Exempli gratia, "urbs" should be pronounced "urps". Was it just nicer on the ears, was the sound borrowed from another language?
gratias vobis in antecessum.
gratias vobis in antecessum.
Why are "bt" and "bs" pronounced "pt" and "ps" in Latin?
- Thread starter Puer Pedens
- Start date
"Bs" and "bt" are hard to pronounce as such—at least for me, and for the Romans. In fact, I think it's generally (maybe universally) hard to pronounce a voiced consonant followed by an unvoiced one. Hence you find phonetic rules like that—where one of the consonants will change to "agree" in voicedness or unvoicedness with the other—in various languages.
By the way, I'll move the thread to the pronunciation section, as it seems to belong there.
By the way, I'll move the thread to the pronunciation section, as it seems to belong there.
Last edited:
In Allen and Greenough you have assimilation (the name of this phonetic change) explained in details:
A dental (t, d) often became s, especially when standing next to t, d, or s.
equestris for †equettris
cāsus for †cadtus (cf. 6., below)
Many instances of assimilation, partial or complete, are found.
cessī for †cedsī
summus for †supmus
scrīptus for †scrībtus (b unvoicing to p before the voiceless t)
16. In compounds with prepositions the final consonant in the preposition was often assimilated to the following consonant, but usage varied considerably.
A dental (t, d) often became s, especially when standing next to t, d, or s.
equestris for †equettris
cāsus for †cadtus (cf. 6., below)
Many instances of assimilation, partial or complete, are found.
cessī for †cedsī
summus for †supmus
scrīptus for †scrībtus (b unvoicing to p before the voiceless t)
16. In compounds with prepositions the final consonant in the preposition was often assimilated to the following consonant, but usage varied considerably.
- There is good authority for many complete or partial assimilations: ad, acc-, agg-, app-, att-, adc-, adg-, etc.
- Before a labial consonant we find com- (comb-, comp-, comm-), but con- is the form before c, d, f, g, consonantal i, q, s, t, and consonantal v; we find conl- or coll-, conr- or corr-; cō- in cōnectō, cōnīveō, cōnītor, cōnūbium.
- In usually changes to im- before b, m.
- Ob and sub may assimilate b to a following c, f, g, or p; before s and t the pronunciation of prepositions ending in b doubtless had p; surr-, summ-, occur for subr-, subm-.
- The inseparable amb- loses b before a consonant.
- Circum often loses its m before i.
- The s of dis becomes r before a vowel and is assimilated to a following f; sometimes this prefix appears as dī-.
- Instead of ex we find ef- before f (also ecf-).
- The d of red and sēd is generally lost before a consonant. The preposition is better left unchanged in most other cases.
IMO it boils down to the ease of pronunciation as the engine that drives all these changes so I think it's fine enough explanation.
Sorry, I had missed this part of your quote:
I thought it was all about assimilation where a letter becomes identical to the other (as in conloqui ---> colloqui).
E
Etaoin Shrdlu
Guest
Are there other languages apart from English that assimilate so as to have final consonant clusters voiced? Voiced consonants even on their own seem rare at the end of a word.
I don't know.
I have always thought that urbs is pronounced urps because Latin s cannot sound like English z, and it is difficult to keep bs from sounding like bz. Urbs might tend to sound like urbz, whereas urps allows the pure s sound to be preserved. The same applies to scribo changing to scripsi in the perfect.
So Since you are at it, observo, obsto, obtineo and obtempero something like "oPSerwo, oPSto, oPTineo and oPTempero" respectively.
If you're talking about eg. cats - dogz, what's going on here is that English has progressive → voicing instead of the regressive ← one of normal human languages like Slavic or Italic or Greek. The canonical plural morpheme is /z/ which is devoiced progressively by the preceding voiceless consonant /t/ and remains voiced otherwise (dogz, crowz, landz). This is related to the trouble @Dumnorix is having keeping the differential voicing. While voicing assimilation in clusters is cross-linguistically expected, languages that don't have absolute-final devoicing might not show regressive devoicing either - standard Ukrainian is often cited in this regard (notice how the downvoted speaker seems devoice it). They still do show regressive voicing.
I'd say a similar variation obtained in Latin since it was likely starting to devoice final stops by the turn of the millennium, with at least some varieties voicing them in return if a vowel followed.
Last edited:
Yes. You even find them actually spelled with p, sometimes. Well, at least some of them; I don't know if all are attested with that variant spelling, but things like opserco and optulit aren't all that rare. Plautus often has opsecro.
E
Etaoin Shrdlu
Guest
Latin S cannot be voiced because it isn't, but it could if it were. That doesn't really answer the question of why Latin, and most languages, when confronted with a word where there is a voiced and unvoiced consonant at the end, choose to make both voiceless rather than voiced.
#
The actual process in most languages, as I've tried to explain, is that the left consonant in a cluster always assimilates in voicing to the right one. When a language devoices all final voiced consonants, this will always result in word-final voiceless clusters, provided all the consonants can be devoiced. If the language allows final voiced consonants and also has regressive voicing assimilation, then final voiced clusters are only possible if the rightmost, absolute-final consonant is voiced by itself; but if the rightmost consonant is by itself voiceless, the language needs to have progressive voicing assimilation in order for final a cluster of two voiced consonants to be possible, which sounds like this (Persian). Thus for example there's no reason to assume that Latin didn't voice /s/ in the cluster /sb/, but voicing it in /bs/ would still be impossible in that language because the voicing assimilation is from right to left, that is regressive: the /s/ devoices the /b/, and the opposite is impossible.
Therefore the reason that final clusters of voiceless clusters are more common cross-linguistically can be reduced to two things: 1) that final devoicing is more frequent than final voicing (tentative counter-examples can be counted on one hand); 2) that regressive assimilation is more frequent than progressive (haven't read mentions about the frequency of counter-examples for this; the English cats-dogz alternation actually seems not to be one because it's of morphemic rather than of phonological nature: cf. bi[gf]oot, never bi[gv]oot).
Last edited: