The supposed "ng" pronunciation.

[Augustus Secondus]

Is it true that Classical Latin words containing "gn" is pronounced as "ng"?

For example: signum (sin-gum).

Much Appreciated.

[FiliusLunae]

Depending on where you look (and hear), some will give the pronunciation [g.n], as in English "significant". Most others, however, render it as a velar nasal (like "ng" in English bang) followed by [n], and that is the one that seems more likely, and the one I use.
Thus, signum » [sing.num].

I'm not quite sure if "gn" represented solely the velar nasal, without the following [n], i.e. signum » [sing-um] (without actually pronouncing the "g", though).

[Bombichka]

Varro talks about an addional character to the alphabet (in fact, an additional sound - the ancient did not distinguish sounds and letters) that the Greeks called agma. It was represented by ng in classical Latin but it was written - in Greek as well as in Old LAtin - as gg, e.g. siggulus, aggelus etc.

It is supposed, then, that the first sound in the ng was in fact a velar nasal, as in thing, and not a pure n. the theory is proposed by W.S.Allen in his Vox Graeca

[Augustus Secondus]

How is this "ng" thought to have sounded during the time of Caesar and Cicero though?

[Cyborg]

I find that funny, but it's true: I cannot really understand that proposed "-ng" sound. Is there somewhere I can hear it?

I can't make sense of it, I cannot imagine how is this really pronounced. Maybe I also pronounce English wrongly, then, because I say "significant" as "sig-neef-cant", and do not see any alteration in its "gn".

[Lucus Eques]

Depending on where you look (and hear), some will give the pronunciation [g.n], as in English "significant". Most others, however, render it as a velar nasal (like "ng" in English bang) followed by [n], and that is the one that seems more likely, and the one I use.
Thus, signum » [sing.num].

I'm not quite sure if "gn" represented solely the velar nasal, without the following [n], i.e. signum » [sing-um] (without actually pronouncing the "g", though).

In my opinion the following 'n' in addition to the velar nasal makes no sense. And it doesn't resolve the "Gnaeus" problem."

Varro talks about an addional character to the alphabet (in fact, an additional sound - the ancient did not distinguish sounds and letters) that the Greeks called agma. It was represented by ng in classical Latin but it was written - in Greek as well as in Old LAtin - as gg, e.g. siggulus, aggelus etc.

It is supposed, then, that the first sound in the ng was in fact a velar nasal, as in thing, and not a pure n. the theory is proposed by W.S.Allen in his Vox Graeca

Well, it's not simply a theory; it's a necessity. Latin n represents more than just the dental nasal, but also the velar and (rarer) palatal nasal, just as it is in English, Italian, German, ad infinitum. The Roman alphabet lacks special characters that differentiate, unlike the Sanskrit devanagari, for example. Take the English "anchor": the "an-" is not pronounced as in the word "an," but more like "ang" as in the "-ang" in "sang." The exact same thing happens in the Italian word for "anchor," ancora.

The reason that the Romans spelled their velar nasal "gn" was a simple matter of orthography: they didn't have a letter to indicate the velar nasal, and so simply used what they had, an "n," and used a "g" to mark the position of the 'n', of the nasal, at the velum, where the sound of 'g' is also made. The Germans did exactly the same thing, except they reversed the order: "ng" instead of "gn."

I find that funny, but it's true: I cannot really understand that proposed "-ng" sound. Is there somewhere I can hear it?

I believe you mean to say gn not ng.

As for hearing it, I am trying to find a way of recording my own voice, for I would happily show you and everyone else.

I can't make sense of it, I cannot imagine how is this really pronounced. Maybe I also pronounce English wrongly, then, because I say "significant" as "sig-neef-cant", and do not see any alteration in its "gn".

"significant" is pronounced "sig-nif-i-cant" (don't forget the other short 'i' sound). Our English pronunciation of Latin is distinctly Germanic: the Germans pronounced Latin words as they saw them: they saw a 'g', and then an 'n', and pronounced them in that order. These harsh phonics are typical of German, but not of Latin. In any case, as far as English "gn" goes, I think you're probably pronouncing that just fine.

[Lucus Eques]

I have recorded my voice pronouncing the following words:

magnus
magna
magnum

signum
signa

gnaeus

wav file here: http://myth.bungie.org/hosted/inmates/gn.wav

Enjoy!

[Cyborg]

Thanks for the sample, Luce Eques!
That explains a lot. Wow, it's really a hard sound to describe.
It seems to me almost a ñ spanish sound, but with a "g" somewhere in it. It really seems to prolong its previous vowel, and nasalize it. then it appears to say "g" slightly and then the proper "n".

That's what I heard (I tried my best):
mãnus, mãga, mãnu, si(n)nu, singa, (g)naeus. I can bearly hear the "click" of the consonants in parentheses.

I find it really hard to pronounce.

[Lucus Eques]

Did you say your native language is Spanish, Cyborg?

[Cyborg]

Did you say your native language is Spanish, Cyborg?

Portuguese. I just used the "ñ" there because it's more recognizable than the Portuguese way "nh".
But when I wrote up what I heard in that last post I did use no "ñ". I'm just stating that as (i think) it might come up wrong in some PC's (the tilde might get above another letter). I've only used tildes above vowels there, and they meant only "nasalization", not "ñ".

[Lucus Eques]

Portuguese, as I recall, does not have the velar nasal; indeed, most Romance languages, ironically, lack this sound, and therefore have trouble at pronouncing the "ng" in "sing" and "king" and the like in languages like English and German; usually they simply drop the 'g'. That explains your confusion, both in producing the sound and hearing it; indeed, if the gn of Latin truly is like the "ng" of English, and I believe it is, the pronunciation of many Latin words is counterintuitive for native English speakers as well (our language begins no word with the the velar nasal, except for gnome, for which most speakers don't use a velar nasal, but a dental nasal — interestingly, "gnome" is a Latin word, from gnomus). It takes practice.

