I don’t know what to make of Juliette Blevins’ ideas about language as an evolving system, not being an evolutionary anthropologist, but anything that “undermines a central tenet of modern Chomskyan linguistics: that Universal Grammar, an innate human cognitive capacity, plays a dominant role in shaping grammars” automatically awakens my interest, and I look forward to learning more about them. (Link via aldiboronti at Wordorigins.)


  1. The notion of using a more evolutionary model of language change is more closely linked to Salikoko Mufwene. He has a whole book on hte concept. He doesn’t go in for the phonology so much.
    Mufwene is a lot more interested in contact linguistics. His approach suggests that displaced communities that have to learn a new language tend to adopt and generalise the features of that language that most match their native language’s structures. I think he makes a pretty good case for it for creoles. Certainly Melanesian creole fits the model, but his case for the Black creoles in the Atlantic and Indian Oceans is also a very compelling account.
    What would be interesting would be to see it applied to French, Spanish and maybe even English in the UK. Two centuries ago, all those countries were a patchwork of distinct local languages – some more closely realted than others, and if syntactic features of those languages persist in regional dialects, it would make a compelling case for generalisation.
    As for phonology, many people have noticed that things like devoicing final consonants have reappeared in parallel all over the world. I am pretty down on universal grammar, but I might be willing to accept a physiological predisposition towards final devoicing wherever the semantic ambiguity this produces is minimal.
    The idea that some kinds of language changes can happen coincidentally in different places, and that similar changes can have similar consequences, isn’t that new either. My historical linguistics prof in Montreal built most of his course on the parallel development of articles in the evolution of vulgar Latin and their development – centuries later – in the evolution of Bulgarian. He claimed that the two language developed almost identical structural solutions to the problem, but I’ve forgotten the details.

  2. Doug Marmion says
  3. Scott: Thanks for the Mufwene information; sounds plausible to me too.
    Doug: Not sure what you mean. The link works for me. I believe the pdf is just a pdf file of the same thing; at least that was the impression I got when I visited it.

  4. Doug Marmion says

    Hi LanguageHat,
    the link didn’t work for me when I tried it (or for some others that tried it around the same time) so I assumed it was incorrect, but it must have been only temporarily down. Sorry.

  5. For another aspect of the evolutionary nature of language — how we acquired it in the first place — I can heartily recommend the excellent Carl Zimmer’s most recent essay, Building Gab. That distant clattering you hear is the sound of horns locking in academe: evolution in action, I suppose.

  6. Natural selection, of course, exists in languages as well as biology… two or more words often compete with each other in a language for the same meaning and in the end one wins out. In Linguistics these kinds of words are called “competing forms.”
    For example, in Old Spanish several words like rapoza “fox”, cán “dog”, hiniestra window” and “Rovrico “Roderick” competed with zorro, perro, ventana and Rodrigo but the latter four eventually won out in Modern Spanish. (In Judaeo-Spanish, however, rapoza is still used for “fox”).
    In English, Northern Middle English “they, their, them” eventually won out over Southern Middle English “hi, hir, hem” but the Southern English “church” prevailed over the North English “kirk”. Nowadays, North English “lad” and “lass” have been almost completely vanquished by Southern English boy and girl in both England and America.
    A Scientific American article published in 1952 said that strong verbs like “beaten” and “sung” and “thrown” are gradually disappearing from the English language and will be gone completely by about the year 2850 A.D. Not too long ago, I heard a little girl say “See the ball? I throwed it up in the air!” and I young boy say “I beated the game!” showing that, so far, the prediction of the Scientific American article is right on target.

  7. will be gone completely by about the year 2850 A.D.
    I wonder how they came up with that date?

  8. Language Hat,
    The Scientific American article was written by a British-born linguistics professor, Joshua Whatmough (1897-1964) who taught at Harvard near as I can tell. In the article he says that the 2850 date was calculated by one of his students, Robert Abernathy, but he doesn’t say how. While Scientific American is a good publication ,just as any other, what you usually get in it is a boiled down version of an original study. Nevertheless, I imagine that Abernathy’s original study is probably filed away somewhere in the dim recesses of the Harvard University archives.

  9. Here’s paper on the decay of English strong verbs: it’s short, has graphs, and doesn’t drown you in technicalities. Strong verbs decay into weak ones exponentially, as radioactive isotopes do, but the rate of decay depends on the log frequency of the verbs.

    Checking a Modern English corpus for frequency of the OE strong verbs (a total of 177, excluding verbs lost altogether) among all verbs in the corpus, the authors found that be and have appear with frequency one in ten or better and remain irregular, 11 more verbs with frequency one in a hundred or better all remain irregular, and 37 further verbs with frequency one in a thousand or better all remain irregular except help, reach, walk, work. At the other end, the verbs that appear less than one in a hundred thousand have all regularized except slink, and slinked does show up even in edited prose.

