19 July 2013

Consider a Hammer

Figure1: Co-opting a natural object
According to Wikipedia, “A hammer is a tool meant to deliver an impact to an object. The most common uses for hammers are to drive nails, fit parts, forge metal and break apart objects. Hammers are often designed for a specific purpose, and vary in their shape and structure.” Hammers have been shaped by the functions they typically perform. A heavy metal head fixed on a light handle stores kinetic energy before the blow is delivered. The length, cross-section and shape of the handle are ergonomically adapted to human handgrip and typical working conditions. There are functionally motivated differences between, say, a light claw hammer used for driving and removing nails and a heavy-duty sledgehammer used for tearing down walls.

Figure 2: Putting a handle on it
The ancestors of all hammers were natural cobbles used as hammerstones by Palaeolithic humans (as well as earlier hominins). They carried out some of the same functions as modern hammers, albeit less efficiently. There was no handle (its function was played by the user’s arm), and hammerstones used for different purposes had the same general shape, differing mostly in size and weight. Small gradual improvements and  occasional major inventions (a wooden handle, the use of bronze or steel instead of stone) transformed the primitive tool visible in Figure 1 into a more sophisticated version (Figure 2), and finally into a fully streamlined  modern hammer (Figure 3).

Figure 3: Shaped by its functions
Of course a hammer can be used for many other purposes beside pounding nails into things or splitting hard objects. It can serve as a makeshift paperweight, a percussion instrument (as in Penderecki’s De Natura Sonoris No. 2), an improvised weapon, and even as a ritual or ceremonial object – for example, the emblem of a smithing god. Such accidental functions do not normally influence the evolution of hammers. If a type of hammer acquires a historically stable secondary function (e.g. removing nails), you can see the characteristic adaptations (a flattened and rounded claw), copied and perfected by new generations of hammer manufacturers. Ad hoc functions have no such consequences. Nobody modifies the shape of a hammer to make it a better paperweight. Its only when a hammer is regularly recruited for a new task that adaptation begins to shape it in a new way. This may lead to the emergence of highly specialised hammers (such as the judge’s gavel or the doctor’s knee mallet).

The first hammers were not designed by anyone. Our distant ancestors learnt to select naturally formed stones. Then they learnt to improve their shape, fix them on a handle to optimise energy transmission, etc. The functional features of a hammer are those that have been consistently selected for in the past. It’s always possible to use a tool in an unconventional way, but such occasional applications don’t explain why the tool looks the way it does. Some features (for example, the colour of the handle) are free to vary. They are non-adaptive, devoid of functional importance.

I hope you can see how this hammer analogy can be applied to linguistic structures. That’s what will be done in the next post.

47 comments:

  1. There is also the interesting fact that English hammer and Russian камень 'stone' are relatives, from PIE *h₂éḱmō. This tells us that there was once a time when hammers were made of stone, quite independently of any archaeological evidence.

    ReplyDelete
  2. It's difficult to say if they are really related. Slavic *kamy/*kamen- is mysteriously distorted (as if by a late metathesis: *h₂áḱ-mon- > PBSl. *Hák-men- > *káH-men-). If it weren't for the existence of regular Baltic reflexes of the PIE word, it would be hard to believe that we are dealing with a real cognate of the Greek and Indo-Iranian words. Germanic *xamaraz is still more problematic. It's a thematic stem, and the *r is hard to explain. Animate nouns do not normally show any *-r/n- heteroclisy. The vowel of the first syllable is short, which rules out the metathetic explanation that works for Slavic. The suffix is more likely a reflex of thematic *-ero- than anything related to *-mVn-. But if so, the *m is part of the root (whatever it is).

    ReplyDelete
    Replies
    1. This is a tangent, but are English 'chimney' and Swedish 'kamin' "stove" related to '*kamy/kamen'?

      Delete
    2. Chimney is a Romance loan, from Lat. camīnus, which itself was borrowed from Greek. Gk. kámīnos 'oven, furnace, kiln' is a bit mysterious. Beekes says (in his Greek etymological dictionary) that comparison with Slavic *kamy "is possible". I don't agree. The short a in the first syllble, the long ī in the suffix, the different stem type and the different meaning all militate against a connection.

      Delete
  3. I seem to recall there being an inherited Balto-Fennic word meaning 'rock' that could be the source of the Germanic word. Can't remember the exact shape of the word now, however.