[benissimus]

Luce, where are you getting your information? You are arguing as though you know as fact that Latin gn is pronounced as English ng, but everything I have ever read on the subject said it should be pronounced ngn. I have no problem pronouncing Gnaeus or gnosco or gnascor; as a matter of fact I find them a lot friendlier than such words as [size=150]πτερον[/size], [size=150]μνημη[/size], [size=150]κνημη[/size], etc. Simply by such Greek examples, I think your argument that a difficult string of consonants must really be standing for another sound is invalid by itself. English itself has far more difficult consonant patterns, but most of us are used to them (like the kstk sound in Textkit). Why was Gnaeus originally begun with a Cn if the first letter was not in fact a velar consonant? I have studied quite a bit about Latin pronunciation and I have not seen your approach before, so I would be interested to see where you found it or if you derived it yourself. This pronunciation would have some virtues to me (it would alleviate at least one mystery of the comparative degree) but I do need some proof.

Cyborg, I always find it interesting to hear how people learn English. I can't believe you weren't taught the -ng sound - English uses it with all present participles/gerunds and in quite a few other words. I seriously doubt anyone pronounces the word "gnome" any differently from "*nome" though.

[Cyborg]

Portuguese, as I recall, does not have the velar nasal;

Well, I think we do have it. In word like "banho" (bath), for example, which would be transliterated to spanish as baño. Is this it?

Cyborg, I always find it interesting to hear how people learn English. I can't believe you weren't taught the -ng sound - English uses it with all present participles/gerunds and in quite a few other words. I seriously doubt anyone pronounces the word "gnome" any differently from "*nome" though.

benissime, I must say I'm finding this rather amusing.
I cannot see how is "ng" in "sing" any different from any other "ng" I've seen in English. Well, ok, maybe "bingo" has it differently, "longer" has it differently... maybe so does "exchanged"... I don't know, it's so subtle it's driving me crazy!
But I watch too much American-English-spoken TV and I really have a somewhat nice accent (or so everybody tells me), so this means I probably am not mispronouncing "sing" or "longer" - I just never really looked at them and noticed "oh, so there's a difference between these two...".
I actually learnt English in a very intensive way - I read (and still do read) too much English, I wrote it much too... And heard it and spoke it a lot, because teachers never speak Portuguese or let us do it in-class at good English schools.
I may not have the largest vocabulary I could have, but then again I don't either in Portuguese (I like KIS - "keep it simple" at all times).
But to say the least I do have some sense of English idiomatic expressions - like "but then again", for instance. I've read a couple of books on this subject.

[FiliusLunae]

Well, I think we do have it. In word like "banho" (bath), for example, which would be transliterated to spanish as baño. Is this it?

<ñ> in Spanish is a palatal nasal, like in Italian and French. However, in Brazil (you're Brazilian, right?), <nh> usually, though not always, tends to be pronounced differently, what to me sounds like a nasalized vowel, followed by /j/, or simply a nasalized /j/. In Portugal, the patalal nasal is, generally, always preserved. The thing is that in Portuguese it is difficult to isolate the velar nasal ("ng") sound because vowels are nasalized before nasal consonants (in closed syllables, and in some regions, even in open syllabes). So while in Spanish it is easy note that the <n> in "inglés" sounds different than that in "mentir", in Portuguese, as you know, the <n> is basically "lost", having been absorved into the nasalization of the previous vowel. Nonetheless, with careful (and I mean, very careful!) analysis, one may hear a dental release in Portuguese "mentir", so that a brief (very brief) <n> is sounded, following the nasalized <e>. The same with "inglês", a brief velar nasal (that is, the <ng> being discussed) can be heard after the nasalized < i >. However, like I said, this requires careful examination, that speakers are highly unlikely to notice them.

everything I have ever read on the subject said it should be pronounced ngn

Me too. I don't have it at hand right now, but I recall that in the book "The Romance Languages", in the section about Latin, this topic is discussed. The minimal pair "agnus" and "annus" is compared, each pronounced, respectively: /ang.nus/ and /annus/. The phonemic status of /ng/ is then dismissed on the basis that the velar nasal (<ng>) in "agnus" is simply an assimilation of the <n> due to the velar consonant (<g>). The way I see it <gn> was used in opposition to <ng>, where the first one represented a velar nasal followed by <n>, and the second one a velar nasal followed by <g>. Therefore:
gn » /ng.n/
ng » /ng.g/
However, I don't dismiss what Lucus says, as I've considered that as well.
So <ng> seems to have been a frequent sound, ocurring as well with <n> before /k/. And, there's also the hypothesis that /m/ in word-final position was pronounced as a velar nasal as well, e.g. filium /filiung/, etiam /etiang/ (which is exactly how this man pronounces them, and how I'm starting to pronounce them myself)

But I watch too much American-English-spoken TV and I really have a somewhat nice accent (or so everybody tells me), so this means I probably am not mispronouncing "sing" or "longer" - I just never really looked at them and noticed "oh, so there's a difference between these two...".

My Italian professor pronounced <n> and <g> separately, so that "singing" sounded like /sin.gin.g/. Especially in word-final position, one could clearly hear a <g> after the <n>.

[Lucus Eques]

Luce, where are you getting your information? You are arguing as though you know as fact that Latin gn is pronounced as English ng,

Not at all; my theory is simply an opinion, as I've stated repeatedly:

indeed, if the gn of Latin truly is like the "ng" of English, and I believe it is,

but everything I have ever read on the subject said it should be pronounced ngn.

If I read the Latin of a Roman who explained as much, I will surely believe it. But until then it is merely philological theory, no? just as my own.

I have no problem pronouncing Gnaeus or gnosco or gnascor; as a matter of fact I find them a lot friendlier than such words as [size=150]πτερον[/size], [size=150]μνημη[/size], [size=150]κνημη[/size], etc.

I do not, personally; the first two for me are distinctly easier to render, while the last is on par with the "ngn" pronunciation of "Gnaeus."

Simply by such Greek examples, I think your argument that a difficult string of consonants must really be standing for another sound is invalid by itself.

Not really. Indeed, in Latin, nc or ng is extremely different from a literal n + c or n + g. And the palatal 'y'-sound that is made by consonantal Latin i is drastically different from the vowel i. They are related, and connected, but a poverty of alphabetical characters is a feature of nearly all languages, excepting for the most part Sanskrit. It's an economy of letters.

English itself has far more difficult consonant patterns, but most of us are used to them (like the kstk sound in Textkit).

