Pñæk grammar (so far)

[chris_notts]

Firstly: How dare you peer into my mind and steal all my best ideas! I’m working on an SVO isolating conlang as well, and Pñæk uses all the best ideas I was looking forward to using!! Now I’ll have to find new ideas so my conlang doesn’t become a Pñæk ripoff!!! Do you have any idea how much work that will be?!!!! You should feel ashamed of yourself for doing such an awful thing to me!!!!!

(Terry Pratchett once said that ‘five exclamation marks are the sure sign of an insane mind’. On reflection, I think he was right.)

Sorry!!!!! I didn't mean to upset you!!!!!

Now on a more serious note: In case I have to say this explicitly, I don’t seriously think you read my mind, nor do I think you should be ashamed of yourself. Rather, I think you should be proud, for making what is one of the most detailed conlangs I’ve ever seen. Parts of it are even more detailed than many natlang grammars!

Thanks. I still have more on my to-do list though.

In terms of the content itself, I don’t have too many comments or criticisms. Here’s all the ones I can think of, which are mostly places where you’ve forgotten to include something:

It would be nice to know if Pñæk is set in a conworld, or if it’s purely a personal artlang.

It was intended to be part of a conworld, but I got a bit distracted (for the best part of a year) on the language itself. I had intended the speakers to be small island peoples, a bit like the Polynesians, but early research proved quite hard. I did a lot of googling on the obvious examples of such peoples, and finding detailed descriptions of their pre-contact cultures, was difficult. It may simply be that much of it wasn't documented that well, or I'm not googling the right terms. Of course, I could just make things up, but I'd really like some real world examples of similar cultures.

Similarly, does Pñæk have any diachronics behind it?

Not really. In the past I've done this, and I even wrote a whole featural sound change applier because I was unhappy with the other options available at the time. But to be honest I just couldn't be bothered this time round. On the other hand, I might well use Pñæk as the ancestor for another conlang at some point, and hopefully having something really well fleshed out will be an advantage there.

The Pñæk linker takes the form ⟨-’r⟩ after a vowel. However, this seems to disobey the phonotactics, which states that a word may not end with a consonant cluster. It would be nice to resolve this apparent contradiction.

Yes, I've been pondering this. Originally the constraints on multiple rhotics in a word weren't so strong, but they evolved for other reasons and the earlier linker 'r was retained. I don't think this is impossible, since 'r is more like a clitic than an affix and it's not uncommon for clitics to violate rules that apply within roots and stems. But I have to admit I also don't like it, I just haven't decided which consonant to replace the 'r with.

On pages 22–23, you give the meaning of the agent nominalisation (‘derives words for those who typically perform the action described by the verb’) and the process/result nominalisation (‘marks nominalisations which describe the process or result of a verb’), but you don’t include the meaning of the instrument/locative nominalisation. I assume this is probably just an accidental omission.

The organisation of section 3.1 (on the nominalisation infixes) seems decidedly odd to me. Normally, if I were writing a grammar, I would first describe the basic meaning of an affix, then its exact semantics, and finally any phonological issues. However, you do nearly the opposite: first you define their semantics (subsection 3.1.1), then their phonology (3.1.2–3.1.3), and only at the very end do you say what their meaning is (3.1.4–3.1.7). I think you should alter the order of these subsections to give a more logical flow.

Good points.

You give a set of nominalisers in section 4.8, which is part of chapter 4 (The Noun Phrase). However, these nominalisers are described as having several functions, only some of which relate to the noun phrase. Due to this, I think you should move section 4.8 into chapter 2 (Word Classes).

Hhhm, I think they should be mentioned there, perhaps, but I'm not sure if that's the right place to describe their function in detail. I viewed Word Classes more as an overview than a detailed description of each class. I'll think about where the detailed description belongs.

Section 5.2 (Basic Argument Structures) seems to be purely a typological overview. You could probably remove it with minimal impact to the rest of the grammar.

I think it mostly just introduces some shorthands used in the following section. It can probably be simplified and merged into 5.3.

Two of the moods of Pñæk are ones you call ‘indicative’ and ‘subjunctive’. However, to me these seem to be closer to realis and irrealis (although admittedly I’m hardly an expert on this topic).