    Anders

    ReplyDelete
    Replies
    1. That would be interesting, especially because the Finno-Ugric 'hammer' word (*waśara) is in turn thought to be a loan from Proto(-Indo?)-Iranian.

      Delete
    2. Now I found it. It's not 'rock', but 'back of a knife/axe'. Finnish hamara 'back of a knife/axe' and Samic šimer, -bmer- 'back, back of a knife'. I just took the data from the StarLing database, so I don't know if the forms are correct. Anyway, if this is the source of the Germanic word for 'hammer', the immediate source must be Balto-Fennic, with *š- > h-.

      The idea may well have been suggested to me by Adam Hyllested, but I don't know if it is his or if he got it from someone else.

      Anders

      Delete
    3. Oh yes, indeed, Adam mentions it here:
      http://www.nostratic.net/books/(174)Hyllested%20-%20Finn%20loan%20Germ.pdf

      Delete
  4. Thanks! I personally like the idea of *hamara- being a loan from a Balto-Fennic 'back of an axe' (which you often use as a hammer anyway). At least when the Germanic word means 'hammmer'. I don't know what to do with the alternative (just North Germanic?) meaning 'rock'. Is it just a metaphorical use of 'hammer'? As I recall, people seem to regard the latter meaning, 'rock', as primary in Germanic, but that may just be because they wanted to connect it with PIE *h2ak'men-. Or is there another reason for doing so?

    Anders

    ReplyDelete
    Replies
    1. Zoëga's glosses for OIc. hamarr are: (1) hammer, (2) back of an axe, (3) crag, precipice (rather than just "stone"; there are also compounds like berghamarr 'rocky precipice', hamarrifa 'rift in a crag'). The first and second match the BF etymology perfectly, and the third is more likely figurative than primary. My gut feeling is that Adam is right about the direction of borrowing.

      Delete
    2. There is of course a small problem with the direction of the borrowing: the when. As far as I can see there is no other Balto-Fennic loanword that is Common Germanic.

      The oldest evidence by the way for the word 'hammer' in Germanic by the way is a peronal name: lat.-gmc. (gen.sg.) Chamari.

      Delete
    3. Youy mean M(arcus) Chamari f(ilius) in the Enzen dedication, right? That's still NWGmc. rather than Common Germanic.

      Delete
    4. The Chamari was just an info because the name is most often overlooked when the Gmc. forms are cited - it was of course not intended as Commen Gmc. evidence ...
      Still when a loanword is considered it is clear that it must have wandered into the Westgmc. area till the 2./3. century AD - a process for which there is just no parallel.

      Just read in NIL that by Benveniste oi. aśmari- f. was once seen as a reflex of an old men/r-heterocliticon (rejected by EWaia) - when one holds up to this idea the hammer-words could perhaps also be seen as a reflex of this (haven't checked this out).

      Delete
    5. Still, very early contacts between Germanic (to all intents and purposes the "Common" stage) and BF are confirmed by demonstrably archaic loans (in the opposite direction) as kuningas and rengas. Asymmetry in language contact is a normal phenomenon. There are lost of Iranian and East Germanic loans in Slavic, for example, versus just a few borrowings in the other direction, like Goth. plinsjan < *plęsati 'dance'.

      The existence of a deverbal neuter like *h₂áḱ-mr̥/-m(e)n- underlying the animate 'stone' word is of course possible but I'm not aware of any surviving reflexes of such a form. Ockham's Razor favours the miundane analysis of aśmari- as *h₂áḱ-mn̥- + -ri- over more exotic explanations.

      Delete
    6. Hyllested's Finnic → Germanic loan proposal does not sound bad at all. However, the standard account in Fennistics is the opposite — loaning from Germanic into Finnic. The proposed West Uralic cognates do not really abide to the typical sound correspondences, and no original Uralic root can be securely set up. (Samic *sɤmērē would incidate former **šimara; Mordvinic *šov would indicate former **šuvV.)

      Factoring in the pre-II reconstruction *h₂áḱmn̥ri seems promising however. A very early (or very archaic) layer of Indo-European loans into West Uralic is known, in which initial laryngeals are substituted by *k-, medial ones by *-š-. Perhaps this one word managed to make it in with initial *š-.