Except that each one of those consonants is not pronounced, not articulated clearly at all. Indeed, the consonants, when spoken at normal velocity and not enunciated for the virtue of spelling, largely lose their natural, unmitigated character, and together make a sound entirely different from a literal "kstk." Such is the nature of our language. I do not disagree that there are plenty of difficult consonant patterns English, for there are, but Latin is not English.

Why was Gnaeus originally begun with a Cn if the first letter was not in fact a velar consonant?

Actually, the fact that the Romans lacked 'g' for some time and used 'c' in the manner you have above only vindicates my proposition. I'm not exactly sure what you're trying to say, but what I'm trying to say is that the (early) Romans understood, to some degree, that they had a sound — the velar nasal — which lacked a character in their writing system. 'n' was thus charged with taking on a double, even tripple role as a nasal from the dentals to the velars. The letter's official pronunciation is dental, of course, but it is not denied by any of us here that it takes on a dynamic role in the course of the Latin language (i.e., "ancora," "ingens," et cetera). The 'n' may represent a velar nasal, therefore. However, presume that this sound, of the velar nasal, comes on its own, as I believe it does in the gn of Gnaeus. What is there to let the reader know that this is not a dental nasal, but a velar one? It seems that the solution was to add c before the 'n', a velar consonant to denote the velar nature of the nasal that was to be expressed, not to be pronounced separately on its own. And later with the invention of g, this letter took the place of the former, except in abbreviation of the above name. Thus to me it seems clear beyond a shadow of a doubt, especially since both velar consonants were employed in the job of "marking the velar," as I have proposed, that my theory on this particular pronunciation is vindicated simply by the Roman orthographical convention.

I have studied quite a bit about Latin pronunciation and I have not seen your approach before, so I would be interested to see where you found it or if you derived it yourself.

I have derived it myself, as I always have said. Heck, you were present in the thread where I first proposed it, a long time ago.

This pronunciation would have some virtues to me (it would alleviate at least one mystery of the comparative degree) but I do need some proof.

I have none but for the proposals of philology. The reason that the "ngn" pronunciation for gn does not make sense to me is because it is two different nasals in succession, a velar nasal and then a dental nasal. As a solitary sound on its own, it seems unnecessary to me.

The other means we have for identifying the nature of gn in Latin is to see how it reacts in its daughter languages, such as Italian. In Italian, this spelling represents the palatal nasal "ñ," as it is spelled in Spanish, and "nh" in Portuguese, as we have learned from our friend Cyborg. In Italian, numerous consonants which in Latin are velar are rendered palatal, such as ce and ci, possessing the Spanish or English "ch" sound, as in "church," with ge and gi right along side them. With this trend in mind, we may reverse the linguistics of time, as it were, moving ce and ci in Italian back from the palate and returning them to the velum as they were in Latin. If we do the same thing for the gn of Italian, we arrive with the velar nasal. Therefore I believe that gn in Latin represents the velar nasal, alone.

[Cyborg]

Well, I think we do have it. In word like "banho" (bath), for example, which would be transliterated to spanish as baño. Is this it?

<ñ> in Spanish is a palatal nasal, like in Italian and French. However, in Brazil (you're Brazilian, right?), <nh> usually, though not always, tends to be pronounced differently, what to me sounds like a nasalized vowel, followed by /j/, or simply a nasalized /j/. In Portugal, the patalal nasal is, generally, always preserved. The thing is that in Portuguese it is difficult to isolate the velar nasal ("ng") sound because vowels are nasalized before nasal consonants (in closed syllables, and in some regions, even in open syllabes). So while in Spanish it is easy note that the <n> in "inglés" sounds different than that in "mentir", in Portuguese, as you know, the <n> is basically "lost", having been absorved into the nasalization of the previous vowel. Nonetheless, with careful (and I mean, very careful!) analysis, one may hear a dental release in Portuguese "mentir", so that a brief (very brief) <n> is sounded, following the nasalized <e>. The same with "inglês", a brief velar nasal (that is, the <ng> being discussed) can be heard after the nasalized < i >. However, like I said, this requires careful examination, that speakers are highly unlikely to notice them.

Wow, that's throughout right; you really know this well.

Luce Eques, I really respect the right you have to propose a hypothesis, just like any other person who is a language researcher. However, I guess that even if I agreed with your hypothesis, I couldn't use it for I cannot pronounce it.

[Misopogon]

Latin pronunciation is a very interesting subject that I should read more about. I find also fascinanting the way the romance languages have devoloped their own pronunciation, like for the sound -gn. Let me know a good textbook about Latin pronunciation, thanks.

My Italian professor pronounced <n> and <g> separately, so that "singing" sounded like /sin.gin.g/. Especially in word-final position, one could clearly hear a <g> after the <n>. [/quote]

I wonder if the "clear" g you can hear at the end of the word really depends on lack of such a sound in Italian: that's in part true of course. Italian alphabet is pretty "phonetic" and I find that some people tend to make mistakes just reading English like it was Italian. I make such mistakes sometimes, especially with new words. I believe that English and Italian are phonetically pretty different, that's why English speakers of Italian (and probably also Italian speakers of English) tend to have a strong accent/intonation. Anyway I am not a linguist, so I could be wrong, feel free to correct me.

[Cyborg]

Let me know a good textbook about Latin pronunciation, thanks.

I think Sidney Allen's "Vox Latina". I only "think" because I'm about to receive the one I bought, so I really haven't checked it out yet. I wonder what does it say about the "gn" issue.

[sisyphus]

My Italian professor pronounced <n> and <g> separately, so that "singing" sounded like /sin.gin.g/. Especially in word-final position, one could clearly hear a <g> after the <n>.

This could well be an artefact of where he or she learned English or resided. It is a feature of certain Northern (England) accents.

[annis]

but everything I have ever read on the subject said it should be pronounced ngn.

If I read the Latin of a Roman who explained as much, I will surely believe it. But until then it is merely philological theory, no? just as my own.

No. Or rather, that depends on how you're using the word theory. The 'merely' philological theory has a wider base of evidence.

(I will represent the 'ng' sound of 'sing' with an upper case 'N'. Not everyone will have the Unicode IPA font.)

A few more points in support of the -gn- = /Nn/ pronunciation.