The subjunctive clearly is an irrealis category. The main reason I didn't call it irrealis is that the "potential", as I've called it, also has a mixture or realis irrealis uses. Similarly, for the realis - the issue I had was the presence of an mood sitting somewhere between the most realis (indicative) and most irrealis (subjunctive). Even in languages with exactly two main non-imperative mood categories, indicative/subjunctive can be used, e.g. in Latin. I'm not wedded to the current labels, the main issue is a set of names which feels right and also includes the potential.

A positive criticism: I really like chapter 8, on SVCs. (Although I may be biased — right now I’m in the middle of Dixon and Aikhenvald’s Serial Verb Constructions: A Cross-Linguistic Typology, which you’ve based your description on.)

Thanks!

I feel that you should reorganise this grammar into a more logical order more suited to a reference grammar. Here’s what I suggest:

1. Phonology (your chapter 1)
2. Word classes (your chapter 2 + 5 + section 4.8)
3. The Noun Phrase (your chapter 4)
4. Basic Clause Structure and Verbal Marking (your chapter 6 + 5.7 + 10)
5. Special Predicates (your chapter 7)
6. Serial Verb Constructions (your chapter 8)
7. Spatial and Temporal Expressions (your chapter 13 + 14)
8. Clause Linking (your chapter 11 + 12)
9. Derivational Processes (your chapter 3)
10. Discourse Pragmatics and Reference (your chapter 9)

To me, this seems like a much more logical flow compared to the current grammar.

I agree that the order might be improved. I'm not sure about some of the merger suggestions, but I think many of your combinations can be at least placed next to each other. Maybe:

PHONOLOGY
1
WORD CLASSES
2+5
CLAUSE INTERNAL STRUCTURE
4
6
7
8
13
14
CLAUSE LINKING
9
11
12
DERIVATION
3

[bradrn]

Firstly: How dare you peer into my mind and steal all my best ideas! I’m working on an SVO isolating conlang as well, and Pñæk uses all the best ideas I was looking forward to using!! Now I’ll have to find new ideas so my conlang doesn’t become a Pñæk ripoff!!! Do you have any idea how much work that will be?!!!! You should feel ashamed of yourself for doing such an awful thing to me!!!!!

(Terry Pratchett once said that ‘five exclamation marks are the sure sign of an insane mind’. On reflection, I think he was right.)

Sorry!!!!! I didn't mean to upset you!!!!!

It’s OK. I forgive you.

Now on a more serious note: In case I have to say this explicitly, I don’t seriously think you read my mind, nor do I think you should be ashamed of yourself. Rather, I think you should be proud, for making what is one of the most detailed conlangs I’ve ever seen. Parts of it are even more detailed than many natlang grammars!

Thanks. I still have more on my to-do list though.

Let me see if I can understand this. You already have 108 pages of grammar, and you’re still not done yet‽

The Pñæk linker takes the form ⟨-’r⟩ after a vowel. However, this seems to disobey the phonotactics, which states that a word may not end with a consonant cluster. It would be nice to resolve this apparent contradiction.

Yes, I've been pondering this. Originally the constraints on multiple rhotics in a word weren't so strong, but they evolved for other reasons and the earlier linker 'r was retained. I don't think this is impossible, since 'r is more like a clitic than an affix and it's not uncommon for clitics to violate rules that apply within roots and stems. But I have to admit I also don't like it, I just haven't decided which consonant to replace the 'r with.

It wasn’t the rhotic I had a problem with — it was the fact that there appears to be a consonant cluster in the coda. (On the other hand, I suppose it could work if you consider it to be a minor syllable with a syllabic rhotic.)

You give a set of nominalisers in section 4.8, which is part of chapter 4 (The Noun Phrase). However, these nominalisers are described as having several functions, only some of which relate to the noun phrase. Due to this, I think you should move section 4.8 into chapter 2 (Word Classes).

Hhhm, I think they should be mentioned there, perhaps, but I'm not sure if that's the right place to describe their function in detail. I viewed Word Classes more as an overview than a detailed description of each class. I'll think about where the detailed description belongs.

I don’t mind particularly their exact location — just that they have so many uses that it doesn’t seem right to put them in the noun phrase chapter.

Section 5.2 (Basic Argument Structures) seems to be purely a typological overview. You could probably remove it with minimal impact to the rest of the grammar.

