zompist bboard

So, I have been working on eliminating or minimizing the noun-verb distinction in my conlang. My main approach thus far has involved treating all roots, even those describing objects, as verbs and putting those verbs into an attributive form used for relative clauses to form arguments of verbs. Thus the equivalent of "my mother" would become something like "she who is a mother to me". The problem with this approach is that words signifying "nouns" all have the same inflectional suffix, resulting in considerable repetition. Imagine how a sentence like "The quick brown fox jumps over the lazy dog" would translate if every noun and adjective became a verb in the attributive form.

malloc wrote:So, I have been working on eliminating or minimizing the noun-verb distinction in my conlang. My main approach thus far has involved treating all roots, even those describing objects, as verbs and putting those verbs into an attributive form used for relative clauses to form arguments of verbs. Thus the equivalent of "my mother" would become something like "she who is a mother to me". The problem with this approach is that words signifying "nouns" all have the same inflectional suffix, resulting in considerable repetition. Imagine how a sentence like "The quick brown fox jumps over the lazy dog" would translate if every noun and adjective became a verb in the attributive form.

Could you make the suffix shorter? If it is just a single phoneme, it would not add much length to a word. E.g. "kwikt braunt fokst overjamp leizit doxt" is not that long. You might also want to consider if you want to actually make it so the finite verb form gets a suffix, and use the bare stem as what you're calling the attributive form. Usually there is only one finite verb in a sentence, while as you point out there are fairly often multiple nouns or adjectives. Also, finite verbs often inflect for other purposes anyway, like for tense or for person, number and/or gender of the subject or of various other participants, so I'd imagine a fair number of finite verb forms will use some kind of suffix anyway.

I'm not sure how this is eliminating the noun-verb distinction. If the 'attributive suffix' is, say, -t, then that's just a roundabout way of saying that all nouns end in -t and verbs don't. Which is hardly eliminating the distinction - if anything, it's just making it more concrete. You could also say, "ahh, but the same roots are used for verbs and nouns!" - but that's not that exciting either - in English pretty much all verb roots can be used as nouns and vice versa - and English often doesn't even need any affixes to mark this! [See the 'garden path' thread in NOTA for many examples of the ambiguities this can cause when carried to excess by headline writers]
Consider, for example, the English sentence, "let's science the shit out of the problem". Look at the subtantive words: let, science, shit, and problem. "Let" is both a noun and a verb, "science" is normally a noun but here is being used as a verb and "shit" is both a noun and a verb"; "problem" isn't normally used as a verb, though it's easy to see it being used that way, and in any case the root can easily be turned into a verb with affixes. And, for good measure, the proposition "out", if it is a proposition here and not an adverb or a part of a verb, is also both a noun and a verb.


More generally: if a certain word "is the equivalent of" saying 'my mother', then that's what the word means: 'my mother'. It only makes sense to say "this word actually means 'she who is a mother to me'" if the language also has a way of saying 'my mother' and makes a distinction between 'she who is a mother to me' and 'my mother' - if the 'relative' is the only way of saying it, it's not a relative, it's just a plain noun. It's exactly the same as the way you can't say "that's not /p/, that's really /b/" in a language where voicing is not phonemic...

Indeed. If there's just one class, we could also easily say all verbs are actually nouns. If nouns also take the additional morphology commonly associated with verbs, you could say it's zero copula, or whatever. I think Zompist described something like that in the LCK, about Vance's The languages of Pao.


JAL

I toyed with replacing the noun-verb distinction with a different one for a language "coloured" by cephalopods. I couldn't think of anything and went back to my human languages, which always had a higher priority anyway. (When I think of something I might like, I'll use it.)

