(L&N) Q&A Thread - Quick questions go here

[GodzillaLouise]

Would Zareian (https://docs.google.com/document/d/1jgS ... sp=sharing), the language of the Zareians from my Ultraman/Godzilla Fanfiction world, be considered a conlang or a natlang? If it’s a natlang, what should I do to elevate it to a conlang?

[Arayaz]

Would Zareian (https://docs.google.com/document/d/1jgS ... sp=sharing), the language of the Zareians from my Ultraman/Godzilla Fanfiction world, be considered a conlang or a natlang? If it’s a natlang, what should I do to elevate it to a conlang?

Any artificial language (aside from things such as Modern Hebrew) is a conlang. There is no entry-level requirement.

[GodzillaLouise]

Would Zareian (https://docs.google.com/document/d/1jgS ... sp=sharing), the language of the Zareians from my Ultraman/Godzilla Fanfiction world, be considered a conlang or a natlang? If it’s a natlang, what should I do to elevate it to a conlang?

Any artificial language (aside from things such as Modern Hebrew) is a conlang. There is no entry-level requirement.

Yeah, I figured there was no entry level requirement. I was asking because I saw that a language is only counted as a conlang here when I don’t just copy English grammar and use different words, which is what I originally did for Zareian. I’m just trying to make sure that I changed Zareian enough from what it was then.

[Arayaz]

Would Zareian (https://docs.google.com/document/d/1jgS ... sp=sharing), the language of the Zareians from my Ultraman/Godzilla Fanfiction world, be considered a conlang or a natlang? If it’s a natlang, what should I do to elevate it to a conlang?

Any artificial language (aside from things such as Modern Hebrew) is a conlang. There is no entry-level requirement.

Yeah, I figured there was no entry level requirement. I was asking because I saw that a language is only counted as a conlang here when I don’t just copy English grammar and use different words, which is what I originally did for Zareian. I’m just trying to make sure that I changed Zareian enough from what it was then.

Ah, I see. That wouldn't be a natlang, but a relex.

You seem to have a few changes from English syntax and grammar, but it is pretty similar. I suggest that you formalize the rules for forming the past tense of a verb and the plural of a noun (unless it's entirely unpredictable and without patterns, which it doesn't seem to be).

And overall, I think you should investigate languages other than English and see how they work and sound. Don't just throw everything you find into Zareian, but instead broaden your horizons of how a language *could* work to beyond just English.

Edit: The board's 323,000th post!

[Ælfwine]

I cant seem to find an answer to this so im asking around, but does anyone know if the Ukrainian dialects in east Ukraine (north of Crimea) share with Russian word final devoicing? I'm aware normally Ukrainian does not, but I'm also aware that the dialect continuum means peripheral varieties sometimes have differences from whats standard.

[Davush]

I have had a question that has been at the back of my mind for a while, so any thoughts, comments, opinions are welcome on the below.

The questions is, basically, can morphemes (or rather, 'morphological process') appear ex nihilo?

Obviously, morphemes can mostly be seen to derive either from previous morphemes (having been inherited, and possibly modified along the way) or from words that have become grammaticalised as morphological affixes. But what about morphological processes that just seem to appear from nowhere, with the function being arbitrarily defined (i.e., not related to an eroded word, or existing morpheme)?

Reduplication is a kind of example of this, I suppose: it just seems to occur, and is used for a wide variety of purposes (plural marking, intensives, diminutives, augmentatives, TAM, etc.). Of course, reduplication is using 'existing material' from the word itself, but it's not like the process of word > affix, where the function of the affix is (usually) somewhat related to the word's original meaning. With reduplication, it just seems to be assigned to whatever the speakers start using it for, hence 'kind of' ex nihilo.

But do morphological processes ever just 'appear' without a phonological/diachronic motivation? Let's use things like vowel lengthening, mora insertion, gemination, and infixes as an example. If we have a language that has the rule 'insert a mora' for the past tense, but this cannot be seen to derive from an earlier morpheme, would we ever find an example where the rule 'insert mora' gained the function of 'past marking' seemingly from nowhere? Assuming there was no clear pathway, it would seem that speakers just 'invented' it.

A particular example may be certain templatic forms in Semitic. Let's take L-stems for example (usually indicating reciprocity or pluractionality of some sort), where the first vowel is lengthened CaCaC > CāCaC. Could this lengthening have just simply (and spontaneously) become associated with the above semantics, or must we postulate a morpheme from which it evolved (for example, an infixed *ʔ or something that caused lengthening)?

I suppose my question is can such things really be ex nihilo or must they actually be derived from something already existing in the language? Thank you.

