropping "in rapid speech in the weak forms of pronouns and the auxiliary verb have"; h-dropping "in the second part of a great many compounds, especially those in which the separate elements are not felt as independent words" (such as Chatham, dunghill, coffeehouse, hedgehog and falsehood); h-dropping "between a strong and a weak vowel" (annihilate, vehement, rehabilitate); dropping initial h before a weakly stressed vowel (e.g. historical, Hungarian);

Jespersen agrees with Sweet that "in all English dialects, except the very northernmost ... [h] is completely lost as a significant part of the sound system, and the same is true of the vulgar speech of the towns." He mentions the insertion of [h] for emphasis or as a hypercorrection, and wisely observes this is normally done "without any regard to whether the word 'ought to' have [h] or not", but that "[t]he observer... to whom [h] or no [h] is significant, fails to notice the words that agree with his own rule, but is struck with the instances of disagreement, deducing from them the impression of a systematic perversion ("Am an' heggs"). "

Jespersen has some interesting things to say about geography and history:

13.684. Initial [h] is preserved in Scotland, Ireland, and America. "The Yankee never makes a mistake in his aspirates," says Lowell ... [note that this means that Adams was atypical or perhaps England-influenced in writing "an hero" -- myl]

13.685. It is not easy to find out how old this English disappearance of [h] is. From the great local extension of the phenomenon, one would be inclined to look upon it as very old, though why should recent sound-changes be unable to spread pretty fast over a large area? As a matter of fact, I have not come across any older mention of it than 1787. Elizabethan and even 18th century authors, who represent vulgarisms so frequently, do not seem to use omissions and misplacings of h's as a characteristic of low class speech. E 1787 (vol. 2.254 ff.) complains of exactly the same errors in this respect as are met with nowadays ... W 1791 speaks of the 'fault of the Londoners: not sounding h where is ought to be sounded, and inversely.' B 1809 p. 29 says: 'the aspirate h ... is often used improperly.' ...

H.C. Wyld, in A Short History of Modern Colloquial English (1936), asks "when did the tendency arise to pronounce 'ill for hill, or 'ome for home, &c., when these and other words occur as independent words in the sentence?" He observes that "Norman scribes are very erratic in their use of h- in copying English manuscripts, and we therefore cannot attach much importance to thirteenth- or even to early fourteenth-century omission of the letter which occur here and there." He says that "I have found comparatively few examples in the fifteenth century of spelling without h-", though he does cite a handful that "seem genuine". The first "fine crop of h-less forms" that he finds is in the middle of the 16th century, in the writings of "the Cockney Machyn".

Wyld also observes that h-dropping is not stigmatized in these earlier periods: "Cooper does not include the loss of initial h- among his traits of 'barbarous dialect'". (This is Christopher Cooper's English Teacher, 1687). Wyld also indicates that the restoration of mute h was continuing through the early 20th century, writing that "[t]he restoration of an aspirate in [humour, humoured] is a trick of yesterday, and I never observed it until a few years ago, and then only among speakers who thought of every word before they uttered it." He also observed that in his day the h was still dropped in the phrase at home "by excellent speakers".

He quotes Elphinston 1787 as writing that "many Ladies, Gentlemen and others have totally discarded" initial h-. He adds that "Walker, 1801, also draws attention to the habit, which he attributes chiefly to Londoners." His conclusion: "it would appear that the present-day vulgarism was not widespread much before the end of the eighteenth century... The practice, which aparently did exist in Machyn's day in London, must have been confined to a limited class."

H. Köberitz (in Shakespeare's Pronunciation, 1953) writes (p. 307 ff.) that

From the 15th century on we can witness a general tendency for initital h to be dropped in fully stressed words of Germanic origin, and conversely, for an inorganic h to be added to such words beginning with a vowel...

The implication that Shakespeare perhaps used to drop his h's has nothing startling or derogatory in it. As a matter of fact, the correct use of h had not yet become a shibboleth of gentility. Its omission was simply a colloquialism comparable to the loss of d and t ..., one that Shakespeare woud almost certainly have picked up anyhow on settling down in London, for that most conspicuous feature of Modern Cockney, the dropping of h's, was then merely the local offshoot of the general tendency just referred to. ... Here colloquialisms jostled for supremacy with conservative or artificial pronunciations inspired by the spelling and inculcated by zealous orthoepists...

One of the pieces of evidence brought forward is Shakespeare's punning, examples of which include "Arden-harden, art-heart, ear-here, eat-hate, heir-hair, heir apparent-here apparent, here-year, hour-whore, and perhaps Hiren-Irene-hiring". Another kind of evidence is provided by the distribution of an in an happy, an hayre, an hundred; there is also (orthographic) elision (examples like t'have, t'hold, th'harmony, th'hoorded); and "inverted spellings" howlet for owlet, histy for yeasty and shagge-ear'd for shag-haired.

[h]-loss should be a particularly interesting phenomenon to investigate further, as a case study in the dynamics of language variation and change. The change went essentially to completion, in England anyhow, before being beaten back in the standard language, but it never took hold in America or in Scotland. The crucial time period (roughly from 1600 to 1950, and especially from 1700 to 1900), is very well documented. Finally, it's unusual in being a sound change that can be reasonably well tracked in the orthography, quite apart from the representation of colloquialisms or the frequency of misspellings, by looking at the distribution of a/an.

Posted by Mark Liberman at March 1, 2004 09:16 AM