GV&SP Discussion: That durned î, revisited

[Sir C. M. Siervicül]

My quoted comments and King John's answers are from the Word docs posted in the main GV&SP thread. References to page and paragraph numbers in my quoted comments refer to the original GV&SP doc, also linked from the main GV&SP thread.

---

REGARDING THE Î VOWEL

My original comments:

Number 7 deals with î. I have a big problem with this section, because it's not really about vowel accents and stress, or about orthography. It's about phonology and etymology, and the proposed changes would, in my opinion, remove a very interesting feature of the language (and not just of how the language is written).

Talossan î is inspired by Romanian î/â – not just in the sound or the orthographic representation, but also how it behaves. Just like in Talossan, the Romanian sound has multiple representations (currently î or â, but in the past ê and û were used as well). Just like in Talossan, that one sound represents multiple sounds in the original Latin words (as the multiple ways of spelling it suggest). Just like in Talossan, it usually (but not always) appears before a nasal consonant. And just like in Talossan, there are plenty of words that look like they ought to have that sound, but don't.

The explanation for that last point is that languages don't have a fixed lexicon from the beginning of their development to the present. Words related to the same Latin root can come into the language at different periods, some missing out on sound changes that others went through. That's why Romanian has both mânie (anger) and manie (mania). Now there is still room for regularisation, and there's still the issue that the sound is difficult for English speakers to pronounce, but I really think that getting rid of î altogether and completely undoing this fascinating Talossan sound change would be throwing the baby out with the bathwater. This proposal goes well beyond the points of Proposal 1 (the Overarching Proposal).

King John's answers:

I wrote a few paragraphs one time arguing that Talossan î isn't a letter (in the alphabetic sense) at all, but rather a leftover grapheme from the days when the language was written by hand, when it meant something like vowel-followed-by-nasal. But whatever. I really think it needs to go. Nobody (Ben included) knew how to pronounce it, it represents not just one or two or three, but SIX Latin source vowels, and (to my mind) it's really ugly.

I remember those paragraphs, but they were based on the premise that the distribution of î is unexplainable without assuming (contrary to all evidence) that î is actually pronounced multiple different ways depending on the Latin vowel it replaces. As we can see from Romanian, that premise is not well founded. For example, your original e-mail message regarding î said the following:

I worked for a while on the idea that Talossan must have a phonological tendency (now or at some earlier date) to take vowels occurring before consonant clusters, especially before nasal consonant clusters, and move them to some single intermediate value regardless of what the original vowel was, and that this intermediate vowel is written (but only sometimes written) ī. This hypothesis, though, is defeated by the fact that no such vowel-shifting tendency is remarked anywhere in the linguistic literature as affecting or having affected the pronunciation of the hundreds of common Talossan words that contain "ant", "ent", " ind", and so on. (And the fact of the existence of a small group – quand, trans, etc. – that are spelled with a but pronounced with ī, is good evidence that other words spelled with a are NOT pronounced with ī.) Nor does the hypothesis explain the huge group of words – all the present participles – that are *spelled* with ī but pronounced with o or a, as if their pronunciation had (incredibly) shifted *back again*, beginning for instance with Latin a (amantis =loving), shifting to the hypothetical ī (amīnd), and then back to a modern pronunciation of -ont.

Your original hypothesis was on exactly the right track, and the observations that you said defeated it are in fact easily explainable. The î sound is the result of a merger of Latin vowels in certain environments. It is based on Romanian î, which works in a very similar way. Here are the observations you made (as I understand them), and the explanations for them:

ISSUE: There are many words (with ant, ent, ind, etc.) that did not undergo the supposed î merger.