1. Epigraphic evidence. Among alternate early spellings we see SINNU 'signum' and, SINGNIFER 'signifer'. The latter type especially, with the duplication of 'n', points to /Nn/.

The other means we have for identifying the nature of gn in Latin is to see how it reacts in its daughter languages, such as Italian.

2. Daughter Languages, such as Romanian. The Romanian reflex of Latin -gn- is -mn-, limn < lignum. In fact, all velar+dental stop combintations became labial+dental (faptu < factum), and -mn- from /Nn/ would parallel that.


Edit: Oops, forgot to cite: Sihler, Section 220.

[Lucus Eques]

No. Or rather, that depends on how you're using the word theory. The 'merely' philological theory has a wider base of evidence.

Seriously, though; what did the Romans say? Surely a Roman like Quintillian had something to say on the matter.

1. Epigraphic evidence. Among alternate early spellings we see SINNU 'signum' and, SINGNIFER 'signifer'. The latter type especially, with the duplication of 'n', points to /Nn/.

I fail to see why either of those indicate the Nn proposition. The "ng" in the "SINGNIFER" resembles our English and German spelling to indicate the velar nasal, but keep in mind that, if "ng" is pronounced phonetically, you do not get the velar nasal, but instead exactly what our resident Italian Misopogon was describing: "n + g," like his Italian professor when mispronouncing English. Therefore the occurrence of the same in this one Latin example does nothing to further a sense of a more "phonetic" representation of the sounds in question, namely the Nn, for if so, the Romans would have had to have had a Germanic orthographic sensibility.

Indeed, both the double-'n' and the "ngn" in those examples could be orthographic representations of the velar nasal, and solely the velar nasal. There is nothing about their nature which implies inherently the Nn conclusion.

Actually the "nn" is even more of a vindication for my own proposal. In Spanish, "ñ" is short for "nn," the palatal nasal. And solely the palatal nasal. Therefore a double-'n' in Latin, such as in the example of "SINNU," could be, by logic of Occam's Razor, also solely the the velar nasal.

The other means we have for identifying the nature of gn in Latin is to see how it reacts in its daughter languages, such as Italian.

2. Daughter Languages, such as Romanian. The Romanian reflex of Latin -gn- is -mn-, limn < lignum. In fact, all velar+dental stop combintations became labial+dental (faptu < factum), and -mn- from /Nn/ would parallel that.

Not to insult the Romanians, but I find that laughable; Romanian is perhaps the most distant Romance language from Latin, ironic as that fact is considering its name. Romanian has been affected by incessant eastern influences, linguistically as well as culturally. It's eminantly clear that those labial-dental combinations are rendered by exposure to Greek or other similar tongues. Compare this to the Italians, who hold the birthright to the land and the blood of the Romans, and whose language is an evolution directly from the Italic vulgate, with a slight Etruscan accent mixed in. And as for "lignum," the Italian is legno, again the gn representing solely the palatal nasal. I conclude that the Latin here possessed solely the velar nasal.

I believe my underlying current of conviction with my postulation is that it is, in my opinion, the simplest explanation. The sound of Nn requires a complexity that is not supported by Latin's daughter languages (not even Romanian; I understand and appreciate the logic of your corollary via the velar-dental -> labial-dental transformation, but to me it seems farther fetched, and litterally farther, from less distant examples). Occam's Razor is an important tool, especially when information is limited.

[Lucus Eques]

My Italian professor pronounced <n> and <g> separately, so that "singing" sounded like /sin.gin.g/. Especially in word-final position, one could clearly hear a <g> after the <n>.

Yes, such a pronunciation is definitely wrong. Ma succede; la lingua nostra è molto difficile pronunciare perfettamente — ma per fortuna, ci piacciono gli accenti stranieri.

I wonder if the "clear" g you can hear at the end of the word really depends on lack of such a sound in Italian: that's in part true of course.

It's absolutely true.

Italian alphabet is pretty "phonetic" and I find that some people tend to make mistakes just reading English like it was Italian. I make such mistakes sometimes, especially with new words. I believe that English and Italian are phonetically pretty different, that's why English speakers of Italian (and probably also Italian speakers of English) tend to have a strong accent/intonation.

All very true.

Anyway I am not a linguist, so I could be wrong, feel free to correct me.

Not at all! you're right on the ball. Cioè, secondo me, hai detto correttamente.

[annis]

I fail to see why either of those indicate the Nn proposition. The "ng" in the "SINGNIFER" resembles our English and German spelling to indicate the velar nasal, but keep in mind that, if "ng" is pronounced phonetically, you do not get the velar nasal, but instead exactly what our resident Italian Misopogon was describing: "n + g," like his Italian professor when mispronouncing English.

So? I do not dispute that Latin had no separate letter to indicate /N/. So they had to find some other way; we both agree -g- is used to indicate that a nearby "n" is a velar nasal. The -NGN- notation is a passable way to do that. -NN- is less clear on articulation, though does point at the possibility of a consonant cluster, which is consistent with metrics which absolutely point at a cluster, not a single nasal. 'Signum' does not scan uu.

Therefore the occurrence of the same in this one Latin example does nothing to further a sense of a more "phonetic" representation of the sounds in question, namely the Nn, for if so, the Romans would have had to have had a Germanic orthographic sensibility.

Please explain what you mean by "Germanic orthographic sensibility."

Indeed, both the double-'n' and the "ngn" in those examples could be orthographic representations of the velar nasal, and solely the velar nasal. There is nothing about their nature which implies inherently the Nn conclusion.

And this is where I feel you are misusing the word "theory." No one point I or others have made is by itself conclusive. All the evidence has to work together before bits get sliced off by Occam. Neither -NN- nor -NGN- argue against the /Nn/ interpretation, nor necessarily for your /N/ idea.

Actually the "nn" is even more of a vindication for my own proposal. In Spanish, "ñ" is short for "nn," the palatal nasal. And solely the palatal nasal.

Again, so? This is just a "Spanish orthographical sensibility." This is no sounder an argument than reference to "Germanic orthographical sensibility."

Therefore a double-'n' in Latin, such as in the example of "SINNU," could be, by logic of Occam's Razor, also solely the the velar nasal.

Could be. But only if the other evidence support this interpretation.

