Triconsonantal glossing conventions

[dewrad]

So, I have a language with Semitic-style nonconcatenative morphology. As far as I can tell, there isn't a standard way to apply the Leipzig glossing conventions to pervasively nonconcatenative languages like the Afroasiatic languages. This is the system I have come up with for my language, and I'm wondering if it is clear enough.

Taking an example sentence:

ʔən həyʕəzzənyə́t mətʼə́t
And I have not just given you the slave-girls.

mətʼə́t 'the slave-girls' is a definite feminine plural accusative noun. Initially, the word ʔən, here glossed 'and', is a semantically empty particle which is required here simply because Proto-Oxaric does not permit a verb in initial position in main clauses.

The actual verb here is həyʕəzzənyə́t: it's the negative polarity, perfect aspect, 1sg>3fp>2 (i.e. 1st person singular subject, 3rd person feminine plural direct object, second person indirect object) form of the verb yəzún 'to give'. The conjugated form here is segmentable, but not easily so. Furthermore, due to the conjugation being effected by infixes, vowel interchange and reduplication of a stem-internal consonant, it doesn't lend itself easily to neat left-right Leipzig-style glossing.

Let us then break this down into morphemes, from the inside out. First, the stem of the verb is y-z-n, picked out in red:

həyʕəzzənyə́t

The perfect aspect is formed by reduplicating the second consonant of the root (in orange):

həyʕəzzənyə́t

Negative polarity is expressed by the infix ʕ (blue):

həyʕəzzənyə́t

The 1st person singular subject is marked by the prefix h (green):

həyʕəzzənyə́t

The 2nd person indirect object is marked by the infix y (purple):

həyʕəzzənyə́t

And finally, the 3rd person feminine plural direct object is marked by the suffix t, and the quality of the vowels used throughout the word (brown):

həyʕəzzənyə́t

I deeply apologise to anyone reading this who is affected by colour blindness. And anyone affected by good taste. Obviously, colour is not going to be a suitable way of isolating and identifying morphemes.

The Leipzig conventions dictate that prefixes and suffixes are marked off from the stem with hyphens, which is easy enough:

h-əyʕəzzənyə́-t

Infixes are enclosed in angle brackets (adding a convention here, where there are multiple infixes, they can be kept distinct with subscript numbers):

h-əy<ʕ>₁əzzən<y>₂ə́-t

Reduplication is properly expressed by separating the reduplicated segment with a swung dash ~, but I'm not sure how well this works when the reduplication is word-internal. Two options suggest themselves: either treat the dash as a bracket, or treat the reduplicated element as an infix:

h-əy<ʕ>₁ə~z~zən<y>₂ə́-t or h-əy<ʕ>₂ə<z>₁zən<y>₃ə́-t

The latter looks more intuitive, I think?

Considering the vowels, I guess they actually mark the number of the direct object: 'I haven't just given you the slave-girl' would be ʔən hayʕazzanyá mətʼí. Rule 4D of the Leipzig glossing conventions states "If a grammatical property in the object-language is signaled by a morphophonological change (ablaut, mutation, tone alternation, etc.), the backslash is used to separate the category label and the rest of the gloss."

So, putting it all together:

h-əy<ʕ>₂ə<z>₁zən<y>₃ə́-t
1SG.NOM-give\PL.ACC<PF>₁<NEG>₂<2.DAT>₃-F.ACC
I have given them to you

Does that make any sense at all, or is it just too over-complex?

[zompist]

I think you've just made an excellent case for color-coding. Let me see if I've got this right...

həyʕəzzənyə́t

1SG.NOM-give\PL.ACC<PF>₁<NEG>₂<2.DAT>₃-F.ACC

I'm not a Semiticist, but I think expecting your readers to be Semiticists would be a stretch. This is fascinating morphology, and the raw gloss is nearly impenetrable; I think you should make it as easy as you can. Color certainly would help. Font variations would help for the color-blind, e.g. placing the root in boldface. If you (say) reserved italics for infixes, you could lose the brackets. (And if you gloss them in order, you should be able to omit the subscripts.)