RESOLUTION: Romanian, which had a similar merger, also has many words with an, etc. which ought to have changed to în – if they had existed in the language prior to the merger. But, of course, they came into the language later, and thus missed the sound changes that effected the merger. For example, Latin gave Romanian angelus, which became înger (angel), but Romanian also has angelic (angelic). Latin mania became mânie (anger – â sounds the same as î in Romanian), but Romanian also has manie (mania). Latin virtus became vîrtos (strong), but Romanian also has virtuos (virtuous). Of course, Talossan probably has somewhat more words that missed the hypothetical merger (more late borrowings or reborrowings) than it ought to, and probably some of the words that ought to be "older" have a, e, i, etc. while some of the words that ought to be "newer" have î. And we should address that. But that's no different from other apparent Talossan sound changes that have clearly been imperfectly executed, like initial d to initial z and initial h to initial þ. There are still hundreds of common words that start with d and h, and they're probably not all the right ones. And the ones that did change to z and þ probably aren't all the right ones. But those are issues of diachronic linguistics, not of orthography, and we should not get into fixing the phonetic/phonological history of Talossan in this Arestadâ. Let's please focus on fixing Talossan orthography right now, like we decided to do by approving the Overarching Proposal.

ISSUE: There is a group of words that are spelled with a and pronounced with î.

RESOLUTION: This is a simple matter of some words having irregular pronunciation. Obviously, there are words in Talossan with spellings that do not match their pronunciation. This is normal in languages when pronunciation changes but spelling has not yet caught up completely. Romanian used to have four different ways to represent the î sound (listed in my original comment above), for etymological reasons. During the Communist era, they officially went down to just î for a while, but a few words (such as "Romania" itself) continued to be commonly spelled with the â version because î "felt" wrong. Eventually they officially approved the use of both â and î, with (if I recall correctly) â to be used internally and î to be used at the ends of words. Quand and trans in Talossan are just like people continuing to spell certain words with "a" in Romanian when the barred-i sound was officially supposed to be spelled with î.

ISSUE: Present participles are spelled –înd but pronounced –ant.

RESOLUTION: The record is clear about what happened here: a bad compromise between two competing forms of the present participle. The CÚG in 1992 ought to have decided on one form or the other, or combined the –ant spelling with the –înd pronunciation (which would have been explainable from a diachronic perspective). Instead, they chose the least logical combination. No hypothesis can survive a decision that bad.

I think that at some point in the not-too-distant future we need to take a serious look at Talossan as a Romance language in a genetic/diachronic sense. We need to try to inventory the sound changes from Vulgar Latin that can be extrapolated from existing words, and work on ironing out the inconsistencies some, with the help of Swadesh lists etc. It'll be a tall order, but Talossan will become much more respectable. Some of the d-initial words will probably become z-initial words, some of the h-initial words will probably become þ-initial words, and vice versa. At that point we might come to the conclusion that the î merger in Talossan just doesn't make sense in the context of Talossan's own grand vision of itself (what a posteriori conlangers like to call a "Grand Master Plan"). But I strongly believe that this Arestadâ is not the right time or place to make such a decision.

The right course of action to take in this Arestadâ is to apply the principles of the Overarching Proposal to î. Which means that in order to distinguish the sound of this vowel from a regular i, we should follow the now-standard diacritic rules and use an umlaut instead of circumflex. So î should become ï ordinarily, and î only when irregularly stressed. Then (in combination with the other changes included in the GV&SP) usage of all diacritics will be completely regular.

[Xhorxh Asmour]

This last option sounds good enough to me

[King John]

Sir Cresti wants to keep î as Yet Another Vowel in Talossan (and spell it ï unstressmarked and î stressmarked). I think keeping it is a bad idea, for several reasons:

1. Talossan already has eight phonemically distinct vowels and a great many polyphthongs — a huge inventory for a Romance language. We don't need a ninth.

2. Even if we want it, we don't know how to pronounce it. One source document says (of the sound of î):

Doesn't occur in English: like the "oo" in "book", "foot" but pronounced with the lips spread wide instead of rounded.

The Introduction to the Treisour says:

This common letter has no equivalent in Standard English. It lies between the vowel sounds of reed and rude, and is pronounced with spread lips. It may be practised by putting the lips in position to say /i/ (as in reed) and, while keeping them spread, raise the centre — not the front or back — of the tongue as high as you can and try to pronounce /i/. The resulting sound should be similar to Talossan î as in the word rîd (“a laugh”). The sound is also found in a handful of words written with â .