2. Daughter Languages, such as Romanian. The Romanian reflex of Latin -gn- is -mn-, limn < lignum. In fact, all velar+dental stop combintations became labial+dental (faptu < factum), and -mn- from /Nn/ would parallel that.

Not to insult the Romanians, but I find that laughable;

Please do explain to me when laughability became a standard measure of philological soundness. By that standard Armenian isn't Indo-European at all.

Romanian is perhaps the most distant Romance language from Latin, ironic as that fact is considering its name.

The -mn- < -gn- equivalence must still be addressed. As must the development of /ñ/ from /N/ in other Romance languages, which is a bit of stunt to manage in distant languages with no phonological conditioning. The transition from /N/ to /n/ takes you right through /ñ/, however.

Romanian has been affected by incessant eastern influences, linguistically as well as culturally. It's eminantly clear that those labial-dental combinations are rendered by exposure to Greek or other similar tongues.

No. Absolutely not. Until you can show me the development of /N/ to /mn/ in other languages in the region this argument has no basis in fact. There is nothing like it in the history of Greek, so you needn't look there.

Compare this to the Italians, who hold the birthright to the land and the blood of the Romans,

Meaningless nativism. Language change operates via language communities, not blood, even if there is often overlap.

The sound of Nn requires a complexity that is not supported by Latin's daughter languages

How is this complex?! In an earlier post you say this:

The reason that the "ngn" pronunciation for gn does not make sense to me is because it is two different nasals in succession, a velar nasal and then a dental nasal.

Do you also have a theory about how 'somnus' or 'amnis' are too difficult because -mn- is two different nasals in succession?

Points all consistent with the interpretation of -gn- as /Nn/:

• metrics: all vowels before -gn- are scanned long
• epigraphy: SINGNIFER
• historical comparative linguistics: both *kn and *gn of the parent language are -gn- in Latin; assimilation of the stop to nasal articulation is common, and is paralleled by other stop+nasal changes: summus < *supmos (cf. 'super'); summitto < submitto (not sumus or sumitto)
• changes in the daughter languages, Rom. -mn- < L. -gn- is consistent with /Nn/ as is -ñ- < L. -gn-

The metrical and comparative points undermine the interpretation of -gn- as a single nasal sound.

[Lucus Eques]

So? I do not dispute that Latin had no separate letter to indicate /N/. So they had to find some other way; we both agree -g- is used to indicate that a nearby "n" is a velar nasal. The -NGN- notation is a passable way to do that. -NN- is less clear on articulation, though does point at the possibility of a consonant cluster, which is consistent with metrics which absolutely point at a cluster, not a single nasal. 'Signum' does not scan uu.

Now that is a good point, and a very interesting one. In that case, is it conceivable that the sole velar nasal that I propose could have a doubled quality to it? that is, represent /NN/. I am reminded of the IPA spellings of Italian words; for instance, the preposition-article combination agli (meaning "to the" [masculine plural, before a vowel]), pronounced vaguely like "alyee," is written like so:

a[size=117]λλ[/size]i

In IPA, the Greek lambda is used to represent the palatal liquid rendered by Italian gli (actually, the real IPA has backwards lambdas, but this is the best I have ). "gli" is spelled "[size=117]λ[/size]i," while a[size=117]λλ[/size]i has a doubled consonant, merely because of its position between vowels, since its nature is inherently doubled in force when pronounced. It is interesting to wonder if the same could be true with Latin gn being /NN/.

Please explain what you mean by "Germanic orthographic sensibility."

I mean that Germanic languages like English and German use "ng" to represent the velar nasal, even though the velar nasal is in truth neither letter. Your postulation that the "ng" in the "NGN" of "SINGNIFER" indicates this same sound, then to be followed by the second "N" (a dental nasal), is based on the spelling conventions of an entirely different language group.

And this is where I feel you are misusing the word "theory." No one point I or others have made is by itself conclusive.

I apologize; I was not using the word "conclusion" so strictly, but more liberally as a general synonym for the word "thought."

Neither -NN- nor -NGN- argue against the /Nn/ interpretation, nor necessarily for your /N/ idea.

Nor necessarily against my /N/ interpretation, nor necessarily for /Nn/.

Again, so? This is just a "Spanish orthographical sensibility." This is no sounder an argument than reference to "Germanic orthographical sensibility."

Ah! so you do know what I mean by "Germanic orthographic sensibility"!

I might be that "Spanish orthographic sensibility" is no sounder, except that Spanish is a Romance language, from Latin, whereas German is not.

I am inclined to agree that there are significant limits to utilizing the daughter languages for understanding Latin, but short of the texts of the Roman grammarians, they are all we have.

Could be. But only if the other evidence support this interpretation.

And it does.

Please do explain to me when laughability became a standard measure of philological soundness. By that standard Armenian isn't Indo-European at all.

If you read the phrase which immediately follows, I explain how to me it seems absurd to utilize Romanian when it is such a distant dialect, heavily influenced by languages not of a Romantic nature. My word "laughable" came from the fact that I literally laughed outloud upon reading the use of Romanian for justification — not that I don't think it should be considered, but simply that it is comparatively less informative than ... almost any other choice.

The -mn- < -gn- equivalence must still be addressed. As must the development of /ñ/ from /N/ in other Romance languages, which is a bit of stunt to manage in distant languages with no phonological conditioning.

I do not follow what you mean in that last phrase.

The transition from /N/ to /n/ takes you right through /ñ/, however.

I have considered this possibility, yes, and it seems plausible.

Meaningless nativism. Language change operates via language communities, not blood, even if there is often overlap.

And where are the most concentrated remaining language communities descendent from the Romans? That Italian and its dialects are the closest living languages to Latin remains true.

How is this complex?! In an earlier post you say this:

The reason that the "ngn" pronunciation for gn does not make sense to me is because it is two different nasals in succession, a velar nasal and then a dental nasal.

Do you also have a theory about how 'somnus' or 'amnis' are too difficult because -mn- is two different nasals in succession?

My original emphasis was misleading. The interpolation of "two different nasals rendered by the tongue in succession" would have clarified the matter. 'm' is a very particular nasal, as it is the only one which does not require the tongue. Indeed, this gives the opportunity for all the lovely combinations, two of which you mentioned. Forming two nasals with the same tongue one after the other, however, is significantly more challenging.

