Pronouncing initial rho with rough breathing

[interprete]

Hello,

I'm a complete beginner in Greek, someone offered me a Greek textbook for Christmas so here we go.
The first lesson introduces the alphabet and the 'breathings' (unsure of the English terminology sorry, my textbook is in French). It says that rough breathing consists in adding an H sound BEFORE the letter bearing such sign.
However, among the examples it goes on to give, are words starting with rho, and transliterated as if there was no H. So it phonetically spells rose (which I would assume to be hrodon) as 'rodon', and what I would assume ought to be read 'Hrodos' is phonetically described as Rodos, no H.
What further surprises me is that modern languages have inherited many such Greek words starting with rough rho, and they are invariably spelled rh- instead of hr- (and yet my book says the rough breathing comes befor, not after the letter). Why is that so? It's a hymn, not an yhm, so why is it a rhino and not a hrino?

Also, I'd be extremely curious to hear how words with multiple aspirated consonants are supposed to be pronounced. Like khthon...

Thanks!

 

[Anbrutal Russicus]

"rough breathing" is a pre-scientific term for aspiration or voiceless onset, which with vowels means the glottal fricatvie /h/, and with consonants a delayed start of voicing, i.e. consonant aspiration /ʰ/. rh and hr are simply digraphs for the voiceless alveolar trill [r̥], so whether to write the h before or after is a matter of convention - they wrote the modern English wh as hw in Middle English (it's pronounced as the voiceless /w̥/ in those accents without the whet-wet merger - kaldim & floridagirl). Your book transliterates it as r because all initial r's are voiceless in AGreek. Here's how χθών was very likely pronounced. Two-stop clusters in AGreek are only both aspirated or both unaspirated, and I suspect it made no difference whether the first consonant in them was aspirated or not. I don't actually know of a language that contrasts aspiration before consonants - in K(h)mer for instance there's no contrast, it's determined by the following consonant, which looks to be the same as in AGreek.

 

[interprete]

Thank you very much Anbrutal Russicus! In the meantime I came across a PhD dissertation on Greek aspirations, in which the author concludes that the most likely assumption is that initial pho was pronounced as a geminated trilled R, much like in Spanish with initial trills vs. flapped R's within words. It also says that most probably all geminated Greek R's were aspirated to some degree, hence rrh transriptions as in diarrhea and hemorrhoids (sorry, I can't come up with anything more sexy), and since initial R was automatically geminated (trilled not flapped), it was aspirated too, hence Rh.
So I'm glad that you are saying roughly the same thing (right? or did I misunderstand your point?), then I can consider the issue as settled. To think that I'm only on lesson 1

 

[Anbrutal Russicus]

Yeah you're saying the same thing, only I wasn't sure if all intervocalic trills were voiceless as well so I only said "initial" (I'm still not sure they were). Also, "geminated" means the same as "trilled" intervocalically but in all other positions where the trill typically occurs (initial, before consonants, variously after consonants that close the previous syllable) it can hardly be called geminate. Spanish illustrates this well: its initial trill is always a trill even when preceeded by a vowel at the end of the previous word, but it's always a single consonant as Spanish has no geminate consonants (apart from ct pronounced as [t:], but that's not a phonemic geminate). It appears that the initial AGreek rho also becomes a tap if a vowel preceeds, but judging by διά-ρροια and poetic scansion it could also remain a trill as in Spanish, but closing the previous syllable unlike in Spanish (where, again, it behaves like all other single consonants). Or actually, more likely always remained a voiceless trill, but whether it left the previous syllable open varied, like with all those AGreek liquid clusters (mn, kr, ps).

 

[Anbrutal Russicus]

Btw forgot an example of how that voicelss trill sounds: Icelandic, Welsh.

 

[interprete]

Also, "geminated" means the same as "trilled" intervocalically but in all other positions where the trill typically occurs (initial, before consonants, variously after consonants that close the previous syllable) it can hardly be called geminate.

