BECKWITH ON INDO-EUROPEAN.

As I said in my first post on his new book Empires of the Silk Road, Beckwith has an appendix on “The Proto-Indo-Europeans and Their Diaspora,” and I was probably one of the only readers to turn to it first and devour it eagerly. I was, of course, interested in what he had to say about Scythians, Turks, and so on, but I had no expertise in those areas and would have to take his word for a lot of things. I spent the better part of the 1970s immersed in the study of Proto-Indo-European (I was at one point the world’s leading expert on zero-grade thematic present-tense formations in the early IE languages—the topic of my unfinished dissertation, which would surely have been one of the more boring dissertations ever), and I figured if he could convince me he knew what he was talking about in that area, I’d be willing to trust him on the other stuff.
Now, the problem with the study of Indo-European is that the groundwork was done over a century ago, and although exciting discoveries have been made since (notably Hittite and Tocharian), the basic story is still what it was then. You can tinker with the decorations, but the framework was set firmly in place by Bopp, Rask, Grimm, and the other punchily named nineteenth-century forefathers. Don’t get me wrong, there’s no shortage of people who want to tear everything up and connect it all differently (usually to Caucasian or Semitic or Ural-Altaic), but those people tend to have either an insufficient knowledge of the linguistic facts or an excessive willingness to throw the rules of historical linguistics overboard. Actual Indo-Europeanists tend to be commendably but boringly conservative.
Now, this guy is not an Indo-Europeanist by trade, but he’s published on the PIE obstruent system in Historische Sprachforschung (which was known in my day as Kuhns Zeitschrift and has been in business since 1852), and he has an admirable respect for the regularity of sound laws that sets him apart from the wild-eyed theorists. Nevertheless, he’s willing to make sweeping changes to the accepted picture. First, he takes on the notorious problem of the PIE stop system (“a typologically unlikely, if not impossible, phonological system”), saying PIE had “only a two-way phonemic opposition of stops”:

It is a fairly simple matter to show that, as a result, all known Indo-European languages belong to one of three Sprachbund-like groups… Group A, the first-wave languages (with only unvoiced stop phonemes, though there is evidence of the former existence of both unvoiced and voiced stops), consists of Anatolian and Tocharian. Group B, the second-wave languages (with unvoiced, voiced, and voiced aspirate phonemes), consists of Germanic, Italic, Greek, Indic, and Armenian. Group C, the third-wave languages (with unvoiced and voiced stop phonemes), consists of Celtic, Slavic, Baltic, Albanian, and Iranian.

This made sense to me. He then goes on to the issue of Avestan, which causes all kinds of problems in trying to establish chronology and interrelationships, and brings up a point that bothered me back when I studied the language: “it has been remarked, ‘The Avestan speech is very closely related to Sanskrit,’ so astonishingly close, in fact, that ‘we are able to transpose any word from one language into the other by the application of special phonetic laws.’ Avestan’s extensive case system and verbal conjugation system is not just similar to that of Vedic Sanskrit, it is almost identical to it. That is extremely odd.” He gives an example of an Avestan sentence rendered mechanically into Sanskrit: təm becomes tam, yazatəm becomes yajatam, and so on. This “astounding closeness” is what makes people reconstruct a separate branch descending from a Proto-Indo-Iranian language. Beckwith has a different solution, one that I found immediately convincing: “Avestan looks less like an Iranian language than like a phonologically Iranized Indic language. The many inexplicable problems of Avestan… can be accounted for as an artifact of Iranians having adopted an oral religious text… from an Old Indic dialect. As required of Indic religious practitioners, they memorized it exactly, but… it underwent specifically Iranian sound shifts in the mouths of the Iranian-speaking oral reciters.” This allows us to get rid of the separate Indo-Iranian branch entirely and makes the overall picture more plausible.
He ends the appendix with an explanation of why the early IE languages should be regarded as creoles and how this affects the historical picture; I won’t try to summarize it, but again, I found it thoroughly plausible, and I decided I was willing to give credence to his approach to subjects I wasn’t as well acquainted with.

Comments

  1. Siganus Sutor says:

    Hmmm, some people might need a bit of explanation about what a “stop phoneme” is — in IE creole if possible.
    And one cannot but wonder how there could be “first-wave”, “second-wave” and a “third-wave” languages. First with respect to what? It sounds as if there was some unmovable block somewhere that just sent wave after wave to the big wide world. How do we decide that this particular block, this particular group, this particular bunch of people, is considered the source of all waves? In my mind it would rather be a dynamically interdependent system of languages. At that time I can’t imagine it to be centralised as for instance Latin might have been during the time of the Roman Empire.

  2. G.P. Advert says:

    I find it hard to see how the different sound changes among the various IE languages happened without some sort of non-IE language contact, whether it was creolization or or something not quite so radical.
    The caste system and grammaticization of religious texts seen in ancient Indian culture, for example, smacks of a purposeful retention of social power by means of linguistic and religious elitism by a nobility in the context of an “other,” possibly conquered, people.
    That this isn’t an already-obvious part of IE studies confuses me, and I fully admit to probably having not read the “right” literature.
    On the linguistic side, however, we have many quite elaborate sound change schemas for different IE-derived language groups, but little (or at least not enough) commentary as to why or how things happened how they did.
    Does anyone have an recommendations for works that cover socio-cultural or archaeological aspects of IE studies in light of phonological change?
    I’m familiar with Green’s Language Change and History in the Early Germanic World and Mallory’s In Search of the Indo-Europeans, as well as a some less academic, more popular looks at the topic, but would love some suggestions of interdisciplinary academic works on different branches of IE.

  3. Interestingly, this is precisely what Spengler (he of Decline of the West), of all people, believed of Avestan, but since he was merely a polymath and not a credentialled linguist, nobody, as far as I know (except yours truly) has ever paid him much mind. As a trained South Asian linguist who has studied both Sanskrit and Avestan, I found Spengler’s offhanded comment to this effect in his discussion of Persian culture intriguing and insightful.

  4. Not sure if you’re there yet, but I wonder what your response will be to the pages and pages of ranting against modern European art. He spends a paragraph on Timurid Herat and huge sections on why Picasso and Eliot ruined the world.

  5. Hang on. If he says PIE had “only a two-way phonemic opposition of stops”, what’s his explanation for the “second-wave languages (with unvoiced, voiced, and voiced aspirate phonemes)”? Where does he think the extra contrast developed from?

  6. I find it hard to see how the different sound changes among the various IE languages happened without some sort of non-IE language contact
    Precisely his point.
    That this isn’t an already-obvious part of IE studies confuses me
    Like I said, IE is an extremely conservative field; when it started out, nobody knew much about linguistics as we now understand it, and it’s pretty much stayed in its own backwater since, largely ignoring what other linguists have taken for granted for a long time now. It’s high time it caught up with at least the late 20th century.
    Hang on. If he says PIE had “only a two-way phonemic opposition of stops”, what’s his explanation for the “second-wave languages (with unvoiced, voiced, and voiced aspirate phonemes)”? Where does he think the extra contrast developed from?
    He says the two voiced series “reflect the history of a temporary allophonic distinction that later became phonemic in some of the daughter languages, though in all attested languages the unnatural system has been changed to a natural two-way or four-way opposition of stops. The distinctions therefore can be reconstructed only for a temporary, convergent group…” For the details you’d presumably want to read his “On the Proto-Indo-European Obstruent System,” Historische Sprachforschung 120 (2007): 1-19.

  7. I’ve known Chris Beckwith nearly 40 years, since his student days when I was junior faculty in the Department of Uralic and Altaic Studies at Indiana. His work has been solid and transformative from beginning to now! He has demonstrated that his selection for one of the MacArthur “genius grants” was most appropriate. That said, I would have liked to have seen more of the Eastern Turks and Uygurs in this current tome, but, nevertheless, it is another important work in bringing world history’s focus to a critical set of cultures long ignored in “great, sedentary culture” historical prejudice!

  8. I’m delighted to hear that, and if anyone deserves a MacArthur grant, it’s certainly him.

  9. Lugubert says:

    Hi Hat,
    Dead languageas are awfully dead in Swedish academia. My late professor of IE comparative philology is still here and there cited on his work on laryngals. Do those have any bearing on any “four-way opposition of stops”?

  10. Wow! Those last two pieces of spam are almost worth keeping. Clearly at least one computational linguist has a well-paid job on the dark side.
    I’d like to take a look at that article sometime and see what conditioning factors he proposes. But just going by the quote, it would seem more natural to suppose that the distinction was phonemic in proto-[IE minus Hittite and Tocharian] than to suppose that a “temporary allophonic distinction” independently became phonemic 5 different times.
    Also, that term “creole” gets used to mean several different things – what does he mean by it? I assume he doesn’t mean that they’re descended from pidgins, which is the most familiar meaning of the word to me.

  11. Lameen: In his other writings Beckwith seems to use “creole” as a term to designate any language which underwent heavy language contact.
    As for his claim that Avestan is basically an “Iranized Indic language”…well, it is certainly interesting, but I’m dubious: from what little I know of Iranian historical linguistics, ALL other Ancient and Modern Iranian languages exhibit a morphosyntax which is either as Indic-like as that of Avestan (Old Persian) or which can be reconstructed as once having been as Indic-like. I’ll suspend judgment until I read the book: if his other writings are any guide, Beckwith will certainly have some powerful arguments to offer.
    Hat: I think you’re a little bit too harsh on Indo-Europeanists, some of whom (William Schmalstieg, Kenneth Shields) have proposed some rather radical new theories: likewise the whole “Glottalic theory” sought to account for the typological oddities of the Indo-European stop system. I don’t remember who it was who claimed that over the past two centuries no language has changed more than Indo-European…

  12. Well, I haven’t kept up with the field in some time, so I’m perfectly willing to grant that my view of it is out of date.

  13. I have just started studying about the history of the Turks and was amazed at how accurate it was.

  14. The Turks are among the most accurate of peoples, and their history is no different.

  15. The Turks are among the most accurate of peoples, and their history is no different.

  16. In short, LH’s view that IE studies is very conservative, is itself very conservative. Zing!

  17. In short, LH’s view that IE studies is very conservative, is itself very conservative. Zing!

  18. marie-lucie says:

    LH, I join the round of applause for those two new postings about Central Asia and the perspectives this scholarship brings to history and linguistics. I can’t wait to get hold of those books. I was vaguely familiar with the name Chris Beckwith but the volume and breadth of his scholarship are incredible.
    I find it hard to see how the different sound changes among the various IE languages happened without some sort of non-IE language contact
    This is my position too. (I consider myself a historical linguist, although not an Indo-European specialist).
    I think that the reason why the position that IE languages were adopted by different populations (rather than spread through the expansion of a unified population) is not the officially favoured, default hypothesis is that it was overused in the early days of historical linguistics, before the study of sound change became stricter with the “Neo-Grammarians” (actually Junggrammatiker) in the late 19th century: many features thus far unexplained used to be attributed to “substrate influence”, something which conveniently avoided other attempts at explanation, while there was no way of determining what the “substrate” (the language spoken before the presumed adoption of IE) was like. However, whether a given phonological (= sound) change (or series of changes) is to be attributed to internal evolution in a particular population or to imperfect learning of a language by people speaking “substrates” with a different phonology (sound system) is irrelevant to the fact that this change occurred according to a “sound law” which is valid for that language.
    For instance, many words which had initial f in Latin have h in Spanish (eg facere > hacer ‘to do or make’, farina > harina ‘flour’), and this change, which was still in progress at the time of Cervantes (Sancho Panza uses h where Don Quixote uses f), is attributed to Basque influence, since the Basque-speaking territory was shrinking as Basques far from the coast were switching to Spanish, and Basque did not have an f but did have an h. Whether this historical origin is true or not (since the change could have occurred indepently) is irrelevant to the demonstrated existence of the historical “sound law” f > h in Spanish. Similarly, whether the Indo-European “families” such as Germanic, Slavic, etc are due to expansion of IE speakers themselves or to the (far more likely) adoption of IE in some form by speakers of other languages (something which points to some form of IE dominance, just as the existence of languages historically attested to derive directly from Latin results from Roman dominance, not mass migrations of Romans) is irrelevant for the existence of the various sound laws connecting the present-day languages to the presumed ancestor. On the other hand, for French, which is very different in sound from Spanish or Italian or even Occitan, which have remained closer to Latin, many sound changes are credibly attributed to the “superstrate” influence of the Germanic language of the dominant Franks after the breakup of the Roman empire: the Franks did not impose their own language but learned the then variety of Latin with a Frankish accent, which I presume became the prestigious one among the population, who by then spoke a form of Latin. But the individual changes themselves are not peculiar to the evolution of French but have occurred in other language families.
    Another reason for the neglect of possible “substrate” influence is that since the Neo-Grammarians (who codified the principles for the study of phonological change) there has been a strong emphasis on sound rather than morphology (word-formation), even though morphological resemblances were a key factor in first determining the relatedness among the languages of the IE group: after that was done (at least for the major features), tackling the evolution of sounds was the main problem as there were so many little details to take into account. In general, morphology changes much more slowly than phonology within the same language, but when a population adopts another language (under political, economic, etc pressure), the speakers bring to it not only their own sound system, which affects their “accent” in the new language, but also their own system of word-formation, which includes not only the way of forming single words (eg happiness or happily from happy) but also the way those words change to form noun plurals, verb tenses, etc. It is not hard to learn new single words, but much harder to get used not only to a new way of forming noun plurals, etc, which new speakers might expect to have to do, but to some of the very concepts embodied in such word forms. For instance, English speakers learning French or Spanish have a hard time learning the many verb forms in those languages, not only because of the large number of such forms, and many of them are irregular, but because the learners are unused to that type of verb formation, while a French person learning Spanish or Italian is not surprised to have to learn those verb forms because the system is largely the same in all three languages. In Russian as in Latin the cases are a stumbling block for people unused to those changes in nouns.
    When many members of a population starts using the new language between themselves, they tend to avoid the more difficult parts of the new language, especially if this language makes fine distinctions which do not exist in their own language (eg between different kinds of past tense, or between dual and plural, etc). Thus the loss of an ancient distinction can be a clue to language shift (but it can also be due to other, unrelated causes).
    Hundreds or thousands of years ago anybody having to learn a version of Indo-European would have faced similar obstacles (and there were no grammars or dictionaries to help the learner at that time), so one could presume that languages which have a drastically changed (and usually simplified) morphology were built over substrates which were quite different from IE in this respect (this is a well-known factor in the formation of creoles, which preserve the vocabulary – with different pronunciation – but not the morphology of the dominant language). However, sound changes internal to the language can interfere with the morphology as well: Old English had cases and verb endings which were lost later because the endings became more and more similar to each other as time went on, so that they no longer had a function and were abandoned altogether. Whether the arrival of Norman French (as a “superstrate”) contributed to this process is debated, and in any case English speakers did not adopt any part of, for instance, the complex French verb system. Even more drastic changes are reputed to have occurred in Chinese, where apparently Old Chinese was more similar to IE in structure, while the modern Chinese language varieties have lost all their morphological “appendages” (I don’t know the details as I don’t know Chinese, but “similarity” is a relative term, not necessarily presuming genetic relationship). China still has a number of ethno-linguistic minorities and no doubt had many more in the past, which have long been absorbed at least linguistically but probably contributed to the regional differentiation of Chinese into several languages.
    All those factors (plus many other possible sources of error) have to be considered when trying to reconstruct the distant linguistic past, and that is one of the reasons for “conservatism”. Many of the challengers (eg Colin Renfrew, an archeologist known for his views on the origin and spread of the IE languages) have limited or non-existent training in historical linguistics and are not aware of all the possible pitfalls (as well as the possibilities) of historical linguistic methods. This does not mean that well-trained Indo-Europeanists cannot be open to new approaches. In addition to the ones who have been working on alternative explanations to the unusual sound-system reconstructed fot PIE, there is for instance the respected historical linguist Theo Vennemann in Germany, who has been trying to reconstruct a substrate for Western Europe. I find his work interesting enough to try to keep an eye on it, although I am not sure about how far I could agree with his conclusions.
    (Sorry if I might seem to preach to the converted or the knowledgeable – I sometimes get carried away).

  19. Please feel free to get carried away more often; that was a very enlightening comment! I’m still wrapping my head around how all this might have worked, so it was a help to me, and I’m sure to others.

  20. marie-lucie says:

    Thank you, LH. I am sorry about the typos – my comment turned to to be so long that I didn’t catch them all.

  21. Yes, thank you, M-L. The kind of thing we hope for here.

  22. Yes, thank you, M-L. The kind of thing we hope for here.

  23. marie-lucie, that was fascinating. More than that, it was very clear and easy to follow for a non-academic. Often it seems that we are presented with a choice: material that is academically rigorous and too dense for general consumption OR material that has been quite literally “dumbed down.” Your comment seemed to be neither. You comparisons and analogies made it easy to understand, but there was no whiff of “historical linguistics for dummies”. Thank you so very much for a fascinating read.

  24. Marie-Lucie–
    I found your comments fascinating, but I must disagree with you on a couple of points: first of all, I have no problem believing that the sound changes which separated the branches of Indo-European from one another took place with no external influence whatsoever: there are many well-known instances of radical sound changes which took place in contexts wherein we may exclude the operation of any external influence: the Great Vowel Shift in the history of English, or the extreme sound changes (loss of diphthongs, affricates, final consonants, final schwa and /h/) which separate Old from Modern French, are excellent examples.
    Second, I think one factor which is forgotten is that attempts to prove substrate influence in the case of language families such as Romance (whose expansion was definitely due to elite dominance rather than demographic displacement) have yielded such disappointing results (the /f/ to /h/ shift in the history of Spanish, for example, is still on life-support, unlike such theories as the [allegedly Gaulish-influenced]Gallo-Romance shift of /u/ to /y/ or the [allegedly Etruscan-influenced] GORGIA TOSCANA, both of which have been quite thoroughly demolished) that Indo-Europeanists can scarcely be blamed for not wishing to explain the break-up of Indo-European through recourse to (basically unknown) substrata, considering how unfruitful such an approach is even when a great deal is known about the substrate languages (as in the Romance case).