      I even suspect that the k/š distinction could be a chronological one rather than positional. The latter layer contains an abundance of words preserved only in Finnish, or maybe a small subset of Finnic (e.g. Finnish tehdas "factory", originally "site, stall"; ← PIE *dʰeh₁tos), while words of the former layer are frequently also found in Samic and Mordvinic (e.g. Finnic *kesä, Samic *keasē, Mordvinic *kizə "summer" ← *kesä ← IE *h₁es-en-.)

      There's incidentally a pre-existing parallel for this kind of an IE → U → IE detour: IIUC the current etymology of Scandinavian *rebaz "fox" is to derive this from Finnic *repo "id."; which in turn is an Indo-Iranian loanword.

      Delete
    7. Would it be possible to explain the irregularities in West Uralic as borrowings from one branch to the other within the language family. Or do we have to assume independent borrowings into the respective branches of West Uralic? If the latter is the case, one wonders if any connection to the IE 'stone; firmament' word is necessary. The primary meaning in Uralic seems to be 'back of a knife, axe', somewhat removed from the IE reflexes. So one wonders if not the source of the Uralic words could be, O Horror, an unknown substratum. Of course this is a last resort and it cannot be proven. But we might keep the possibility in mind.

      Anders

      Delete
    8. A very early (or very archaic) layer of Indo-European loans into West Uralic is known, in which initial laryngeals are substituted by *k-, medial ones by *-š-. Perhaps this one word managed to make it in with initial *š-.

      One thing that I've never understood about Finnic etymologies involving putative Indo-European *H > Finnic *š is how they account for the seeming cross-linguistic rarity of the change [back fricative] > [palatal fricative], whether language-internally or in the process of loaning. Perhaps this change is more common than I realize, but the only clear examples I've been able to find so far are cases where the back fricative is palatalized to [ç] or similar by a nearby front vowel/glide, whereas in the Finnic etymological proposals it seems that this change is supposed to have been independent of the phonetic environment.

      Delete
    9. One other thing, in response to Roland Schumann and Piotr:

      Just read in NIL that by Benveniste oi. aśmari- f. was once seen as a reflex of an old men/r-heterocliticon (rejected by EWaia) - when one holds up to this idea the hammer-words could perhaps also be seen as a reflex of this (haven't checked this out).

      Are there any other examples of the heteroclitic n/r-alternation in nasal-final stems? I thought (though I'm not sure where I originally got the idea) that the heteroclitic alternation was blocked by a preceding nasal, and that this was reflected in the many words ending in *-mn / *-men.

      Delete
    10. The normal declension of neuter verbal substantives in *-m(e)n- is non-heteroclitic (whatever underlies the 'hammer' word would have to be one of them if based on the verb root 'sharpen'). Hittite is an apparent exception, but there -mar/-men- is a dissimilated allomorph of -war/-wen- used after stems in -u-, not a PIE relict form. Of course animate nouns in *-mon- are not heteroclitic.

      Delete
    11. The normal declension of neuter verbal substantives in *-m(e)n- is non-heteroclitic (whatever underlies the 'hammer' word would have to be one of them if based on the verb root 'sharpen').

      Right, but I thought that one of the explanations for the *n/*r alternation in the first place was that *-n shifted to *-r word-finally, except after a preceding nasal, as in the case of the *-m(e)n words. Is this explanation considered problematic?

      Delete
    12. Not in my book. I'm pretty sure the explanation of *n/r heteroclisy must be phonological, even if we can't recover the pre-PIE details of the process). Many IEsts prefer a morphological solution (suppletion involving independently formed derivatives in *-n and *-r), but it's precisely the fact that there are non-heteroclitic neuters with a phonological environment that might block word-final rhotacism that decides the case for me.

      Delete
    13. One thing that I've never understood about Finnic etymologies involving putative Indo-European *H > Finnic *š is how they account for the seeming cross-linguistic rarity of the change [back fricative] > [palatal fricative], whether language-internally or in the process of loaning. Perhaps this change is more common than I realize, but the only clear examples I've been able to find so far are cases where the back fricative is palatalized to [ç] or similar by a nearby front vowel/glide, whereas in the Finnic etymological proposals it seems that this change is supposed to have been independent of the phonetic environment.
      TBH, I find this substitution kind of handwaved as well, particularly when in East Uralic, both *k and *ɣ also existed. What I suspect is that this is related to two sound changes of Finnic:
      1) *kt → ht. A word like *tehtas could come from older *tektas just as well as *teštas.
      2) *š → h. If the very few examples where an actual sibilant is found in non-Finnic languages were actually adopted not directly, but with Finnic as the mediator (this has been suggested for several other old IE loans), then *š might be a hypercorrection or nativization standing for Finnic *h.

