Can Latin vowel length be inferred from inherited descendants without attested diacritics?

Lusus Naturae

Senior Member
Cantonese
For example, instead of consulting ancient manuscripts or inscriptions, can one determine the length of each stressed i in piscis, scrinium, ficus by looking at their descendants such as pesce, scrigno, figo? Since the e is a result of short i, the i is a result of long i.

Can one infer the length of each stressed o in rota, totus from their descendants such as rueda, ruota, tutto, todo? Since ue and uo are expected from short o.

How (un)reliable is such inference? How often are the results of Latin long vowels and short vowels unexpected when inherited?

Is the vast majority of Latin long vowels and short vowels attested rather than inferred?
 
  • Sobakus

    Senior Member
    Yes, it can, just as your examples show.

    There was more than one standard of Latin, both written and spoken; some vowels varied between them. In non-Urban dialects of Latin the vowels didn't differ in quality, so Imperial Urban Latin vīlla [i:] < Republican veilla corresponded to vēlla [e:] in those, mūtus sounded like mōtus [o:], and sērō [ē:-o:] like saerō [ɛ:-ɔ:]; the Urban /au/ merged with the Urban /ō/ as [ɔ:].

    Sometimes Romance reflects these disagreements, and you get It. fieno < faenum but Fr. foin and Sp. heno < fēnum - the Urban variant was exported, while the dialectal one remained in Italy. Other times these dialectal variants entered the standard language, as was the case with the name Clōdius and the word cōda "tail". Then there were the Greeks, where Η/Ω were open for a while, and thus their /ē, /ō/ were the same as in dialectal Latin and different from the close Urban vowels. The Sardinian vowel system merges all long and short pairs, and might very well go back to some combination of the above.

    The vast majority of Latin vowels are attested in poetry, or can be forwards-reconstructed based on etymology. Not many are surmised only from backwards reconstructions from Romance, mostly the stuff of uncertain etymology that didn't make it into poetry.

    The cases where most dictionaries will cite a shortening of the Classical length in some variety (Fr. onze for ūndecim, It. undici) may be nothing other than a dialectal monophthongisation outcome.
     
    Last edited:

    Penyafort

    Senior Member
    Catalan (Catalonia), Spanish (Spain)
    It is relatively possible, but the number of inconsistencies is too high, because it's mainly spoken unattested Latin that Romance languages developed from.

    Looking at your examples:

    In the Romance languages, fish is with a close-mid e (Portuguese pejʃɨ, Spanish peθ or peχe, Catalan peʃ, Old French peis, Italian peʃ:e, Romanian peʃte). That makes us think the word came from a Vulgar Latin/Proto-Romance close-mid e (pescem), but was it the result of a long e or of a short i? Why not *pēscis instead of pĭscis?

    Figo is an easier case. The fact that it’s i in all of them (Portuguese figo, Spanish higo, Catalan fic, Old French fie and Italian fico) makes it clear that the original was a long i, fīcus. Generally speaking, long i's and u's are more possible to infer.

    Totus may not be the best of examples, as it was long in Latin but many Romance forms probably come from an intensive form *tottu, which would explain some irregularities in evolution of the modern forms.

    Diphthongs can be trickier, because they depend on phonemic contexts. Take the case of yods:

    - In Spanish, they prevent o's from diphthongizing: hoja, ojo, noche . However this does not affect diphthongization in neighbouring Asturian and Aragonese: fuella, uello, nueit(e).
    - In Old Oc and Old Oil, a different diphthongization took place, but led to different monophthongs in some cases. That explains Catalan i and u in words coming from Ò before yod: FÒLIA > *fuóiʎa > *fúuʎa > fulla /'fuʎə/, *ÒCLU > *uóiʎ > *úuʎ > ull /uʎ/, NÒCTE > *nuóit/*nuéit > nwit > nit

    The same yod cases seem to close the vowel in Portuguese (the o in ollo and noite is not open).

    Other factors may leave some vowels to not diphthongize too, or the other way round, to diphthongize by analogy, metathesis, etc. Not to mention that some cases seem to mirror the fact that both lengths already existed in Latin forms.