  25. marie-lucie says:

    Etienne: thank you for your well-informed comments. I agree with you on some of them, not entirely on others.
    I have no problem believing that the sound changes which separated the branches of Indo-European from one another took place with no external influence whatsoever
    This is (and theoretically has to be) the default position in the absence of proof of the contrary, and it is true (as I did mention) that the various sound changes are individually attested in all sorts of other languages, simply because of the structure and functioning of the human vocal tract, which is predisposed toward certain articulations, so that there no need to appeal to other influences to explain the phonological evolution of particular languages away from the PIE ancestor: there is no such thing as a typically Indo-European sound-change, only changes which have happened within IE.
    However, this default position, although justified from the point of view of purely linguistic history and the techniques of proto-language reconstruction (as I mentioned also), is very unrealistic in sociohistorical terms since it would imply that the expansion of IE took place in a region devoid of inhabitants, or from which all the previous inhabitants had swiftly disappeared, or that those inhabitants had survived and adopted the new language without retaining any trace of their former linguistic habits. A recent example of substrate influence is the typical pronunciation(s) of English in India and Africa (also a product of elite dominance rather than mass migration) which definitely reflect(s) in greater or lesser detail the phonological patterns of the local languages. A less well-known example is the retention of Spanish (written) ll with a pronunciation distinct from that of y in South American countries where Quechua (which makes this distinction) is widely spoken: here the pronunciation of the local or substrate language has prevented a change (the merger of ll and y which has affected the dominant language in most or all other areas where it is spoken.
    there are many well-known instances of radical sound changes which took place in contexts wherein we may exclude the operation of any external influence: the Great Vowel Shift in the history of English, or the extreme sound changes (loss of diphthongs, affricates, final consonants, final schwa and /h/) which separate Old from Modern French, are excellent examples.
    It is true that there is no reason or possibility for appealing to external influences in these cases, but the changes only seem extreme if seen in the aggregate: individually they are not particularly exotic or extreme. Old French complex segments such as diphthongs or affricates were not lost from the words in which they occurred, they just simplified to single segments. The genuine OF losses are not extreme or rare either: Spanish has also lost /h/, and some Spanish dialects are on their way to losing final or preconsonantal s and unstressed final o. The English Great Vowel Shift involved most of the long vowels: different evolution of long and short vowels is common (for instance between Latin and most Romance languages). Again, in this case the process of vowel change was carried further than in many other languages (since it affected more of the vowels), but vowel changes following a certain general direction are not rare. The generalized nature of phonological change is the main reason why the classification of languages into families cannot rely simply on phonological criteria.
    attempts to prove substrate influence in the case of language families such as Romance … have yielded such disappointing results
    True, but there are reasons for this (see below).
    the /f/ to /h/ shift in the history of Spanish, for example, is still on life-support
    The theory of a Basque origin for the Spanish change is bolstered by the fact that the same shift occurred in Gascon, the Occitan variety closest to the present-day Basque country, as in Fr fait, Oc feit, fech (etc), but Gasc heit (with a true [h[), Sp hecho 'done, made, fact'. The word Gascon (earlier Guascon) comes from the Latinization Vascones (where v = [w]), where the [wask] root = “Basque”, and similarly Vasconia > Fr Gascogne for their territory (in the region called Aquitaine, a triangle between the Pyrenees, the Atlantic and the Garonne river. A small portion of this earlier territory, on the Atlantic coast next to the Spanish Basque country, is still unofficially Basque-speaking).
    The other two examples you give were inspired more by nationalistic than linguistic factors, I think. The supposedly “Gaulish” shift may have been an attempt to find a homegrown French origin for the existence of /y/ [written u] as opposed to the origin of the same sound [written ü] in German. For the gorgia toscana* (attested only from a few centuries ago), a stigmatized dialectal feature in the otherwise prestigious Tuscan speech area, Etruscan would have provided an illustrious precedent, even though the explanation would have required a shift dating from the Latin of early Roman times and continuing unchanged through many more centuries, involving consonants (fricatives) which are prone to weaken and disappear between vowels.
    *la gorgia toscana: this term refers to a feature of the Italian dialect spoken in and around Florence, the capital of Tuscany, which was formerly Etruscan territory. In this dialect the consonants /p, t or k/ between vowels are pronounced as “fricatives”, most typically /k/ like “h” or German “ach”, less commonly also t like English “th”. Etruscan was thought to have such consonants, but now they are generally considered to have been aspirated stops instead, eg [k-h] not German “ch” for general Italian /k/. [Wikipedia gives a very technical explanation. The English article is translated from Italian and gives Italian references, but there is also a very good book in English by Herbert Izzo].

  26. McWhorter argues that you can spot the traces of comparatively recent creolisation by the combination of three prototypical features:
    1. few or no inflectional affixes;
    2. no use of tone for morphosyntactic functions, nor for distinguishing monosyllabic words;
    3. little or no non-compositional derivation.
    If correct, this would mean that we can tell that a language must have been through one type of contact situation even without knowing anything about its substrate/s or history. I can’t think of any early Indo-European language that fits this bill (2 yes, 1 and 3 no.)
    Otherwise, it’s not easy to think of ways that you could convincingly demonstrate contact effects without knowing anything about the substrate. One thing you could do is find a phoneme every occurrence of which is in a word with no cognates among its more geographically distant relatives, like ħ and ʕ in Kurdish (though other factors can easily mess that up.) Has this been observed in any early IE language?

  27. The southern “dialects” of Chinese like Hokkien, Hakka, etc. have also been explained by contact with a non-Chinese language. This makes sense to me because the newest areas of Chinese (in the South) have the most diversity, whereas in English (and Spanish?) the oldest areas (earliest-settled areas) have the most diversity, which commonsensically seems like it should be the norm.
    On the other hand, the North of China has been depopulated and repopulated many times by invasions and civil wars, so maybe diversification was impeded by the lack of stable settlement areas.
    I’ve mentioned this before, but it’s relevant here, so I’ll repeat. I’m not a linguist at all, as most people here know.

  28. The southern “dialects” of Chinese like Hokkien, Hakka, etc. have also been explained by contact with a non-Chinese language. This makes sense to me because the newest areas of Chinese (in the South) have the most diversity, whereas in English (and Spanish?) the oldest areas (earliest-settled areas) have the most diversity, which commonsensically seems like it should be the norm.
    On the other hand, the North of China has been depopulated and repopulated many times by invasions and civil wars, so maybe diversification was impeded by the lack of stable settlement areas.
    I’ve mentioned this before, but it’s relevant here, so I’ll repeat. I’m not a linguist at all, as most people here know.

  29. “The supposedly “Gaulish” shift may have been an attempt to find a homegrown French origin for the existence of /y/ [written u]”
    M-L, there is something similar in both Welsh on the one hand and also Irish and Scots Gaelic – it’s not exactly the same, but there is fronting. It can’t very well be an inherited trait because it’s fairly new (in this context) – is it just an areal feature? Is there any way to determine where this shift started in French – Brittany and down the coast maybe?
    “so maybe diversification was impeded by the lack of stable settlement areas….”
    Diversification wouldn’t be impeded, just not maintained, under these conditions.
    I saw a good article detailing Hmongic influence in the dialects in Hunan or whatever provinces take in the area of the old Chu kingdom, and the cherry on the top was the non-linguistic evidence – in classical times there was this sense that Chu wasn’t really, truly Chinese. And then here comes this linguistic evidence that shows this substrate. Is that crcular or mutually supporting?

  30. Boy, is this a wonderful thread, taking the mind zinging in so many directions it’s hard to follow the arguments. I’ve read it twice now and I will read it again, but in the meantime:
    1) Marie-Lucie: I’ve long wondered about the Latin f > Spanish h change (I’ve forgotten the use of various kinds of parentheses): was the h velar, pharyngeal or glottal on the one hand, or was it bilabial (as in Japanese) and related to bilabial fricative (affricate?) b/v?
    2) Why are North American linguists so prejudiced against the strata explanation (which seems obvious to me, possibly through Menendez Pidal’s influence), when Americanists have had it staring them in the face when American aboriginals speak English?
    3) Is it possible we have all had the wrong idea about creolization and should consider Derek Bickerton’s discoveries instead of reacting in prejudice?

  31. I think Etienne and Marie-Lucie are a little hasty to declare unilaterally that the expansion of Latin was due to elite dominance rather than demographic displacement. Most of Romance speaking Europe consists of areas – Gaul, Cisalpine Gaul, Iberia, the Danube Valley – to which Roman (or Italian) colonists migrated in fairly large numbers as administrators, merchants and farmers. And don’t underestimate the amount of devastation the Romans wrought on the native populations of those areas over the course of the last two centuries BC. The expansion of Latin in Italy itself was arguably mostly due to elite dominance but I think population change may have well played a significant role in other parts of the Western Roman Empire.

  32. I’ll try to give an answer to everyone, and apologize in advance for forgetting that “brevity is the soul of wit”.
    Vanya: while the initial Roman conquest was indeed often bloody, it must be remembered that the “Roman” colonists were often not themselves native speakers of Latin: also, the shift to Latin plainly took place very gradually, allowing whatever demographic losses the conquest had entailed to be erased.
    Marie-Lucie: the /f/ to /h/ shift is also found in many varieties of Aromanian and Southern Italian, which cannot have had a Basque substrate, obviously: furthermore, there is a matter of chronology: loans from Germanic with initial /h/ have zero in Castillian, which suggests that at the time of those loans the /f/ to /h/ shift had not yet taken place. Moreover, the shift in Spanish did not take place if /f/ was followed by /ue/ (from Latin stressed short /o/), but the diphthongization of this Latin phoneme is a specifically Castillian innovation, not found in its neighbors (Gallego-Portuguese, Catalan or Gascon), which again suggests the shift to /h/ took place too late to be explicable through substrate influence. Also, Basque itself replaces Latin /f/ by /b/, not /h/, in Latin loans (hence BABA “bean” from Latin FABA), which is especially problematic from the vantage point of substrate explanations, since this sound change is alien to any Romance dialect spoken in the area.
    Don’t forget: the above is one of the *stronger* substrate hypotheses in Romance!
    As for the /u/ to /y/ shift as being due to Gaulish, one its stoutest defenders, Von Wartburg, was a Swiss German scholar: the theory was widely believed by many scholars who were not themselves French. I’m glad you mentioned Izzo, since he has shown in his book (“Tuscan and Etruscan”)that the Etruscan substrate hypothesis is utterly groundless.
    Jim: there indeed are similar phenomena in Insular Celtic languages, but the question is whether there is any evidence for Gaulish having such a sound, and the answer is a resounding “NO”: Gaulish had a vowel system which was nearly identical to the Classical Latin one, except that it had more diphthongs (the fact that Vulgar Latin in Gaul eliminated the Latin diphthongs, despite the presence of a “diphthong-rich” substrate, is something substratophiles have simply ignored). Chronologically, too, there is considerable evidence that the fronting took place far too late (between the sixth and ninth centuries) for it to be explicable by a Gaulish substrate.
    Lameen: to the best of my knowledge there is no Early Indo-European language with a phoneme found only in (etymologically) non-Indo-European morphemes. As for McWhorter’s “prototype” theory, he himself accepts that a creole will, over time, change like any other language and develop such features as tone, opaque derivational morphology and inflection, so that failure of a language to adhere to his prototype doesn’t mean it was never creolized, only that, if it was creolized, said creolization took place long enough ago for its more visible traces to have been eliminated.
    Iakon (your point 2): I think that like many you are extrapolating from the micro-social to the macro-social. What makes the substrate hypothesis attractive is the fact that, indeed, we have all heard L2 speakers of various languages transfer features of their L1 to their L2. But the same prestige difference, whereby speakers of Language A shift to Language B, also means that the next generation of Language B speakers will continually modify their speech, shedding stigmatized features, *especially ones deriving from Language A*. It would only take a few generations for there not to be any significant trace of Language A left.
    And this to my mind explains why so many instances of expected substrate influence are simply not there. Russian expanded at the expense of Uralic over most of European Russia: why aren’t there any Russian dialects which have lost grammatical gender, which have lost voicing in consonants, which have postpositions to the exclusion of prepositions? (To name just a few conspicuous differences between Russian and Early Uralic). Where is the Basque-influenced Romance dialect which lacks grammatical gender, word-initial clusters, word-initial voiced/unvoiced distinctions, and grammatical gender, but which has ergative case-marking and postpositions to the exclusion of prepositions? (To name just a few conspicuous differences between Romance and Early Basque).
    This is why I am skeptical about claims of heavy substrate influence in the history of Indo-European as well as Chinese (Jim, John Emerson): if similarities are observed between Hmongic or Tai-Kadai and neighboring varieties of Southern Chinese, how do we know this is a case of substrate influence? For ought we know these might be instances of local innovations which began in Southern Chinese and thence spread to neighboring languages (along with the huge number of Sinitic loanwords found in these languages).
    Okay, I’ll get off my soapbox now. Thank you for your patience.

  33. Fascinating—I love this kind of discussion! It makes me wish I was still in academia and had kept up with all this stuff.

  34. Bill Walderman says:

    Don’t the homologies in nominal and verbal morphology among the Indo-Iranian, Greek, Slavic, Germanic and Celtic branches of Indo-European, argue against a creolization phase, and maybe also against extensive substrate influence, for at least these branches? Don’t these homologies, suggest continuous development of quite complex systems from a common origin, unbroken by a creolization phase or strong influence of alien grammatical structures? I’m not a linguist (and forgive me if I’ve expressed myself ineptly) but my question is addressed to those who are.

  35. If similarities are observed between Hmongic or Tai-Kadai and neighboring varieties of Southern Chinese, how do we know this is a case of substrate influence? For ought we know these might be instances of local innovations which began in Southern Chinese and thence spread to neighboring languages (along with the huge number of Sinitic loanwords found in these languages).
    That explanation doesn’t seem impossible but I don’t see why it needs to be assumed. But as I said, I’m not an expert on these things. The diversity of the Sinitic languages in the last-Sinified area strikes me as odd.
    A repeated alternation of creolization (simplification with contact) and “decreolization” (elaboration in isolation) would at best make the problem of the search for ultimate “origins” much more difficult, but history does generally tend to be sort of like that (proliferation and decimation).

  36. If similarities are observed between Hmongic or Tai-Kadai and neighboring varieties of Southern Chinese, how do we know this is a case of substrate influence? For ought we know these might be instances of local innovations which began in Southern Chinese and thence spread to neighboring languages (along with the huge number of Sinitic loanwords found in these languages).
    That explanation doesn’t seem impossible but I don’t see why it needs to be assumed. But as I said, I’m not an expert on these things. The diversity of the Sinitic languages in the last-Sinified area strikes me as odd.
    A repeated alternation of creolization (simplification with contact) and “decreolization” (elaboration in isolation) would at best make the problem of the search for ultimate “origins” much more difficult, but history does generally tend to be sort of like that (proliferation and decimation).

  37. marie-lucie says:

    Etienne: my turn to get back on the soapbox.
    I won’t try to argue with Etienne about details of Romance phonological evolution, as he is a Romanist, while I have a general familiarity with the subject but I am not cognizant of all the details or the scholarship. So, readers, on disputed points of fact about Romance languages you are on safer ground with Etienne.
    Just a few other comments:
    la gorgia toscana: After reading a very favorable review of Izzo’s book, I found the book fun to read as well as absolutely convincing. I would not recommend it to people totally uninterested in Latin or linguistics but it might appeal to those with some background in these areas.
    we have all heard L2 speakers of various languages transfer features of their L1 to their L2. But the same prestige difference, whereby speakers of Language A shift to Language B, also means that the next generation of Language B speakers will continually modify their speech, shedding stigmatized features, *especially ones deriving from Language A*. It would only take a few generations for there not to be any significant trace of Language A left.
    Here I think there is a difference according to whether the dominant language is that of a majority or a minority. Indeed minority speakers who adopt the language of a dominant majority will generally get closer to the dominant norm as the generations pass. But if the dominant language is that of a small minority, the new speakers belonging to the majority might develop their own brand of the dominant language, still influenced by the non-dominant majority language, especially if there is no renewed incoming pool of dominant language speakers.
    Take English in India: the language has been spoken for several generations by a part of the population which was in close contact with the dominant British minority but which has not abandoned its own language(s). The case therefore is not language switch but stable bilingualism. Before independence there was a steady renewal of the British minority in administrative, educational, etc positions, but that is no longer the case, and it does not seem that the new English-speaking generations are trying to get closer to a non-Indian-like pronunciation. Instead there is a move toward an “Indian English” norm in both pronunciation and vocabulary, rather than a desire to emulate British, American or other norms from outside India. If for some reason India became isolated from the rest of the English-speaking world, and English continued to be used as a lingua franca throughout the country and later became the dominant language, the Indian English norm would continue to develop on its own, without reference to the once-dominant British norm but probably influenced at least in phonology by the local languages.
    why aren’t there any Russian dialects which have lost grammatical gender, which have lost voicing in consonants, which have postpositions to the exclusion of prepositions?… Where is the Basque-influenced Romance dialect which lacks grammatical gender, word-initial clusters, word-initial voiced/unvoiced distinctions, and grammatical gender, but which has ergative case-marking and postpositions to the exclusion of prepositions?
    To rephrase these questions: why aren’t there mixed languages which are using Russian or Romance words together with wholly Uralic or Basque phonology and morphology? The main reason must be the differential prestige and power of Russian or Latin (or its descendants) versus the Uralic languages or Basque: the non-dominant people need and want to learn the dominant language as well as possible, not to create a language of their own with a mix of the two languages. It might be different with two languages of more or less equal status. The rare examples of mixed languages such as Mitchif (Cree + French)* in Canada or Mednyj Aleut (Aleut + Russian) in the Aleutian islands developed in very specific situations of intermarriage between male outsiders (in the absence of whole communities of such outsiders) and local women, and consequent bilingualism of a sort within the mixed families thus constituted, in which the children felt equally attached to both languages and under no pressure to choose one over the other. This kind of situation is not comparable to that of an imperial or colonizing language superimposed on subjugated populations.
    The attested mixed languages used structural elements of both languages, they did not preserve the entire structure of the local language together with the vocabulary of the imported language, as in Etienne’s hypothetical examples. For simple language shift, “substrate influence” would not necessarily mean preserving key elements of the non-dominant language. It is perhaps more likely to be manifested through a lack or loss of key elements of the dominant language, which new speakers may have failed to acquire if they were too distinct from what they were used to, or if they were felt to be redundant: for instance, persons used to a singular/plural distinction in their own language may fail to see the need for a singular/plural/dual distinction existing in their new language.
    [*Mitchif: this mixed language still spoken among the Metis people of the Canadian Prairies uses French nouns (together with adjectives and articles) but Cree verbs, which have a complex morphology based on quite different principles from French verbs.]