      Delete
  5. Isn't (NW?)Germ. *selxa- 'seal' (the animal) taken to be a borrowing from Balto-Fennic? I seem to remember Schindler arguing that in an old Die Sprache. I don't have any books with me here, so I can't check it right now.

    ReplyDelete
    Replies
    1. It was in Die Sprache 12 (1966), 65f. There do of course exist other etymologies for this word ...

      Delete
  6. One thing that I've never understood about Finnic etymologies involving putative Indo-European *H > Finnic *š is how they account for the seeming cross-linguistic rarity of the change [back fricative] > [palatal fricative]

    Not being a Uralicist or anywhere near, I wonder how safe the sound value of the Proto-Uralic (and -West-Uralic and so on) *š actually is. Is there a chance it was actually [ɬ]? You know, like the Proto-Semitic (and later) *ś?

    I'd expect a language that has [ɬ] but lacks [x] to borrow [x] as /ɬ/.

    ReplyDelete
    Replies
    1. Reconstructing *š as a postalveolar sibilant is quite safe. *š surfaces essentially intact as /š/ ~ /ž/ in all the central branches: Mordvinic, Mari and Permic (the realization varies in the [ʃ] to [ʂ] range, I've not seen a detailed phonetic overview study). Samic has /s/ in inherited vocabulary (in parallel with *č → /ts/), but /ś/ in old loans from Finnic. The latter point, and Finnic *h frequently standing for Germanic *s or Baltic *š, *ž in loanwords, make it quite clear that Finnic *h also came from older *š.

      Two examples:
      • *mekšə "bee" (loaned from Indo-Iranian) → Finnish mehiläinen, Estonian mehiläne, Erzya /mekš/, Mokša /meš/, Mari /mükš/, Udmurt /muš/, Komi /moš/, Hungarian méh
      • *šiŋərə "mouse" → Finnish hiiri, Estonian hiir, Erzya /čejer/, Mokša /šejər/, Udmurt and Komi /šɨr/, Hungarian egér, Southern Mansi /täŋkər/, Northern Mansi /taŋkər/, Surgut Khanty /ɬăŋkʷər/, Northern Khanty /loŋkər/

      *š does show up as *ɬ (traditionally reconstructed *θ) in East Uralic. Offhand, it actually does not seem impossible that this WAS actually [ɬ] to begin with, and became *š in Finno-Permic… but, this would be of no help for loanword issues in the west. (And once we go that far back, PIE laryngeals start corresponding to Uralic *x anyway.)

      Delete
  7. I'd expect a language that has [ɬ] but lacks [x] to borrow [x] as /ɬ/.

    It depends on what other possible substitutes can be found in the inventory of the receptor language. For example, Chukchi meets your description: it has /ɬ/ but lacks /x/; it also has a voiceless alveopalatal fricative (~ affricate) /ɕ/ and a voiced velar /ɣ/. Yet Russian /x/ in loanwords is substituted by the uvular stop /q/ rather than any of the above, while /ɕ/ may replace Russian /s/ and /ʧ/ (and presumably any other coronal sibilant). For example, сахар ['saxər] 'sugar' becomes Chukchi [ɕaqaɹ].

    ReplyDelete
  8. I just came across this discussion of my etymology for Gmc. *hamara-. Three comments:

    1) Roland writes that "As far as I can see there is no other Balto-Fennic loanword that is Common Germanic." However, at a paper given i Leiden, November 2012, I argued that the 'hammer'-word is only one out of quite a few, probably as many as 25, Germanic lexemes borrowed from Balto-Fennic. Many of the candidates begin with h- because these words are the most easily detectable and cannot be loanwords in the reverse direction since the Balto-Fennic words with h- in these cases have cognates elsewhere in Uralic beginning with š- or č-. Among the other proposals are these two:

    PGmc. *halba- '1. incomplete, partial, reduced; 2. a significant portion of; 3. exactly ½’ <- Balto-Fennic *halpa-, gen. *halβan- 'reduced, cheap of prices';

    PGmc. *hama(n)- ’shape, physical form’, in the individual languages also ’skin (of animal); apparition, ghost; covering; net’ <- Balto-Fennic *hahmV ’shape’ (Fi. hahmo, haamu ’apparition, ghost’, Lule Saami sjipmō ’similarity’, Mordvin M šama, E čama ’face’); traditionally reconstructed as Fenno-Volgaic *šama; but cf. FU *čamčV ’skin (of animal), membrane’ (N Saami cuoži ’membrane, fleshy fibres on the inner side of the skin’, Khanty čunč ’skin, hide’). Balto-Fennic *hahmV could be metathesis of *čamčV, and the Mordvin forms might just as well come from *čama.