I think it mostly just introduces some shorthands used in the following section. It can probably be simplified and merged into 5.3.

I suppose that’s pretty much what I was trying to say.

I feel that you should reorganise this grammar into a more logical order more suited to a reference grammar. Here’s what I suggest:

1. Phonology (your chapter 1)
2. Word classes (your chapter 2 + 5 + section 4.8)
3. The Noun Phrase (your chapter 4)
4. Basic Clause Structure and Verbal Marking (your chapter 6 + 5.7 + 10)
5. Special Predicates (your chapter 7)
6. Serial Verb Constructions (your chapter 8)
7. Spatial and Temporal Expressions (your chapter 13 + 14)
8. Clause Linking (your chapter 11 + 12)
9. Derivational Processes (your chapter 3)
10. Discourse Pragmatics and Reference (your chapter 9)

To me, this seems like a much more logical flow compared to the current grammar.

I agree that the order might be improved. I'm not sure about some of the merger suggestions, but I think many of your combinations can be at least placed next to each other. Maybe:

PHONOLOGY
1
WORD CLASSES
2+5
CLAUSE INTERNAL STRUCTURE
4
6
7
8
13
14
CLAUSE LINKING
9
11
12
DERIVATION
3

I like this order better than mine, actually. (I think I was afraid of merging too many chapters and getting a result which is too long, but it works if you retain all the individual chapters and merely add another level of structure above them as you suggest.)

[chris_notts]

Let me see if I can understand this. You already have 108 pages of grammar, and you’re still not done yet‽

Nope, still more to go. I'm seriously thinking of diverting into conculture for a while now though. I probably also need to change a few items of vocabulary, since a semi-tropical island isn't likely to have many apples or pears.

It wasn’t the rhotic I had a problem with — it was the fact that there appears to be a consonant cluster in the coda. (On the other hand, I suppose it could work if you consider it to be a minor syllable with a syllabic rhotic.)

At the end of words, 'C just marks a contraction or a single consonant clitic. Similarly for 'k etc. I thought I said that somewhere...

[bradrn]

It wasn’t the rhotic I had a problem with — it was the fact that there appears to be a consonant cluster in the coda. (On the other hand, I suppose it could work if you consider it to be a minor syllable with a syllabic rhotic.)

At the end of words, 'C just marks a contraction or a single consonant clitic. Similarly for 'k etc. I thought I said that somewhere...

Oh, right. I thought the apostrophe represents a glottal stop. You should probably mention that somewhere.

[bradrn]

I was reading through the Pñæk grammar to check a few things, and I noticed something: apart from a couple of cursory references to them, there doesn’t seem to be any mention of adpositions. Does Pñæk really not have any adpositions at all, or did you just forget to mention them in the grammar (however unlikely that seems)? If there really are no adpositions, you should probably clarify this in the ‘Word Classes’ section.

EDIT: Also, while I’m thinking about Pñæk, another formatting-related question: What package did you use for glossing? The only one I’m familiar with is expex, but the glosses it produces by default look a bit different to yours.

[chris_notts]

I was reading through the Pñæk grammar to check a few things, and I noticed something: apart from a couple of cursory references to them, there doesn’t seem to be any mention of adpositions. Does Pñæk really not have any adpositions at all, or did you just forget to mention them in the grammar (however unlikely that seems)? If there really are no adpositions, you should probably clarify this in the ‘Word Classes’ section.

No, there aren't really any. They are covered by SVCs, by other multi-verb constructions, and by unmarked location and temporal adjuncts. I think this is a bit unrealistic or at last very marked, since almost all languages have at least a few adpositions or case suffixes, but I was trying to shift as much as possible into more open or general categories like verbs and adverbs. You'll also notice that there are almost no subordinators or dedicated clause linkers apart from the generic ka, the consecutive, (maybe) the headless relativisers, and ongoing grammaticalisation of the quotative construction.

Regarding whether to overtly mention it, this was an interesting question that was discussed in dewrad's thread recently. Do you mention the stuff that's not there, or in a detailed grammar is absence of evidence also evidence of absence?

EDIT: Also, while I’m thinking about Pñæk, another formatting-related question: What package did you use for glossing? The only one I’m familiar with is expex, but the glosses it produces by default look a bit different to yours.

