18 Mar 2008

Semitic and IE in the Neolithic: How intensive was the language contact?

Okay, let's tackle this issue once and for all. The "borrowing hierarchy" mentioned in Elsik and Matras' Markedness and Language Change: The Romani Sample (2006) has got my mind turning and churning ever since I read about it. While I was always aware of the widely accepted early IE-Semitic contacts[1] and presumed that it was the result of direct contact, I never thought about quantifying the level of contact between these two protolanguages until this past year. That is to say, whether it was a light contact or whether it was an "intensive" contact one shade removed from creolization. Despite the naysaying nihilist badnewsbears who say the past can never be figured out, I for one have no interest in desperate people's grandiose and arrogant prophecies about what our human species somehow cannot know in the future. I want to know exactly what happened. Scientific inquiry is about asking questions and seeking answers and those who say "I give up!" should be left behind in the dirt to rot in their depression.

About the borrowing hierarchy

The borrowing hierarchy presented here prioritizes which kinds of words are more likely to be borrowed in a language depending on the intensity of contact. So let's start with verbs. It's suggested on page 320 of Markedness and Language Change that the following borrowing hierarchy exists:
content verbs > modal verbs > existential verbs
So what this is saying is that verbs from general vocabulary (content verbs) are likely to be borrowed first before other verbs that express moods such as "can", "would", etc. If contact is intensive enough we should even observe borrowed verbs denoting existence such as "to be". Later on in the same page of this book, it specifies the hierarchy followed by nouns:
content noun > pronoun
So again, we expect ordinary nouns borrowed into a language before loaned pronouns find their way into a language. If we see a borrowed pronoun, it can be a sign of rather intensive language contact, as happened in Old English when the language borrowed the Norse pronoun "they". To add to this, I personally suspect that third person pronouns are more likely to be borrowed from one language to the next than first or second person pronouns and I might be tempted to modify the chain to: content noun > 3p pronoun > 1p/2p pronoun. But I digress. That assertion isn't important to the evidence I'm about to shamelessly suggest to demonstrate intensive contacts between an early stage of Proto-Indo-European (PIE) and Proto-Semitic (PSem) around 5500 BCE.

Evidence for intensive contacts between Proto-IE & Proto-Semitic

As already explained with a footnote, the idea that early Proto-IE speakers and Proto-Semitic speakers were in contact with each other isn't controversial, but debate continues on as to how, where and when they were in contact with each other. The neolithic is really the most sensible time for this to have occurred since it gives widening trade as a motive for this cultural and linguistic exchange. The how is clear then: growing trading networks in the Neolithic and exchange of goods. The relations at this time between the Balkans and Syria are reflected firmly in archaeology which show commonalities in ceramics and material culture. However, Neolithic trade only gives us a window between 6000 and 4000 BCE, so as far as I'm concerned, the precise "when" is too obvious to feign confusion. The matter of where this contact occurred continues to be a senseless debate, egged on by sensationalists like Ivanov and Gamkrelidze who attempt desperately to place Indo-European in Anatolia. It doesn't fly. The least controversial hypothesis is that PIE was centered around modernday Ukraine and there's no reason to rebel against this economical view. However, it must be understood that if this contact started a millenium or more before PIE proper, there's no guarantee that earlier stages of the language existed further to the south as I suggested earlier for the stage of PIE I refer to as Mid IE (MIE).
So let's get straight to the chase and start mapping the possible loans from PSem into PIE and see how intensive this contact might have been:

content verbs:
PSem *šáðrawu 'he causes to scatter (caus.)' (root *[ðrw])
=> PIE *streu- 'to scatter' (via MIE *satréwa-)
modal verbs:
existential verbs:
PSem *yiθ 'there is'
=> PIE *ʔes- 'to be' (via MIE *es-)

PSem *báwiʔu 'it is come (stat.)'
=> (?) PIE *bʰeuh₂- 'to appear, to become' (via MIE *béuxa-)
content nouns:
PSem *θáwru 'bull'
=> PIE *táuro- (via early Late IE *tä́urə-)

PSem *gádyu 'kid, young goat'
=> PIE *ǵʰáido- 'goat' (via early Late IE *gä́id̰ə-)

PSem *sábʕatum 'seven'
=> PIE *septḿ̥ (via MIE *séptam)
PSem *šu 'him, himself'
=> PIE *swe 'oneself' (via MIE *sʷa).
So I hope you all enjoy these ideas. I'm still stuck on finding an example for "modal verbs" because PIE morphology expressed mood by way of inflection not seperate words but I'll be continuing to adapt my argument in the future. If I goofed up on the Semitic verbal forms, please let me know and correct me because I had to derive them from their root forms and I'm more familiar with IE and Aegean languages than Proto-Semitic which I've sadly neglected far too long.