  38. Russian expanded at the expense of Uralic over most of European Russia: why aren’t there any Russian dialects which have lost grammatical gender…?
    I had a Chuvash nanny when I was a kid, and she’s never mastered the use of grammatical gender in Russian, although she has lived in Moscow since the age of fifteen or so. [I know Chuvash is Turkic, but the point is, it doesn't have a grammatical gender.] I strongly suspect that this was the way people spoke Russian in her native village – but not in a city like Cheboksary. Which sort of illustrates Marie-Lucie’s argument that “the non-dominant people need and want to learn the dominant language as well as possible, not to create a language of their own with a mix of the two languages.”

  39. “the /f/ to /h/ shift is also found in many varieties of Aromanian and Southern Italian”
    and in Japanese, as noted, and elsewhere no doubt. But contact-induced changes usually follow the same paths as ones not induced by contact (as Heine and Kuteva argue for grammatical change). In general, an individual contact-induced change can be detected only by compararison to neighbouring languages, not by its “unnaturalness”.
    “which again suggests the shift to /h/ took place too late to be explicable through substrate influence.”
    Why? Surely language shift from Basque to Romance was continuing throughout the Middle Ages, just as it is now?
    “Also, Basque itself replaces Latin /f/ by /b/, not /h/, in Latin loans”
    Good point.
    “if similarities are observed between Hmongic or Tai-Kadai and neighboring varieties of Southern Chinese, how do we know this is a case of substrate influence? For ought we know these might be instances of local innovations which began in Southern Chinese and thence spread to neighboring languages”
    Reminds me of North Africa. Maghrebi Arabic and Northern Berber have nearly identical phonologies – not because either adopted the other’s phonology, but because of multiple different changes. The pharyngeals in both came from Arabic, the pharyngealised ẓ in both from Berber, and the shared vowel system (basically a, i, u, plus ə only in closed syllables) is innovative in both languages. Here, I suspect the primary driver for convergence is not so much a one-time language shift as continued coexistence and widespread bilingualism – in other words, the absence of complete language shift! – over the past 1400 years.

  40. If effects are mutual it doesn’t exactly disprove the language-contact theory. In other words, if Hmong and Chinese meet and converge, producing Hokkien and something like Sino-Hmong, from the point of view of each language it’s a change caused by contact, even though the hierarchal analysis in terms of substrate and superstrate is less convincing since it’s more mutual.

  41. If effects are mutual it doesn’t exactly disprove the language-contact theory. In other words, if Hmong and Chinese meet and converge, producing Hokkien and something like Sino-Hmong, from the point of view of each language it’s a change caused by contact, even though the hierarchal analysis in terms of substrate and superstrate is less convincing since it’s more mutual.

  42. Etienne says:

    All-
    If we keep this up I strongy suspect some lazy professor of historical linguistics will simply print out this thread and have students read it: less trouble than organizing a lecture, and dare I say, more informative and better-organized than many a lecture I’ve attended.
    Bill Walderman (by the way, I can tell you’re no linguist: your question is both logical and clearly stated): you are quite right, the common morphosyntax of early Indo-European languages quite precludes any of them having been a “radical” creole. But creolization needn’t be an “all or nothing” phenomenon: there are languages known as “semi-creoles” (Afrikaans for example), which exhibit an unusual degree of morphological loss but have preserved some inherited morphology. In like fashion, some losses in various branches of Indo-European have been explained as being due to creolization “light” or language contact phenomena: Proto-Germanic, for example, has very simple-looking verb morphology compared to Latin or Greek, and some have argued on this basis that Proto-Germanic must have undergone *a degree* of creolization: Grimm’s law, likewise, has been explained as being due to a speakers of a fricative-rich substrate acquiring an Indo-European language. I’m dubious about both hypotheses, truth be told.
    Marie-Lucie: it seems to me that it is not so much the demographics as the long-term stability of the prestige of the expanding language that is key here. Suppose language A expands at the expense of language B, and that you start out with one percent of the population speaking language A natively, and the ninety-nine percent remaining speaking language B natively. Assume that for every mixed marriage between a speaker of A and a speaker of B you get child native speakers of A, who, having acquired the language from a parent, will have acquired it fully. Assuming speakers of A and B mix freely, it would take less than eight generations for language A to eliminate language B, *without language B influencing language A significantly*. Assuming one generation = thirty years, this means that in less than two and a half centuries language A will be the only one left. Since pre-modern language spreads took place over a longer period of time, the above model shows very clearly that there is no need to assume significant substrate influence just because a language has expanded through non-demographic factors. And it answers my own question as to why there isn’t any Uralic-like Russian (or Turkic-like! Thank you for that piece of information, Sredni Vashtar) or Basque-like Romance dialect.
    Lameen: I entirely agree with you, and am struck by how often people assume this to have been a case of one-way influence, i.e. of a Berber substrate transforming Arabic into Maghreban Arabic. The central difficulty in North Africa is that we have no direct evidence of what Berber looked like before it came into contact with Arabic, just in the same way that we have no direct evidence of what Hmong-Mien and Tai-Kadai languages of Southern China were like before they came into contact with Chinese: I certainly agree with John Emerson that language contact played a major role, but I very much doubt it was a case of Hmong-Mien and Tai-Kadai languages influencing Chinese without themselves being heavily influenced by Chinese. Disentangling the contributions made by each language in these contact situations may well prove impossible: even in the case of the famous Balkan SPRACHBUND, whose constituent languages have far better-known histories than Berber or Hmong-Mien or Tai-Kadai, there remains a lot of uncertainty regarding the origin of various features.

  43. “This is why I am skeptical about claims of heavy substrate influence in the history of Indo-European as well as Chinese (Jim, John Emerson): if similarities are observed between Hmongic or Tai-Kadai and neighboring varieties of Southern Chinese, how do we know this is a case of substrate influence? For ought we know these might be instances of local innovations which began in Southern Chinese and thence spread to neighboring languages (along with the huge number of Sinitic loanwords found in these languages).”
    Right, Etienne. The answers to that kind of question usually come form extra-linguistic sources. in the case of a Hmongic substrate, it’s known for a fact that there were Hmong populations in Central China in the areas where these similarities are found. It’s known for a fact that these lexical similarities are in the non-learned, low perestige areas of the lexicon. Occam’s razor.
    The same goes for Daic influence in Cantonese, or innovations in a variety of Chinese in the South spreading into Daic or Hmongic languages. I can see how there might be some room for debate with lexical items for oh, say…. cooking techniques. Just maybe, but there had better be evidence in Chinese from earlier than the contact. I’m a lot less ready to see it in the case of names of species Daic and Hmongic people have been familiar with, like lychee, linji etc. saying that such a term arose as an innovation in some variety of Chinsese and spread to indigenous langugaes is no different that if you find a word ‘hikkvri’ in Maskoki and a word’hakli’ in Cherokee, or some such, that they are borowings form English ‘hickory’. I think the response that suggestion would get is “Pfoar!”
    “The diversity of the Sinitic languages in the last-Sinified area strikes me as odd.”
    John, that’s not the only diversity in Southern China. There is an intense ethnic animosity between Cantonese and Southern Fujian people. It’s a joke in law enforcement here that if you see a Cantonese and a Fujianese involved in a deal, that’s prima facie evidence that it’s criminal, since decent busisnessmen from those areas will have nothing to do with each other. Northern Fujian and Southern Fujian get along about as well too. So what’s the basis for that? The simplest explanation is that it is simply a continuation of an earlier political situation.
    “Jim: there indeed are similar phenomena in Insular Celtic languages, but the question is whether there is any evidence for Gaulish having such a sound, and the answer is a resounding “NO”: Gaulish had a vowel system which was nearly identical to the Classical Latin one, except that it had more diphthongs (the fact that Vulgar Latin in Gaul eliminated the Latin diphthongs, despite the presence of a “diphthong-rich” substrate, is something substratophiles have simply ignored). Chronologically, too, there is considerable evidence that the fronting took place far too late (between the sixth and ninth centuries) for it to be explicable by a Gaulish substrate. ”
    Etienne, that fits with my question about an areal influence. Those sounds arose late in Welsh, in the middle ages. The direction of influence could be either way.
    “Where is the Basque-influenced Romance dialect which lacks grammatical gender, word-initial clusters, word-initial voiced/unvoiced distinctions, and grammatical gender, but which has ergative case-marking and postpositions to the exclusion of prepositions? ”
    Would you be satisfied with a load of indigenous lexical marerial? Mexican Spanish is more or less grammatically Spanish, sort of, but it is full of local terms from a range of languages, for foods and plants and insects and even socail relationships, and this is true for the rest of the culture too.

  44. A great discussion. On the “creole” issue, I should point out that Beckwith defines the term very broadly: “In this book, ‘creole’ is used to refer to languages that have undergone significant changes due to convergence with other languages, but not the kind of radical simplification of structure that is stereotypically said to characterize creoles”; he later adds “It has been said that ‘all mature languages are creoles’”—a formula that appeals to me.
    more informative and better-organized than many a lecture I’ve attended.
    Indeed, and this is one of those threads that makes me intensely glad I decided to start a blog.
    (by the way, I can tell you’re no linguist: your question is both logical and clearly stated)
    Zing!

  45. “On the “creole” issue, I should point out that Beckwith defines the term very broadly”
    The best definition I have ever heard was that “Creole” means it takes two chickens to feed one family and “Cajun” means you take one chicken a feed two families.
    “but I very much doubt it was a case of Hmong-Mien and Tai-Kadai languages influencing Chinese without themselves being heavily influenced by Chinese.”
    This makes sense to me. I think it pretty weel describes the situation, and I don’t think reference to Tai-Kadai langugaes that suppsoedly have never been in contatc isgoing to help much because that would beg the question of how simialr they had to be to the languages in question.
    Sprachbund – I thought the definition of a Sprachbund was that you couldn’t determine the direction of borrowing. Except when you can guess….
    M-L or somebody, help me with this one. Take a well agreed-upon Sprachbund, Mosan. Let’s look at one defining feature, VSO. Well, Makah verbs have all their junk in the trunk – all their administrative crap – tense/aspect, and so on – on the back, the most awkward place for it to be in a VSO set-up. In North Straits and Lushootseed and maybe most Salish languages, that stuff is either on the front or internal to the verb, where it doesn’t get in the way of the case/voice morphology – in other words, where it belongs. Isn’t it likely that VSO is an overlay in Makah, with the internal structure of the verbs inherited and retained since there is no other conditioning factor apparent, and then on the other hand original to the system in Salish?

  46. Bill Walderman says:

    Beckwith defines the term very broadly: “In this book, ‘creole’ is used to refer to languages that have undergone significant changes due to convergence with other languages, but not the kind of radical simplification of structure that is stereotypically said to characterize creoles”; he later adds “It has been said that ‘all mature languages are creoles’
    Doesn’t this expand the term “creole” to the point where it has no practical utility as a concept?

  47. marie-lucie says:

    Lameen:
    McWhorter argues that you can spot the traces of comparatively recent creolisation by the combination of three prototypical features:
    1. few or no inflectional affixes;
    2. no use of tone for morphosyntactic functions, nor for distinguishing monosyllabic words;
    3. little or no non-compositional derivation.

    Actually this list applies to English:
    1. it has fewer inflectional affixes than other European languages: only one on nouns (plural), two on verbs (-ed and -ing), two on some adjectives (-er and -est);
    2. no use of tones;
    3. little or no non-compositional derivation: this would refer to instances like song from sing, heat from hot, where the difference between noun and verb or adjective is shown by internal vowel change rather than the addition of an affix; this sort of word-formation is very limited in English, and totally non-productive: no new words are formed this way, instead they add suffixes, as in nerd > nerdy > nerdiness.
    Bill W: Don’t the homologies in nominal and verbal morphology among the Indo-Iranian, Greek, Slavic, Germanic and Celtic branches of Indo-European, argue against a creolization phase, and maybe also against extensive substrate influence, for at least these branches?
    The homologies in question are what suggested that the languages were genetically related, that is, descended from a common ancestor. “Creolization” seems to mean different things to different people, and the scope of the word seems to be constantly widening (into insignificance even), but in its stricter sense I personally don’t think that it applies to IE. Substrate influence does not have to be massive to exist, as I mentioned earlier.
    Don’t these homologies, suggest continuous development of quite complex systems from a common origin, unbroken by a creolization phase or strong influence of alien grammatical structures? In the main, yes. Substrate influence could be indicated by more subtle features, but those are very difficult to prove since they could also have developed independently, as both Etienne and I wrote.

  48. marie-lucie says:

    Why did half of my comment disappear? I had written twice as much, but half of it disappeared.
    Here is some of what got chopped off.
    Jim:

    Take a well agreed-upon Sprachbund, Mosan. Let’s look at one defining feature, VSO. Well, Makah verbs have all their junk in the trunk – all their administrative crap – tense/aspect, and so on – on the back, the most awkward place for it to be in a VSO set-up. In North Straits and Lushootseed and maybe most Salish languages, that stuff is either on the front or internal to the verb, where it doesn’t get in the way of the case/voice morphology – in other words, where it belongs. Isn’t it likely that VSO is an overlay in Makah, with the internal structure of the verbs inherited and retained since there is no other conditioning factor apparent, and then on the other hand original to the system in Salish?

    [Mosan is a group of 3 indigenous language families in Washington, Idaho, Montana and British Columbia. Makah is at the tip of the Olympic Peninsula.]
    I don’t know that much about these languages, but I suppose you mean that Lushootseed, etc have the order Prefix-Prefix-Verb-Suffix– Subject noun–Object noun, while Makah has Verb-Suffix-Suffix-Suffix–Subject noun–Object noun. Could it be that the current Makah basic VSO order was originally an alternate word order which became more prevalent under the influence of neighbouring Salishan languages?
    I don’t want to go into technicalities here, but a possible example in English could be if the basic order “I am here” were to be replaced by the alternate, less common order “Here I am” under the influence of a prestigious language where “Here I am” was the basic order. (I see this sort of thing happening in current written French in France under the influence of too-literal translations from English).

  49. marie-lucie says:

    two on verbs (-ed and -ing)
    Sorry, I should have said three on verbs (-ed, -ing and -s). A few verbs also have -en as in given. But three or four verbal suffixes are nothing compared to the number of those of Latin, Spanish or Italian, or even French.

  50. Bill Walderman says:

    “three or four verbal suffixes are nothing compared to the number of those of Latin, Spanish or Italian, or even French.”
    Not to speak of ancient Greek!

  51. Marie-Lucie: Actually, by “little or no non-compositional derivation” McWhorter means semantic non-compositionality, cases where the meaning of the morphologically complex word cannot be determined just from its parts, like return

  52. [lost some of a paragraph due to angle brackets]:
    …cases where the meaning of the morphologically complex word cannot be determined just from its parts, like return from turn or beget from get. But the loss of so many inflectional affixes in English at a period without much contact does suggest contact is not a necessary cause of the properties McWhorter suggests.

  53. marie-lucie says:

    Lameen: One of my disappeared comments made the same point about Etienne’s model of language mixing (or rather, mixed language families) and children raised in the dominant language, etc. Children in such families might speak only the dominant language at home, but unless the family remains totally isolated, an unrealistic situation, they will no doubt acquire the local language just by playing by other children, some of whom will be their own relatives on the side of the parent speaking the local language. Stable bilingualism is just as likely to result under the initial circumstances, even if more and more people acquire the dominant language, as long as enough children are raised as native speakers of the local language. For instance, in Southern France until about 100 years ago there were many French-Occitan bilinguals, especially in the cities, while most of the rural population spoke only Occitan. This situation lasted for centuries: French made inroads among adults, especially men, but all rural children were raised in Occitan until compulsory schooling (in French) made an entire generation of children bilingual, and many of those children later chose to raise their own children in French. Even then, as long as there were living Occitan speakers to interact with (and there still are, as discussed on an earlier thread), the rural population did not become exclusively French-speaking.

  54. marie-lucie says:

    Lameen: Determining where the innovation started is usually harder with grammatical similarities, which by their nature are unlikely to be culture-specific.
    By “grammatical”, do you mean syntactic? syntactic convergence among unrelated languages is quite well documented. Morphological convergence would be less likely unless the languages in contact were already fairly similar in their morphology.

  55. Morphology does tend to be more resistant, but it can be borrowed too. A remarkable case is Siwi, where the Arabic method of forming comparatives from adjectives by imposing a template ‘aCCaC got borrowed (> CCəC); some Cappadocian Greek varieties have even borrowed 1pl and 2pl subject agreement markers, and of course plural morphemes are comparatively easily borrowed, like English -us > -i. Influence can also affect the semantic content of morphology – thus Bulgarian has evidentials probably because of Turkic influence, even though they’re formed from native resources.

  56. “Could it be that the current Makah basic VSO order was originally an alternate word order which became more prevalent under the influence of neighbouring Salishan languages?”
    Maybe. In fact, maybe it had to be that way. Maybe that had to be an acceptable word order in order for the feature to catch on in the first place. Anyway, thanks for your answer.
    Lameen, English is full of similar examples. we had stomachs long before we came in contact with Greek (again). Matisoff wrote a cool monograph on boorowing of body part words.
    Alexandra Aikhenvald has written a to on borrowing and language shift and she says there are lots of levels to this that go way beyond relative prestige of the languages. For instance she says that if some langauges in a region have evidentiary systems, they spread to all the other langauges because bilingual speakers get used to them, they become a general cultural norm in the region, and if you don’t use them, whatever the laguage, you sound like you’re being cagey about your information. She notes that the passive is spreading too, for less obvious reasons.
    She also talks about instances where languages resist borrowings. One example is the way that Iroquoian langauges will accpet some English nouns but very few verbs. The reason – I think it’s her – is that it’s just too much bother to crunch English verbs into Mohawk or Onandaga syntax. Don’t know why – syllables are too heavy, or too many polysyllabic verbs that jam up the process of building a verb complex?