    2) Protouralic writes that "The proposed West Uralic cognates do not really abide to the typical sound correspondences, and no original Uralic root can be securely set up. (Samic *sɤmērē would incidate former **šimara; Mordvinic *šov would indicate former **šuvV.)"

    First of all, it should be noted that the Indo-European cognates, accepted by most do not correspond to each other regularly either. Historically, there has been a tendency to ignore irregularities on the IE side and, conversely, to put too much weight on irregularities on the Uralic side (e.g. instead of trying to explain the irregularities as conditioned reflexes or results of analogy or contamination). This, among other things, has led scholars to assume that similarities are always due to the loan direction IE > Uralic.

    Second, the Saami forms with ši- might very well represent the regular outcome of Fenno-Ugric/-Permic/-Volgaic *ša- in the position before nasal. The only other example, in fact, is Lule Saami sjipmō, mentioned above, which would then also be regular from šama.

    Note that the Fenno-Volgaic form *šama-ra might, in turn, be borrowed from some stage of Balto-Slavic? language with regular satem reflex but the same metathesis or ablaut form as Slavic kamy, Slovincian kamor. Thus, there is a possibility that *hamara could be connected to *ak^-men- (I reconstruct it without h2) after all, however not as a direct reflex of its protoform.

    3) Finally, you touch upon the reflexes (substitutions) of IE loanwords in Uralic as sketched by Jorma Koivulehto: Allegedly k- in initial position,-š- in medial position. Elsewhere, namely at a paper given at the UCLA Indo-European conference in 2008 (see abstract with material at http://www.pies.ucla.edu/IECarchive.html ), I have tried to show that the latter substitution did not take place, and that the only secure cases are from Balto-Fennic in the position before stop (in fact, only participial -t-), where Balto-Fennic *h rather represent a direct reflex of the laryngeal (these cases, by the way, seem to confirm an aspirated pronunciation of PIE *h1 as we assert here in Copenhagen). There is no particular reason to expect that all the laryngeals behaved the same way, and certainly not in all positions, since they were probably phonetically quite dissimilar.

    ReplyDelete
    Replies
    1. I forgot to note the crucial point that, as is the case with *hamara- and *haman-, *halba- cannot be a loan in the reverse direction either, because the Balto-Fennic words such as Finnish halpa are cognate with Mari (Cheremis) *šul-δo and thus come from Fenno-Volgaic *šala-.

      Delete
    2. Thanks for the comments, Adam! PGmc. *xalβa- is a particularly good example, given that there are no plausible cognates whatsoever on the IE side (Gk. kólpos, etc. can't be taken seriously), and 'half, cheeper' would surely have been an impotant word in trading contexts.

      Delete
    3. Hello Adam!

      I agree wholeheartedly that in Finnic etymology, difficulties in Uralic sound correspondences have been overemphasized in comparision to ones on the IE side. Both should be taken into account. Your critique of the substitution *H → *š is also quite interesting. (I would have a few comments on it too, but I'll restrain myself from going further off-topic than we are currently.)

      The comparision of Finnic *haamo with a Uralic root meaning "skin" seems unlikely to be correct though. For one, the reconstruction of the latter root with *-m- is based solely on Kildin Sami /tsuːmts/, despite all other Samic varieties pointing to *cōncë. (A more reliable example of original *mč is found in Samic *cōmcē "rotten", reflected across Samic essentially analogously to the better-known *mt.) For two, medial single *č becomes Finnic *t (e.g. *kočka → Fi. kotka "eagle"; *künčä- → Fi. kyntää "to plow"), so *mč would be predicted to yield **nt. And for three, the medial -h- in standard Finnish hahmo is unetymological - the variants haamo, haamu already show this, but compare also Karelian hoamo, Votic aamo, both indicating original *haamo. (The -h- is probably either by influence of the initial /h/, as also in huhmar ~ huumar "mortar"; via metathesis from a derivative *haamo-h; or less likely, by transmission via an extinct Sami language, as in kahlata ~ kaalata "to wade" ← Samic *kālē- ← PU *kälä-).