    In other words, we may infer the length grosso modo, but it will always be far from being conclusive, rather complementary.
     

    berndf

    Moderator
    German (Germany)
    Totus may not be the best of examples, as it was long in Latin but many Romance forms probably come from an intensive form *tottu, which would explain some irregularities in evolution of the modern forms.
    You are probably thinking of things like French tout or pour. But those are probably later developments. The oldest attestations are tot and por.
     

    Sobakus

    Senior Member
    That makes us think the word came from a Vulgar Latin/Proto-Romance close-mid e (pescem), but was it the result of a long e or of a short i? Why not *pēscis instead of pĭscis?
    But the question was
    can one determine the length of each stressed i in piscis, scrinium, ficus by looking at their descendants such as pesce, scrigno, figo?
    The answer is a comprehensive yes: if given a word spelt with the letter I, we can look at its Romance descendants and figure out whether that letter represented a short or a long vowel.
    You are probably thinking of things like French tout or pour. But those are probably later developments. The oldest attestations are tot and por.
    I think they mean the development of the T which surfaces as /t/ in leniting varieties where a /d/ or a Ø is expected. I think it's more complicated than this, because most of Italy goes back to /tūttu/, where the most obvious explanation is a crossing with cūnctum. There are other examples of such "expressive gemination" floating around, but if you remove the putain-related terms (which are themselves probably a cross with an imitative root for "little", but ultimately regularly reflect pūtida), I can't think of any other examples ATM. Also in both of these cases the /ū/ remains unexplained by any expressive gemination.

    tot and por are simply spellings for the mid-high vowel /o/, now spelt ou. Gallic Late Latin used V and O promiscuously to spell that sound due to etymology, and the Old French orthography inherts this; eventually they just decided to write both xD Oh, and also because it dipthongised into an actual /ow/ when in stressed open syllables.
     

    fdb

    Senior Member
    French (France)
    The vast majority of Latin vowels are attested in poetry, or can be forwards-reconstructed based on etymology. Not many are surmised only from backwards reconstructions from Romance, mostly the stuff of uncertain etymology that didn't make it into poetry.

    The length of the vowels in closed syllables cannot be deduced from poetry. This is the whole problem.

    For example, the quantity of the first vowel in stella is not clear from scansion, nor from IE etymology (Hster-na-? Hstēr-na?). It is known only from Romance forms like étoile (with ē > oi).
     
    Last edited:

    Sobakus

    Senior Member
    The length of the vowels in closed syllables cannot be deduced from poetry. This is the whole problem.

    For example, the quantity of the first vowel in stella is not clear from scansion, nor from IE etymology (Hster-na-? Hstēr-na?). It is known only from Romance forms like étoile (with ē > oi).
    Except in those cases the explanation "oh it was shortened in the closed syllable: Latin was continuously simplifying its syllable structure" is always available, or a case for conflation with another word can be made: both are possible for example in the outcomes of jū̆xtā +- jūstum. Other times you see Latin inscriptions with SEIGNVM, SÍGNVM and then look at its Romance reflexes, all of which point to sĭgnum. This in addition to the monophthongisation outcome issue that I described. That's why I personally never take these as conclusive.

    As for stēlla, the suspicious /r/ in Ibero-Romance hints at conflation with Gaulish *sterā - /e/ was a close vowel there far as I can tell. Besides, French and others point to a form with single /l/.
     

    Lusus Naturae

    Senior Member
    Cantonese
    How to understand the outcomes of, for example, officium, olīva, aequalis?In Italian, standard, archaic, and alternative forms, uffizio officio, oliva uliva, uguale eguale?
     

    Penyafort

    Senior Member
    Catalan (Catalonia), Spanish (Spain)
    You are probably thinking of things like French tout or pour. But those are probably later developments. The oldest attestations are tot and por.

    As Sobakus explained, it's much more than that. Both forms with o and u exist in Portuguese. French and Catalan have a -t- for the feminine where lenition would have been expected. And so on.

    But the question was...

    Granted. If we know which Latin vowel is in advance, then I agree, things simplify.

    It goes without saying that many reconstructions can be perfectly done too withou much hesitation, as in the case of *retina for rein.
     
    Top