  57. David Marjanović says:

    Ouch. I came way too late, and the comments are not numbered*, so I can’t scroll up 20 times to make sure I’m adding my 2 cents to everything I can say something on. Where to begin…
    * Maybe someone should invite you to ScienceBlogs.com. Unfortunately that has to be done by someone who already has a ScienceBlog.
    – Could someone send me the pdf of Beckwith’s paper? I’m interested in what conditioning factor he suggests for the split he assumes between plain and aspirated voiced consonants.
    – In return, I can send a recent Diachronica paper that explains in some detail what has happened to English grammar: basically, the Vikings came and burninated it. Or rather, large numbers of Norsemen settled in northwestern England and spoke bad English to their children. Both Rather Late Old English and Late Old Norse had grammatical gender, but not the same genders for the same nouns, so the compromise was to basically get rid of it — with the remarkable exception of the personal pronouns; I don’t know if there are other languages without grammatical gender in nouns that distinguish it in personal pronouns. Similar things hold for the drastic slashing of person endings on verbs, while the tense system with most or all of its irregularities was retained because both languages more or less shared it.
    – I’ve heard before, and not from a linguist, that the French [y] is supposed to come from the Franks. I don’t believe it. In German, [y] is part of the Umlaut system, which French never had and which AFAIK only developed during Old High German or something (and, in parallel or in some Sprachbund way, in Old Norse); the Franks can’t have had much of it. A wholesale shift that turns every /u/ into an [y] (or [ʉ] or something) is something completely different – and a very common sound shift, at least in Europe: it has happened in Dutch, Swedish, Icelandic, and Welsh within the last few hundred years, and in Greek long before. The simplest explanation is that it simply happened once more in French.
    – I’m told that various northern Russian dialects have lots of Uralic words. I forgot if some grammatical features have crept in, too (but clearly nothing major like lack of gender).
    – Different people mean different things by “Proto-Indo-European”: sometimes the last common ancestor of the living IE languages, sometimes the last common ancestor of those plus the Tocharian and the Anatolian languages. These two may well have been quite different. I get the impression that not much attention is being paid to this possibility; indeed, sometimes the same historical linguist means both when talking about “PIE”!
    – And indeed, the Anatolian languages had a two-way contrast that was probably a matter of length (like in Swiss German), and Tocharian had no contrast at all. Still, however, that doesn’t mean that the three-way contrast only evolved after Anatolian and Tocharian had split off. I’ve read that Tocharian is full of Uralic features, and most Uralic languages have just three plosives (no voice/aspiration/whatever contrasts), just like Tocharian.
    – While it is true that the plosive system usually reconstructed for “PIE” (see above for what that can mean…) is highly unusual in having voiceless, plain voiced, and aspirated voiced consonants, it is by no means impossible. Over a dozen languages ranging from western Africa to Indonesia are reported to have just such a system – and Sanskrit almost had it: its fourth series, the voiceless aspirates, were very rare. Furthermore, most Sanskrit words with voiceless aspirates are clearly loans, and practically all the rest can be explained as being derived of clusters from plosives + “laryngeals”; the (voiceless) aspirates of modern Korean are thought to have been derived the same way, from clusters of plosives with /h/. The latest word I’ve read on the subject is in Alan Bomhard’s 2007 book on Nostratic, which says that voiceless aspirates were present in PIE in a small number of onomatopoietic words and otherwise absent.
    – Somewhere on his blog, the wrathful Glen Gordon has suggested a hypothesis on how this three-way contrast could have evolved from another three-way contrast, the globally common one between plain, aspirated and ejective plosives (all three voiceless): the ejectives became creaky-voiced, the plain ones became fully voiced, and the aspirated ones became plain voiceless. Later, the new creaky-voiced plosives became fully voiced, and, to stay distinct, the previously fully voiced plosives became breathy-voiced = voiced aspirated – except when they didn’t and instead merged with the new fully voiced plosives, as happened in Celtic and Balto-Slavic, off the top of my head.
    – Does Beckwith say anything about the other oddity of the PIE plosive system? I mean the three-way contrast between palatalized, plain and labialized velars (…which, together with the voice and aspiration contrasts, means there were no less than nine velar plosives). This, too, is rare enough that many IEists are uncomfortable about it, and many have tried to explain it away; but it occurs in Hausa, for example, not to mention in several languages of the Pacific Northwest and the Caucasus.
    – It would surprise me if Grimm’s Law could be attributed to substrate influence. That’s because it seems to have happened very late, based on various evidence: for example, every single Celtic loanword in Proto-Germanic has undergone that shift, and the tribe after which Hessen is named was called Chatti by Tacitus, which indicates [kʰ] or maybe [x], but clearly not [h]. The AFAIK only evidence to the contrary, the inscription on the “Harigast helmet” from Slovenia, is more probably Rhaetic than Germanic… Also, a few hundred years later, basically the same thing happened again: the High German consonant shift, which again turned voiceless plosives into fricatives (or affricates) and voiced plosives into voiceless ones, and this time clearly without any substrate influence. There’s even evidence it started among the Longobards in Italy, where any substrate would (one should think) have had the opposite effect!
    – Basque and Spanish seem to have made two other sound shifts together: the merger of /w/ or /v/ with /b/, and the change of voiced plosives between vowels (and in… most other places) to approximants. Concerning the former, Basque must once have had a /w/, as shown by the Latin spelling (Vascones) and even more drastically by the Ancient Greek one (which started with Ουασκ-; I forgot the rest and can’t bother to get the breathing on the υ).
    – Basque must also be the reason why Spanish, which has lost the contrast in length in all other consonants (preserved in Italian), has kept it for /r/. In Basque, this same contrast between a (very) long and a short /r/ developed when /l/ between vowels changed to (short) [r] but stayed distinct from /r/ (which was emphasized by making it longer); and this change is apparently shared by Gascon, where de la has become dera (to cite the one example I know).
    – This is not to say that convergence doesn’t happen in geographically close languages. If you kindly ignore the fact that Hungarian has vowel length (so prominent that hexameters have been composed in Hungarian), the vowel systems of southeastern German and the neighboring Hungarian are quite similar, despite the almost complete lack of contact except in 19th-century Budapest and Vienna. But things like the Spanish and Basque /r/-/rː/ contrast are just too weird.
    – The geneticists have shown that there was a pretty large immigration from Italy to Spain that was part of the Roman conquest. So, it looks like the development of Spain started with lots of native speakers of Latin (…and maybe other Italic languages…) right from the very beginning.
    – While I am at it, they’ve also found out that both the people and the cattle of Tuscany have come from Asia Minor. Herodotus was right.
    – German provides examples of simplification in the tense system without substrate influence. Standard German has dead obvious homologues of the English past tense and the present-perfect tense, but they are exactly synonymous; the only rules on when to use which one are stylistic and might conceivably have been imported from Latin (no idea if any research has been done on that). In spoken northern German, the past tense is used because it’s shorter, and the present-perfect basically isn’t used; in the south, a round of apocope rendered the past of regular verbs identical to the present (in some persons), so it died out (except for “be” and all but one person of “want”), and the present-perfect is used instead – this goes so far that the past-perfect, which elsewhere looks like in English, was recomposed and now consists of three words instead of two: instead of “I had already done that” we say “I have already had done that” (except of course for the word order).
    – Forget stomach. German Kopf “head” is from Latin cuppa, the ancestor of English cup. Think “noggin”. The expected word, Haupt, survives only in poetic uses and as a prefix that means “main”.
    – And that is part of a wave of dysphemisms that separates Latin from the Romance languages. In this word it spilled over into German. Isn’t that cool?
    – /b/ is (…yet another oddity, though part of the one about the voice and aspiration contrasts…) reconstructed as having been extremely rare, maybe even absent, in PIE. All or practically all Sanskrit words that contain /b/ are clear loans, as far as I know, or at least unique to Indic and absent from the rest of IE.

  58. I thought the pejoratives ‘barbarian’ and ‘savage’ went when the series ‘savage, barbarian and civilized peoples’ was discarded in the same era that ‘frigid, temperate and torrid zones’ was, before my time. It’s the word ‘Berber’, which has the same origin, that should probably go now, as those of that ilk consider it to be pejorative still. It appears to be still used, though, because the suggested substitute ‘Tamazight’ applies to one of those languages. What say, Lameen?

  59. marie-lucie says:

    Lameen: plural morphemes are comparatively easily borrowed, like English -us > -i.
    I disagree that in words like cactus, cacti the plural morpheme is actually borrowed: the entire word is borrowed, in both its singular and its plural forms, but the plural -i has not spread to native English words. Instead, outside of some specialized domains many people use the most general English plural morpheme, hence cactus, cactuses, and similarly for words like uterus and hippopotamus. Similarly, the Greek word skeleton had a plural skeleta, but if that form is used at all in English it must be restricted to technical papers in biology. In general usage the plural is skeletons, just as in buttons. If the plural suffixes -i and -a had truly been borrowed, they would have been generalized to other forms.
    If a word borrowed together with its plural is used mostly in the plural, that plural may be used in English as a singular: common examples are data, bacteria and criteria, which are commonly used as singulars, either mass nouns (data) or count nouns. In Canada we have learned that the correct name of the northernmost indigenous population is the plural Inuit, corresponding to the singular Inuk which is found in the name of the language Inuktitut, but in spite of knowing this theoretically, even well-informed people will most often say an Inuit or an Inuit woman for an individual.
    As an example of a true borrowing, the feminine noun suffix -ess was borrowed from Old French into Middle English. At first the morpheme occurred only in French borrowings such as princess, mistress (OF maistresse) or empress (OF emperesse), but only when -ess was added to completely English words, as in waitress from waiter, could it be said to have been truly borrowed.

  60. David Marjanović says:

    if that form is used at all in English

    It isn’t. In spite of femora, humeri, scapulae, pelves, vertebrae, and so on.

  61. I love this thread!
    Everyone, remember to write &lt; instead of < when you want a < to appear, or not only will the < not appear, the rest of your post may well disappear.
    Substrate hypotheses come in two varieties, the sensible and the loopy. Sensible hypotheses suggest that when a population changes language, it may retain characteristics of the old language in the new. There are ethnically Welsh people, for example, whose ancestors have been speaking nothing but English for centuries, and they have on the whole more substrate features than modern bilingual Welsh/English speakers do (because the latter tend to learn English from standard models, not local ones).
    Loopy hypotheses suppose that when part of a population speaking A adopts a new language B, sound changes that A undergoes subsequently somehow parallel, as if by some sort of shared mystical genetics, the B-speakers whose ancestors spoke A. This of course is absurd when baldly expressed, but it generally isn’t. And substrate-ists are not too careful to divide their proposals into sensible and loopy; what’s more, it’s not unknown for the same person to hold several hypotheses of each type. One could construct a fine loopy hypothesis, for example, for the late 20th-century fronting of /u/ in American English based on a French substrate in New England, Louisiana, and the upper Mississippi.
    As for Northern and Southern Chinese, the traditional explanation is the four-character idiom, “南船北马 nán chuán běi mǎ”: in the South you go by boat, in the North by horse. Southern China is very broken up by its watercourses, and locations that are quite close together physically would have been quite inaccessible from one another until modern times. The gradual breakup of Southern Sinitic, on the other hand, is a fairly normal diversification pattern for unwritten languages.
    In truth, what really needs explaining is the amazing spread, persistence, and uniformity of Mandarin over the whole North and even in Sichuan, where it probably arrived due to population replacement after some depopulating event in the 16th century. There is really nothing remotely like it in human history until the settlements of North America and Australia in the 19th century.
    David: There is nothing, absolutely nothing, to beat how the Germans use their inherited superlative suffix to mark high-value ordinals too: fleissig:fleissigste::dreissig:dreissigste”. Apparently this is a calque on the Latin superlative ending -issimus and the frozen e:simus ending of vicesimus, tresimus, etc.; in Galloromance they had fallen together as -iesme.
    Marie-Lucie: There really is quite a lot of productive Romance morphology in English, though. In Dutch, the Romance bound morphemes are only attached to Romance-derived stems, but English has rewrite and remake as well as return and redress, printable as well as checkable.

  62. marie-lucie says:

    David: thank you for your erudite comments.
    Late Old English and Late Old Norse had grammatical gender, but not the same genders for the same nouns, so the compromise was to basically get rid of it — with the remarkable exception of the personal pronouns; I don’t know if there are other languages without grammatical gender in nouns that distinguish it in personal pronouns.
    The vast majority of nouns refers to inanimates, where grammatical gender does not have real-world counterparts as masculine and feminine do among sexed beings (humans and animals), so there is no necessary compatibility between genders of ordinary nouns in different languages. But pronouns referring to people are in constant use, and the attribution of names and terms of address would not vary in gender from language to language, so even though grammatical gender disappeared from nouns, it was kept for the pronouns which most often referred to people, thereby enforcing the previously loose link between sex and gender. For a (grammatically) genderless language to develop gender for pronouns where it did not have gender before could probably only happen through contact, but for instance Hungarian did not borrow such a system from its gender-ful neighbours, nor did Basque from Spanish.
    [geneticists]: they’ve also found out that both the people and the cattle of Tuscany have come from Asia Minor. Herodotus was right.
    Not just Herodotus: a number of ancient sources were saying the same thing, against just one that contradicted it, yet that single one was for a long time officially considered the correct one. It is one thing to say that at some point the Etruscans developed a taste for “Oriental” imports (brought from Phoenicia), but how to explain the sudden, independent development in Italy of such improbable, esoteric customs as divination through the observation of fresh livers (as in Babylonia, with the same liver models in clay), unless the users brought the custom with them?
    A wholesale shift that turns every /u/ into an [y] (or [ʉ] or something) is … a very common sound shift, at least in Europe: it has happened in Dutch, Swedish, Icelandic, and Welsh within the last few hundred years, and in Greek long before. The simplest explanation is that it simply happened once more in French.
    It also happened in Occitan, even though Southern Gaul was not much affected by the Franks who had such an impact on Northern Gaul (later renamed “France” because of them). This is one of the many sound changes which could happen in just about any language.
    The same shift is actually creeping into some English varieties without any substrate help: I often hear almost “güüd füüd” for “good food”, for instance. Some time ago a woman on Canadian radio was reporting from Baghdad about bombs going off: “Suddenly we heard a big [büüm]“, she said, the decidedly front vowel totally defeating the onomatopeic effect of the intended word “Boom”.
    skeleta, if that form is used at all in English
    - It isn’t. In spite of femora, humeri, scapulae, pelves, vertebrae, and so on.
    But these forms are hardly part of general vocabulary. The average English speaker would say femurs not femora, which is used only by the medical profession and other anatomists. I have read vertebrae for both singular and plural, but I don’t read medical journals.

  63. marie-lucie says:

    Siganus: It is high time someone should respond to your questions.
    some people might need a bit of explanation about what a “stop phoneme” is — in IE creole if possible.
    (sorry, no IE creole available at this point)
    A “stop” (also called “plosive”) is a consonant sound produced from the sudden opening of the mouth cavity after the exhaled air has been stopped from leaving the mouth. It might seem that this would only occur before opening one’s lips, producing the sounds p, b, but the tongue can also completely block the outgoing air for the sounds t, d and k, g, among others. Some other consonant articulations let the air leave the mouth in spite of some interruption of the air flow, like the “fricatives” f, s, sh, th and others (there are yet more consonant types, look up “phonetics”).
    A phoneme is a sound that is distinctive from other sounds in a language, for the purpose of forming words: in French you would be understood whether you said une rose or une rothe (using English th), but in English it matters whether you say sing or thing: English has the two phonemes /s/ and /th/ (a sound represented by the proper symbol, the Greek “theta”) but French only has /s/, and saying th instead does not interfere with comprehension, it just “sounds funny”. In contrast, pronouncing f instead of s, or the opposite, in either language would lead to serious misunderstanding, because /s/ and /f/ are distinct phonemes in both.
    And one cannot but wonder how there could be “first-wave”, “second-wave” and a “third-wave” languages. First with respect to what? It sounds as if there was some unmovable block somewhere that just sent wave after wave to the big wide world.
    If there was a single language that differentiated into several others through migration of at least some of its speakers in different directions, several such migrations could have occurred, first from the original location, then from that one again or from secondary locations. The latter happened for instance in Polynesia. For the Indo-European languages, the location of the original centre of dispersal is still disputed but the most common position mentions somewhere in or near the Ukraine, while another, not as common one favours Anatolia (other less likely places have also been advanced).
    How do we decide that this particular block, this particular group, this particular bunch of people, is considered the source of all waves?
    It is very difficult, and one problem in IE studies is that linguistics and archeology, geography or history do not always give the same answers. (See for instance Mallory: In search of the Indo-Europeans). But at present there is no language or group of people that is considered closer to the original PIE language or speakers than another.
    In my mind it would rather be a dynamically interdependent system of languages. At that time I can’t imagine it to be centralised as for instance Latin might have been during the time of the Roman Empire.
    Many linguists think that Proto-Indo-European (the reconstructed ancestor) was not a single language but a complex of dialects (= language varieties), since as you say there was no centralized authority at that time. Whether PIE was also a variety of some supra-familial form of language is also debated.

  64. marie-lucie says:

    JC: There really is quite a lot of productive Romance morphology in English, though. In Dutch, the Romance bound morphemes are only attached to Romance-derived stems, but English has rewrite and remake as well as return and redress, printable as well as checkable.
    Absolutely, but a plural morpheme is part of inflectional morphology (making different forms of the same word) rather than derivational morphology (making different words), as with the suffixes -ess and -able and the prefix re-. Inflectional morphology is much less likely to be borrowed than derivational morphology, of which there are many examples. That’s why it was hard to find an inflectional example of a borrowed morpheme.
    There is one actually in English: the borrowing of Norse -s for the 3rd person singular of words, in replacement of Old English -th (still used in the KJ Bible: The Lord giveth …, etc). I think that this occurred because a) the languages were very close to each other at the time and b) the sound s is more audible than the sound of th.