You might also try the glossing convention Randolph Valentine uses for Nishnaabemwin. I think he'd render your sentence thus:

ʔən conj 'and' həyʕəzzənyə́t v perf neg 1s>3pf>2 '1s have not given 3pf to 2' mətʼə́t n def fp.acc 'the slave girls'

This doesn't identify specific bits, but it also doesn't require deep knowledge of the glossing system!

If I were writing a grammar of this language, I'd probably have a meaty section on the verb itself with full glosses, and then something simpler for the syntax system. So, if you were using this sentences to talk about how verbs work, you give everything, but if you're just talking about negation, you just highlight the negative infix.

[cedh]

So, putting it all together:

h-əy<ʕ>₂ə<z>₁zən<y>₃ə́-t
1SG.NOM-give\PL.ACC<PF>₁<NEG>₂<2.DAT>₃-F.ACC
I have given them to you

Does that make any sense at all, or is it just too over-complex?

Yes, it does make sense, and the complexity is probably necessary.

A few ideas:
- I would probably gloss the infixes in the same order as they appear in the word.
- You could treat gemination of a root consonant as a morphophonological change as well, giving the following:

h-əy<ʕ>₁əzzən<y>₂ə́-t
1SG.NOM-give\PF\PL.ACC<NEG>₁<2.DAT>₂-F.ACC
I have given them to you

- Or you could treat the C₁VC₂C₂VC₃ variant of the root as a fusional allomorph with [+PF] value, giving the following:

h-əy<ʕ>₁əzzən<y>₂ə́-t
1SG.NOM-give.PF\PL.ACC<NEG>₁<2.DAT>₂-F.ACC
I have given them to you

[Richard W]

The 2nd person indirect object is marked by the infix y (purple):

həyʕəzzənyə́t

Why do you call this an infix? I would expect an infix to be inserted within another morpheme, and this doesn't seem to qualify any more than the stem //yzn//, which you could claim is enclosed 3rd person feminine plural direct object marker (an ubiquifix?).

[Moose-tache]

Instead of color, you could use a second dimension of space to break a word apart:

[Vardelm]

Instead of color, you could use a second dimension of space to break a word apart

I think the chart could use a bit more work, but this is borderline brilliant IMO.

A few ideas:
- I would probably gloss the infixes in the same order as they appear in the word.
- You could treat gemination of a root consonant as a morphophonological change as well, giving the following:

I agree with this. I don't think the "infixes" need to be separated by <>. Either way, if the order between the 1st & 2nd lines is the same and the punctuation used to separate the elements is the same between them, it makes it much more readable, especially for someone less versed in linguistics. You can then see the 1-1 correspondence, or at least closer.

An idea to toss out there: you could maybe think about indicating the root consonsants somehow. That could be with <>, (), or just capital letters.

If I were writing a grammar of this language, I'd probably have a meaty section on the verb itself with full glosses, and then something simpler for the syntax system. So, if you were using this sentences to talk about how verbs work, you give everything, but if you're just talking about negation, you just highlight the negative infix.

I definitely think this is important too (added emphasis is mine). I think there's value in showing 2 examples that are identical except for the 1 morphological/syntactic feature being illustrated. I sometimes find grammars confusing & frustrating when there are several different examples given that show multiple variations between them, making it hard to pinpoint exactly how the 1 feature supposedly being discussed is actually formed & works.

P.S. - As Zompist said, cool morphology. #thumbsup

[Kuchigakatai]

Triconsonantal glossing conventions

In terms of conventions, actual works on Semitic languages normally don't bother with morphological breakdowns in glosses. People just gloss whole words without worrying about the identification of morphemes (if they gloss them at all...). If you're talking about the appearance of a specific morpheme, then you can highlight it with bold weight if you want.

ʔən həyʕəzzənyə́t mətʼə́t
and 1S-not-give.PRF-3FPl>2S slave-girls.DEF.ACC

So you're getting into fairly uncharted territory with morphological breakdowns of such languages. I don't remember how Yng used to do his Standard Arabic glosses exactly, but I recall he'd separately give the triconsonantal root in the gloss and the conventional number of the valency derivational affix. So he'd write something like,

yatakaatabna
ya-takaatab-na
3-√ktb.6(RECIP.APPL)~write.ACT.PRES-3.FEM.PL
'They (the women) write to each other.'

From "√ktb" you were supposed to deduce that taCaaCaC is the transfix that stands for "reciprocal applicative (for indirect object), active-voice, present-tense" in this word. Here I haven't made it clear that the ta- of taCaaCaC is the reciprocal part though, as CaaCaC by itself is "indirect-object-applicative, active-voice, present-tense"; maybe he would've clarified that too.

I'm not sure what he would've done with idiomatic valency derivations, like pseudo-passive* inkataba 'to be subscribed, registered [for some service, etc.]'. Maybe he'd just gloss it as "subscribe": inkatab-na √ktb.7~subscribe.ACT.PST-3.FEM.PL 'they (the women) were/appeared registered'.

* By "pseudo-" I mean that it often expresses notions in and around the passive, like the state of the result an action, or a labile sort of meaning (where an inanimate "does" an action like 'it opens, it breaks'), although sometimes it's basically a real passive.

[Nortaneous]

Not sure if this is good but my first pass would be something like:

h-√y<ʕ>z⪝z⪞n<y>-ᵊt
1SG.S-give<NEG>⪝PRF⪞<2SG.I>-3PL.F.O

Angle bracket + tilde for incopyfixation, superscript for "the quality of the vowels used throughout the word" by analogy to the Chadic prosodies.

[dewrad]

Thank you guys, there's plenty for me to think about here. I particularly like Moose-tache's idea, but I think it would be a bit of a pain to implement in html.

If I were writing a grammar of this language, I'd probably have a meaty section on the verb itself with full glosses, and then something simpler for the syntax system. So, if you were using this sentences to talk about how verbs work, you give everything, but if you're just talking about negation, you just highlight the negative infix.

I definitely think this is important too (added emphasis is mine). I think there's value in showing 2 examples that are identical except for the 1 morphological/syntactic feature being illustrated. I sometimes find grammars confusing & frustrating when there are several different examples given that show multiple variations between them, making it hard to pinpoint exactly how the 1 feature supposedly being discussed is actually formed & works.

On which note then, let's talk about the infix marking the 2nd person indirect object that Richard W was talking about.

The grammatical category marked here isn't necessarily the indirect object. In fact, the more I think about it (and kind of considering how I've used the morphemes' cognates in other languages of this family), this is something more along the lines of verbal deixis. We have three options in this category: x~š~Ø marks the first person, y the second person, and z~t the third person. With with ditransitive verbs, they mark the recipient, as seen in the example in my first post.

However, they can also occur with intransitive, reflexive or ditransitive verbs, where they index an adjunct. For example:

θʼi yəkʼə́ʕ šəktə́za
with razor:POSS shave:REFL:1SG3IO
"I shave myself with a razor."

Here, the third person index -z- essentially cross-references with the prepositional phrase θʼi yəkʼə́ʕ "with a razor".

With other verbs, the indexes have a andative/venitive meaning. Take, for example, the verb ɣətúw "to carry":

taxnát ɣətwíš mɨtítʼ
brother:NOM.DEF carry:3S>3PL-1IO slave.girl:ACC.DEF
"The brother brings the slave-girls" (lit "the brother carries the slave-girls to me")

vs

taxnát ɣətwít mɨtítʼ
brother:NOM.DEF carry:3S>3PL-3IO slave.girl:ACC.DEF
"The brother takes the slave-girls away" (lit "the brother carries the slave-girls to him/her/it")

(Something like that, you mean?)

[Vardelm]

(Something like that, you mean?)

Yep! That makes it pretty clear to me what those morphemes are for. If there was also an example with the same verb for just an intransitive (middle? reflexive?) that didn't have that infix (suffix?) at all, I think it would be complete.

I just realized that I like grammars that present the material in the way that I was taught to write training materials (for software, in my case). You start with the simplest example (probably an intransitive) and build up from there, 1 piece at a time so that the student can easily see how the new element interacts with everything that has been learned before. If you're an expert in something (like linguistics) it's really easy to overwhelm a newbie with information, so building up in clear, logical, & digestible chunks can help a ton.

[Bob]

...

This thread is a lot of fun.

Recently I see glosses like dog:NOMINATIVE and it must be this sort of thing to which it's referring. Otherwise, I wouldn't gloss something like that unless I had to. What a mess.