[1] Hock/Joseph, Language History, Language Change, and Language Relationship: An Introduction to Historical and Comparative Linguistics (1996), p.513 (see link): "Given the evidence of 'bull' and other pastoral/agricultural words in western Indo-European languages that are likely to be borrowed from Semitic or Afro-Asiatic, one may begin to wonder whether the western Indo-European agricultural words for 'field', 'plough', and 'sow, seed' may likewise owe their origin to Semitic or Afro-Asiatic influence. Archeologists find much evidence indicating that many important aspects of European agriculture, perhaps even all of agriculture, spread from the ancient Near East."

(Mar 19 2008) I decided to throw PSem *báwiʔu and PIE *bʰeuh₂- into the list as a point of discussion. The verb 'to become' is considered an existential verb in the cited book, and hence would be more evidence of the strongest kind of language contact. I add a question mark here because I'm undecided as to which of two competing proposals work best. Nostraticists often link PIE *bʰeuh₂- to Uralic *puxi 'tree' with an underlying meaning of 'to grow up from the ground' and hence later 'to appear' and 'to become' in PIE. This is plausible and I must admit that in order to establish a clear link between the PIE and PSem words, the fact that the laryngeals are incongruent (i.e. a glottal stop in PSem but a uvular fricative in PIE) needs to be addressed. Otherwise however there is an alluring semantic and phonetic link between the two that would postdate the etymology that many Nostraticists provide. So chew on that for a while. ;)


  1. *táuro- had always been one that was way too suspicious to me too. I always figured greek borrowed it from Semitic or the other way around
    But I just realised Dutch has 'stier' going back to proto-germanic *stiuraz which should go back to PIE *(s)teuro- I guess. The vocalism of Greek is pretty odd though.

    I'm not known to the PSem. *šu, but if it is as you say; I think that one is pretty convincing.

    Nor was I aware of the causative prefix *ša- is that related somehow to the Arabic future marker *sa-?

    And just as a quick musing, might this *ša- prefix somehow be the origin of that mysterious s-mobilé in PIE?

  2. Phoenix: "*táuro- had always been one that was way too suspicious to me too. [...] Dutch has 'stier' going back to proto-germanic *stiuraz which should go back to PIE *(s)teuro- I guess."

    I'm a little skeptical of this word for different reasons. How sure are we that the forms in *st- really go back to the same word or even to PIE? Apparently some are skeptical like me and don't include Germanic's words in the cognate set, leaving it to just the Hellenic, Italic and Balto-Slavic branches to prove a stricter correspondence.

    Phoenix: "I'm not known to the PSem. *šu, but if it is as you say; I think that one is pretty convincing."

    No, don't take my word for it. I'm too cocky for my own good so you need to question everything I say! Life isn't fun without a fight. :) I base this connection on the fact that Proto-Semitic didn't necessarily employ reflexive pronouns for reflexive nuances but often used regular pronouns and affixes to refer back to the same 3rd person. I also notice a pattern of PIE *sw- for PSem *š- which would particularly make sense if Proto-Semitic *š- was pronounced with lip-rounding, something that occurs in many other languages including English, and I suspect for internal reasons that Mid IE had a labialized sibilant, *sʷ, that was dephonemicized into two phonemes, *sw, in Late IE.

    Phoenix: "Nor was I aware of the causative prefix *ša- is that related somehow to the Arabic future marker *sa-?"

    Yep, Semitic causative *ša- is real and also found just outside the Semitic family in Ancient Egyptian s- hence Egyptian s-mn 'to cause to remain; to establish'.

    As for this Arabic future construction which I wasn't aware of until you mentioned it, the particle sa- may be an abbreviated form of sawfa which is also used for the future tense apparently. The causative on the other hand pops up in Arabic as ʔ- (< *h- < *š-).

