The American Language as I knew it in my callow days was the third edition, 1923, revised and enlarged. It ran to 500 pages, counting pp. i-x. There have since been a fourth edition, further revised and enlarged, and two supplementary volumes and some magazine pieces. What is now before us is an abridgment of all that, and an abridgment not uncalled for. For all its abridgment, the volume exceeds the 1923 enlarged edition by a factor of 2 1/3. (To verify, add xxv and 777 and, all over again, cxxiv in back, and then compare print and format.)

Along with his abridging, the editor has made corrections and, in brackets, judicious supplementations. Between author and editor the 1923 mistakes have dwindled.

Thus in 1923 (p. 164) Mencken reported R. G. White (fl. 1868-81) as deploring the American presidential and favoring presidental, “following the example of incidental, regimental, monumental, governmental, oriental, experimental, and so on.” Mencken failed to observe that four of these six are irrelevant, being built not on Latin participles in -ens but on -mentum. Also he failed to observe that presidential, paralleling referential, is impeccable as an adjective for presidency. Now the new abridged volume still mentions that there had been disapproval of presidential; it mentions it six times (as against twice in 1923, which helps explain why this is 2 1/3 times as long as the 1923 edition). But happily it skips the reasoning.

This may seem to you like sweeping the dirt under the rug. Similarly for the next case. Take p. 182 of 1923 on bust: “This…has come into a dignity that even grammarians will soon hesitate to question. Who in America, would dare to speak of bursting a broncho or of a trustburster?” Appreciation that bust means break would have prevented this remark. Now in the new abridged volume I still find no equating of bust to break. But the remark is gone.

I see it rather as editorial restraint, reluctance to meddle beyond necessity. My view is encouraged by some passages containing the technical term back-tormation. In 1923, p. 190, Mencken misapplied the term of prof, co ea, dorm, and the like. In this new abridged edition, the mistake persists (p 203) and even recurs (p. 213). Yet the editor knows better. In the course of one of his bracketed inserts (p 205) he uses the term himself and uses it right.

Many errors are gone. In the 1923 edition dead was called preterite (p. 285), you in How do you do? was called objective (p. 305), Rainier was called the greatest American peak south of Alaska (p. 357), the verb house-clean was listed among nouns used as verbs (p. 198), the open o sound in standard sauce was misidentified (p. 323), the British prounuciation et for ate was taken for distinctive American (pp. 275, 280, 284), and rench was said to be the invariable American for rinse (p. 281). These errors, hence no doubt also many others, have disappeared.

Cases of mere bad judgment have been eliminated too. Thus in 1923 Mencken attributed the “raciness” of Where are we at? to “the somewhat absurd text-book prohibition of terminal prepositions” (p. 187); surely it is due mainly to the redundancy. He saw the noun try as an apocopation of trial (p. 191); surely it is rather a freshly nominalized verb. He saw kindergarden as of a piece with pardner (p. 325); surely folk etymology is more to the point. In each case better judgment has now prevailed—not by substitution, just by deletion of the injudicious passage.

Cases of bad judgment are also preserved. The vernacular tole for told was lamely explained in the 1923 edition (p. 288) assimilation of d to l. “So also, perhaps, in swole,” Mencken continued, “which is fast displacing swelled.” It should be evident that standard told is (like sold) anomaious: it resembles a strong verb in the drastic vowel change from tell, but is weak in taking -d. The dropping of this -d is mere resistance to anomaly. As for swole, the fact is that swell used to be a strong verb, preterite swoll, participle swollen; adjectivally this old participle is still going strong; small wonder, then, if there is still a little life in the old preterite. Assimilation indeed. Happily the swole tale is missing from the new edition, again by simple curtailment; but the tole tale still hangs on (p. 531).

Some strong-verb trouble even emerges since the 1923 edition. Thus in the new edition an Irish pronunciation ped of paid is cited to illustrate “a tendency…toward strong conjugations” (p. 529). It is parallel to standard said, and weak. Also in the new edition we read, apropos of a contrary tendency toward weak conjugations, that “Even when a compound has as its last member a verb ordinarily strong, it is often weak itself. Thus the preterite of to joyride is…joyrided” (p. 532). This again is no proper illustration; to joyride is not a compound of to ride, but a compound noun joyride gone verbal, and nouns newly gone verbal always make weak verbs. Here the editor’s touch is arrestingly light: he inserts the bracketed remark “And no baseball player ever flew out to end the inning; he always flied out.” This example slyly shows that he has properly in mind the point about nouns gone verbal, and no nonsense about compounds; still he leaves Mencken’s remark intact.