Points all consistent with the interpretation of -gn- as /Nn/:

• metrics: all vowels before -gn- are scanned long

Excepting the possibility of an inherent doubling.

• epigraphy: SINGNIFER

I've already shown how this is deceptive logic; "ng" in English does not automatically prove the same pronunciation in Latin. Indeed, this epigraphy is as consistent with my theory as with yours.

• historical comparative linguistics: both *kn and *gn of the parent language are -gn- in Latin; assimilation of the stop to nasal articulation is common, and is paralleled by other stop+nasal changes: summus < *supmos (cf. 'super'); summitto < submitto (not sumus or sumitto)

I do not deny this. However, those are all single or double nasals of the same type: namely, 'm'. This supports my theory.

[*] changes in the daughter languages, Rom. -mn- < L. -gn- is consistent with /Nn/ as is -ñ- < L. -gn- [/list]

You're still using Romanian. Romanian is ridiculously distant from Latin, Italian, Spanish, even Portuguese and French. It's so deeply imbued with Slavic pronunciation and vocabulary Stalin tried to prove it was a Slavic language.

The metrical and comparative points undermine the interpretation of -gn- as a single nasal sound.

The metrical point dooms a solitary velar nasal, potentially, unless it always comes doubled.

[annis]

It is interesting to wonder if the same could be true with Latin gn being /NN/.

It's not implausible.

Just to be clear, you're now saying that -gn- represented /NN/, not /N/?

I mean that Germanic languages like English and German use "ng" to represent the velar nasal, even though the velar nasal is in truth neither letter. Your postulation that the "ng" in the "NGN" of "SINGNIFER" indicates this same sound, then to be followed by the second "N" (a dental nasal), is based on the spelling conventions of an entirely different language group.

I still fail to see the relevance of this. If English had chosen 'pt' to represent the sound /N/, maybe, but using a cluster of -n- and -g- to represent a feature blend, /N/, isn't that much of a stretch. Your own interpretation assumes it. Why is it somehow fundamentally Germanic to write it -NG- and Latinate -GN-? This Germanic sensibility argument seems a distraction.

If you read the phrase which immediately follows, I explain how to me it seems absurd to utilize Romanian when it is such a distant dialect, heavily influenced by languages not of a Romantic nature.

It is nonetheless transparently a Romance language. You can't just toss the evidence it offers.

The -mn- < -gn- equivalence must still be addressed. As must the development of /ñ/ from /N/ in other Romance languages, which is a bit of stunt to manage in distant languages with no phonological conditioning.

I do not follow what you mean in that last phrase.

Sorry. By conditioning I mean sound changes that take place in a particular environment. For example, in the evironment 'vowel' _ 'stop' the sound 'k' underwent different changes in the different Romance languages.

nocte-: Fr. nuit, Sp. noche, It. notte
octo-: Fr. huit, Sp. ocho, It. otto

So the change of /k/ to /i/ in French is condiditioned by being after a vowel and before the /t/.

You're positing a universal change from /NN/ to /ñ/ in the western Romance languages. That /NN/ might become /ñ/ in one language is reasonable enough. If you were saying /ñ/ < /NN/ near /i/ or some other front vowel, I could see it. But you're saying it happened everywhere, from French to Italian, etc. (except of course Romanian), regardless of environment, and this seems less likely.

[*] changes in the daughter languages, Rom. -mn- < L. -gn- is consistent with /Nn/ as is -ñ- < L. -gn- [/list]

You're still using Romanian. Romanian is ridiculously distant from Latin, Italian, Spanish, even Portuguese and French. It's so deeply imbued with Slavic pronunciation and vocabulary Stalin tried to prove it was a Slavic language.

I don't much care what Stalin thought. He wasn't a linguist. Until you can account for the -mn- from /N/ (or now I suppose it's /NN/) I will continue to use Romanian for the evidence. As will, no doubt, practicing linguists.

[Lucus Eques]

It's not implausible.

Just to be clear, you're now saying that -gn- represented /NN/, not /N/?

I'm suggesting it as a possibility, and favor one or the other.

I still fail to see the relevance of this. If English had chosen 'pt' to represent the sound /N/, maybe, but using a cluster of -n- and -g- to represent a feature blend, /N/, isn't that much of a stretch. Your own interpretation assumes it. Why is it somehow fundamentally Germanic to write it -NG- and Latinate -GN-? This Germanic sensibility argument seems a distraction.

My position is that it is potentially misleading, and that "ngn" could just as well mean /N/ (or /NN/) as it could /Nn/. Spellings of strange sounds are drastically different between languages. In German, sch always makes the sound of sh in English, while in Italian this same spelling always makes the sound of sk in English, whereas in Norwegian sk makes the sound of English sh, which in Italian is achieved by sce or sci, which in Old English required merely sc, and is found in French as ch, which in English is the sound of the ch in "church," which German forms by tsch, and Italian ce or ci, and Norwegian by kj ... You get the picture. Our own spellings are deceptive. And assuming they must apply to a very rare and possibly non-standard Latin source does not take this into consideration. Otherwise "ngn" supports my theory as much as your own.

It is nonetheless transparently a Romance language. You can't just toss the evidence it offers.

I agree, and my original ridicule was a bit extreme, for everything descendent or related to Latin has a part to play. My argument is that it should not come first.

The -mn- < -gn- equivalence must still be addressed. As must the development of /ñ/ from /N/ in other Romance languages, which is a bit of stunt to manage in distant languages with no phonological conditioning.

I do not follow what you mean in that last phrase.

Sorry. By conditioning I mean sound changes that take place in a particular environment. For example, in the evironment 'vowel' _ 'stop' the sound 'k' underwent different changes in the different Romance languages.

nocte-: Fr. nuit, Sp. noche, It. notte
octo-: Fr. huit, Sp. ocho, It. otto

So the change of /k/ to /i/ in French is condiditioned by being after a vowel and before the /t/.

You're positing a universal change from /NN/ to /ñ/ in the western Romance languages. That /NN/ might become /ñ/ in one language is reasonable enough. If you were saying /ñ/ < /NN/ near /i/ or some other front vowel, I could see it. But you're saying it happened everywhere, from French to Italian, etc. (except of course Romanian), regardless of environment, and this seems less likely.