  65. Both Rather Late Old English and Late Old Norse had grammatical gender, but not the same genders for the same nouns, so the compromise was to basically get rid of it — with the remarkable exception of the personal pronouns.
    My memory of the pronoun systems of Middle English is that they were totally confused from one ME dialect to another and from one century to another.
    I am reminded of the “tundish” (= “funnel”) in Joyce’s POTAAAYM, which first figures as an Irish word unknown to the English, but turns out to be an Anglo=Saxon word which survived in Irish English, but not English English.

  66. Both Rather Late Old English and Late Old Norse had grammatical gender, but not the same genders for the same nouns, so the compromise was to basically get rid of it — with the remarkable exception of the personal pronouns.
    My memory of the pronoun systems of Middle English is that they were totally confused from one ME dialect to another and from one century to another.
    I am reminded of the “tundish” (= “funnel”) in Joyce’s POTAAAYM, which first figures as an Irish word unknown to the English, but turns out to be an Anglo=Saxon word which survived in Irish English, but not English English.

  67. I don’t know if there are other languages without grammatical gender in nouns that distinguish it in personal pronouns.
    WALS doesn’t seem to have the right features to call any others out. Closest I can see is cross-plotting Gender Distinctions in Independent Personal Pronouns and Number of Genders and looking at the interesting hits away from the diagonal. Hindi has gender, but not in pronouns. I’m not clear what pronoun distinction is meant for Burmese, Mandarin and Persian in the 3rd person; I didn’t think there was any gender at all. I believe Burmese has different 1st and 2nd person words to be used by female speakers, but that’s more a facet of its politeness scheme.

  68. Etienne says:

    I’m afraid I can’t respond to everyone’s comments, as I do want to write something of manageable length.
    David: to your list of languages which have turned /u/ to /y/ we may add certain lects of Azorean Portuguese, which, considering that the islands were uninhabited when the colonists arrived, makes a substrate explanation rather difficult to entertain.
    Jim: Mexican Spanish is actually a great argument AGAINST substrate influence in general. You’ve a large number of (non-basic) loanwords from various Native languages, Nahuatl most importantly: but the morphosyntax of Mexican Spanish differs very little from (Andalusian) Spanish. Yet we know that native Spanish speakers were a small minority in Mexico: Mexican Spanish was born and became the dominant language of Mexico in a context of mass illiteracy, making its spread very reminescent of the expansion of Latin in the Roman provinces. The main difference, indeed, is that the substrate languages were MORE prestigious in Mexico than in the Roman Empire: the local friars did after all use them and compose grammars of Nahuatl and other major languages (which does make it easier to identify loans and [lack of, in this instance]substrate features). Hence one would, if anything, expect there to be even *fewer* traces of the substrate in the Roman Empire than in Mexico.
    (I think Herbert Izzo made the above point somewhere, don’t quote me on that)
    Also: why couldn’t a word found in Hmong as well as Southern Chinese have entered Hmong from Chinese, *which itself may have borrowed the term from some third, now extinct language?* This is by no means implausible: it was Luis Michelena who stressed that non-Romance words in Basque needn’t be indigenous, that they in fact may have entered Basque as Romance loans, having first entered Latin/Romance from some now-extinct pre-roman language: and based on his reconstruction of Basque diachronic phonology Michelena identified a number of such words.
    Here’s a more recent example: take the term “squaw”, which is found practically all across North America, in Indian English as well as in many Indian languages: if a hundred years from now the only Indian language left in North America is, say, Navajo, it would be very easy for a future scholar to pronounce “squaw” a native Navajo word and conclude, from its widespread presence in North America (“squaw bread”, “squaw dance”), that most of the continent must once have spoken Navajo-like languages, or at the very least that “squaw” was part of a pre-european North American SPRACHBUND. Of course we know this is false: “squaw” entered English from Wampanoag in early colonial days, and spread over North America *as an English word*. I think many scholars, in automatically attributing Southern Chinese loanwords to Hmong/Tai-Kadai/whatever, may be making the same mistake as our hypothetical future Navajo scholar…
    This brings up what I consider to be THE trap of historical linguistics: because as a group we are fascinated by the past, we tend to over-interpret the evidence: we all WANT to know as much as possible about as deep a time-period as possible, and thus tend to suspend critical judgment and disbelief when encountering data which PRIMA FACIE seemingly gives us a glimpse into such a past…
    Marie-Lucie: actually, the history of Occitan/French contact in Southern France illustrates my point, since in urban settings especially the Occitan-influenced French of the older generation is making way to much more Parisian-like French among the younger generation (I noticed this the last time I was in Marseilles): the same social forces which caused French to replace Occitan are still in place, causing Occitan-influenced French forms and pronunications to be stigmatized and, hence, not passed on to the next generation. If present trends continue I see no reason to doubt that a few generations hence Occitan will have been entirely replaced by a form of French which will show few if any traces of Occitan.

  69. thymios says:

    This has been a fascinating discussion. I can’t add anything on the main topic, but about
    marie-lucie: If the plural suffixes -i and -a had truly been borrowed, they would have been generalized to other forms.
    What about those Latinate words that have acquired the -i even though it’s not justified etymologically, like apparati, octopi, peni (< penis) and ignorami? (Has this by the way occurred in other languages?) Plus, I don’t have a problem saying a plural of (Toyota) Prius is Prii. But there’s also Elvii (< Elvis) which is out there and I’ve heard (and said) things like Guini (< Guiness). None of these forms are “correct” but it does that -i does exist as a plural suffix, even if it’s fairly marginal.

  70. iakon: Many Berbers do object to the word, but many others don’t; the noted Kabyle linguist Salem Chaker has written in defense of it, for example. In Arabic, I normally use أمازيغية rather than بربرية; but in English, I’ll stick with the term people are least unlikely to have heard of.
    On the productivity of -i, what Thymios said.

  71. Also, on plural affix borrowings: English -s into Welsh; Arabic -āt into Persian. In both cases the extension to native words is marginal, but significant. A few other cases are listed in Grant 2008.

  72. Just to make things more complicated, let’s talk about origins. I don’t have specific citations or even specific publications, but I remember that when writing about the language and rhymes of the Book of Songs, Karlgren derived the “dialects of Chinese” (= “Sinitic languages”, I guess — for example Hokkien) from the court language of the Chou dynasty. Then when he was writing about the language and rhymes of T’ang poetry he sketched a different history, deriving Hokkien et al from the official language of the T’ang court.
    Take that all with a grain of salt — Karlgren probably didn’t write that naively, he was probably just trying to show the kinds of transformations that would lead from the language he was studying to, on the one hand Mandarin, and oin the other, Hokkien. The historical schema might have been just a manner of speaking.
    Nonetheless, a lot of study of history talks about origins, and pure original forms, but whenever we look at the supposedly pure original forms themselves, we find that they are impure and not original (i.e., have a past). So it’s most reasonable to assume that Ur-IE had something behind it and probably a “creole” in Beckwith’s loose sense.
    In other words, Ur-IE’s Ur-ness is a convention of investigation and presentation, and is not real at all.
    I have a nice quote somewhere by a philologist (in “Seth: God of Confusion”, which everyone should read anyway) to the effect that it’s normal to assume that the historical background or origin was simpler than what rose from it, but that’s a function of what explanation is:

    Explanandum and explanans are two forms of the same latin word. The explanans is the participle and means: ‘flattening out’ or ‘making plain’; The explanandum is the grammatical object and refers to that which might undergo a ‘flattening out’ or a ‘making plain’.

    In other words, if the explanandum is as complicated as the explanans, you really haven’t done much explanatory work, and there isn’t much explanatory payoff.
    From this point of view, all you have is a succession of creole languages (Beckwith’s weak definition) going back into the past until our records disappear less than 10,000 years ago. Where records are sufficient we can observe many sorts of transformations at work and write complicated histories which are quite illuminating, and where the records are insufficient we can conjecture and reconstruct but before about 5,000 years ago everything we say is doubtful indeed.
    Or put it this way: suppose before Ur-IE there was Ur-nostratic around ? 10,000 BC. Ur-nostratic was a “creole” influenced by neighboring language and with its own history. In 20,000 BC there was Ur-super-nostratic, and in 30,000 BC there was Ur-super-super-nostratic. And both of them were “creoles” with histories. And this chain goes back at least to 60,000 BC, and some say 300,000 BC.
    Now add in the idea that languages go through irregularly alternating processes of creolization (like simplification) and anti-creolization (like complexification) while both receiving various influences from other languages and also having endogenous internal processes of change. The outcome to me seems to be that you can’t really extrapolate back into the past any where near as much as you’d wish; the system has been stirred too much and is too jumbled.

  73. Just to make things more complicated, let’s talk about origins. I don’t have specific citations or even specific publications, but I remember that when writing about the language and rhymes of the Book of Songs, Karlgren derived the “dialects of Chinese” (= “Sinitic languages”, I guess — for example Hokkien) from the court language of the Chou dynasty. Then when he was writing about the language and rhymes of T’ang poetry he sketched a different history, deriving Hokkien et al from the official language of the T’ang court.
    Take that all with a grain of salt — Karlgren probably didn’t write that naively, he was probably just trying to show the kinds of transformations that would lead from the language he was studying to, on the one hand Mandarin, and oin the other, Hokkien. The historical schema might have been just a manner of speaking.
    Nonetheless, a lot of study of history talks about origins, and pure original forms, but whenever we look at the supposedly pure original forms themselves, we find that they are impure and not original (i.e., have a past). So it’s most reasonable to assume that Ur-IE had something behind it and probably a “creole” in Beckwith’s loose sense.
    In other words, Ur-IE’s Ur-ness is a convention of investigation and presentation, and is not real at all.
    I have a nice quote somewhere by a philologist (in “Seth: God of Confusion”, which everyone should read anyway) to the effect that it’s normal to assume that the historical background or origin was simpler than what rose from it, but that’s a function of what explanation is:

    Explanandum and explanans are two forms of the same latin word. The explanans is the participle and means: ‘flattening out’ or ‘making plain’; The explanandum is the grammatical object and refers to that which might undergo a ‘flattening out’ or a ‘making plain’.

    In other words, if the explanandum is as complicated as the explanans, you really haven’t done much explanatory work, and there isn’t much explanatory payoff.
    From this point of view, all you have is a succession of creole languages (Beckwith’s weak definition) going back into the past until our records disappear less than 10,000 years ago. Where records are sufficient we can observe many sorts of transformations at work and write complicated histories which are quite illuminating, and where the records are insufficient we can conjecture and reconstruct but before about 5,000 years ago everything we say is doubtful indeed.
    Or put it this way: suppose before Ur-IE there was Ur-nostratic around ? 10,000 BC. Ur-nostratic was a “creole” influenced by neighboring language and with its own history. In 20,000 BC there was Ur-super-nostratic, and in 30,000 BC there was Ur-super-super-nostratic. And both of them were “creoles” with histories. And this chain goes back at least to 60,000 BC, and some say 300,000 BC.
    Now add in the idea that languages go through irregularly alternating processes of creolization (like simplification) and anti-creolization (like complexification) while both receiving various influences from other languages and also having endogenous internal processes of change. The outcome to me seems to be that you can’t really extrapolate back into the past any where near as much as you’d wish; the system has been stirred too much and is too jumbled.

  74. marie-lucie says:

    Etienne:

    actually, the history of Occitan/French contact in Southern France illustrates my point, since in urban settings especially the Occitan-influenced French of the older generation is making way to much more Parisian-like French among the younger generation (I noticed this the last time I was in Marseilles): the same social forces which caused French to replace Occitan are still in place, causing Occitan-influenced French forms and pronunications to be stigmatized and, hence, not passed on to the next generation.

    Here there is a difference from more natural situations in that French is becoming more homogeneous, not only because of the mass media, but also because teachers and other state employees migrate throughout the country, especially from the South to the North. It is not that Occitan-influenced French is not “passed on” by the parents (who do not speak another form of French), it is that the younger generations are consciously learning another form of French from speakers other than their own parents and neighbours. (Typical [lower-class] Marseille speech is indeed stigmatized, but it is so within the Occitan-influenced Southern French area, just like the [lower-class Right Bank] accent parisien within the Northern area).

    If present trends continue I see no reason to doubt that a few generations hence Occitan will have been entirely replaced by a form of French which will show few if any traces of Occitan.

    Actually, current Standard French already shows significant traces of indirect Occitan influence, because of the spread of Southern French throughout the country by state employees including teachers: the simplification of the vowel system, including the merger of long and short vowels, most noticeable in that of a and â, the generalization of the complementary distribution of high-mid and low-mid vowels in open and closed syllables (so that for instance the lower vowel of faite ‘done, made (fem)’ is different from the higher vowel of fait ‘done, made (masc), fact’), the reintroduction of schwa in many cases where Northern French (including Canadian French) omitted it (eg the pronunciation of maintenant ‘now’ instead of maint’nant), seem to me to be due to recent Occitan influence via Southern French. (I am in a unique position, having had bilingual Occitan- and Southern-French-speaking maternal grandparents, Left Bank Parisian parents and paternal grandparents, but being raised in Northwestern France, a linguistically very conservative area).
    I mentioned teachers, a profession I am very familiar with: because the South is a poorer area than the North, many Southerners (including my Occitan-speaking grandparents) look(ed) for state positions, which in France include the educational system. These positions bring them and their pronunciations into many other areas of the country, and schoolteachers do influence primary school children learning to read and write. Teachers who do not make a difference between tache ‘spot, stain’ and tâche ‘task’ and therefore are not sure which one takes the circumflex (which represents a distinct pronunciation in the more conservative forms of the language), cannot use the pronunciation of their own Northern students as a clue to which is which, and confuse those students (who do have a difference in their speech and therefore would have no problem with either the pronunciation or spelling of the words) by pronouncing both words identically. Southern influence also at least contributes to the identical pronunciation of faite ‘done, made (fem)’ and fête ‘festival’ (formerly a length difference), and I think more definitely in the case of cote (a legal term) and côte ‘coast or rib’ (formerly a difference of both length and quality). Examinations in the educational system do not take pronunciation into account, but for radio or television a Southern accent was long unacceptable (except for sports!), but nowadays one hears speakers with slight Southern accents, who have all these simplifying vowel features as well as many more schwas in non-final position than is still normal in the North.

  75. marie-lucie says:

    thymios: What about those Latinate words that have acquired the -i even though it’s not justified etymologically, like apparati, octopi, peni (
    Your examples are not
    Latinate, they are Latin, and the -i suffix is being used in a generalization within Latin (because of limited knowledge of Latin).
    Plus, I don’t have a problem saying a plural of (Toyota) Prius is Prii. It’s up to you! Is the usage common among Prius owners? Do you pronounce it pry-eye?
    But there’s also Elvii (
    Again, those words
    sound to some people like Latin words, and some of those plurals sound facetious. It would be different if -i was also used in words which are not reminiscent of Latin.
    NOTE: The rest of your sentences, after the left arrow sign, were deleted along with it, and I did not want to retype the missing words. Thanks to John Cowan :
    Everyone, remember to write < instead of . [read the rest above, at the beginning of his post]

  76. What a great discussion!
    From this point of view, all you have is a succession of creole languages (Beckwith’s weak definition) going back into the past until our records disappear less than 10,000 years ago. Where records are sufficient we can observe many sorts of transformations at work and write complicated histories which are quite illuminating, and where the records are insufficient we can conjecture and reconstruct but before about 5,000 years ago everything we say is doubtful indeed.
    This is very much the way I look at it, and one of the reasons (besides burnout, lack of jobs, and having to take on debt) that I got out of academia: I was studying something that in a sense couldn’t really be studied, only guessed at.
    Incidentally, I have no idea why Lameen’s last link (“Grant 2008″) won’t work; I’ve tried redoing it and rebuilding the post half a dozen times with no success. The URL is http://cgi.server.uni-frankfurt.de/fb09/ifas/JLCCMS/issues/THEMA_2/09_Anthony_Grant.pdf (and I’ve visited it, so I know it works; it’s Anthony P. Grant, “Contact-Induced Change and the Openness of ‘Closed’ Morphological Systems: Some Cases from Native America,” from Journal of Language Contact 2, 2008), but when you try to make a link of it there’s a glitch. Here‘s a Google cache; maybe that will work.
    On preview: yup, the Google cache works.

  77. “Now add in the idea that languages go through irregularly alternating processes of creolization (like simplification) and anti-creolization (like complexification) while both receiving various influences from other languages and also having endogenous internal processes of change.”
    Yeah, but just to complicate things: creolisation (in Beckwith’s very broad sense) definitely does not imply simplification. As I understand it, he would consider languages like Michif or Tasawaq as “creoles”, and in them contact has led to substantial complexification if anything.
    “Octopus” is Greek.

  78. David Marjanović says:

    - Interesting hypothesis about the German ordinals. Could a generalization from erst- “first” be involved?
    - What is Tasawaq?
    - Forgot to mention: it’s very easy to tell that hickory is not native to English, because the word and the tree are absent from Europe (the latter since one of the first ice ages – the combination of ice ages with east-west-trending mountain ranges and the Mediterranean Sea has greatly impoverished the European fauna and flora, which is a great pity because Europe used to preserve interesting animals that North America and, where applicable, the rest of the world had lost long before… but I digress…). So this example is a bit unfair. :-)

    My memory of the pronoun systems of Middle English is that they were totally confused from one ME dialect to another and from one century to another.

    Yes, because most 3rd-person pronouns were collapsing towards hem/him or something, leading to the borrowing of she and they/them from Norse. Personal pronouns are very rarely borrowed.

    skeleta, if that form is used at all in English
    - It isn’t. In spite of femora, humeri, scapulae, pelves, vertebrae, and so on.

    But these forms are hardly part of general vocabulary.

    That’s my point: even though anatomists use all those exotic forms daily, they never use *skeleta. It’s always skeletons.

    A “stop” (also called “plosive”)

    If I may get pedantic, nasals (m, n, ng) are also stops, but not plosives. However, this distinction is usually ignored.

    I’m not clear what pronoun distinction is meant for Burmese, Mandarin and Persian in the 3rd person; I didn’t think there was any gender at all.