      You might also want to note that while there's indeed still wiggle space for explaining instances of Proto-Samic *i, initial *š- in Samic is basically a loanword phoneme, consistently indicating loans from Finnic/Baltic/post-palatalization Scandinavian/etc. Inherited postalveolars were lost entirely early on (*č, *š → *c, *s). Though this probably only helps your later Balto-Slavic loaning proposal.

      (I am also not particularly impressed with the Mari cognates for *halpa, but I also don't believe this would really hinder a Finnic → Germanic etymology.)

      Delete
    4. Thanks for your reply! Those are interesting points, although actually in favor of what I am saying. That -h- is indeed secondary in hahmo (as also in huhmar) makes the shape of the word even closer to the Germanic one. That š- is mainly of foreign origin in Saami, as you indicate yourself, does not preclude a Fennic-Saami origin of a Germanic word. As for Kildin Saami /tsu:mts/: That a reflex of a sequence *-mD- (i.e. -m- + dental) is only preserved in one daughter-language and assimilated everywhere else is typologically common because dental assimilation of m in such a sequence is in itself very common. The situation is parallel to Indo-European where usually only Baltic preserves *-mD- whereas all other language groups assimilate (not necessarily by a rule-governed development). Note also the assimilation of Uralic *-mt- > Fenno-Ugric *-nt- in the ordinal suffix (and in lexemes as well). In fact, a reconstruction *cōncë would leave the Kildin form unexplained, while *cōmčë accounts for all of them (by preservation or assimilation).

      Delete
    5. I am aware that (Pre-)Proto-Germanic probably also preserved *-mD- for some time, as indicated by N Saami ruovda 'edge' <-- Pre-Proto-Gmc. *ramdo:-, but the point is that, eventually, in the documented languages only Baltic preserved them. Anatolian, contrary to Baltic, seems to have preserved *-mG- as in hama(n)kzi 'binds' < PIE *h2emg^h-. But again, we might simply be dealing with irregular assimilation (of a typologically extremely common kind) rather than actual rule-governed developments.

      Delete
    6. I did not comment on your second point:

      "For two, medial single *č becomes Finnic *t (e.g. *kočka → Fi. kotka "eagle"; *künčä- → Fi. kyntää "to plow"), so *mč would be predicted to yield **nt."

      If -h- is secondary in Balto-Fennic, there is no need to account for it. We are already operating with two forms, *čama or *šama, and *čamčV. Finnish haamo etc. could come from the former. Both are reconstructed for stages later than Proto-Uralic, and their variation may be due to their identity as Eurasian culture-words, cf. e.g. Written Mongolian čamča 'shirt' ~ Persian ja:ma 'apparel'.

      Delete
  9. Interesting discussions here! I have some comments on the Germanic evidence. As to Sam. ruovda, note that Germanic preserves m only before þ, not before d: the source is therefore likely to have been PGm. *ramþa-, cf. MHG ranft ramft m. 'edge', rather than *randa-. On *hamara- itself, if it is of IE origin, it is most likely to continue *k'h2-moro-, with zero grade of the root and with metathesis as in Slavic and - not least - as in Gr. kamára 'vault' < k'h2-mr-. I suspect the meanings 'hammer', 'vault' and 'stone' to have diverged from 'meteorite'. To the Indo-Europeans, meteorites provided an early source of iron (for tools and weapons), and were considered to have been chips hammered off the heavenly vault.

    ReplyDelete
  10. As to Sam. ruovda, note that Germanic preserves m only before þ, not before d...

    Doesn't it, though? What about OHG sambt ~ sampt < *samd < PGmc. *samða- 'sand'?

    ReplyDelete
  11. Nope. Cf. *hunda- '100' from < *dk'mt-o- and *sunda- 'sound' < *sumtó- (related to 'swim'). Sand is a later loanword into Proto-Germanic, as also follows from the irregularities in the other IE languages, cf. Gr hamathos, psammathos < *(X)sam-ndh-o- and Lat. sabulum < *sadh-lo-.