Early and late there is a puzzling insensitivity to the orthography of hard and soft g and c. In 1923 (p. 232) and again in the new abridged edition (pp. 483 f), we read: “The superiority of jail to gaol is manifest by the common mispronunciation of the latter by the Americans who find it in print, making it rhyme with coal.” What is really glaring about the English gaol escapes mention, the soft g before a. Nor do I find in either edition any notice of the one other example I know: the frequent American pronunciation of margarine. Correspondingly, where skeptic and sceptic were compared (1923, p. 233), the anomaly of hard c before e went unmentioned. And in the new edition Passaicite and Quebecer are exhibited (pp 681 f) with obviously no thought of a soft c.

The handling of foreign languages is postpossessing. In the new edition Mencken speaks of the sermo vulgus. Look who’s talking. Read sermo vulgaris, or, following Cicero, vulgi sermo. Between editions mistakes in foreign words (1923, pp. 256, 338, 358, 364) have been caught, but one would welcome wider perspective. Thus in the new edition (pp. 374 f) Polack and Chinaman are given under Terms of Abuse, and their histories in English are enlarged upon, with never a hint that Polak is the Polish for Pole and Chinaman is a translation of the Chinese term. Or again take a as in He musta been. We read: “The OED describes this reducing of the OE habban (Ger. haben) to a as the ne plus ultra of the wearing-down tendency among English words” (p. 534). If we are to dwell on the point, French bears notice for its a form habet.

“Mencken always insisted, with what seems to most linguists an excessive modesty, that he was not a scholar himself”—so writes the editor, to everyone’s credit. It is a credit to Mencken to have insisted, a credit to the editor to remark that he did, and a credit to the hearts of most linguists to have protested. As for the book, in attainment and in evident aspiration it is less linguistic treatise than fun book. So be it Vive le sport. And it is less fun book than, if I have found an adequately neutral word, compendium. It is for all its abridgment a big compendium of varied material, varied in entertainment value, varied in degree of inconsequence. Thanks to its thick index it is admirably suited to sporadic reading.

It is not primarily a manifesto, but it savors of that too. The American Language early and late casts an image of its author as vulgi defensor, champion of the low-faulting. There is an air of indiscriminate forthrightness and no nonsense. It is murky air, and it blankets conflation. Through it darkly we seem to descry two gathered hosts opposed: regular fellows on the left and a mealy-mouthed ruck of schoolmarms, Englishmen, and displaced Latin grammarians on the right.

Now this illusion of a simple contrast is a confusion of five separate contrasts that are pertinent to Mencken’s remarks and quotations. One, assuredly, is the contrast between English in the United States and English in the United Kingdom. A second is the contrast between English grammar efficiently described as by Jespersen in expressly devised categories, and English grammar clumsily described in earlier decades in categories inherited from: Latin grammarians. A third is the contrast between speech as a basic trait of the human species and writing as a recent derivative cultural quirk. A fourth is the contrast between colloquial and literary style. And a fifth is the contrast between the descriptive and the normative treatment of language.

Mencken quotes E. H. Sturtevant thus. “Whether we think of the history of human speech in general or of the linguistic experience of the individual speaker, spoken language is the primary phenomenon, and writing is only a more or less imperfect reflection of it” (p 517). In his next sentence Mencken invokes Jespersen, and in the next he quotes H. E. Palmer thus: “[Spoken English is] that variety which is generally used by educated people in the course of ordinary conversation or when writing letters to intimate friends.” Now Sturtevant was alluding to the third contrast in my list of five; Palmer was concerned with the fourth, and Jespersen always primarily with the second. Mencken pictured all three linguists joined as in crusade against a common mawkish host.


Whatever it was that Mencken stood for may seem, for all its softness of focus, to have prospered; witness the new permissiveness of the Merriam-Webster dictionary. “This data and like I said are all right now,” the schoolboy protests when his theme is marked down. “The new dictionary says so.” That fixes the schoolmarm.

I shall not argue that this is a scene best calculated to have gratified Mencken. But it does suggest that the last in my list of five contrasts is the one that merits most thought: that between descriptive and normative.

Schoolmarmism comes in part from tampering with facts to accomodate poor theory. Insofar it is bad. And schoolmarmism is normative. Scientific linguistics is descriptive, and good. The new permissiveness of the dictionary, a waiving of the normative in favor of the descriptive, is therefore good. So now anything (if it is already going) goes.

Let us sort this out. Scientific linguistics is indeed good. In particular a purely descriptive, non-normative Merriam Webster was a good thing to make, if the work was competent and the job had not previously been adequately done, which is as may be. But there is a fallacy in calling the result permissive if the book is not normative it no more permits than forbids. And it would be a fallacy also to conclude from the virtues of descriptive linguistics an the faults of schoolmarmism that the normative must be bad. This would be a normative conclusion and a false one.

Behind the schoolboy’s illusion there is a feeling that nothing in language is wrong save as a rule book makes it wrong. People fail to reflect that there remain values in language even if all dictionaries go descriptive, and conversely that normative dictionaries and other manuals of good usage when they do exist are purely advisory like cookbook. In this capacity they are useful even against bad schoolmarms.

This Issue

January 9, 1964