Convington, I think:

https://www1.essex.ac.uk/linguistics/ex ... dex_4.html

[bradrn]

I was reading through the Pñæk grammar to check a few things, and I noticed something: apart from a couple of cursory references to them, there doesn’t seem to be any mention of adpositions. Does Pñæk really not have any adpositions at all, or did you just forget to mention them in the grammar (however unlikely that seems)? If there really are no adpositions, you should probably clarify this in the ‘Word Classes’ section.

No, there aren't really any. They are covered by SVCs, by other multi-verb constructions, and by unmarked location and temporal adjuncts. I think this is a bit unrealistic or at last very marked, since almost all languages have at least a few adpositions or case suffixes, but I was trying to shift as much as possible into more open or general categories like verbs and adverbs.

That’s a pretty interesting idea. I think there’s plenty of languages which shift adpositional meanings into SVCs etc., but I am a bit worried that — as you say — their total absence seems a bit unrealistic (albeit not excessively so).

You'll also notice that there are almost no subordinators or dedicated clause linkers apart from the generic ka, the consecutive, (maybe) the headless relativisers, and ongoing grammaticalisation of the quotative construction.

I actually haven’t noticed that, probably since I haven’t read the sections on clause linking yet. Certainly, I believe that the absence of dedicated subordinates is attested (although I’d probably need to check to be sure).

(Interestingly, my previous conlang also didn’t have any conjunctions or subordinates per se — their functions were taken up by various clitics and adverbs instead.)

Regarding whether to overtly mention it, this was an interesting question that was discussed in dewrad's thread recently. Do you mention the stuff that's not there, or in a detailed grammar is absence of evidence also evidence of absence?

Personally, I don’t really know. Certainly you’re not going to write about every single feature which isn’t there. But equally, I think that adpositions are so common that it’s probably useful to mention their absence somewhere.

EDIT: Also, while I’m thinking about Pñæk, another formatting-related question: What package did you use for glossing? The only one I’m familiar with is expex, but the glosses it produces by default look a bit different to yours.

Convington, I think:

https://www1.essex.ac.uk/linguistics/ex ... dex_4.html

Thanks! The examples from the documentation seem to match up with your glosses, so I agree that it’s probably Covington. I’ll have to have a look at it then.

[quinterbeck]

This is a great read, really interesting! Thanks for sharing.

I'm curious how speakers would describe a motion event with attention to the path or route, since you say the motion verbs are punctual, and boundary crossing verbs are instantaneous. So for e.g. "We traveled through the forest", it seems like these wouldn't be appropriate, since a forest is a bit big to conceptualise as a boundary. How does Pñæk handle that?

[chris_notts]

That’s a pretty interesting idea. I think there’s plenty of languages which shift adpositional meanings into SVCs etc., but I am a bit worried that — as you say — their total absence seems a bit unrealistic (albeit not excessively so).

The languages which spring to mind as possibly being closest to this are not just languages with SVCs, but the most verby polysynthetic languages, some of which IIRC strongly restrict the number of NPs per clause and tend to favour applicatives and multiverb sequences over case/adposition marking. Whether any can be said to lack adpositions completely I'm not sure. Then there are languages like some of the Salishan languages and Mayan languages which have a generic, semantically empty adposition that covers most oblique roles, so the distinction is basically core vs non-core. To be more specific about the role of something you need to use a verb and maybe an applicative.

Personally, I don’t really know. Certainly you’re not going to write about every single feature which isn’t there. But equally, I think that adpositions are so common that it’s probably useful to mention their absence somewhere.

I suppose there's no harm in mentioning it. I was also going to write a chapter to synthesize the various role marking strategies in one place.

[chris_notts]

This is a great read, really interesting! Thanks for sharing.

I'm curious how speakers would describe a motion event with attention to the path or route, since you say the motion verbs are punctual, and boundary crossing verbs are instantaneous. So for e.g. "We traveled through the forest", it seems like these wouldn't be appropriate, since a forest is a bit big to conceptualise as a boundary. How does Pñæk handle that?