    Mandarin does now (since the late 19th century) have separate characters for “he” and “she”, but the pronunciation is identical. Did someone goof here?

    but when you try to make a link of it there’s a glitch.

    This link has “www.languagehat.com/archive” between “http://” and the rest of the URL. Interesting paper, BTW.

  79. link has “www.languagehat.com/archive” between “http://” and the rest of the URL.
    I know that, but there appears to be no way to get rid of it. Trust me, it’s not there in the HTML as posted.

  80. David Marjanović says:

    Deeply bizarre.
    Anyway, here’s a borrowed bound morpheme that’s freely, in fact mostly, used with native vocabulary: the -er of English and German comes from Latin -arius.

  81. You know, I don’t like Beckwith’s notion of “creole” because it’s not a natural class. The only thing that “languages that have undergone significant changes due to convergence with other languages” share is a fact about their history. They can be agglutinating or inflectional or isolating or polysynthetic (like modern Nahuatl); they can have huge or tiny phoneme inventories, big or small vocabularies, any or no basic word order… If you tell me a language is a creole in the usual sense, I can be pretty sure it will be unusually regular, irrespective of the language(s) it emerged from. But knowing that it’s undergone significant contact-based changes tells me nothing unless I know which language it was in contact with. And as noted above, all languages fit in this category anyway!
    Tasawaq is the heavily Tuareg-ised Songhay language of In-Gall, Niger. Rather like Kwarandzyey, actually, and closely related.

  82. David Marjanović says:

    The one for agents, like maker, of course, not the ending for adjectives in the comparative or something…

  83. You know, I don’t like Beckwith’s notion of “creole” because it’s not a natural class. … And as noted above, all languages fit in this category anyway!
    Right, but he’s making an ideological point, not using it to categorize languages. If you look on all languages as creoles, it promotes a more useful kind of thinking than categorizing some as (strictly defined) creoles and others (implicitly, by contrast) as “pure.” (Although this is not something explicitly claimed by modern Indo-Europeanists, there’s still a hangover from the 19th-century overt racism embodied in the term “Aryan.”) It’s like saying “We humans are all mongrels”—the point is not to make a scientific classification but to try to get people to stop thinking in terms of “pure bloodlines” that can be “defiled” by “lesser breeds.” Or that’s how I interpret it, anyway.

  84. marie-lucie says:

    David: here’s a borrowed bound morpheme that’s freely, in fact mostly, used with native vocabulary: the -er of English and German comes from Latin -arius.
    Again, this morpheme is derivational, not inflectional. The latter are much less likely to be borrowed.

  85. Everyone may be a mongrel (me more than most!), but when it comes to genetic illnesses, ethnic heritage is still a great predictor (eg sickle-cell anemia or Tay-Sachs.) Likewise, as Schuchardt said, there is no entirely unmixed language, but some languages exhibit much more recent and intense contact effects than others – and interesting generalisations can be made about several subtypes, such as creoles-in-the-usual-sense or mixed-languages-in-Bakker’s-sense. I appreciate the need to correct naive and discriminatory ideas of “purity” in either sphere, and the urgent need to get historical linguists to pay more attention to contact effects (Ehret’s proto-Nilo-Saharan contains a number of undetected Arabic loans, for example, and I hate to think how many intra-Semitic loans must get mistakenly reconstructed as proto-West Semitic.) But just rejecting classification altogether is throwing out the baby with the bathwater.

  86. But just rejecting classification altogether is throwing out the baby with the bathwater.
    Sure, but he’s not doing that, he loves classification—he’s just using “creole” in a particular way to make a point. I imagine if the point he was trying to make were widely acknowledged, he wouldn’t feel the need to do that.

  87. I’m pretty sure no one who says “we’re all mongrels” wants to cut off research into sickle-cell anemia or Tay-Sachs; I certainly don’t!

  88. marie-lucie says:

    I agree with Lameen’s comments. Using technical terms “in a particular way to make a point” is not particularly useful when it is carried over to technical discussions. Seeing “periods of creolization” in the formation of Proto-Indo-European ignores the specific, well-documented circumstances of the formation of present-day creoles, which circumstances are unlikely to have existed thousands of years ago. And equating “creolization” only with language simplification is itself a gross simplification of the situation.
    From a more general point of view, I find that “language contact” is also a very vague concept. There are many types of “language contact” and it is not surprising that they should lead to distinct types of influence on the languages in so-called “contact”: the speakers are in contact, not the languages, and the circumstances of their “contact” influence the linguistic results.
    “Borrowing” is also a vague word. Adoption of foreign words generally occurs whether there is very slight or even no contact between speakers (as when the name of a fruit originating in a specific country arrives along with the fruit shipment to another continent), or intimate contact through frequent intermarriage and stable bilingualism, or through pervasive, largely unidirectional translation (as with English to French translations in Canada and France, or from Latin to local languages in Europe in earlier centuries). In these cases, words are “borrowed” (or rather adopted) quite consciously, and syntactic constructions (types of sentence formation) may also be adopted more or less consciously. But the “borrowing” or rather the spread of sound features requires actual and frequent contact between populations in a situation of bilingualism, and it is not usually conscious: speakers learning another language always try to reproduce what they hear, but that is processed through the phonological system of their own language and the muscular movements associated with the actualization of that system, so that in most cases the reproduction is imperfect, causing an “accent” on the part of learners. Even when those learners try to “lose their accent” under social pressure, the results are usually not perfect.
    For instance, in a recent textbook on historical linguistics I found a statement to the effect that in a certain city in Brittany, French has “borrowed from Breton” a particular feature of pronunciation. I find this statement very misleading: it is not that local French speakers have consciously adopted a feature from Breton, it is rather than those speakers retain a feature passed on by their Breton-speaking grandparents who learned French in school. Similarly, Southern French speakers have not “borrowed” features from Occitan, but preserve features of the Occitan speech of their forefathers.

  89. marie-lucie says:

    LH: If you look on all languages as creoles, it promotes a more useful kind of thinking than categorizing some as (strictly defined) creoles and others (implicitly, by contrast) as “pure.”
    This suggests that languages are either creole or pure. It would be more useful to say that all languages are “mixed”, since that is not a very specific term, and everyone understands that a mixture can include elements in various numbers and proportions.

  90. Etienne says:

    Marie-Lucie, Lameen, LH–
    Along with “Creole”, “mixed language” is so ambiguous that we should always very carefully define what we mean when we use either term (which is what Beckwith does). Thus I would accept that there is a gap between Creoles and non-Creoles, defining “Creole” as a language which is a nativized form of an earlier pidgin: from this point of view there is a real difference between dialects of English and English Creoles: Rural Appalachian being an undeniable example of the former, Sranan Tongo of the latter.
    Marie-Lucie: I’m afraid I can’t accept your claim about Occitan being the cause of the loss of length distinctions as well as of the mid-high/mid-low vowel distinctions in contemporary (European) French.
    Regarding the latter phenomenon: the distinction between final open mid-high and mid-low back vowels was already lost in standard French by the seventeenth century, long before Occitan influence could have made itself felt (hence the homophony in Standard Fench today of SAUT and SOT, both /so/, as opposed to the distinct vowel phonemes in SAUTE and SOTTE), a change which seems to have happened quite independently in Canadian French, which also lost the length distinctions in the high vowels well before most of France did (for other readers of this thread: Occitan speakers played next to no role in the genesis of Canadian French).
    The fact that *all* French Creoles have undergone the exact same changes (loss of inherited mid-high/mid-low vowel distinctions and of vowel length) makes me suspect that the recent evolution of French vowels has less to do with Occitan influence than with a general “drift” in that direction in the history of French, especially since most Occitan varieties do have separate mid-high and mid-low front unrounded phonemes.
    I’m not denying that Occitan has had some influence on French today, but such influence is probably of a very subtle nature. I suspect that, as a rule, the influence of various substrates (Gaulish, Oscan, Celtiberian…) on Latin as it became the various Romance languages we know and love today was likewise of a subtle nature, so subtle that I very much doubt we could detect it today.

  91. David Marjanović says:

    Again, this morpheme is derivational

    ARGH! I keep overlooking that.

    Ehret’s proto-Nilo-Saharan contains a number of undetected Arabic loans, for example

    Horror!
    Isn’t that Ehret someone who has worked on Afroasiatic, so presumably he knows Arabic to some extent? Or am I confusing him with someone else?

  92. David Marjanović says:

    so presumably he knows Arabic to some extent

    Or at least, more importantly, has a couple of Arabic dictionaries and knows how to use them.

  93. Thanks, all. This was tremendous fun for a non-linguist. I’m like a six-year-old on a bike being pushed by his mom: it’s as if I really knew how to ride! Except I don’t have to pedal and I know I won’t fall :-)

  94. “But there’s also Elvii (
    Again, those words sound to some people like Latin words, and some of those plurals sound facetious. It would be different if -i was also used in words which are not reminiscent of Latin.”
    Marie Lucie – you’re moving the goalposts. Wasn’t soemone here talking about a prejudice against mixing of Greek and Latin and Germanic roots in word formation? derivational versus inflectional – why would a general speaker trip over that disitinction necessarily?
    “- Forgot to mention: it’s very easy to tell that hickory is not native to English, because the word and the tree are absent from Europe ”
    That was my point. America is full of species that English obviously had not native words for and it is a little strange to say that an intrusive language is going to be the source of etyma of those unfamiliar species, and reaching for on-site innnovations in that intrusive language as the source of those etyma is, well, reaching. The same holds for the botany of Southern China.
    “Similarly, Southern French speakers have not “borrowed” features from Occitan, but preserve features of the Occitan speech of their forefathers. ”
    Yes, it’s easy to anthropomorphize languages. Too bad English grammar allows that; it leads us into temptation. Oops, too Whorfian? Anyway, speakers are more real than their languages.
    “Jim: Mexican Spanish is actually a great argument AGAINST substrate influence in general. You’ve a large number of (non-basic) loanwords from various Native languages, Nahuatl most importantly: but the morphosyntax of Mexican Spanish differs very little from (Andalusian) Spanish. ”
    Etienne, why do you privelege morphosyntax above lexical material? Afterall all, which is likelier to change faster, which is most resistant to change? which is more inherent to the speech behavior of the speakers?
    You’re point about “squ@w” (there are ladies in this board) is very well taken and it is a good example. I wonder if the same general thing is hapening when people look at what is supposedly pre-IE Europe. I am thinking of the root for the word ‘alder” which suppodely only shows up in Western Europe and surpsise surprise, in Basques – mnow why would any language supposedly from the Ukraine oe even further east need a new word for that species? Whatever – that’s why I think your next point, about the temptation to drive the data too far, is also very well taken.

  95. marie-lucie says:

    Jim: derivational versus inflectional – why would a general speaker trip over that distinction necessarily?
    A general reader would not consciously think about that distinction, but since derivation produces new words while inflection just modifies words in order to add grammatical meanings on the model of most or all words of the same category, a speaker is much more likely to consciously produce or recognize those words as new ones than to consciously add an inflectional morpheme to a new word, since the latter formation tends to be automatic: in a rapidly developing section of the vocabulary, for instance computer technology, new words are formed all the time, and these words receive plural suffixes or verb endings on the model of the most regular formations in the language. Nobody had to wonder how to form a plural for the new word “byte”, for instance: “bytes” was the automatic, default plural. Conversely, a person encountering the word for the first time in its plural form “bytes” would be conscious that this was a new word and might wonder what it meant, but not what the singular form could be: just chop off the obviously plural suffix and you have the singular.
    the root for the word ‘alder” which supposedly only shows up in Western Europe and surprise surprise, in Basque
    What root is that?
    now why would any language supposedly from the Ukraine or even further east need a new word for that species?
    In addition to the possibility of retaining such words from a substrate, many nouns are originally descriptive, and descriptions of the same thing may vary from one culture or region to another (and from one time period to another). An example is the tree known as “oak” in English: even in the single Romance family, which became differentiated relatively recently, there are four or five distinct origins for the names of this tree in different languages, and sometimes within the same language (see a map in Mallory’s In search of the Indo-Europeans). Another example concerns the names of smaller, less noticeable or useful plants (which would not be traded even with neighbouring communities, for instance): the same flower can have quite a number of local names, even within a small geographical area, or the same name can refer to quite different plants in different places, which is why Linnaeus sought to introduce a consistent naming practice according to which the same plant would be known in scientific discourse by a single Latin designation.
    Another example of unexpected differentiation in vocabulary is cited by Henriette Walter in her book on the languages of Europe: although there has been a lot of borrowing between European languages for words referring to technical innovations, the self-lighting match (for starting a fire) invented in the 19th century has distinct, unrelated names in all twelve of the languages considered in the book.

  96. marie-lucie says:

    (Etienne): “Jim: Mexican Spanish is actually a great argument AGAINST substrate influence in general. You’ve a large number of (non-basic) loanwords from various Native languages, Nahuatl most importantly: but the morphosyntax of Mexican Spanish differs very little from (Andalusian) Spanish. ”
    It might be different if there had been only Nahuatl (or another single language) spoken throught Mexico.
    I read a book by an author from another Latin American country (sorry, that was a while ago) where a character who travels to a remote area of his country encounters a native man who can speak Spanish but treats all nouns as uniformly masculine (the default option). This feature (shown in the quoted speech of the man) is not explained, but it is logical to conclude that in the local indigenous language there is no grammatical gender.
    I lived for several years in a native community in Canada where the influence of the native language (even though no longer spoken by young people) on local English was very obvious, not only in pronunciation but in a large number of details of morphosyntax.

  97. marie-lucie says:

    Etienne: the term “squaw”, which is found practically all across North America, in Indian English as well as in many Indian languages … “squaw” entered English from Wampanoag in early colonial days, and spread over North America *as an English word*
    Jim:
    Your point about “squ@w” (there are ladies in this board)…
    Lady linguists are not so easily flustered.
    The word in question is the ordinary word for “woman” in several Algonquian languages, such as Wampanoag. Some people have spread the rumour that it referred to women’s genitals and was therefore a derogatory word, but there is no shred of evidence for any original meaning beside that of just “woman” (and no other word actually meaning “woman” is attested from the same languages). If it was used derogatorily, it was for other reasons, unrelated to any other supposed original meaning.

  98. “Isn’t that Ehret someone who has worked on Afroasiatic, so presumably he knows Arabic to some extent? Or am I confusing him with someone else?”
    Yeah, he’s written an etymological dictionary of Afroasiatic too, and has drawn all kinds of interesting historical conclusions from early loanwords in East Africa. You’d think that would have stopped him from including (for example) Songhay sóote “to whip, flog” (سوط), Nobiin šìmm- “to smell, stink” (شمّ), Kanuri làn “to abuse, curse” (لعن), Fur wassiye “wide” and Songhay wásà “to be wide” (واسع), but it didn’t. His reconstructions tend to be unreliable for semantic reasons anyway.

  99. Some people have spread the rumour that it referred to women’s genitals and was therefore a derogatory word, but there is no shred of evidence for any original meaning beside that of just “woman” (and no other word actually meaning “woman” is attested from the same languages).
    Yes, the whole “squaw” ruckus is one of the more annoying examples of over-the-top PC in today’s ultrasensitive America. (Not knocking sensitivity, and I’m very glad certain words have become taboo, but I wish we could do these things without going overboard.)

  100. “That was my point. America is full of species that English obviously had not native words for and it is a little strange to say that an intrusive language is going to be the source of etyma of those unfamiliar species, and reaching for on-site innnovations in that intrusive language as the source of those etyma is, well, reaching.”
    Bobcat. Mountain lion. Robin (in the American sense.) Buffalo. Corn. Springbok. Eucalyptus…

  101. Big Dave M.: it’s very easy to tell that hickory is not native to English, because the word and the tree are absent from Europe

    Hickory Dickory Dock:
    The earliest recorded version of the rhyme is in Tommy Thumb’s Pretty Song Book, published in London about 1744, which uses the opening line: ‘Hickere, Dickere Dock’. (Wiki)

  102. “Lady linguists are not so easily flustered.”
    No, indeed, M-L. I was just joking. Hat is right about the motivation for the distortion of the term.
    “What root is that? ”
    ‘al’. This may be totally wrong, so bear with me. Apparently it occurs in Basque, in Spanish the word is ‘aliso’, I think it occurs in French ‘aune” and it may be the ‘er’ in ‘Erle’ in German. In Swedish the word is ‘al’. The ‘der’ in the English is pretty stairghtforward. Anyway that was the proposal. Can’t remember where I saw it. For all I know that really is an IE root anyway.
    M-L – you mentioned plant names and how they are often descriptive.
    “Bobcat. Mountain lion. Robin (in the American sense.) Buffalo. Corn. Springbok. Eucalyptus…”
    Lameen, the point is not that English doesn’t come up with its own words for new species, even when they really aren’t new species but that these words don’t pass into indigenous languages, with reference to the direction of borrowing between Chinese and indigenous languages in South China. I doubt that any language in North America, Africa or Australia borrowed any of those words. In fact I can’t imagine the concept “eucalyptus” even making any sense to an aboriginal Australian unless he grew up speaking English – it’s an almost uselessly broad term in that botanical context.

  103. “Rural Appalachian being an undeniable example of the former, Sranan Tongo of the latter.”
    Etienne, most Appalachian English speakers descend from people who lived on the “Celtic Fringe” in Britain, so there may be some creolization going on, or whatever we are going to call this kind of influence. Wait – I know there is – they overuse the progressive tenses in places where it’s obligatory in Gaelic.

  104. marie-lucie says:

    AJP: Hickory Dickory Dock: your quote is attested from 1744, long after English speakers in the Americas had had time to become familiar with local trees and write about them in works that reached England.
    Jim: Appalachian English speakers … overuse the progressive tenses in places where it’s obligatory in Gaelic.
    I believe this is also done in Irish English and it has been suggested that Celtic is the source of the progressive formation in English. But similar formations are used in Spanish and Italian, though not in French, so a Celtic origin seems doubtful (unless it was nipped in the bud in very early French by Frankish influence, since it does not seem to exist in German; Etienne and David would probably know).
    However, there is what seems to be a related French version in archaic expressions found in some old songs, using forms of the verb aller “to go” and the present participle, as in le fils du roi s’en va chassant “the king’s son is going hunting” or perhaps better and also archaic “the king’s son is a-hunting.” (English also had the verb “to go” in Froggy went a-courting).