[Creyeditor]

I think the question is a good one but I don't have a general answer.
Full reduplication can in principle derive from syntactic doubling and lengthening by a mora from a loss of a vocalic affix.
Something cool that I recently read about is grammatical tone coming from fossilized intonation.
Another source of morphological constructions - apart from combinations of syntax and phonology - is iconicity. Several languages lengthen a vowel to produce an intensive form.
Languages use reduplication to indicate plural or repetition.
If I recall correctly, David Gil found that several languages in SEA/Papuanesia express great spatial distance by lengthening, high pitch, and a certain hand gesture.
I guess that's the closest I can get to ex nihilo.

[Omzinesý]

I'm not sure of what you count as ex nihilo, but reanalysis is very common in the history of languages. Workaholic is the usual example. alcohol-ic (of course Arabic al-kuhul) -> alco-holic, -holic being a derivational suffix for 'addiction'.

If there was a more or less coherent group of verbs that usually have a past reading (kill, die ...) and they happened to have a long vowel, and a group of verbs that usually have a present reading (sleel, sing ...) with a short vowel, it could be possible for vowel length be associated with tense.

kaak 'die(d)'
tak 'sing(s)'
-> by analogy
kak 'dies'
taak 'sang'

Of course the groups should be phonemically different by accident.

[Khemehekis]

If there was a more or less coherent group of verbs that usually have a past reading (kill, die ...) and they happened to have a long vowel, and a group of verbs that usually have a present reading (sleel, sing ...) with a short vowel, it could be possible for vowel length be associated with tense.

Sleel?

Steal? Sleep?

[Arayaz]

If there was a more or less coherent group of verbs that usually have a past reading (kill, die ...) and they happened to have a long vowel, and a group of verbs that usually have a present reading (sleel, sing ...) with a short vowel, it could be possible for vowel length be associated with tense.

Sleel?

Steal? Sleep?

Sleep seems likely, since it has a present reading and L and P are close to each other.

[Nel Fie]

So, I just randomly stumbled across the Wikipedia page of Wandala, a Chadic language - and the article claims that under some analyses, Wandala does not have phonemic vowels (a slideshow linked as a source also mentions Central Chadic as an example, if I'm reading right).

Has anyone heard of this before, and any further information/thoughts about it?

[VaptuantaDoi]

So, I just randomly stumbled across the Wikipedia page of Wandala, a Chadic language - and the article claims that under some analyses, Wandala does not have phonemic vowels (a slideshow linked as a source also mentions Central Chadic as an example, if I'm reading right).

Has anyone heard of this before, and any further information/thoughts about it?

There's been a few languages analysed with zero vowels – most famously Kabardian, but also Mandarin and pre-Proto-Indo-European (and I think some Papuan language or other). Generally it comes down to either (1) analysing /i u/ as syllabic variants of /j w/ and then trying to claim that remaining /ə/ or /a/ is epenthetic, or (2) in vertical vowel system languages, treating all vowel variation as allophony produced by consonants and then try and explain the rest with epenthesis. Generally it's accepted that this is nothing more than a parlour trick; while it's certainly possible to analyse some languages as having no vowels, it's invariably simpler to posit at least one or two vowel phonemes. It seems the case with Chadic languages is that [​i u a] are treated as syllabic variants of /j w ʔ/ and [ə] is explained through epenthesis – i.e. a variant of (1) – but I'd only accept that analysis if there's very convincing reasons for not just positing /i u ə a/.

There are most certainly some languages with only two vowel phonemes, mostly VVS languages. You can analyse PIE as having only two vowels */e o/ (which some say are from earlier */a aː/), but there's a few cases of */a/ which can't really be explained away, and it's a protolang so it doesn't really count. Mandarin can be pretty neatly analysed with only two vowels too. There's a fairly good analysis of Moloko with one vowel, although that requires some deft prosodic wrangling, but on the plus side it doesn't need any bludgey syllabic consonants.

[eldin raigmore]

What is a VVS language? Can’t find it via Google nor Wikipedia!

[Creyeditor]

maybe vertical vowel system?

[VaptuantaDoi]

What is a VVS language? Can’t find it via Google nor Wikipedia!

As Crey said, Vertical Vowel System.

[Nel Fie]

There's been a few languages analysed with zero vowels – most famously Kabardian, but also Mandarin and pre-Proto-Indo-European (and I think some Papuan language or other). Generally it comes down to either (1) analysing /i u/ as syllabic variants of /j w/ and then trying to claim that remaining /ə/ or /a/ is epenthetic, or (2) in vertical vowel system languages, treating all vowel variation as allophony produced by consonants and then try and explain the rest with epenthesis. Generally it's accepted that this is nothing more than a parlour trick; while it's certainly possible to analyse some languages as having no vowels, it's invariably simpler to posit at least one or two vowel phonemes. It seems the case with Chadic languages is that [​i u a] are treated as syllabic variants of /j w ʔ/ and [ə] is explained through epenthesis – i.e. a variant of (1) – but I'd only accept that analysis if there's very convincing reasons for not just positing /i u ə a/.

