PIE *gh > latin [h] word initially and [g] word internally but why veho:, traho: and incoho: with h?

Erkattäññe

Member
Spanish - Argentina
Pretty much the tittle, this treatment seems to be isolated to this three stems with a PIE etymology but I could be missing more instances. The regular outcome "should" have been [g]. Perhaps we are dealing with old latin dialectalism, this treatment looks like sabellic or maybe faliscan which is closer to latin.
 
  • Pretty much the tittle, this treatment seems to be isolated to this three stems with a PIE etymology but I could be missing more instances. The regular outcome "should" have been [g]. Perhaps we are dealing with old latin dialectalism, this treatment looks like sabellic or maybe faliscan which is closer to latin.
    Stuart-Smith J · 2004 · Sound change in Italic: 44
    More difficult is the treatment of intervocalic *-gh-, where we find both <H> (or ∅) and <G>:³¹ e.g. uehō<*wegh-, cf. uectus, meiō 'urinate'<*H₃eigh-, Gk ὀμείχω, Skt mehati; figūra ' figure'<*dheigh-, cf. figulus. Here we accept the traditional assumption that <H> was the regular outcome of *-gh- intervocalically (as e.g. Leumann 1977: 165; Meiser 1998: 104). Forms with <G> are then explained as analogous formations after those where <G> occurred as a regular reflex. Thus <G> in figūra, for example, may occur after analogy with expected <G> in related forms such as fingō and figulus.³²
    ³² Assuming <G> as the regular outcome of intervocalic *-gh- (as e.g. Pisani 1940, whose evidence is refuted by Szemerényi (1952/3 [1987]: 641–2) requires awkward explanation of the forms with <H> , for example, the assumption that <H> in uehō is an orthographic marker of vowel hiatus in a form /weō/ formed from the perfect ue-xī, after analogy with struō, struxī (Pisani 1932).

    My own contribution: the above assumption that the regular intervocalic reflex is h is in line with the development -*gʷʰ->u̯.
     
    de Vaan MAC · 2008 · Etymological dictionary of Latin and the other Italic languages: 818–825 contains a list of reconstructed PIE protoforms of the Italic lemmata mentioned in the dictionary. I have found only the following ones to contain intervocalic *-gʰ- and *-gʲʰ-:

    *dʰ(e)ı̯gʲʰ-, *dʰeı̯gʲʰ-(o)- > figūra, effigia; figulus<*dʰigʰlos
    *hₑlegʰ-u- > *lexu̯is > levis
    *hₑmegʲʰı̯o > mihei > mihī
    *hₒmeı̯gʰ-(ı̯)e-
    > meiō
    *legʰ-e/o-
    — only in lectus
    *segʲʰ-ur/-un-
    > *sexʷēros > sevērus
    *smihₐgʲʰesl-ihₐ
    >*smīxeslī > mīlle
    *u̯egʲʰ-, *u̯egʲʰ-e/o-
    > vehō, vehis

    Cohum
    is Italic, Celtic and Germanic; trahō is Italo-Celtic; T…Dʰ is a combination prohibited in PIE roots, so both are likely loanwords (pp. 123–124; 626–627). Nevertheless, the correspondences point at an early period, and the consonants agree in *gʰ or rather something like as the likely intermediate stage for all three branches.

    So, leaving aside *xı̯>j and *xu̯>v, we have the above figūra~effigia (explained through fingō and figulus), mihī, vehō, plus cohum and trahō. There seems thus to be no good evidence to postulate -g- as the normal intervocalic reflex. I think mihī is crucial for the interpretation of h as the basic outcome.
     

    Zec

    Senior Member
    Croatian
    As for explaining why intervocalic *gʰ does not give a stop consonant, while *bʰ and *dʰ do, I believe there is no need to postulate dialectal influence. Out of all voiced stops, [g] is the most likely to become a fricative - such a thing happened in many Slavic languages, it was the case in early Germanic, and still is in today Dutch.

    Therefore, what probably happened is that, in pre-Latin, when *β and *ð became b and d, *ɣ instead remained a fricative long enough to merge with initial *x into h.
     
    As for explaining why intervocalic *gʰ does not give a stop consonant, while *bʰ and *dʰ do, I believe there is no need to postulate dialectal influence. Out of all voiced stops, [g] is the most likely to become a fricative - such a thing happened in many Slavic languages, it was the case in early Germanic, and still is in today Dutch.

    Therefore, what probably happened is that, in pre-Latin, when *β and *ð became b and d, *ɣ instead remained a fricative long enough to merge with initial *x into h.
    However, millennia later, after another turn of phonetic changes, g and d remained stable (piaga, legare; nudo, sudore), whereas it was b that experienced lenition (cavallo, scrivere).
     

    Zec

    Senior Member
    Croatian
    Indeed, the tendency I mentioned is only statistical, not a necessary law. It would have been better if I had said that [g] is the most likely to spontaneously become a fricative.

    In this case, I am actually envisioning a unconditioned fortition: Assuming PIE aspirates gave fricatives in Italic (Voiceless initially, voiced intervocalically - as assumed for Sabellic), they would have had to fortify to stops in Latin.

    The later change is a conditioned, intervocalic lenition - though that still doesn't explain why only b lenited (in Italian - in Spanish and French, all voiced stops lenite and disappear). I suppose the change is linked to the change of earlier Latin [w] into [v], at which time the letters begin to be exchanged in inscriptions. After this change, there was now a voiced labial fricative, while voiced dental and velar fricatives didn't exist. The reason only b lenited might have been that only it had a corresponding fricative already in the language.

    One might also consider that [v] had already existed in Latin as an intervocalic allophone of /b/. Then, when phoneme /w/ changed to /v/, it took over those former allophones of /b/.

    But of course, these are all just speculations based on some common tendencies - one can rarely prove the cause of a sound change.
     

    Erkattäññe

    Member
    Spanish - Argentina
    Well, the examples almost explain a predictable distribution, I also forgot the most important example, mihī:

    /h/ before vowels, same in the beginning of words
    /g/ before liquids, same in the beginning of words
    /0/ before glides
    /k/ obviously the neutralization before unvoiced stops, mostly /t/ and before /s/

    Thanks for the elucidations.
     
    Top