I thought so too but then the author goes on to give period examples of inscriptions with initial double-rho (as in, two rho's written in a row at the beginning of words) which she uses as a clear indication about how initial rho was pronounced. Thanks for the other clarifications. I'm not sure I understood everything, but maybe I will later on

 

[Anbrutal Russicus]

I mean yeah, seeing as it can close the previous syllable, this means it can be a geminate - but when it doesn't, it's a singleton. It could have been pronounced the same way - as a trill - in both cases, it's only the syllabification/number of phonemes that differs.

 

[interprete]

Thanks @Anbrutal Russicus , by the way if this could be of interest, here is the source I was mentioning: https://www.sfl.cnrs.fr/sites/default/files/images/2016-11_jatteau_sta_1.pdf
And the gemination thing is discussed inter alia at page 356 of the document itself (i.e. p. 388 in Acrobat Reader). Maybe I didn't understand it, the concepts are way above my paygrade.

 

[Anbrutal Russicus]

Thanks for linking the source, I found it very useful. So it seems there's been a development over time from treating it as a geminate to treating it as a singleton (p.382):

Les redoublements en e-rr- montrent donc qu’en grec classique, il faut partir d’une racine à géminée initiale. Si la racine commençait par une simple, on attendrait une autre forme de redoublement, en rer-.

I'm always suspicious of such studies though because their authors tend to focus on narrow areas of evidence, such as focussing on word-formation (how the perfect is formed) and paying little attention to poetic scansion, the statistics of closed vs. open syllable scansion before the word-initial rho. Allen 1987, p. 44 says it normally scans as closing the syllable (= geminate) in Attic scenic verse, and optionally in epic (where this also affects initial λ, μ, ν), but I haven't found any more specifics on this.

 

[interprete]

Thanks for linking the source, I found it very useful. So it seems there's been a development over time from treating it as a geminate to treating it as a singleton (p.382):

I'm always suspicious of such studies though because their authors tend to focus on narrow areas of evidence, such as focussing on word-formation (how the perfect is formed) and paying little attention to poetic scansion, the statistics of closed vs. open syllable scansion before the word-initial rho. Allen 1987, p. 44 says it normally scans as closing the syllable (= geminate) in Attic scenic verse, and optionally in epic (where this also affects initial λ, μ, ν), but I haven't found any more specifics on this.

I'm completely out of my depth here but, again, if this can be of interest, she discusses at length the issues of meter, scansion and theater with the tragedy vs comedy distinction on pp. 331-334 and then from p.340 to p.356.
A teaser here:

Outre le caractère apparemment sporadique de #rr-, le phénomène se heurte également à un obstacle théorique : à plusieurs reprises, l’idée d’une géminée à l’initiale de mot est explicitement et a priori rejetée par les grammaires. Nous avons ainsi rappelé l’insistance de Lejeune sur ce point dans l’introduction de cette section (p. ).⁴⁵ Pourtant les géminées initiales, quoique rares, existent bel et bien. On en trouve par exemple, au sein du domaine grec, en grec chypriote moderne.⁴⁶ En grec classique, en tout cas en attique, nous voyons deux arguments pour défendre l’idée que #rr- est un phénomène vivant à l’époque classique. Le premier est l’examen précis du corpus métrique, que nous développons ci-après. Le deuxième est le fait que <ρρ-> est bien attesté dans les inscriptions ; il est même plus fréquent que les attestations de l’aspiration (cf. infra section ()). Nous verrons ces attestations juste après, en section ..

 

[Anbrutal Russicus]

Thank you again, this was most definitely of interest! I often tend keep on searching in google instead of thoroughly looking through each paper I find (especially in a language I can sort of understand when written, but can't speak ). The author does present some pretty thorough statistics while also mentioning two previous studies, but her conclusions on p.355 don't gel well with me. In particular, she says "Dans le passage des Guêpes ci-dessus, la resyllabation de r- n’échouerait pas à cause d’un abrègement absolu de r-, mais simplement relatif : la vibrante serait toujours vibrante, mais elle ne paraîtrait plus si longue en comparaison des autres consonnes initiales."