    ReplyDelete
    Replies
    1. We don't know, really, if they are irregularities, and if all those 'sand' words are really related (by the way, the Greek one should be ámathos 'sandy ground'; the aspitation, if originally present, would have been lost by Grassmann's Law). The alternative hypothesis is that psámmathos 'sea-sand' and ámmos ~ hámmos (= ámathos) resulted from cross-contamination (ámathos x psámmos) between two words unrelated to each other. In its turn, psámmos can be connected with psêphos (Doric psâphos) 'pebble' and further with Lat. sabulum 'gravel, coarse sand' (if from *psabʰ-(e)lo-). Even if *psa(:)bʰ- comes from some unknown Mediterranean substrate, there is no reason to insist on a non-IE etymology for PGmc. *samða- and Gk. ámathos. They may easily reflect *som(h₂)-dʰh₁-o-/*sm̥h₂-dʰh₁-o- (for a similar apophonic pattern, cf. PGmc. *wurða- vs. Lith. var͂das 'word', from the root *werh₁- 'speak').

      The slightly inconsistent behaviour of heterorganic nasals with respect to place assimilation in Germanic, it may reflect a typologically common natural tendency, as suggested by Adam, rather than the operation a regular "sound law".

      Delete
  12. Thanks for the interesting replies; I can't comment here often because logging into Google is a PITA.

    There is no particular reason to expect that all the laryngeals behaved the same way, and certainly not in all positions, since they were probably phonetically quite dissimilar.

    ...Yes, if you're assuming a loan straight from PIE proper. But once Anatolian and the rest had split, the laryngeals seem to have merged fairly quickly. In Greek, syllabic laryngeals became e/a/o depending on the laryngeal, IIRC, and I don't know what has come of the idea that *h1y- turned into /h/ while *y- turned into /dz/; in Armenian and Albanian, *h2 and *h3 seem to have become /h/ in front of stressed vowels; but all other branches, IIRC, treat all laryngeals the same, implying a merger in a common ancestor of "core kentum" and "core satəm" that incidentally left *a phonemic if it wasn't already.

    (There's Cowgill's law in Germanic... except it probably isn't there: Google Books preview sez Rasmussen has suggested /g/ was inserted into /jw/ clusters sometime before Grimm's law, using German Speichel "saliva" as an example where there never was a laryngeal.)

    Even if *psa(:)bʰ- comes from some unknown Mediterranean substrate

    That would imply that this substrate had voiced aspirates, right?

    ReplyDelete
    Replies
    1. [i]That would imply that this substrate had voiced aspirates, right?[/i]

      Correct me if I'm wrong (not an expert), but isn't the (near) lack of a simple *b a chief criticism of proponents of glottalic or glottalic-like proposals (or criticisms of traditional reconstruction in general)?

      Though I consider myself agnostic on the issue (i.e. on glottalic-like theories), it seems to me that we understand far less about the exactitudes of PIE phonology than the traditional orthography would have us believe. Data like this should be understood in that context, then; borrowings like this are probably less about rare substrate phonemes and borrowing irregularities and maybe more to do with the actual incompleteness of our understanding. I wouldn't take the presence of a *bʰ in a borrowed term as evidence of much of anything or a counterargument to the borrowing theory (on its own).

      My beliefs at least; take them within a grain of salt.

      Delete
  13. But once Anatolian and the rest had split, the laryngeals seem to have merged fairly quickly.

    I wouldn't say so. For example, the "colouring" laryngeals *h₂ and *h₃ (but not *h₁) caused high-vowel breaking ("Francis-Normier's Law") In Greek, Tocharian and Armenian (of which Greek has contrasting vowels resulting from the breaking). *h₂ (but neither of the other two) caused stop aspiration in Indo-Iranian.

    Some of those branch-specific laryngeal reflexes remain somewhat controversial anmd still disputed. The most popular view concerning initial h- vs. z- in Greek is that the voiceless reflex derives from *h₁/₂j-, and the voiced one from *(h₃)j-. The eventual parallel merger or loss of all laryngeals is definitely a "dialectal" phenomenon, postdating the protolanguages of some of the "branches".

    The Albanian and Armenian aitches are tricky, and don't seem to have much to do with the IE laryngeals. There are too many counterexamples both ways.

    That would imply that this substrate had voiced aspirates, right?

    Not necessarily. If we are talking of a pre-Greek and pre-Italic stage when neither lineage had a *b phoneme, * (or whatever came from it at the time) was perhaps the closest thing to a plain voiced stop from the donor language.

    ReplyDelete
  14. Correct me if I'm wrong (not an expert), but isn't the (near) lack of a simple *b a chief criticism of proponents of glottalic or glottalic-like proposals (or criticisms of traditional reconstruction in general)?