    They see the same exponential decay pattern in classes of strong verbs as in individual verbs (money quote: “Like a Cheshire cat, dying rules vanish one instance at a time, leaving behind a unimodal frown.”) Their estimate for the loss of all the strong verbs is 14,400 years from now, but since there is no regularization data for the top 13 verbs, they admit the extrapolation is meaningless, and most probably the high-frequency strong verbs will remain irregular forever.

    Some minor nits:

    They correctly exclude sneak, spit from the historic strong verbs, but fail to exclude dig, dive, which were weak in OE and ME (“He made a pit, and digged it, and is fallen into the ditch which he made”, KJV Ps 7:15).

    Worse, they sweep under the rug such verbs as sweep, which are treated as survivals because they were strong in OE (swápan, swéop, swápen), became weak, and then became irregular due to later shortening and assimilation, giving sweep, swept, swept. In fact, it isn’t even clear that sweep is the direct descendant of swápan, which regularly became swope in early ME before sweep displaced it; the OED says it may be from ON svipa (with short vowel) ‘move swiftly and suddenly’, but then again the vocalism may be due to analogy with creep, sleep or from the old strong preterite.

    Lastly, they predict that wed will be the next irregular verb to go, but as it has always been weak, a study of strong verbs only is not predictive of its behavior; they say so just because it is very rare and wedded competes with it. Indeed, the OED says that preterite wed is now only dialectal, and participial wed is rare except in poetry. I’m betting on smite, because though it is less than one in ten thousand, smote is much rarer in proportion than other strong preterites/participles.

  10. Thanks!

  11. marie-lucie says

    Brian: I heard a little girl say “See the ball? I throwed it up in the air!” and I young boy say “I beated the game!” showing that, so far, the prediction of the Scientific American article is right on target.

    You do not say how old these children were. In the mouth of pre-school children this is not an instance of permanent language change, but of “overgeneralization” – or rather generalization of the most common past tense formation (addition of -ed) to “irregular” verbs. Usually the forms produced by children at that stage of language acquisition are individual and do not stay with them, they are soon replaced by the adult forms, especially the most common ones that the children hear from older children and adults.

  12. Irregular nouns are fewer than irregular verbs (only about 30, excluding Latin and Greek plurals adopted as-is), and oxen may well be the next noun to regularize. Oxes is already current in speech, ox is a rare word to begin with except when used metaphorically, and oxen is much rarer than ox.

  13. David Marjanović says

    Smote seems to be gone among people who don’t regularly read the KJV, even if they talk about or allude to it; smitten ~ “fallen in love at first sight” is no longer associated with smite.

    Is build excluded (I’m on a tenuous Internet connection at the moment)? Builded is all over 19th-century inscriptions on buildings in the US, not to mention the KJV.

    Checking a Modern English corpus for frequency of the OE strong verbs (a total of 177, excluding verbs lost altogether)

    Are strong verbs preferentially lost? Of the German verbs that used to require the genitive, it seems to me that half now accept the dative (at varying degrees of acceptance as standard), while the other half has been falling out of use altogether.

    ox is a rare word to begin with except when used metaphorically

    Even bull is being lost, and cow is going back (to pre-West-Germanic times) to meaning “head of cattle” in general.

  14. smitten ~ “fallen in love at first sight” is no longer associated with smite.

    It happens that I read something online the other day that used smitten and smited contrastively. The writer clearly drew a connection between the two—it was a bit of wordplay—but evidently didn’t find it possible to use smitten as the past participle of smite.

  15. Builded is all over 19th-century inscriptions on buildings in the US, not to mention the KJV.

    I think of it exclusively in connection with the KJV, and to me it’s utterly obsolete — if someone used “builded” other than as a Biblical quote, I wouldn’t think they were being old-fashioned, I’d think they didn’t know the past tense of “build.”

  16. And what about on bended knee?

  17. Used only in that fixed expression, as far as I know.

  18. If you’re off to Philadelphia in the morning,
      You mustn’t take my stories for a guide.
    There’s little left, indeed, of the city you will read of,
      And all the folk I write about have died.
    Now few will understand if you mention Talleyrand,
      Or remember what his cunning and his skill did;
    And the cabmen at the wharf do not know Count Zinzendorf,
      Nor the Church in Philadelphia he builded.

    —Kipling, “Philadelphia” (1910)

  19. Builded strikes me as unremarkable in a poetic register.

    I have seen Him in the watch fires of a hundred circling camps.
    They have builded Him an altar in the evening dews and damps.
    I can read His righteous sentence in the dim and flaring lamps.
    His day is marching on!

  20. Builded strikes me as unremarkable in a poetic register.

    For the pre-modern period, yeah.

