Best way to classify verbs as “impersonal“ and “phrasal“

[josh_s...@sil.org]

Greetings all,

In the language I am working in I have certain verbs which I want to classify as “impersonal“ (meaning they take a dummy subject). I also have verbs which I want to classify as “phrasal“ (meaning they take a syntactically independent verbal particle which in combination with a root forms a new lexeme). So far I have been using “grammatical info“ to define these, so I have these categories:

Verb
-- Intransitive Verb
---- Impersonal Intransitive Verb
---- Phrasal Intransitive Verb
-- Transitive Verb
---- Impersonal Transitive Verb
---- Phrasal Transitive Verb
-- Ditransitive Verb
---- Impersonal Ditransitive Verb
---- Phrasal Ditransitive Verb

This is kind of redundant, however, since the basic template for the three levels of transitivity is essentially the same (phrasal verbs represented by the particle (PART)):

Intransitive: SBJ=(ASP-)STEM(-DIR)(-SBRD) (PART)
Transitive: SBJ=(ASP-)STEM(-DIR)(-SBRD)=OBJ (PART)
Ditransitive: SBJ=(ASP-)STEM(-DIR)(-SBRD)=I.OBJ=D.OBJ (PART)

I am now realizing that some verbs are both impersonal and phrasal (should have seen that coming). I don‘t want to add yet another grammatical category... in fact I would like to reduce them if I could. So my question is:

- What other options do I have for tagging an entry/sense as “impersonal“
- What other options do I have for tagging an entry/sense as “phrasal“
- Does it make sense to keep either of these as a category under grammatical info?

Thank you

[Andreas_Joswig]

Hi Josh,

My first hunch would be to get rid of the transitivity-based verb classification. The templates for intransitive, transitive and ditransitive verbs appear to be identical, except for the object or indirect-object clitics. Did you create separate templates in order for you to have fewer optional slots in your templates (they seem to be the only slots next to the stem that are not optional)? Just for that I don't think it is worth it to distinguish according to transitivity. In my experience, although transitivity of course makes a big difference semantically, in very few languages it actually has any formal consequences, such as having to use a completely different verbal template with an entirely different set of affixes. If it does not have any formal impact, there is not much point to trying to model transitivity differences in FLEx. All this is of course further complicated by verbs that are not easily classified as transitive or intransitive, as all too many languages see this with a lot more flexibility than we'd like them to. So, if there is not any other compelling reason, I would certainly ditch transitivity from your verb classifications. If you really want to make any helpful claims about the transitivity of a verb in the lexicon, you can use the grammar note with snippets such as "usually transitive" or "always found intransitive".

As for the impersonal verbs, I would go the same way - make it clear in the grammar note that they are impersonal, as otherwise there doesn't seem to be any morphological effect on the verb itself. Phrasal verbs are a much more complicated matter - of course you'll want each of them listed as a separate entry in the lexicon, but as the automatic parsers only ever look at individual words (and never beyond), I'm not quite sure how to model them. You will almost certainly have to break them up into their parts in the lexicon so that the parsers can handle them, except if someone else here has a better idea (which I would love to hear).

Andreas

Thank you --
"FLEx list" messages are public. Only members can post.
flex_d...@sil.org
http://groups.google.com/group/flex-list.
---
You received this message because you are subscribed to the Google Groups "FLEx list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to flex-list+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/flex-list/aab53bc9-bfa8-493c-91b0-4927de863be1n%40googlegroups.com.

[Ron Moe]

Hi Josh,

There is a general rule in lexicography--keep the technical terms to a minimum. Most users will not understand them, especially if they are abbreviated.

I would highly recommend creating a sense level custom field (I call mine "Sense Form") to indicate the syntactic case frame of a verb (and of nouns and adjectives when appropriate). I also use this field to indicate diagnostic forms of the word (active vs passive verbs, plural nouns, etc).

I put this on the sense level because that is where the vast majority of this information applies. I have found indicating this information in the definition in a second analysis language is almost impossible and it is highly disruptive.

So what I do is to indicate the syntactic case frame (for instance) and then follow it in the Definition field with a literal translation of the Sense Form field, and then a longer explanation of the semantics of the word. I also put in example sentences (which I have left out in the entries below).

If there is an inflectional restriction for the sense, I put it in the Restrictions field. In the second sense of the first entry below you will see "(passive)", which is an example of this. The form of the word is also in the passive in the Sense Form field. So the two match and reinforce each other.

ἀναμιμνῄσκω  ἀναμνη-  V.

  1) ἀναμιμνῄσκω σε τι/ποιεῖν τι. I remind you (about) something. I remind you to do something.

  2) ἀναμιμνῄσκομαί τι. (passive) I remember something.

ἐκδικέω  ἐκδικη-  V.

  1) ἐκδικῶ σε ἀπὸ αὐτοῦ. I give you justice from him. I avenge you. ἐκδικέω refers to a situation where someone has committed a crime or done something wrong against you. I (a judge or God) punish him or make him pay for his crime. I gain justice for you from him, for example by executing him or making him pay you for your loss.

  2) ἐκδικῶ ἐμαυτόν. I avenge myself. I get revenge by doing something bad to someone because he did something bad to me.

  3) ἐκδικῶ τι/παρακοήν. I avenge a crime. I punish someone for committing a crime or sin.

The great advantage of this system is that it has very few technical terms, and those I use are basic and important to the inflection. Perhaps even better is that the information is in the vernacular using simple basic vocabulary that the reader is likely to already know. (The dictionary is designed for language learners.)