    Yes; but these proposals have fallen out of fashion rather severely in the last 20 years. Even Vennemann gave up in 2006; Bomhard may be the last glottalist. It has been suggested (though only vaguely, without specific examples, AFAIK) that the missing *b might have merged into *m or *w...

    ...and of course, everything is the way it is because it got that way (J. B. S. Haldane, embryologist). The Moscow School finds PIE *p/t/k, *b/d/g and *bʱ/dʱ/gʱ to be derived from Proto-Nostratic *pʼ/tʼ/kʼ (ejectives), *p/t/k and *b/d/g, respectively; [p] is a less weird sound to lack than [b].

    Finally, b is very rare indeed in Vedic Sanskrit, proving that such a language isn't outright impossible.

    it seems to me that we understand far less about the exactitudes of PIE phonology than the traditional orthography would have us believe. Data like this should be understood in that context, then; borrowings like this are probably less about rare substrate phonemes and borrowing irregularities and maybe more to do with the actual incompleteness of our understanding.

    Borrowings can definitely illuminate such issues. However, on the voiced aspirates, I'm sure they were indeed voiced and aspirated: no other combination of features I can think of can explain the observed reflexes. While a sound system with voiced but no voiceless aspirates is rare, a few dozen languages in the world do have it, Vedic Sanskrit is almost there (the voiceless ones were very rare), and PIE seems to have contained a very small number of onomatopoietic roots with voiceless aspirates.

    *h₂ (but neither of the other two) caused stop aspiration in Indo-Iranian.

    Oh! I didn't know that! Indo-Iranian is deeply nested in IE, so this argues for a long retention of separate laryngeals.

    The Albanian and Armenian aitches are tricky, and don't seem to have much to do with the IE laryngeals. There are too many counterexamples both ways.

    Both ways? So there are examples where *h₂ or *h₃ disappeared in front of a stressed vowel?

    I know that /h/ can spread in weird ways by analogy or whatever. All Greek words with y- ended up with it, so that the letter y was called y psilon, "naked y"; and the /h/ of Greek híppos lacks an explanation (as does the /i/). Outside IE, Proto-Altaic initial *pʰ corresponds to 0 in Turkic languages except for Khalaj, which has /h/- there, but /h/- also occurs for unexplained reasons in some (not all) words that must have begun with *0 or */ŋ/; yet, it is apparently never absent from roots with *pʰ-.

    If we are talking of a pre-Greek and pre-Italic stage when neither lineage had a *b phoneme, *bʰ (or whatever came from it at the time) was perhaps the closest thing to a plain voiced stop from the donor language.

    Point taken!

    ReplyDelete
  15. Oh! I didn't know that! Indo-Iranian is deeply nested in IE, so this argues for a long retention of separate laryngeals.

    All the uncontroversial examples of *TH becoming in Indo-Iranian involve *h₂.

    Both ways? So there are examples where *h₂ or *h₃ disappeared in front of a stressed vowel?

    Of course, e.g. Armenian arawr 'plough' < *h₂árh₃-trom, ayc 'goat' < *h₂áiǵ-, acem 'lead' < *h₂áǵ-e/o-. There are, moreover, internal inconsistencies in both languages, dialectal variants with and without /h-/, as well as Armenian and Albanian examples contradicting each other (e.g. Alb. herdhe vs. Arm. orjikʰ 'testicles'). Ad hoc explanations work for some of them, but not for all.

    ReplyDelete
  16. ...Oh. Yeah. I think I was confusing the "*h₂ or *h₃ are preserved in Armenian under whichever conditions" hypothesis with the "there was an *h4 which explains away the remaining cases of PIE *a and is preserved in Albanian when word-initial before an originally stressed vowel, particularly in herdhe" hypothesis.

    Also, the textbook example for Armenian is hoviw "shepherd" < *howi-pā < *h₃owi-pah₂ – but "sheep" itself is just awdikʰ!

    Of course, perhaps something is seriously wrong with the reconstruction of PIE stress, but nobody has made the proposed new reconstruction yet, so I can only speculate...

    ReplyDelete
  17. I'm still hoping that you will continue to write your blog!

    Best,
    Hakan

    ReplyDelete
  18. I will soon. Sorry, the first month of the academic year is not a stroll in the park :)

    ReplyDelete