  105. marie-lucie says:

    Jim: (the root “al”) Apparently it occurs in Basque, in Spanish the word is ‘aliso’, I think it occurs in French ‘aune” and it may be the ‘er’ in ‘Erle’ in German.
    French aune in this sense has an older form aulne where the root is more recognizable. The title of the famous German poem and lied Erlkönig “the Alder King” is translated as Le roi des aulnes.
    Spanish aliso: The Gaulish chief and hero of the resistance against the Romans known as Vercingétorix was defeated in a place known by the Latinized Celtic name Alesia, and the name is preserved in French by the local village of Alise-Sainte-Reine. The name is believed to refer to alder trees, and the place is in Burgundy, quite far from the Basque country. Could Spanish aliso be from Celtiberian, the ancient Celtic language of the Northwest of the Iberian peninsula?
    The alder tree seems to have played an important role in Celtic and perhaps also pre-Celtic religion, so that may be why the root for the name of this tree is found in Western Europe but not further East: it could be pre-Indo-European, what Theo Vennemann would call “Old European” (one feature of TV’s hypothetical pre-IE ancestor – or possibly substrate – is the frequent occurrence of vowel-initial roots). (I mentioned TV earlier).

  106. marie-lucie says:

    isn’t Ehret someone who has worked on Afroasiatic
    Lameen: Yeah, he’s written an etymological dictionary of Afroasiatic … His reconstructions tend to be unreliable for semantic reasons anyway.
    I know someone who is well-trained in historical linguistics and has done extensive fieldwork in Africa with some of the lesser-known Chadic languages (which are considered Afro-Asiatic), and who also considers Ehret’s work to be unreliable.

  107. m-l:Hickory Dickory Dock: your quote is attested from 1744, long after English speakers in the Americas had had time to become familiar with local trees and write about them in works that reached England.
    Ok, but that doesn’t strike me as a very likely way for a nursery rhyme to originate, even though it can’t be proved either way. The point is that the word was not absent from Europe, even if the tree was — and I very much doubt that ‘hickory dickory dock’ has anything to do with the hickory tree.

  108. Here is a more complete version of the North Country sheep-scoring numerals to which the Opies (quoted in Wikipedia) attribute it.

  109. Wow, look at that. Of course it’s not quite the same words, but it might be pre-Anglo Saxon.
    Yes, the Opies explained away many rhymes by saying they sounded like Cumbrian sheep counting. Since we don’t need to go higher than three with our goats I’ve never learned any, but it is interesting — like Danish numbers.

  110. “I believe this is also done in Irish English and it has been suggested that Celtic is the source of the progressive formation in English. But similar formations are used in Spanish and Italian, though not in French, so a Celtic origin seems doubtful (unless it was nipped in the bud in very early French by Frankish influence, since it does not seem to exist in German; Etienne and David would probably know). ”
    Yes. I am not saying that the Englsih progressive tense has a Celtic origin; I am saying that Scotch-Irish Americans’ use/misuse of it has Celtic origins.
    The progressive in Englsih may have a mixed ancestry. I saw somewhere that someone had found adjectival forms used to make progresives very early. The progresive in Spanish uses unequivocally adjectival forms. But then in Englsih as in Irish and Welsh, you see these adjectival forms treated like nouns:
    “(English also had the verb “to go” in Froggy went a-courting).
    That ‘a’ is a preposition, and that is exactly the constrruction used in Irish and in Scots Gaelic. And guess what, that is a sterotypically Appalachian shibboleth.
    The alder was the royal tree, supposely because the twigs are purple and from a distance they form a purple haze. (I can’t think or any other reason sucha weedy little tree would be considered royal.) That’s true for red and white alder here on the West Coast, but I don’t know if that’s true for brown alder in Europe. In all my years of dragging throught mud and sleeping in the woods in Germany I don’t remember seeing anything that looked like an alder. They must have cut them all out for cropland or their managed forests.

  111. Etienne says:

    All: the hypothetical lazy linguist who might prefer printing this out to organizing a lecture might prefer printing this out to writing a textbook, if present trends continue…
    Jim: I tend to focus on morphosyntax because, whereas adult L2 speakers who acquire a (wholly unrelated) language informally seldom if ever master morphosyntax, they frequently do acquire basic vocabulary and phonology. And while Appalachian English does exhibit some Gaelic influence, this does not make it a “Creole”, in the sense I gave above.
    Also, I realize my Mexican Spanish example, as a comparandum to the expansion of Latin, is even better than I thought, inasmuch as, whereas Spanish and the various substrate languages of Mexico were genetically unrelated to and typologically quite unlike Spanish, Latin in the Roman Empire mostly replaced Indo-European languages whose structure was quite similar to that of Latin: thus, all other things being equal, you’d expect more substrate influence in Mexican Spanish than in the Romance languages. Considering how little there is in the former, this certainly suggests there may be VERY little in the latter.
    Marie-Lucie: perhaps there indeed would have been deeper substrate influence had Spanish encountered Nahuatl only in Mexico: but I somehow doubt it. The reason being that Nahuatl and its neighbors do form a SPRACHBUND, with a large number of common features (vigesimal counting system, use of genitives to express possessives, i.e. “the girl her father” to express “the girl’s father”) which somehow ALL failed to make their way into Mexican Spanish.
    That individuals speaking an L2 will transfer features from their L1 therein is indubitable, but I can only repeat my point about distinguishing micro-social from macro-social realities: the same power differential whereby one language (A) is deemed more prestigious than another (B) also means that features of B in A will be stigmatized and thus are likely to be lost.
    For example, in the French spoken by the Metis of Western Canada a “Mesoamerican”-like structure (“Le Bon Dieu son fils” instead of “Le fils du Bon Dieu”) exists and is clearly due to Cree and/or Ojibwe influence: the reason it is still alive in Metis French is because, scattered as they were (until comparatively recently) in isolated settlements throughout the prairies, they were cut off from any further contact with French, making it impossible for later generations to modify their speech and make it “less Metis”.
    Tellingly, the Metis are shifting to English today, and I see no reason to doubt that, had French become the dominant language of the Canadian prairies, the younger generation would be shedding specifically Metis features of their French (such as the substrate-derived feature presenteed above).
    (By the way, in Metis French this feature, and frequent confusion of masculine and feminine pronouns, are the ONLY instances of morphosyntactic influence: considering that the substrate consisted of two very similar languages [Cree and Ojibwe] whose structure is quite unlike any European language, I think this strengthens my contention that if, in an alternate universe, Nahuatl has been the only language of Mexico at the time of the Spanish conquest, Mexican Spanish would still exhibit little in terms of substrate influence).
    I am quite certain that any similar such structure in various lects of early Mexican Spanish did not survive because the same stigma which drove speakers of Native languages to shift to Spanish drove the next generation(s) to rid their Spanish of anything Native-like.

  112. David Marjanović says:

    Tasawaq is the heavily Tuareg-ised Songhay language of In-Gall, Niger. Rather like Kwarandzyey, actually, and closely related.

    Thanks!

    the self-lighting match (for starting a fire) invented in the 19th century has distinct, unrelated names in all twelve of the languages considered in the book.

    And two in German: Streichholz and Zündholz.

    Some people have spread the rumour that it referred to women’s genitals

    …which would not in itself be remarkable: English wife is apparently cognate to Tocharian (B, I think) kwipe, which plainly meant “vulva”.
    On a smaller scale, German Weib is nowadays derogatory, despite the complete lack of any sexual (or in fact any other) connotations whatsoever. It just means “woman I don’t like” (of any age – schoolboys use it to talk about schoolgirls, especially collectively). BTW, isn’t that where bitch is going these days?

    Bobcat. Mountain lion. Robin (in the American sense.) Buffalo. Corn. Springbok. Eucalyptus…

    Oh yeah. A week or two ago I cyber-ran across someone who complained that corn is mentioned in the Bible. Evidently he had no idea that that word had ever meant anything other than Zea mays.

    ‘al’. This may be totally wrong, so bear with me. Apparently it occurs in Basque, in Spanish the word is ‘aliso’, I think it occurs in French ‘aune” and it may be the ‘er’ in ‘Erle’ in German. In Swedish the word is ‘al’. The ‘der’ in the English is pretty stairghtforward.

    Ah, so Erle is one of the words that escaped being spelled with ä…
    Anyway, alnus in Latin, if I can trust the botanists.

    Anyway that was the proposal. Can’t remember where I saw it. For all I know that really is an IE root anyway.

    If you can manage to put an h2 in front of it. Otherwise the /a/ gets you into trouble – which is Vennemann’s point in many other examples (no idea about this one).

    [the progressive] does not seem to exist in German

    Not the slightest trace of it.
    In general, German is remarkable for being an aspect-free language. All there is are true tenses and what you might call lexicalized aspect (many verbs can have a wide range of prefixes, and the difference between these can often be considered one of aspect, but each prefix has different meanings with different verbs… it’s not even like in Russian).
    However, for the Franks, Dutch is probably more relevant. I’ve never seen anything looking like a progressive in Dutch, but it’s not like I had ever learned that language…

    le fils du roi s’en va chassant

    No trace of that in German either: Der Königssohn geht jagen – infinitive.

    “the king’s son is a-hunting.”

    No trace of that in German either.
    Well, actually…
    In my dialect (central Upper Austria) there are the fixed phrases “if/when it becomes raining/burning”, which mean things like “when it starts raining” and “if this ever catches fire”. Together with “stinking”, which could simply be considered an adjective, they comprise almost all cases where the present participle is ever used; I’d say most verbs lack it altogether (…rummaging through vocabulary… “sit”, “stand”, “write” lack it… “sleep” has it…).
    Make of that what you will.

    Yes, the Opies explained away many rhymes by saying they sounded like Cumbrian sheep counting.

    And Vennemann explains English eeney meeney miney moe and German ene mene muh as pre-IE numbers… I wonder what he’d make of Serb(ocroat?)ian en den dinu, though…

    The title of the famous German poem and lied Erlkönig “the Alder King” is translated as Le roi des aulnes.

    Which, incidentally, was a misunderstanding on Schubert’s part. He should have called that guy Elfenkönig. What exactly have the Danes done to their consonants? – That said, Elfe (pre-LOTR post-Dietrich von Bern* female only) also has a rather different meaning in English than in, apparently, all other Germanic languages; it designates a kind of fairy. Fairytale fairy. Peter Pan fairy.
    * Alberich, the king of the dwarfs, apparently brought to you by the Department of Redundancy Department. Also, Dietrich is Theodoric, king of the Ostrogoths, and Bern is Verona. – Allegedly to Tolkien’s delight, the LOTR elfs were actually translated with the neologism Elb(en).

    In all my years of dragging throught mud and sleeping in the woods in Germany I don’t remember seeing anything that looked like an alder. They must have cut them all out for cropland or their managed forests.

    One species is very common along rivers. (And if I get the moron who wrote it with a hyphen, I’ll CENSORED.)

    “Le Bon Dieu son fils”

    Interesting. This (with dative) is one of the methods to replace the extinct genitive in nonstandard German. A prescriptivist recently wrote a book about it: Der Dativ ist dem Genitiv sein Tod (emphasis added).
    =================
    It does happen that very conspicuous substrate features survive. Large parts of Carinthia have shifted from Slovene to German in the past 100 years. All other Bavarian-Austrian dialects have an extra phoneme /ɒ/, voiceless /b d g/*, and while there are a short and a somewhat longer version** of /s/ and the short version can go all the way to a lenis between vowels, it never becomes voiced. Slovene has no such additional vowel, voiced /b d g/, a /s/ and a /z/. Carinthian German replaces /ɒ/ by [o] (interestingly not [ɔ], which would be very close, but doesn’t exist in most or all Austrian dialects, although it does in Slovene), has voiced /b d g/, and has a single /s/ that always comes across as the long version. I don’t know how widespread these features are in Carinthia, but every single Carinthian I’ve encountered in person or on TV has them; they are clearly not stigmatized within Carinthia. Outside of Carinthia, they aren’t stigmatized either, they’re just considered stereotypically Carinthian. Only when speaking Standard German do Carinthians try to devoice their lenis plosives (…and regularly either forget or overdo it, resulting in fortes that are heard as /p t k/); the /s/ issue tends to stay, though.
    * Distinct from the likewise voiceless and likewise unaspirated /p t k/. And no, unlike in Switzerland it’s not a length difference. Utterly neglected in textbooks, as far as I know. Seems to be unique in the world except for those online audiofiles of Haida where j and ch seem to be distinguished the same way.
    ** This longer version also occurs in Standard German and is (now) written ß behind long vowels and ss elsewhere.

  113. David: The reason that erst and first and pri:mus have superlative-like endings is that they ARE superlatives, as are letzt and last and ultimus; they all represent the extrema of any finite series. Most, the English word for making superlatives, is also itself a superlative whose positive form is lost.
    The separate characters for he and she were a deliberate introduction, based on the notion that “advanced” (i.e. European) languages made the distinction, and Chinese should too, at least in writing. There is now also a character for it, reflecting the grammatical change around 1960 that allowed ta1 to be used for ‘it’ as well as ‘he’ and ‘she’; previously the usual form of ‘it’ in the spoken language was zero.
    m-l: A few newly irregular noun plurals have been introduced, at least marginally, into English: boxen in the sense ‘computers’, in deliberate and playful imitation of oxen, soon followed by VAXen, Unixen, and Macintoshen.
    In order to tell people to type &lt; instead of < in their text, I myself had to type “&amp;lt; instead of &lt;”. And in order to type the previous sentence, I instead had to type “&amp;amp;lt; instead of &amp;lt”. And so on. Y’all are not expected to understand this.

  114. David: NHD Elf is apparently a borrowing from English, under the influence of Shakespeare’s A Midsummer Night’s Dream, which is why Tolkien wanted to avoid it in the German translation: presumably English elf somehow became a sort of calque for fée. He actually wanted AHD Alb for ‘elf’; Elb was a compromise form. As he notes, elf and oaf are actually doublets, so the sense ‘incubus’ of Alp, while regrettable, is not as problematic as the sense ‘fairy’ in English. Hence also the importance of elvish, elven rather than elfish, elfin (and by analogy dwarvish rather than dwarfish).

  115. marie-lucie says:

    Etienne: in the French spoken by the Metis of Western Canada a “Mesoamerican”-like structure (“Le Bon Dieu son fils” instead of “Le fils du Bon Dieu”) exists and is clearly due to Cree and/or Ojibwe influence
    Not necessarily: this sort of structure is common in non-standard French in France, especially in children’s speech, eg “mon papa, sa voiture est au garage”, not “la voiture de mon papa est au garage” which the child would learn to write in school (“my Dad’s car is at the garage”). Is this unknown in Canada outside Metis society?

  116. marie-lucie says:

    JC: A few newly irregular noun plurals have been introduced, at least marginally, into English: boxen in the sense ‘computers’, in deliberate and playful imitation of oxen, soon followed by VAXen, Unixen, and Macintoshen.
    But that is exactly my point: as with Guiness: guini, etc, this is facetious, “deliberate and playful imitation” of an irregularity, not run of the mill, unremarkable formation as is the case with the most common morpheme: “byte: bytes” did not need to be listed as a novelty. Note also that “oxen” has never led to an analogical plural “foxen” which would be obvious and unremarkable if the -en suffix was indeed still productive. As another, slightly different analogy, the fact that an antique store might be called Ye Olde Shoppe does not mean that Middle English is still alive and well.

  117. marie-lucie says:

    English elf somehow became a sort of calque for fée
    I find the apparent equivalence between fairy and elf, and the non-equivalence of English fairy and French fée, quite puzzling.
    An elf (Fr un elfe, a Germanic borrowing) I understand to be a tiny little human-like supernatural creature living among plants, but a fairy seems to be just as small, provided with wings, and otherwise mostly female. In French une fée looks like a normal woman in size and shape, her “supernature” shown by her exceptional beauty and magical powers. She seems to be outside of nature rather than within nature like the English fairy. In Midsummer night’s dream the fairies seem to exist in male and female versions (like Titania and Oberon) and to be of human size. I am very confused about what the terms refer to.

  118. David, thanks – A. glutinosa is the species I was thinking of. I never saw much of it in Germany, either becuase we avoided rivers except in built up areas or for some other reason.
    Etinne, I see what you mean. Thanks.
    Does anyone here know what ‘kelp’ or ‘laver’ or ‘dulse’ are in Basque? Betcha there’s no PIE wiords for any of those.
    Just kidding. ‘Night all.

  119. m-l: this sort of structure is common in non-standard French in France, especially in children’s speech, eg “mon papa, sa voiture est au garage”,
    Wow, that’s also a Norwegian form: “papas sin bil står i garasjen” or “papas bil står i garasjen” or “bilen til papa står i garasjen” — in Norwegian you can use all these forms without seeming colloquial or non-standard or childlike. You lot probably picked it up off the Vikings, I’d say. It never made its way across the Channel, though.

  120. marie-lucie says:

    The title of the famous German poem and lied Erlkönig “the Alder King” ….
    David: Which, incidentally, was a misunderstanding on Schubert’s part. He should have called that guy Elfenkönig.
    Schubert wrote the music, Goethe had written the poem. If there was a misunderstanding it was Goethe’s.

  121. marie-lucie says:

    AJP: The difference between French and Norwegian is that French does not have an equivalent to Norwegian and English s on the possessor, so there is no redundancy as there is in papas sin bil.

  122. marie-lucie says:

    Jim: red and white alder here on the West Coast
    From the pictures on Wikipedia for “alder” I think that the one I am familiar with from the West Coast must be “green alder” or “Sitka alder”, but perhaps you know a more northerly area than I do.
    I think that all (or most) of the alders produce a reddish dye which was widely used before chemical dyes were invented.

  123. Marie-Lucie, Sitka alder is to the north of us. Red alder is mostly coastal an white alder grows in the Central valley and sierra nevada, in Californai. All alders have a load of different tannins, and that may be where the red dye comes form. Red alder is called that because the tannin in the wood turns it a nice cherry-wood color.
    It just strikes me as rather counter-intuitive that a weedy little thing like brown/back alder would be considered the royal tree in Celtic literature. But then again, it’s not at all obvious why the rowan would be considered a tree that blocks external influences.