No no, I'm hardly suggesting "everywhere;" I am suggesting that these languages, especially the Italian ones, have a place that comes before Romanian.

Moreover, there's nothing impossible about all the Western Romance languages having something in common; the vulgate had different branches, and their resultant modern languages show the traits in common; pure 'h' virtually disappeared from the Western daughter tongues, while it remains in Romanian; palatalization and aspiration occurred to an extraordinary degree, etc. Consonantal u became v universally, with some exceptions, regardless of the specific vowel that followed or came before.

You're still using Romanian. Romanian is ridiculously distant from Latin, Italian, Spanish, even Portuguese and French. It's so deeply imbued with Slavic pronunciation and vocabulary Stalin tried to prove it was a Slavic language.

I don't much care what Stalin thought. He wasn't a linguist. Until you can account for the -mn- from /N/ (or now I suppose it's /NN/) I will continue to use Romanian for the evidence. As will, no doubt, practicing linguists.

As you should, as we all should. Besides, who here hasn't fallen in love with this little Romanian melody, as performed by this handsome gentleman:

http://uploads.ungrounded.net/content.p ... =500&h=375

Endless replay value.

Also, I've recently come upon the origin of "Lorraine," which is from the German name "Lothringen," from the Latin Lothari regnum. In this exchange, we see how the Germans spelled "gn" as "ng" to render the velar nasal in their own fashion. Germanic orthographic sensibility indeed. Nothing conclusive, but still interesting.

[Cyborg]

Oh my... I'm lost now!
Have you both decided anything yet?
annis seems to be very well informed on the current theories on the GN pronunciation. So, tell me, annis (I'm really lost) in one short phrase: how should I pronounce GN in "magnus", for instance?

[annis]

Have you both decided anything yet?

I think we've both decided that 1) everybody loves Romanian-language boy-bands and 2) each us is going to stick with our own preferred pronunciation of -gn-.

annis seems to be very well informed on the current theories on the GN pronunciation. So, tell me, annis (I'm really lost) in one short phrase: how should I pronounce GN in "magnus", for instance?

Continuing to use 'N' to indicate the sound of -ng- in 'sing' (siN), I will pronounce 'magnus' as /maN nus/. (How charming... it appears the standard English example for that sound combination is 'hangnail.)

[Cyborg]

Continuing to use 'N' to indicate the sound of -ng- in 'sing' (siN), I will pronounce 'magnus' as /maN nus/. (How charming... it appears the standard English example for that sound combination is 'hangnail.)

Feeble G. That's great, the example helps a lot!
Thanks, annis.

[Lucus Eques]

So <ng> seems to have been a frequent sound, ocurring as well with <n> before /k/. And, there's also the hypothesis that /m/ in word-final position was pronounced as a velar nasal as well, e.g. filium /filiung/, etiam /etiang/ (which is exactly how this man pronounces them, and how I'm starting to pronounce them myself)

Indeed, it seems there's a great deal of evidence and Roman testimony to demonstrate that final m had a very weak nature. I don't think that final m should always been a velar nasal, however; and yes, heh, that gentleman does an admirable job of attempting the reconstructed pronunciation of Latin. His main problem is the equilocution of English short 'i' with Latin short 'i' — the proposition that Latin vowels changed so substantially in quality along with quantity to me is absolutely ridiculous. (Plus, he sounds dreadfully unnatural— I personally opt for the Vox Latina style).

In any case, what that gentleman is trying to do is nasalize every single final m into the preceding vowel. I believe this is a mistake. Quintillian remarked that when final m is followed by a word which begins with a vowel, then the m will fully nasalize with the preceding vowel, and the two vowels on either side of the "m" will elide. The example most often given is "multum ille" from the beginning of the Aeneid, which sounds more like "multu˜ille."

Sandhi is a syntactic concept which explains the blending of words. A good example is what happens to cum when it is affixed to most words. If we assume the archaic com from Old Latin, joining this to a word like tendere results in contendere, where the 'm' becomes the dental nasal 'n'. If we do the same with cedere, we arrive at concedere, where 'n' represents the velar nasal.

Sandhi, however, applies to more than just compound words, but also to the syntax of words in a sentence. Indeed, in the phrase Quid dicis? one would not pronounce the 'd' in "quid" as the final letter of the sentence, but both 'd's together like a doubled consonant; spelling thus the phrase in the Roman way, we have quiddicis, where the first 'd' is implosive, and the second explosive. It is very similar to the pronunciation of "What do you say?" where the 't' of "what" is blended with the 'd' of "do," or, as in many dialects, the 't' becomes a glottal stop — the trademark of a consonant combination.

Applying this to the final m of cum in the Classical Latin, we can imagine that a phrase like cum te eo might more accurately be rendered as cunte eo, where the final 'm' of "cum" has become the dental nasal 'n'. I find eliminating spaces in all Latin sentences a very valuable learning tool for understanding this fluidity of pronunciation between words inherent to all languages, and even in this example, assuming there is a macron over the 'e' of "te," we find that cunteeo can be helpful in understanding the nature of this liaison and elision. In urbem gero might be clearer as inurbengero, where the 'm' has become the velar nasal 'n'. We find evidence for these possibilities in the alternate spelling for quandocumque: quandocunque — both were probably pronounced exactly the same.

I believe therefore that final m possesses a generic nasal quality, not nearly so strong as that of French, as is common in the reconstructed pronuniciation, but significant enough that its sound transforms to accomodate whatever letter follows it in the next word. This would be a pattern consistent with the tendencies of other languages as well.

[sauvagenoble]

I've been following this thread since it was pointed out to me. I publicly acknowledge Annis's efforts and evidence. To which I add minor points.

To clarify, sandhi is not a process of syntax, but of phonology. Perhaps what was meant was phonotactic? Now, just because m can assimilate to a following dental within the same word, doesn't mean it does so across word boundary. Sanskrit, whence we get the term sandhi, had both external and internal sandhi. I'd like to see inscriptional examples with -n##T- (T = {t,d}).