However, in a language like Italian, a trilled R word-initially between vowels will always be perceived as a geminate as its length is clearly different from a tapped R and longer than any singleton consonant. The difference is categorical - you trill it however briefly, you get a geminate. There are some grumpy-sounding languages like Finnish and Icelandic where the R is often trilled even intervocalically and still contrasts with a geminate RR. In Russian, where I don't think there are any geminate RRs (except in borrowings and when two Rs meet at word/stem boundary) but singletons are normally taps word-initially, a trilled intervocalic R sounds very emphatic, but it isn't perceived as a geminate.

In addition, the fact that the geminate RR appears in closely-bound phrases such as after clitics, but normally not between two lexical words, means the domain of gemination is the clitic group (or whatever phonological unit you distinguish bigger than prosodic word and smaller than prosodic phrase). Which means gemination of a singleton R is the relevant process; the opposite explanation would be de-gemination of the geminate R between lexical words, which has no precedents to my knowledge and so makes no sense. On the other hand, such domain-limited gemination is well-known from Italian (raddoppiamento fonosintattico) and maybe also Finnish (not sure if it's domain-limited there).

The author of the disseration rejects that the R was inherently single, but doesn't adopt the de-gemination explanation either. Instead she says that vague thing about the (geminate word-initial) RR being relatively shorter than other (geminate? singleton?) word-initial consonants. This again does not hold as an explanation in my opinion. It's much more intuitive to say that we're dealing with domain-limited gemination (in epic sometimes extended to the other liquids) like in Italian, and that inside words (διάρροια) it either always automatically applied, or the stem-initial geminate simply became fixed as part of the word, and so we don't need to say that the stem-initial R is inherently geminate. The author also mentions that there's quite a bit of variation in spelling at a later period, which doesn't have to mean that geminate consonants were being lost - rather, stem-initial gemination in compound words (incl. prefixes) became variable, like in certain cases in Italian.

 

[interprete]

There are some grumpy-sounding languages like Finnish and Icelandic where the R is often trilled even intervocalically and still contrasts with a geminate RR. In Russian, where I don't think there are any geminate RRs but singletons are mostly taps word-initially, this sounds very emphatic, but isn't perceived as a geminate.

That's very interesting because I did notice trilled R in Russian in unexpected places when people get angry. It's funny how this would lead Russians to feel that people doing this in other languages sound grumpy

but her conclusions on p.355 don't gel well with me. In particular, she says "Dans le passage des Guêpes ci-dessus, la resyllabation de r- n’échouerait pas à cause d’un abrègement absolu de r-, mais simplement relatif : la vibrante serait toujours vibrante, mais elle ne paraîtrait plus si longue en comparaison des autres consonnes initiales."

To be fair, that wasn't her conclusion, she was merely citing yet another hypothesis by Kraehenmann et Lahiri, and she seems to imply that it is interesting but unsubstantiated. And to be fair to you, I was under the impression that your French was better than mine, sorry then for forcing you into such a cumbersome exchange and thanks for your patience

 

[Anbrutal Russicus]

Oh I didn't feel being forced into it at all, on the contrary I'm very happy that I've had the opportunity to read about this and discuss it, as I've never actually studied AGreek in general or read about its phonology in such detail (though I have a good general idea).

I know she's discussing the conclusions of another paper there, but doesn't she say in her own words on the same page that "Quelle que soit l’origine de cette géminée initiale, elle est aussi vivante en synchronie à l’initiale de mot qu’à la frontière de morphème. Le caractère géminé de #rr- ne dépend pas de facteurs lexicaux ni morphologiques identifiables : contrairement à d’autres phénomènes de gémination initiale, #rr- n’est pas provoqué par le mot qui précède. C’est donc nécessairement une propriété des mots à r- initial."? She even specifically mentiones the Italian raddoppiamento in this connection and excludes it (and she makes the same argument in several other sections).