I'm not sure these two descriptions agree with each other, but I am sure that neither conveys to me any clear idea of what sound I ought to be making when I see a Talossan î.

And we have Sir Tomás's recollection that even Ben was inconsistent about pronouncing î, often making it a simple i.

3. It seems incredible to me that all six Latin vowels would in a great many cases transmute into the same vowel in Talossan, but in a great many other cases would remain unchanged or follow more usual changes (like o -> u). We've argued about this, and I'm not a bit convinced.

Why would Latin fons become Talossan funtagna, Latin mons become Talossan munt or montagna, but Latin pons and punctum both become Talossan pînt? This is incoherent, and we can do a lot right now to fix it.

— John R

[Xhorxh Asmour]

OK, you win. So let's eliminate both î and â !

[Sir C. M. Siervicül]

john said:

1. Talossan already has eight phonemically distinct vowels and a great many polyphthongs — a huge inventory for a Romance language. We don't need a ninth.

That's on the high side for Romance as a whole, but right in line with European Portuguese (nine basic vowel phonemes) and French (somewhere from nine to twelve, depending on dialect). And Portuguese and French also have phonemic nasalisation.

2. Even if we want it, we don't know how to pronounce it.

Right, they shoulda used IPA. It's hard to describe because it's an unfamiliar sound for English speakers. I'm pretty confident that Ben was trying to describe [ɨ], or "barred-i," though.

And we have Sir Tomás's recollection that even Ben was inconsistent about pronouncing î, often making it a simple i.

Sir Tomás recalled Ben having trouble pronouncing other things as they are described, too. Let's not make Ben's own pronunciation determinative.

3. It seems incredible to me that all six Latin vowels would in a great many cases transmute into the same vowel in Talossan, but in a great many other cases would remain unchanged or follow more usual changes (like o -> u). We've argued about this, and I'm not a bit convinced.

Why would Latin fons become Talossan funtagna, Latin mons become Talossan munt or montagna, but Latin pons and punctum both become Talossan pînt? This is incoherent, and we can do a lot right now to fix it.

A fairly straightforward explanation would be that pînt is the oldest Talossan word, while the others were borrowed later from other Romance languages, or are "learned" later borrowings from Latin. Note that munt and montagna are inconsistent, too. Any explanation of Talossan (even without î) will have to rely heavily on the assumption that Talossan has done a lot of borrowing from other languages over the centuries.

If that's "incoherent," what do you make of the fact that French has both avoué and avocat? Both words for types of lawyers, both descended from Latin advocatum. Avocat exhibits sound changes like the simplification of Latin dv to just v, but avoué went through more changes with the disappearance of the intervocalic c and vowel changes. Or what about the Romanian examples I cited, which are also similar to your Talossan examples? English has tons of sets of words like that, too. I'm sure you can find words that were borrowed from Latin or Old French during the Old English period, borrowed again from Norman French during the Middle English period, and borrowed a third time from Latin or French into early Modern English, all three surviving to the present day but with obviously different sound changes. Poor/pauper/paucity? Ben didn't design French, Romanian, and English.

The linguistic history of Talossan is a bit of a mess, granted. I agree that it needs some work. Maybe when we get it sorted out, the best thing will be to eliminate î. But I think we should be careful and do it right, which means that î really needs to be considered as part of an overall assessment of the sound changes that led to modern Talossan, not as part of an orthographic regularisation. We aren't ready to do that right now. I don't want to hold up necessary orthographic reforms or jump the gun on evaluating the history of Talossan, which is why I urge the separation of those issues.

Honestly, sometimes it seems like Talossan already exists in multiple dialects, and the grammar and dictionary just lump them all together. "Mountain," for example, is muntagna in Campidanese Sardinian and monte in Logudorese Sardinian -- almost the reverse of the Talossan words munt and montagna. That would also explain the alternative forms of things like the imperative and past participles. It's tempting to try to tease out those differences.