  21. The pre-modern period ended a long time before the “Battle Hymn of the Republic” was written, surely?

  22. Well, obviously there a million definitions of “modern.” To me, the 1860s are definitely pre-modern in terms of linguistic usage. To cite the Battle Hymn of the Republic as evidence that “builded” is part of modern usage doesn’t work for me.

  23. Marja Erwin says

    I used to consider the plague of Pelusium the beginning of the modern world. Now I would say the plagues of contact in the Americas, and the plague of Kaffa and the destruction of the medieval world in Europe and surrounding areas.

  24. At first I read that as “the plague of Kafka” and I was like “Hey now!”

  25. Interesting that “slew”, past tense of the obsolescent verb “slay”, now takes the form of “slayed” in its modern revival.

  26. marie-lucie says

    bended, builded, etc

    In 18th century English writing many verb forms now written with final -ed or -t but pronounced with simply [t] are written with an apostrophe before the final consonant, reflecting the pronunciation of the time, as for instance in work’d or work’t while later spelling has regularized worked for the same pronunciations, perhaps because the two-syllable pronunciation still existed at the time. It seems to me that the period must have been one of evolving pronunciation, with some people (or regions, or registers) possibly using both variants depending on various conditions, therefore bended could alternate with bend’t or bent in the speech and therefore the writing of the same person. As in many other cases, the older, longer pronunciation was the one kept in formal registers such as poetry (even if not of a very high level) while the shorter one became the most common in everyday speech and therefore morphed into the standard one, regardless of spelling. I don’t remember reading about this, but someone must have written about it!

  27. marie-lucie says

    on bended knee

    This phrase originally refers to a specific posture to be assumed by a person of inferior status when facing one of superior status, especially when asking for a favour. Theoretically one could say On bent knee instead, but bent here is not a possible variant of bended, instead it suggests that there is something wrong with the knee in question, whether naturally or as a result of some accident.

  28. A quick look at Google confirms that the fixed expression “on bended knee” is a lot more common than “on bent knee”; however, the latter definitely exists. I know I have seen it in the past and not felt that there was anything wrong with it.

  29. marie-lucie says

    I wrote about my own reaction, which may not be that of a native speaker.

  30. I have the same reaction as m-l: I would consider “on bent knee” an oddity if not an error.

  31. “On bent knee” is the sort of thing I associate with journalists (and/or their editors), probably on no firmer grounds than those of Britons who call everything they don’t like an Americanism.

  32. I wonder if anyone is familiar with shend.

  33. Never heard of it.

  34. I have heard of it, but not heard it in live usage, as it appears as an example in one of Steven Pinker’s books. In an article by Gary F. Marcus et al. on overregularization in children’s speech (Pinker is the second author), I find this:

    The blocking-and-retrieval-failure hypothesis is appealing because it can be deduced from the very logic of irregularity, supplemented only by an uncontroversial fact about human memory known since Ebbinghaus. What is the past tense form of the verb to shend, meaning “to shame”? If you answered shended, then you have overregularized; the correct form is
    shent (Bybee & Slobin, 1982b). Of course, this “error” is not surprising. Irregular forms are not predictable (that is what “irregular” means), so the only way you could have produced shent was if you had previously heard it and remembered it. But you have heard it zero times, hence cannot have remembered it. Now, if in two years you were asked the question again and overregularized it once more, it would still not be surprising, because you would have heard it only once. Since memory storage and retrieval are probabilistic, with a higher probability of retrieval for items that have been presented to the learner more often, hearing an irregular a small number of times should be only somewhat better than not hearing it at all. Thus,
    low-frequency irregulars are inherently prone to overregularization (MacWhinney, 1978; Pinker, 1984; Slobin, 1971).

  35. I’ve seen “latecomers are shent (=ruined)” in the Penguin dictionary of proverbs.
    (In Russian it would be “opozdavshemu kosti.” (Bones are all latecomers can hope for.)

  36. Trond Engen says

    Norw. has a doublet skjenne “reprimand, nag” and skjende “desecrate, defile; ravish”. The pronounced d of the latter marks it as literary. The origin of both is LG schenden.

  37. David Marjanović says

    In Russian it would be “opozdavshemu kosti.”

    Older Than Dirt: tarde venientibus ossa.

  38. Cognate with German Schande “disgrace”?

  39. Trond Engen says

    Looks like a causative *-(i)ja- verb. *skandijan- “make shameful”. Now I wonder about *Ska(n)dinaujō.

    I thought Schande must be related to both Scham and Schade, but apparently not.

  40. I had a nagging thought I had come across shend somewhere in the wild.
    Here it is:

    Thou likenest a woman’s love to hell;
    To barren land where water may not dwell.
    Thou likenest it also to wild fire;
    The more it burns, the more it hath desire
    To consume every thing that burnt will be.
    Thou sayest, right as wormes shend* a tree, *destroy
    Right so a wife destroyeth her husbond;
    This know they well that be to wives bond.”

  41. @Hans: Yes, shend is a cognate. It is a native English word—according to the OED, attested continuously since Anglo-Saxon in the sense To put to shame or confusion; to confound, disgrace (which is pretty close to the modern German meaning). The later sense of To destroy, ruin, bring to destruction. Also, in milder sense, to injure, damage, spoil is attested from the twelfth century. There is also an obsolete noun form, shond. Shame is a parallel development from the same Germanic root.

    Shanda is a very common word in Yiddish, of course. A shanda fur die goyim is something embarrassing done by a Jew in public—a shame before the gentiles.

  42. I’m pretty sure it’s far, though Dr. Google does find fur, as well as English shanda for the goyim.

  43. Trond Engen says

    Hans: Cognate with German Schande “disgrace”?

    I made a short comment on this yesterday, but it seems not to have made it out of the spam trap. My points were 1) Yes. 2) There’s apparently no etymological connection between Schande, Schade, and Scham. 3) We could make an etymology for ‘Scandinavia’ out of this.

  44. Argh! Thanks for letting me know; I searched for and approved it.

  45. David Marjanović says

    *skandijan- “make shameful”

    Indeed, schänden exists and mostly means “rape”. Mostly extinct, but Kinderschänder is still found.

    “Damage” is Schaden, though “that’s a pity/too bad” is indeed schade.

  46. Trond Engen says

    Me: *skandijan- “make shameful”

    Of course, ‘shame’ as decided by the wider society.

    David M.: Indeed, schänden exists and mostly means “rape”. Mostly extinct, but Kinderschänder is still found.

    Wider usage in Scandinavian, then. In Norwegian gravskjender “grave desecrator” and likskjender “corpse desecrator” are quite common and don’t (necessarily) imply sexual actions. The same goes for gravskjending “grave desecration” and likskjending “corpse desecration”. Flaggskjending “desecration of the flag” used to be a thing. It can also be used to form words like kirkeskjender “church desecrator” and decriptions like skjending av nattverden “desecrating of the eucharist”, skjending av klubbdrakta “desecrating the club jersey”, and (somewhat hyperbolically) å skjende offentlig eiendom “to desecrate public property”.

  47. marie-lucie says

    Trond: violate, violation would cover both sexual and non-sexual meanings.

  48. The most common Danish occurrence of the morpheme is in “skændes” (quarrel, argue), which looks like a fossilized mediopassive form of “skænde”. It’s pronounced with stød, though, which isn’t the case for the regular mediopassive form (also “skændes”), and I’m not sure how it got it.

  49. Trond Engen says

    Yes, the corresponding tone 2 would be odd in Norwegian. A lost German prefix (be-, er-, ge-)? Contamination from the verbal noun skænd “scolding”? Two suggestions, and I don’t like either of them.

  50. Lars (the original one) says

    mediopassive — true, according to the ODS. [sg̥ɛnəs] or colloquially [sg̥ɛnˀs], they say, and indeed I think that I would have a stød in most such syncopated forms if they aren’t just a product of allegro speech.

    Danish does not have the cognate to G Schande (it was the LG verb that was borrowed) but does have a zero-derived verbal noun (få) skænd = ‘(receive) a scolding’ — and another, skændsel, which does equate to Schande and shame.

  51. David Marjanović says

    Wider usage in Scandinavian, then.

    The wider usage occurs in literary German of earlier centuries.

  52. Grabschändung, Kirchenschändung und Leichenschändung are also still current as legal terms. But the verb is indeed rare, you’d normally say “x hat Grabschändung begangen” (x has committed desecration of graves), not “x hat ein Grab geschändet”, which would sound bookish / outdated.

  53. Trond Engen says: Yes, the corresponding tone 2 would be odd in Norwegian

    Hold on, I need this explained like I’m a 3-year-old… tone 1 and tone 2 usually corresponds to stød and no stød in Danish respectively, right? So you’re saying Norwegian has a form with tone 1, “skjendes”?

    Lars (the original one) says: true, according to the ODS. [sg̥ɛnəs] or colloquially [sg̥ɛnˀs], they say

    So without stød when the schwa is retained? Interesting, that’s not what ddo ( and my internal lexicon say.
    > [ˈsgεnˀəs] eller [ˈsgεnˀs]

  54. Trond Engen says

    No, sorry, I mistyped. The corresponding tone 1 would sound odd in Norwegian.

    As a stressed monosyllabic word, Norw. colloq. skjenns would be pronounced as tone 1.

  55. Lars (the original one) says

    @dainichi, you can have stød in a short open syllable? I don’t think I can. How old are you? (1960 here).

Speak Your Mind