There are most certainly some languages with only two vowel phonemes, mostly VVS languages. You can analyse PIE as having only two vowels */e o/ (which some say are from earlier */a aː/), but there's a few cases of */a/ which can't really be explained away, and it's a protolang so it doesn't really count. Mandarin can be pretty neatly analysed with only two vowels too. There's a fairly good analysis of Moloko with one vowel, although that requires some deft prosodic wrangling, but on the plus side it doesn't need any bludgey syllabic consonants.

Right - thank you very much for the extensive explanation! I'll have to give that Moloko grammar a read-through.

Out of curiosity, and in theory, what would you consider "very convincing reasons for not just positing /i u ə a/"? I mean in general, not necessarily reasons specific to Chadic languages. Your previous sentence implies there aren't really any - would you make any exceptions?

[VaptuantaDoi]

Out of curiosity, and in theory, what would you consider "very convincing reasons for not just positing /i u ə a/"? I mean in general, not necessarily reasons specific to Chadic languages. Your previous sentence implies there aren't really any - would you make any exceptions?

Well, to convince me you'd have to show completely predictable syllabification rules (e.g. no contrastive [ja] vs. [iʔ]) without resorting to non-standard features, as well as consistent (probably morphological) alternations between the two sets. Ideally native speakers would identify them as the same phoneme, but that's sometimes hard to evaluate. I don't know if these are the case in any Chadic languages; I can't find any papers which thoroughly describe a zero-vowel system so I'll reserve judgement, but I remain sceptical.

[eldin raigmore]

maybe vertical vowel system?

Thanks! I take it that’s not widely adopted yet?

[Nel Fie]

Well, to convince me you'd have to show completely predictable syllabification rules (e.g. no contrastive [ja] vs. [iʔ]) without resorting to non-standard features, as well as consistent (probably morphological) alternations between the two sets. Ideally native speakers would identify them as the same phoneme, but that's sometimes hard to evaluate. I don't know if these are the case in any Chadic languages; I can't find any papers which thoroughly describe a zero-vowel system so I'll reserve judgement, but I remain sceptical.

Good to know, thank you! That matches more or less what I expected.

[Salmoneus]

I'd also suggest that widespread allomorphic alternations would help build a good case for two allophones being the same phoneme.

For instance, if [ i ] only occurs when not adjacent to a vowel, and there are widespread alternations between forms with [ i ] adjacent to consonants or word boundaries and forms with [j] when adjacent to a vowel (eg past [takita] and [satata] alternate with future [takjepa] and [takaepa]), that would encourage me to see [ i ] and [j] as underlyingly the same phoneme.

This is the main reason why many people don't see PIE as having /i/. Because in a huge number of cases [CiC] sequences alternate with [CejC] and [CojC] sequences, in exactly the same way that [CrC] and [CnC] sequences alternate with [CerC]/[CorC] and [CenC]/[ConC] sequences, and in the same way that [CC] sequences often alternate with [CeC] and [CoC] sequences. It just seems to make sense to see the alternation as a single process of the addition of /e/ or /o/ (or their loss) into the syllable with all the other phonemes remaining the same (which requires [CiC] to represent [CjC]), rather than saying that that's the alternation going on in ALL syllables OTHER than those with /i/ (or /u/, vis a vis /w/), which instead show random replacement of /i/ with the diphthong /ej/. The /j/ analysis is further reinforced by the way that /j/ appears to pattern with consonants (or, at least, liquids and nasals) in terms of syllable structure (there are, afair, no roots of the form CeirC - the i and the r have to compete for the same slot).

[iirc there may also be some alternations of the form /CweC/ - /CuC/, reinforcing the 'vowel dropping' analysis, because you get the same outcome whichever side of the vowel the glide starts out.]

[it's also debatable whether syllabic /j/ and /w/ lengthened before a laryngeal as the other vowels did; reflexes of *wiHros (man) outside of Balto-Slavic, for instance, show a short vowel]

The key I think is to remember that phonemic analysis isn't a search for some one underlying truth, but simply an attempt to simplify descriptions and notation for ease of use. And in the case of PIE the way that the majority of cases of [ i ] act like /ej/ sequences that have been subjected to vowel deletion makes it much more convenient to treat it as an allophone of /j/.

--------------

The other thing I'd stress is the question of whether an analysis is merely neat, or actually simplifies things.

For instance, any language can be neatly analysed as having no vowels, if sufficient facts are instead held to be features of the consonantal phonemes. But usually this is pointless. Instead of having, say, five vowels, you've just multiplied the number of consonants by five (or you've introduced a bunch of complicated rules that make surface vowels dependent upon combinations of underlyingly adjacent consonants). That's a neat trick, but doesn't actually make your analysis simpler.

Whereas if you can analyse /j/ and /i/ as being the same phoneme, you've reduced the number of phonemes you need to work with. Provided that the rules for determining the surface allophone are reasonably simple (like 'only a full vowel when not adjacent to another vowel, otherwise a glide') it wouldn't take much to make such an analysis look simpler than the alternative.