Good question. The durative thing I took from Levinson's books Space in Language and Cognition and Grammars of Space. The claim is made that boundary crossing verbs in some languages are not durative:

On the basis of the kind of description in the chapters above, we can suggest tentatively that there are perhaps three rather different styles of conceptualisation involved in the coding of motion events cross-linguistically, as in figure 14.6 (contrasting, incidentally, with this set of notions is another one might call 'internal motion' or 'manner of motion without change of locative state' - this is what is apparently coded in the manner verbs of Yukatek and Tamil). In this typology, translocation, that is a durative event involving passage though an indefinite series of points in space over time, is only one possibility. Motion can instead be thought of solely as a change of state without transitional phases: at time t1 figure F is in state S1, at time t2 F is in state S2 - what happened in between may be immaterial. The simplest case of this is to think of motion as change of location: at time t1, figure F is at source S, at time t2, figure F is no longer at S; alternatively, at time t1 F is not yet at goal G, at time t2 F is now at G. This kind of analysis suggests that motion verbs of this kind should never collocate with both a source and a goal. Bohnemeyer and Stolz argue that Yukatek motion verbs show both these telltale symptoms - punctual aspect, and no specification of both source and goal. For verbs with this kind of semantics, how the figute got from source to goal is not relevant - details of the trajectory, the manner of motion, the medium and the instruments involved are out of focus as it were. Languages that code motion semantics in verbs as change of location in this way are thus not likely to fuse manner (as in crawl), or medium (as in swim) or instrument (as in drive) into a genuine motion verb. Again in Yukatek, there are verbs meaning, for example, 'swim' or 'fly', but these do not take source or goal specification (a location will instead be understood as the place within which the swimming or flying activity took place). The facts in Tzetzal and Yeli Dnye are at least suggestive of a similar analysis.

The question is how this fits with cross in particular, because 'cross' implies the presence of two boundary crossing events, unlike enter or exit which only imply one. Levinson claims a hierarchy of verbs likely to have change of state semantics:

exit > enter, cross > go

If you'll look at table 13.4 in my grammar you'll see that go is durative, but I grouped cross with exit and enter. I'm now wondering if cross should be more durative because, unlike the others, it implies two sub-events.

To answer the question about boundary crossing verbs in general, if you want durative semantics the answer is to serialise the boundary crossing verb with a durative verb:

mi biil gbe m può
PROG wander arrive ART village
He's wandering to the village

?? mi gbe m può <-- odd because arrive is a transition so you can't really be in the middle of doing it
PROG arrive ART village

[Yalensky]

EDIT: Also, while I’m thinking about Pñæk, another formatting-related question: What package did you use for glossing? The only one I’m familiar with is expex, but the glosses it produces by default look a bit different to yours.

Convington, I think:

https://www1.essex.ac.uk/linguistics/ex ... dex_4.html

Does Covington have a non-tedious way of formatting grammatical labels as small caps, or did you have to format each label yourself individually?

Since I've gotten into LaTeX as a quarantine hobby, I've been using typgloss for glosses.

[bradrn]

EDIT: Also, while I’m thinking about Pñæk, another formatting-related question: What package did you use for glossing? The only one I’m familiar with is expex, but the glosses it produces by default look a bit different to yours.

Convington, I think:

https://www1.essex.ac.uk/linguistics/ex ... dex_4.html

Does Covington have a non-tedious way of formatting grammatical labels as small caps, or did you have to format each label yourself individually?

Most of the glossing packages don’t do that. I believe that the usual approach is to use whichever glossing package you like in combination with leipzig, which manages all the abbreviations for you. It lets you do things like defining (or redefining) new abbreviations, collecting them into glossaries etc. (EDIT: It looks like the documentation for typgloss already mentions leipzig, so sorry if you already know about it!) For instance, here’s a gloss from one of my conlangs (from the section on subordination):

Code: Select all

\pex
\a\begingl
\gla Woog-∅ woo-tsap. Zhaag-ek kaang-∅ yaazhlaam. //
\glb 2s-\Abs{} 2s-tall. 1s-\Erg{} that-\Abs{} 3s-1s-know. //
\glft You are tall. I know that. //
\endgl
\a\begingl
\gla Zhaag-ek yaazhlaam mesh= @ [woog-∅ woo-tsap] //
\glb 1s-\Erg{} 3s-1s-know \Comp= [2s-\Abs{} 2s-tall] //
\glft I know that [you are tall]. //
\endgl
\xe

Here all the \pex and \gl stuff is from my preferred glossing package expex, while the abbreviation commands are from leipzig. Here’s how it ends up looking:

Since I've gotten into LaTeX as a quarantine hobby, I've been using typgloss for glosses.