    Phoenix: "And just as a quick musing, might this *ša- prefix somehow be the origin of that mysterious s-mobile in PIE?"

    I personally suspect that PIE retains fossilized Semitic morphology as seen in *septm. So I would expect that if there were lots of loans coming into Mid IE from Semitic then there should be lots of verb pairs showing the same Semitic verb with different prefixes like the causative in *š-, reflexive in *n-, participial in *m-, etc. However, I don't think that the added matter of the "disappearing act" of PIE *s- can be blamed squarely on Semitic but is probably either something internal to PIE grammar and phonetics or is an incredible misanalysis on the part of Indoeuropeanists. But I wouldn't dare suggest that Indoeuropeanists are fallible or anything less than god-like superbeings, hehe. Personally, I have no idea what would be behind s-mobiles other than sporadic changes provoked by Post-IE phonotactics.

  3. Why is PS *ð turning into MIE t in "scatter"? Is there Slavic or similar evidence to not simply reconstruct the word as LIE *s-dreu- with the initial /s/ causing devoicing?

    Also: are you reconstructing a labialized *s for MIE in itself or just to explain the Semitic connections? Because if it's the former, I'm not so sure you need the assumption of labialization in Semitic too. This dissertation mentions also that secondary labialization is not rare in retroflexes - but as a part of a discussion noting that this is due to both features having a similar acoostic output. So if MIE had an inherited unitary *sʷ, even a non-labialized retroflex sibilant would probably be transformed into that when loaning words!

    Wikipedia (yeah yeah, pinch of salt…) notes that PS *š is hypothetized go back to a plain *s - and for such a shift, a retroflex intermediate would not be an unreasonable assumption. The shift is from apical alveolar to laminal postalveolar, and a retroflex mid-stage would require only the contact point to shift first, the tongue shape 2nd. (CF also ruki.)

  4. Tropylium: "Why is PS *ð turning into MIE t in 'scatter'?"

    PIE *streu- predicts MIE *saT(a)réwa- (where T is one of *t, *d̰ or *d). Given *t in PIE, *satréwa- is most economical unless this is indeed a loan from Proto-Semitic, in which case *sadréwa- is more natural with devoicing of resultant *sd- to *st- during Syncope, as you suggested. However, your Slavic ramble appears to be a non sequitur.

    Tropylium: "Also: are you reconstructing a labialized *s for MIE in itself or just to explain the Semitic connections?"

    It's based on internal considerations, not Proto-Semitic. Although, Proto-Semitic does confirm the phonetics of that time period that I theorize so I consider it merely "icing on the cake".

    For reasons specific to Indo-European itself, I reconstruct a dephonemicization of all of MIE's labialized dental sounds because this process affects a-Epenthesis such that I can show that *oktou originally meant "fours" in Old IE. That is, PIE *ʔoktóu < eLIE *ʔaktwáu < MIE *kʷatʷáxʷa < OIE *kʷátʷa "four" plus collective suffix *-xʷa (> PIE *-h2). The a-Epenthesis would not have been triggered unless *tʷ had first eroded into two phonemes *tw and thus the same development is prescribed for all labialized dental sounds including *sʷ. So MIE *sʷ becomes PIE *sw.

    Labialized dentals help fill out the gap in labialization of the IE sound inventory. The ultimate origin of labialized phonemes in PIE is no doubt due to the erosion of the vowel system and transfer of the salient characteristics of vowel to neighbouring consonants. Many have also theorized this as the source of the labialized phonemes and impoverished vowel system in neighbouring Abkhaz-Adhyghe languages.

    Tropylium: "Wikipedia (yeah yeah, pinch of salt…) notes that PS *š is hypothetized go back to a plain *s"

    Whoopdeedo. If you can't cite an academic source of information which is not edited by children and a random assortment of possible lunatics, then silence is better than debasing yourself with such poor argumentation and defense.

    Semitic phonology can only be determined through the careful examination of *Semitic* languages, not through Egyptian, not through Cushitic and certainly not through a faceless wikipedian's sourceless pet theory about Afro-Asiatic. Your reasoning is now becoming erratic. I suggest that you radically improve your argument and elevate your standards.

  5. Slavic languages contain a fair share of /z/ + obstruent onsets, in a similar distribution as with /s/. I'm under the impression that these result from /s/ before a voiced stop, and so Slavic evidence would be able to tell apart whether the onset was originally voiced or not. Is this incorrect?