  124. marie-lucie says:

    Jim, about alders, I am talking about the NW coast of Canada, near the Alaska panhandle as indicated by “Sitka”. I thought you might be in Alaska, but you are obviously farther South.

  125. there is no redundancy as there is in papas sin bil
    Silly old Vikings, wasting esses. It’s because Norway’s an oil-producing nation; they’ve got no common sense any more.

  126. I think Belfast airport is called Alder Grove, so they must have appeared in groups in some places in Europe.

  127. Etienne says:

    Marie-Lucie: in Canadian French the structure “Ma soeur son ami” (instead of “L’ami de ma soeur”) is quite marginal at best (in non-standard/child speech the more common structure is the use of the preposition “a” instead of “de”: “L’ami a ma soeur”). Its status as a Cree/Ojibwe substrate feature in Metis French is confirmed to my mind by the fact that in Quebec this structure is widely used only in the (informally acquired!) L2 French used by native speakers of various Native languages which are quite similar to Cree and Ojibwe. Tellingly, the feature is absent from the younger generation’s L2 French, which is much closer to the standard.
    David: your Carinthian example is interesting, but far more interesting to my mind is the fact that German, as you say, is a nearly aspect-free language: considering how much of what is today German-speaking territory was once Slavic-speaking (basically everything East of the Elba) it is at first glance surprising that there is so little Slavic influence on German (especially morphosyntax), especially since, at the time speakers were shifting to German, the Slavic substrate was fairly uniform (more of a dialect continuum than a collection of clearly differentiated languages): the fact that even such a homogenenous substrate failed to have much of an impact on German does strengthen my belief (as does the very weak impact of substrate languages on Romance) that Indo-European (whose substrate, as a rule, must have been quite heterogeneous) was little affected by its substrate languages.

  128. David Marjanović says:

    A prescriptivist recently wrote a book about it:

    Of course not, there’s simply not enough a prescriptivist could say about it. There’s just one chapter of two or three pages. But it gives its name to the book.

    so the sense ‘incubus’ of Alp

    Naaah. I’ve read a 19th-century explanation of it as “the sort of dwarf monster that sits on your chest, the pressure of its weight causing nightmares, and when you think its name, it vanishes”. In other words, it’s the mare in nightmare.
    Interestingly, the spelling Albtraum has become a lot more common than Alptraum since the spelling reform (1998 – 2005), even though I don’t think the reform rules don’t actually mention it, and even though the pronunciation is not identical (where I come from anyway).

    this sort of structure is common in non-standard French in France, especially in children’s speech, eg “mon papa, sa voiture est au garage”, not “la voiture de mon papa est au garage” which the child would learn to write in school

    While I can see how you could get from there to le Bon Dieu son fils, it’s not the same thing – it’s a Chinese/Japanese/Quechua-style topic-and-comment sentence (a construction that is extremely common in spoken French and almost absent from Standard Average European, though note the “Standard” part).

    Schubert wrote the music, Goethe had written the poem.

    RRRRAAARRRGH!
    <headdesk>
    <headdesk>
    <headdesk>
    <headdesk>
    CRASH
    <headfloor>
    <headfloor>
    <headfloor>
    <headfloor>
    When will I learn not to write comments after, say, 1 at night… <sob>

    a weedy little thing like brown/back alder

    No, it can get pretty tall.

    a nearly aspect-free language

    “Nearly” may still be too much; being a complete autodidact in such things, I may have exaggerated the aspect-replacing functions of parts of the verb prefix system.
    Take the very common question “are you done eating yet”. It is hast du schon fertiggegessen. Of course, fertig- looks like a prefix that indicates perfective aspect. But how grammaticalized is it?
    It’s just an adjective/adverb, also used in ich bin schon fertig “I’m done (and ready for the next task or for having spare time)” and das fertige Endprodukt “the finished end product”.
    Sure, I spelled the whole five-syllable monster as a single word (even though the spelling reform AFAIK recommends against it, one of its more stupid traits), and indeed it’s stressed on the first of the five syllables. But I can’t distinguish this from noun incorporation, like in autofahren “to go by car” or the fact that I prefer saying ich habe schon haaregewaschen over ich habe mir schon die Haare gewaschen for saying “I’ve already washed my hair”.

    considering how much of what is today German-speaking territory was once Slavic-speaking (basically everything East of the Elba) it is at first glance surprising that there is so little Slavic influence on German (especially morphosyntax)

    The territory is huge, yes, but the population density was ridiculous. The eastern expansion of German started with a population explosion due to… I forgot, probably alternating crops so that a field had to lie fallow only once every four instead of three years…, and that population explosion started in what is today the Netherlands. To this day the dialects of the western fringe of eastern Germany have far-western features. So, to a large degree, it was really the people and not the language that moved.
    It might (!) also be relevant that the Slavic aspect system underwent changes sometime around the relevant period. Perfective and imperfective aspect used to be expressed by the aorist and the imperfect tense, then that collapsed to some extent while various prefixes and other features took over these functions and became generalized; in Russian and AFAIK Polish the aorist and the imperfect have completely died out, and the only past tense actually consists of a participle. – But that’s spontaneous speculation on my part.
    (BTW, aren’t West and East Slavic still a dialect continuum or nearly so?)

  129. marie-lucie says:

    Etienne: the genitive with à instead of Standard de is also very common in non-standard French French, as in la bande à Bonnot, “the Bonnot gang”, a criminal group well-known in my parents’ time.
    David: this sort of structure … eg “mon papa, sa voiture est au garage”…
    - While I can see how you could get from there to le Bon Dieu son fils, it’s not the same thing
    I know that it is not the same thing and does not have the same origin, but as you say it can easily lead to being interpreted as a genitive construction of a type which seems fairly common (in older English people mistakenly interpreted “John’s book” as “John his book”). That’s why I suggested that perhaps the Metis construction was not necessarily from Cree/Ojibwe. On the other hand, if the structure had existed in local French (but Etienne says it did not), its potential genitive meaning would have been reinforced by the existence of the Cree/Objiwe genitive construction.

  130. Etienne says:

    David: I’ll grant that the expansion of German eastwards was due to demographic expansion. Still, one would expect some of the more isolated areas to shift to German rather than to be colonized by Germans. And whereas we have to this day a Slavic enclave in German-speaking territory (Sorbian), I am unaware of any isolated German dialect (in Germany!) which exhibits the sort of Slavic features which other languages in contact with Slavic have to show: where is the German dialect in Germany which, like Romani in Russia, has shed its definite articles and borrowed a set of aspect-marking morphemes?
    Your point about Slavic at the time not having a verb system as aspect-centered as most Modern Slavic languages today is well-taken: still, the aspect-marking morphemes, being more readily segmentable than the imperfect- and aorist-marking ones, would have been easier to borrow thereby. And it cuts both ways: the Slavic substrate differed from both (most) Modern Slavic languages and Modern German (standard + dialectal) in important ways: for example, the dual was certainly alive and kicking at the time, but this fact seems not to have prevented German dialects (without any exception, to my knowledge) from shedding the dual as thoroughly as its Germanic sisters.

  131. Thanks, Lameen, for answering my query re Berber.
    My comment, back there on May 1, re ‘barbarian’ should have been on the Beckwith thread re Central Asia. Sorry, folks.

  132. bruessel says:

    “Interestingly, the spelling Albtraum has become a lot more common than Alptraum since the spelling reform (1998 – 2005), even though I don’t think the reform rules [don't] actually mention it”.
    My Duden certainly mentions it and also I recall it being discussed at the time the reform was introduced.

  133. Hat, forgive me for this inordinately long and not even original comment, but I felt it belonged here.
    Marie-Lucie, here’s Tolkien on elf and fairy, from “On Fairy-Stories”. Footnotes so marked are his; other bracketed insertions are mine; I’ve broken up some of his long paragraphs.

    [Tolkien starts by quoting OED1 s.v. fairies:] “supernatural beings of diminutive size, in popular belief supposed to possess magical powers and to have great influence for good or evil over the affairs of man.”
    Supernatural is a dangerous and difficult word in any of its senses, looser or stricter. But to fairies it can hardly be applied, unless super is taken merely as a superlative prefix. For it is man who is, in contrast to fairies, supernatural (and often of diminutive stature); whereas they are natural, far more natural than he. Such is their doom. The road to fairyland is not the road to Heaven; nor even to Hell, I believe, though some have held that it may lead thither indirectly by the Devil’s tithe.
    O see ye not yon narrow road
    So thick beset wi’ thorns and briers?
    That is the path of Righteousness,
    Though after it but few inquires.

    And see ye not yon braid, braid road
    That lies across the lily leven [lea]?
    That is the path of Wickedness,
    Though some call it the Road to Heaven.

    And see ye not yon bonny road
    That winds about yon fernie brae [hill]?
    That is the road to fair Elfland,
    Where thou and I this night maun gae [must go].

    [This word leven, by the way, is a hapax legomenon, according to the marvelous (and freely available) Dictionary of the Scots Language.]
    As for diminutive size: I do not deny that the notion is a leading one in modern use. I have often thought that it would be interesting to try to find out how that has come to be so; but my knowledge is not sufficient for a certain answer. Of old there were indeed some inhabitants of Faerie that were small (though hardly diminutive), but smallness was not characteristic of that people as a whole. The diminutive being, elf or fairy, is (I guess) in England largely a sophisticated product of literary fancy. [Footnote: I am speaking of developments before the growth of interest in the folk-lore of other
    countries. The English words, such as elf, have long been influenced by French (from which fay and Faërie, fairy are derived); but in later times, through their use in translation, both fairy and elf have acquired much of the atmosphere of German, Scandinavian, and Celtic tales, and many characteristics of the hulda-folk, the daoine sidhe, and the tylwyth teg.] [To which we must now add, and of Tolkien's own elves.]
    It is perhaps not unnatural that in England, the land where the love of the delicate and fine has often reappeared in art, fancy should in this matter turn towards the dainty and diminutive, as in France it went to court and put on powder and diamonds. Yet I suspect that this flower-and-butterfly minuteness was also a product of “rationalization”, which transformed the glamour of Elfland into mere finesse, and invisibility into a fragility that could hide in a cowslip or shrink behind a blade of grass. It seems to become fashionable soon after the great voyages had begun to make the world seem too narrow to hold both men and elves; when the magic land of Hy Breasail in the West had become the mere Brazils, the land of red-dye-wood. [Footnote: For the probability that the Irish Hy Breasail played a part in the naming of Brazil see Nansen, In Northern Mists, ii, 223-30.]
    In any case it was largely a literary business in which William Shakespeare and Michael Drayton played a part. [Footnote: Their influence was not confined to England. German Elf, Elfe appears to be derived from A Midsummer-night’s Dream, in Wieland’s translation (1764).] Drayton’s Nymphidia is one ancestor of that long line of flower-fairies and fluttering sprites with antennae that I so disliked as a child, and which my children in their turn detested. Andrew Lang had similar feelings. In the preface to the Lilac Fairy Book he refers to the tales of tiresome contemporary authors: “they always begin with a little boy or girl who goes out and meets the fairies of polyanthuses and gardenias and apple-blossom. . . . These fairies try to be funny and fail; or they try to preach and succeed.”
    But the business began, as I have said, long before the nineteenth century, and long ago achieved tiresomeness, certainly the tiresomeness of trying to be funny and failing. Drayton’s Nymphidia is, considered as a fairy-story (a story about fairies), one of the worst ever written. The palace of Oberon has walls of spider’s legs,
    And windows of the eyes of cats,
    And for the roof, instead of slats,
    Is covered with the wings of bats.

    The knight Pigwiggen rides on a frisky earwig, and sends his love, Queen Mab, a bracelet of emmets’ eyes, making an assignation in a cowslip-flower. But the tale that is told amid all this prettiness is a dull story of intrigue and sly go-betweens; the gallant knight and angry husband fall into the mire, and their wrath is stilled by a draught of the waters of Lethe. It would have been better if Lethe had swallowed the whole affair. Oberon, Mab, and Pigwiggen may be diminutive elves or fairies, as Arthur, Guinevere, and Lancelot are not; but the good and evil story of Arthur’s court is a “fairy-story” rather than this tale of Oberon.
    Fairy, as a noun more or less equivalent to elf, is a relatively modern word, hardly used until the Tudor period. The first quotation in the Oxford Dictionary (the only one before A.D. 1450) is significant. It is taken from the poet Gower: as he were a faierie. But this Gower did not say. He wrote as he were of faierie, “as if he were come from Faërie.” Gower was describing a young gallant who seeks to bewitch the hearts of the maidens in church.
    His croket [hair-curl] kembd [combed] and thereon set
    A Nouche [projecting part] with a chapelet [wreath],
    Or elles one of grene leves
    Which late com out of the greves [groves],
    Al for [So that] he sholde seme freissh;
    And thus he loketh on [appears in] the fleissh,
    Riht as [Just like] an hauk which hath a sihte
    Upon the foul ther [where] he schal lihte,
    And as he were of faierie
    He scheweth him tofore here yhe [their eyes].

    [Footnote: Confessio Amantis, v. 7065 ff.]
    This is a young man of mortal blood and bone; but he gives a much better picture of the inhabitants of Elf-land than the definition of a “fairy” under which he is, by a double error, placed. For the trouble with the real folk of Faërie is that they do not always look like what they are; and they put on the pride and beauty that we would fain wear ourselves. At least part of the magic that they wield for the good or evil of man is power to play on the desires of his body and his heart. The Queen of Elfland, who carried off Thomas the Rhymer upon her milk-white steed swifter than the wind, came riding by the Eildon Tree [Scots: 'aged tree'] as a lady, if one of enchanting beauty. So that Spenser was in the true tradition when he called the knights of his Faërie by the name of Elfe. It belonged to such knights as Sir Guyon rather than to Pigwiggen armed with a hornet’s sting.

  134. No forgiveness required—the more the merrier!

  135. marie-lucie says:

    JC, thank you so much, this quotation is great, and not too long at all. It makes me want to read Tolkien, something that had not attracted me before.
    It must be through a misunderstanding of the French word féerie, originally “supernatural power or world of the fairies”, that the word “fairy” came to be applied to one of the beings in question, in replacement of earlier “fay” (as in “Morgan le Fay”, from la fée Morgane, after English had lost much of the gender distinction in nouns). So Spenser’s “Faerie Queene” perhaps did not mean the queen who was a fairy but the queen of Fairyland.
    I said earlier that une fée was exceptionally beautiful, but that is only true of the good fairies (like the one who enables Cinderella to attend the ball); the nasty ones (such as the one who wants Sleeping Beauty to be put to death) are represented as old hags. Perhaps the Victorian miniaturization of fairies was a way of diminishing the actual power of those feminine characters: such tiny beings would not be capable of wielding magic wands and change the destinies of humans.
    Note however that in Walt Disney’s version of Cinderella the fairy godmother appeared as a stout, grandmotherly figure of normal size, not as a tiny little thing that could live in a flower, or a beautiful woman in the prime of her youth.

  136. So Spenser’s “Faerie Queene” perhaps did not mean the queen who was a fairy but the queen of Fairyland.
    I’m not sure that’s the best way to look at it. He took the archaic faerie / -ie and applied it to his own world. In as much as it means something beyond “what Spenser wrote about,” that would I suppose just be “imaginary.” But then, I believe he used it lexically in all its ways: adj. for unreal, noun for the land, collective for the inhabitants, and individually for one of them. (All but the last going back to Chaucer.) Even in his short introductory letter to Raleigh, he uses “Faerie Queene,” “Queene of Faery land” and “Queene of the Faeries” more or less interchangeably. He established a strong enough brand that the spelling with ae continues today in more literary contexts.

  137. Marie-Lucie:
    Tolkien’s public addresses, including this one, can mostly be found in the posthumous book The Monsters and the Critics and Other Essays. The Tolkien Journal, available on MUSE, has recently reprinted some of his more narrowly philological pieces, one on the god-name Nodens and two on the obscure word Sigelwara in Beowulf, usually glossed ‘Ethiopia’ but by his analysis referring to the mythical offspring of a pagan fire-god corresponding to Norse Muspell.
    The man wasn’t just a great writer of fantasy, he was a brilliant linguist as well, though undoubtedly old-school (now, perhaps, to be called very-old-school). You can see some of his characteristic turns of phrase above: the quiet irony of “unless super is taken merely as a superlative prefix”; the archaisms like “of old” and “that we would fain wear ourselves” that really work because he knows how to use them and how not to, the use of “glamour” in the modern sense and its old sense of “sorcery” simultaneously. Read him, you’ll like him.

  138. (All but the last going back to Chaucer.) … That last being the disputed use in Gower from the Tolkien quotation John Cowan gave. There are editions matching the OED’s version, but the definitive ones seem to agree on of.
    Incidentally, Tolkien worked as an assistant at the OED after the War, on the W‘s. See the walnut etymology there for some of his work. Or read Ring of Words: Tolkien and the Oxford English Dictionary.

  139. marie-lucie says:

    Thank you, JC and MMcM, I will keep an eye out for those books. I did notice the virtuosity and charm of Tolkien’s pointedly old-fashioned style.

  140. David Marjanović says:

    And it cuts both ways: the Slavic substrate differed from both (most) Modern Slavic languages and Modern German (standard + dialectal) in important ways: for example, the dual was certainly alive and kicking at the time, but this fact seems not to have prevented German dialects (without any exception, to my knowledge) from shedding the dual as thoroughly as its Germanic sisters.

    That’s because the dual was already gone, I suppose. It’s very sparsely attested in Old High German, and I don’t know where geographically; in Middle High German, around the beginning of which lies the relevant period, it was absent.
    In many Bavarian-Austrian dialects, the 2nd person plural pronouns are derived from the ancient dual form, but that’s all I know of. OK, the word “both” (which is the dual of “all”, when you think about it) survives in Standard German and some (but not all) dialects, but that’s it. (In those that have lost “both”, you say “all two” like in French.) No equivalent to “either” exists in German.

Speak Your Mind

*