I would highly recommend this system of indicating inflectional patterns related to the sense and for indicating the syntactic case frame.

Ron Moe

To view this discussion on the web visit https://groups.google.com/d/msgid/flex-list/d314462f-6bbf-f13d-8745-6e78dce22ef4%40sil.org.

[Michael Maxwell]

On 3/11/2021 2:03 PM, Andreas_Joswig wrote:
> ...although transitivity of course makes a big difference

> semantically, in very few languages it actually has any formal
> consequences, such as having to use a completely different verbal
> template with an entirely different set of affixes

YMMV--your mileage may of course vary. In languages where both subjects
and objects are marked by inflection on the verb, FLEx will need to know
the transitivity for morphological templates. Languages that do that
include Nahuatl (Aztec) and Mayan languages; the former have accusative
marking, while the latter have ergative marking, but both mark subject
on all verbs, and object on transitive verbs.

If I understand your email correctly, your language does inflectionally
mark a transitivity distinction, in fact a 3-way distinction, on the verb:
--------------

Intransitive: SBJ=(ASP-)STEM(-DIR)(-SBRD) (PART)
Transitive: SBJ=(ASP-)STEM(-DIR)(-SBRD)=OBJ (PART)
Ditransitive: SBJ=(ASP-)STEM(-DIR)(-SBRD)=I.OBJ=D.OBJ (PART)

-------------
Now it looks like you're making a distinction between affixes (your '-'
boundary marker) and clitics (the '=' boundary marker). But if you're
writing those clitics "solid" with their host (no space), then at least
for purposes of morphological parsing you'll need to make a 3-way
transitivity distinction. Clitic vs. affix is a distinction without
meaning, as far as morphological parsing is concerned.

About impersonal verbs: my guess (assuming subjects are inflectionally
marked) is that such verbs will take only a 3rd person subject
inflectional affix, right? Which is a restriction on the morphology, so
you don't get things like *"you rain" or *"you seem that it will work".

So *if* the person of both subjects and objects is inflectionally marked
on transitive verbs (and of course only the subject person on
intransitive verbs), *and if* impersonal verbs can only take 3rd person
subject marking, then it seems to me you would indeed need to
distinguish both 3-way transitivity and the personal vs. impersonal on
verbs in your lexical categories.

About phrasal verbs: I'm assuming this is the 'PART' (particle). It
looks like you're not writing that solid with the verb, that is you have
a space in the written forms. If that's the case, then you don't need
to worry about the particle in the morphology, only in the syntax: the
particle is effectively invisible to the morphological parser. Of
course if you want to do syntactic parsing (maybe using Cheryl Black's
syntactic parsing tools), that's a different question.

Languages are complicated, aren't they? Amazing that children learn them!

What language is this?
--
Mike Maxwell
"Digital objects last forever--or five years,
whichever comes first." --Jeff Rothenberg

[josh_s...@sil.org]

Thanks all,

I will play around with some of your suggestions. I am not keen on the idea of not distinguishing transitivity at least, since I want to preserve that information for my lexicon (I am using one project for both a technical linguistics database and local dictionary project, so I basically want as much info as possible). I do like the idea of a custom field under "sense". 

Ron, could you send a screenshot of what a lexical entry looks like in FLEx so I can see your custom fields, where you put them, and what you put in them?

Mike, the language is Opo (a.k.a. T'apo, Opuuo, Ciita, T'azita) from western Ethiopia and South Sudan.

The phrasal verbs are a doozie for sure. They are unique lexemes formed with a syntactically free particle. A parser would have to be able to handle the following syntax (for clarity it is the combination of the verb root + the particle 'body' that give the sense 'separate' or 'divide'):

ʊ̄npʰátés
ʊ̄n=pʰátá=ēs
3pl=divide=body
'They separated (vi.)' - The particle cliticizes to the verb root and coalesces with the final vowel

ʊ̄npʰátán ēs
ʊ̄n=pʰátá=ān ēs
3pl=divide=3n body
'They divided it (vt.)'

ʊ̄npʰátá ɡʊ́nān ēs
ʊ̄n=pʰátá gá=ʊ̄n=ān ēs
3pl=divide BEN=3pl=3n body
'They divided it for them (vt. + obl. beneficiary)'

Negative, the particle shifts around.

ʊ̄nɗʊ̄kʼʊ́ēs pʰátá
ʊ̄n=ɗʊ̄kʼʊ́=ēs pʰátá
3pl=NEG=body divide
'They didn't separate (vi.)'

ʊ̄nɗʊ̄kʼʊ́ ɡʊ́nān ēs pʰátá
ʊ̄n=ɗʊ̄kʼʊ́  gá=ʊ̄n=ān  ēs pʰátá
3pl=NEG BEN=3pl=3n body divide
'They didn't divided it for them'

[Ron Moe]

I've attached two screenshots of the entry ἀναμιμνῄσκω (anamimnjskw) 'remind; remember' that I was talking about in my previous post. I only included the relevant fields, but I couldn't fit both senses on the screen at the same time. So the second screen shot is for the Sense 2 fields. There are a lot of other hidden fields with other information that is irrelevant to the discussion.

Ron Moe

--

"FLEx list" messages are public. Only members can post.
flex_d...@sil.org
http://groups.google.com/group/flex-list.
---
You received this message because you are subscribed to the Google Groups "FLEx list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to flex-list+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/flex-list/62f537a4-cfd8-4280-bdbd-3b95c38f7a50n%40googlegroups.com.