As for gn, there's no systematic way gn could ever be interpreted as N(N). One would have to believe that gnatura /NNatura/ > /Natura/ > /natur-/ to get the Romance reflexes spelled natur-. In other words, nasalization of g, assimilation of n, degemination of the resulting NN, change back of N to n. The most straightforward interpretation is Nn > (N)n. In other words, Nn after vowel, n in initial position. Of course, Nn in archaizing contexts.

And one should be very careful not to take ancient testimony at face value. The Roman grammarians were notoriously Greek wannabes, operating under the assumption that Latin was a Greek dialect, that Latin had the three accents of Greek.

I've seen a lot of modern language comparison here in support of the minority controversial view. Data from Old Romance sources should be adduced: how did OFr./OSp./OPort./ORom. spell reflexes of Latin gn? Non-standard spellings will be most telling.

[Lucus Eques]

I think we've both decided that 1) everybody loves Romanian-language boy-bands

And how!

and 2) each us is going to stick with our own preferred pronunciation of -gn-.

Indeed. Actually, I feel quite convinced by many of your arguments, annis. However, I still feel certain that some Roman had something to say on the subject. I haven't found anything yet, but I did find this (link in PDF):

[page 4]Before n, however, g spelled the velar nasal. This is suggested by spellings such as INGNES (CIL 4.3121), the evidence of the sound change of dignus from *dek-nos, cf. decet.

with this footnote (9):

It is also probably that [Nn] was the OL phonetic value of initial <gn->. It is a strong (universal?) tedency that if a language has an initial voiced aobstruent plus nasal cluster it will also have at least one initial voiceless obstruent plus nasal cluster. Since Latin has no clusters of this latter sort, it is unlikely that <gn-> could have represented [gn].

Then on page 6 onto 7:

3. / n / before a velar was pronounced as the nasal [N]. But it also seems that g before n was realized as [N]. The arguments for this pronunciation are the following.

• [a.] *e becomes i before what is written <gn>.
  deknos 'fitting'> dignus. Cf. decet 'is fitting.'
  
  If <g> represented [N] then this sound change is simply a part of a more general change rasing *e to i before [N], e.g., *tengo > tingo.
  
  [b.] We occasionally find spellings like SINGNIFER (CIL 6.3637) and DINGNISSIME (CIL 14.1386), which suggest that the consonant before the n was also nasalized.[21]

Footnote 21 says: "Sihler's argument (208 ) about the Romanian outcomes of -gn- is false."

Ah-hah! I knew the Romanian analogy was fishy.

• [c.] Since labial and dental stops assimilate to nasals before n, but retain their place of articulation (e.g. *swepnos > somnus; adnuo > annuo) it would be nice if velars behaved in the same way.

I find the last argument particularly convincing. However, I'm at a loss to discover any other examples of this nature (other than dental ones — in short, I'm looking for more labial 'mn' examples).

I have found a wondrous compilation and resource of Roman grammarians, however, and their commentary on their language (parts of which I will comment on in another thread), and there was this part, which I believe was quoted earlier in this thread:

GN in the terminations _gnus_, _gna_, _gnum_, has, according to Priscian, the power to lengthen the penultimate vowel.

[Prisc. I.] _Gnus_ quoque, vel _gna_, vel _gnum_, terminantia, longam habent vocalem penultimam; ut a _regno_, _regnum_; a _sto_, _stagnum_; a _bene_, _benignus_; a _male_, _malignus_; ab _abiete_, _abiegnus_; _privignus_; _Pelignus_.

(Perhaps the liquid sound, as in cañon.)

In addition to letting us understand why the 'e' of *deknos should change to dignus, and admitting my own lack of instruction in Latin meter, I believe this also explains the problem of the "doubled" nature of "gn" when scanned in poetry, does it not?

This reminds me of the quality of 'n' before 's' or 'f', such as in insula or infelix where the 'n' is not pronounced true, but as a nasalization of the preceding vowel, in both cases 'i', lengthening it.

For now, I believe the mainstream (if there is such a thing when it comes to Latin pronunciation) is where I shall ride, and pronounce gn as the 'ngn' of "hangnail." In the meantime, I shall keep in mind the following verses:


You wish to part, and yet you do not take me along ...
Your visage and love from the linden trees,
And I recall your eyes.

[annis]

Footnote 21 says: "Sihler's argument (208 ) about the Romanian outcomes of -gn- is false."

Ah-hah! I knew the Romanian analogy was fishy.

No.

Or rather, not necessarily. Since the paper is a summary the author gives no evidence for this assertion. I need to see that before I make a decision.

Inspiried by Angelo's (sauvagenoble) post, I had a close look at Old Occitan. The outcome of L. -ng- and -gn- are identical (OOc uses -nh- for the palatal nasal):

renhar < regnare
ponh < pugnum
senhar < signare

tanher < tangere
fenher < fingere

-nh- is also the regular outcome of L. -nct-: planh < planctum. (!)

None of this is conclusive on the L. pronunciation of -gn-, but I thought it was interesting anyway.

You wish to part, and yet you do not take me along ...

That's one of the stranger animations for that song I've seen.

[Lucus Eques]

Inspiried by Angelo's (sauvagenoble) post, I had a close look at Old Occitan. The outcome of L. -ng- and -gn- are identical (OOc uses -nh- for the palatal nasal):

renhar < regnare
ponh < pugnum
senhar < signare

tanher < tangere
fenher < fingere

-nh- is also the regular outcome of L. -nct-: planh < planctum. (!)

None of this is conclusive on the L. pronunciation of -gn-, but I thought it was interesting anyway.

It is, but I'm not sure what it may or may not prove.

That's one of the stranger animations for that song I've seen.

It's actually the first I saw, posted here on Textkit, about a year ago at least. Those albino black sheep ... man, nothing cooler.

[FiliusLunae]

Ah, I have been absent for about two weeks now. There are lot of posts! Hehe...

Anyway, annis, thanks to your comparisons to Romanian, I am now learning Romanian! Haha... Yeah, no kidding, look at my latest blog entry.

Like I wrote once on my blog, I knew I would learn Romanian eventually, just as I knew I would learn Latin someday. But it was almost prophetic, if you read the entry where someone complains about my not including it in the blog, to which I replied that, who knew, I might just begin learning it this year. And look!

Bună ziua! Acum vreau să învăţ româneşte!
Cine este din România???

Hehe...