    I disagree that quoting (not citing; note the difference) Wikipedia is worse than silence. NPOV or similar edit wars are not hard to spot, and misinformation that remains in addition to this is a small percentage and randomly distributed. So I take a statistical stance: on non-vital issues and in absence of counterarguments, I think it's generally reasonable to accept what Wikipedia says (but with the mentioned metaphorical pinch of salt.)

    And yes, this still does not make a solid argument for *š < **s. However, I'm not arguing for the change, but just going over what it would imply. Do you take issue with speculation?

    -On another note, you say comparision with other Afro-Asiatic languages can not provide us with any information about the phonology of Semitic; should I infer that you do not accept the validity of Afro-Asiatic as a language family?

  6. Tropylium: "Slavic languages contain a fair share of /z/ + obstruent onsets, in a similar distribution as with /s/."

    Irrelevant. This is what's actually reconstructed: *streu-. You could have googled that yourself.

    Tropylium: "I disagree that quoting (not citing; note the difference) Wikipedia is worse than silence."

    Your blind, cultish faith in Wikipedia is irrelevant. This blog isn't a democracy. Since it's clear that trolls attack Wikipedia 24 hours a day, there's no guarantee in its accuracy at any given moment. You don't even need to cite it when there are so many other sources of information!

    You will simply refrain in the future from citing Wikipedia or your comments will be thrown in the trash. End of discussion.

    Tropylium: "should I infer that you do not accept the validity of Afro-Asiatic as a language family?"

    No, you shouldn't. It would in fact be illogical to do so since the self-explanatory insistence on using Semitic data to reconstruct Proto-Semitic has no bearing whatsoever on the validity of the Afro-Asiatic family. Your logic is failing you.


    I'm beginning to get annoyed by your inability to keep focussed on facts. Counterarguments based only on tentative long-range theories and Wikipedia warrants a strict warning: Your comments may be deleted in the future if you cannot, or are unwilling to, elevate your standards of reasoning.

  7. As an added piece of information, Tropylium, both Semitic *s and are reflected in Egyptian as s so naturally Egyptian data can have no bearing on Semitic phonology anyway, even if we lower our standards of logic to suit your insufficient argument. As every non-relativist can see, your "speculation" when it's so clearly ignorant of facts is just wasting everyone's time.

  8. Oh, fine, I'll stop even mentioning Wikipedia. I'd appreciate if you in turn stopped assuming things such as "blind, cultish faith" of me.

    Regardless, the point I was trying to make still stands, far as I can see: namely, labialization of *š is not the only possibility that would explain why it's reflected as *sw in theorized loanwords into IE. Plain retroflexion would be another choice.

    So what does this still entirely hypothetic possibility matter as long as we don't have data in direct support of either? Well, previously, an assumption of labialization was required. Now, labialization or retroflexion. This is a less specific assumption and therefore the loanwords in question now have an (admittedly marginally) better footing.

    (Altho if every point I make requires three or more replies to communicate, I'm not sure if it's worth the time of either of us.)


    The Afro-Asiatic question was a bit of a stretch, based on your wording can not provide rather than does not, since it is not absolutely impossible to gain information about a proto-language's phonetics/phonology from the language's sister groups. (For example, let's say we have in a reconstructed phoneme of uncertain realization, but in its place in sister languages we consistently find a normal consonant segment of very certain realization. From this information we could gather that this is also the previously uncertain phoneme's most likely pronunciation.)

    ((Disclaimer: All resemblance of this example to actual language families is coincidental.))

  9. Tropylium: "[...]labialization of *š is not the only possibility that would explain why it's reflected as *sw in theorized loanwords into IE."

    Okay, so you could have tried to support your claim simply with something from a standard book on phonetics, rather than using unverified snips from Wikipedia or someone's Nostratic voodoo theory. Now hopefully you see where you erred and why it appeared cultish after a while. Cut to the chase, dude!

    However, retroflexion isn't a necessary factor in lip-rounding at all. The corresponding Mandarin retroflex esh is **not** rounded. So perhaps you give too much weight to retroflexion. Also retroflexion is not a distinct feature in the Proto-Semitic inventory, so you're delving into secondary co-articulations of a specific sound.