That looks pretty interesting, especially the automatic abbreviation formatting! (Though I must say that I like using commands for them.) Maybe I’ll try using it — expex (which I’m using right now) is good, but a little strange.

EDIT: It looks like typgloss isn’t a standalone glossing package — it relies on gb4e for the glossing itself, and merely takes care of the capitalisation. So it’s not a replacement for expex etc., like I thought it was.

[chris_notts]

Does Covington have a non-tedious way of formatting grammatical labels as small caps, or did you have to format each label yourself individually?

Most of the glossing packages don’t do that. I believe that the usual approach is to use whichever glossing package you like in combination with leipzig, which manages all the abbreviations for you. It lets you do things like defining (or redefining) new abbreviations, collecting them into glossaries etc. (EDIT: It looks like the documentation for typgloss already mentions leipzig, so sorry if you already know about it!)

I should probably use leipzig, since so far I've just been doing it manually. One thing I don't like much is the need to put {} after each command to prevent it suppressing the following space - in general I don't understand why LaTeX was designed to work that way. To me it seems obvious that if a command isn't followed by any brackets it shouldn't consume anything by default. I know it won't be fixed now because it would probably break other LaTeX code somewhere, but I feel a twinge of annoyance every time I have to write something like \Abs{} to protect my spaces from rogue deletion.

On another topic, I do always use XeTeX now, not plain LaTeX, after a conlang grammar I wrote once in plain LaTeX that had stacked diacritics. Never again. Not being able to handle unicode is now a deal breaker for me in a document formatting system.

Oh, and typgloss looks like a cool enhancement to leipzig.

[bradrn]

Does Covington have a non-tedious way of formatting grammatical labels as small caps, or did you have to format each label yourself individually?

Most of the glossing packages don’t do that. I believe that the usual approach is to use whichever glossing package you like in combination with leipzig, which manages all the abbreviations for you. It lets you do things like defining (or redefining) new abbreviations, collecting them into glossaries etc. (EDIT: It looks like the documentation for typgloss already mentions leipzig, so sorry if you already know about it!)

I should probably use leipzig, since so far I've just been doing it manually. One thing I don't like much is the need to put {} after each command to prevent it suppressing the following space - in general I don't understand why LaTeX was designed to work that way. To me it seems obvious that if a command isn't followed by any brackets it shouldn't consume anything by default. I know it won't be fixed now because it would probably break other LaTeX code somewhere, but I feel a twinge of annoyance every time I have to write something like \Abs{} to protect my spaces from rogue deletion.

I find that pretty annoying as well. (It’s actually better with more agglutinative languages, since sequences like \Fsg-say-\Antip-\Pres.\Pfv-\Neg{} only need one {} at the end.) As for why, this claims it’s so you can write e.g. \foo bar if you want a command followed immediately by a letter. (Though I’m not convinced that \foo{}bar is all that bad.)

On another topic, I do always use XeTeX now, not plain LaTeX, after a conlang grammar I wrote once in plain LaTeX that had stacked diacritics. Never again. Not being able to handle unicode is now a deal breaker for me in a document formatting system.

I’ve never had that problem, since I’ve always just used XeTeX for my grammars. (Mostly it’s for IPA support, actually: most of my conlangs don’t use many diacritics, so plain LaTeX isn’t a problem, but I don’t want to combine TIPA’s Computer Modern-esque IPA font with whichever other font I decide to use. It’s easier to just use XeTeX so I can use the same font for both IPA and text.)

[chris_notts]

I find that pretty annoying as well. (It’s actually better with more agglutinative languages, since sequences like \Fsg-say-\Antip-\Pres.\Pfv-\Neg{} only need one {} at the end.) As for why, this claims it’s so you can write e.g. \foo bar if you want a command followed immediately by a letter. (Though I’m not convinced that \foo{}bar is all that bad.)

Exactly. I would have chosen to force \foo{}bar if you really want a command embedded in text. The other alternative would be to have a command terminator of some kind if you really hate having to write two characters instead of one to terminate your command, e.g. \foo|bar maybe. Or maybe, if we're really redesigning LaTeX, repurpose some brackets to execute commands so [foo] means the same as \foo{}.