Salmoneus wrote:I'm not sure how this is eliminating the noun-verb distinction. If the 'attributive suffix' is, say, -t, then that's just a roundabout way of saying that all nouns end in -t and verbs don't. Which is hardly eliminating the distinction - if anything, it's just making it more concrete. You could also say, "ahh, but the same roots are used for verbs and nouns!" - but that's not that exciting either - in English pretty much all verb roots can be used as nouns and vice versa - and English often doesn't even need any affixes to mark this! [See the 'garden path' thread in NOTA for many examples of the ambiguities this can cause when carried to excess by headline writers]

My reasoning is that the suffix would constitute an inflection rather than derivation because it applies productively and transparently to any verbal stem. Verbs simply have an inflectional form which indicates that they are functioning as arguments rather than predicates. For comparison, it would hardly make sense to call verbs in Latin inflected for the subjunctive mood part of a distinct lexical class from verbs in the indicative mood.

More generally: if a certain word "is the equivalent of" saying 'my mother', then that's what the word means: 'my mother'. It only makes sense to say "this word actually means 'she who is a mother to me'" if the language also has a way of saying 'my mother' and makes a distinction between 'she who is a mother to me' and 'my mother' - if the 'relative' is the only way of saying it, it's not a relative, it's just a plain noun. It's exactly the same as the way you can't say "that's not /p/, that's really /b/" in a language where voicing is not phonemic...

But the word for "she who is a mother to me" breaks down morphologically as be.mother.to-attributive-3>1 and parallels the predicative form be.mother.to-indicative-3>1 for "she is a mother to me". Given that it derives transparently from the verb root and differs in syntactic role but not semantics, does it really make sense to classify it as a distinct lexical class?

Salmoneus wrote:I'm not sure how this is eliminating the noun-verb distinction. If the 'attributive suffix' is, say, -t, then that's just a roundabout way of saying that all nouns end in -t and verbs don't. Which is hardly eliminating the distinction - if anything, it's just making it more concrete. You could also say, "ahh, but the same roots are used for verbs and nouns!" - but that's not that exciting either - in English pretty much all verb roots can be used as nouns and vice versa - and English often doesn't even need any affixes to mark this! [See the 'garden path' thread in NOTA for many examples of the ambiguities this can cause when carried to excess by headline writers]

My reasoning is that the suffix would constitute an inflection rather than derivation because it applies productively and transparently to any verbal stem.[/quote] First, that assumes that they are "verbal stems" to begin with. And secondly: English has plenty of affixes that apply productively and (barring irregularities) transparently to any verbal (or nominal) stem. But we don't say that "duckise" ('cause to be or be infested with a duck') is a inflection...

It is indeed true that if a language used all stems completely freely for both nouns and verbs, you could say it had no lexical distinction - English is a long way in that direction, but still has a way to go. However:
a) it's not clear that languages can actually work like that, because it seems inevitable that some concepts will never in practice be found in verbal or nominal forms without the need for exceptional marking;
b) it's not clear that it's particularly interesting in itself; it's been debated whether a great many languages fall into this category, but it doesn't appear to matter all that much; (fwiw, you may want to look into the Austronesian family, which has been extensively argued to lack the distinction)
c) one reason why it's not that interesting is that lacking a lexical distinction doesn't mean there is not a syntactic distinction (though it's not unusual for that distinction to be blurred in some cases). And because it has no syntactic significance, there's not a lot else to say about it, except that dictionary entries may be longer and fewer in number.

Verbs simply have an inflectional form which indicates that they are functioning as arguments rather than predicates. For comparison, it would hardly make sense to call verbs in Latin inflected for the subjunctive mood part of a distinct lexical class from verbs in the indicative mood.

We would call them subjunctive verbs. That's the word for verbs in the subjunctive. And verbs that have been turned into nouns are called nouns.

More generally: if a certain word "is the equivalent of" saying 'my mother', then that's what the word means: 'my mother'. It only makes sense to say "this word actually means 'she who is a mother to me'" if the language also has a way of saying 'my mother' and makes a distinction between 'she who is a mother to me' and 'my mother' - if the 'relative' is the only way of saying it, it's not a relative, it's just a plain noun. It's exactly the same as the way you can't say "that's not /p/, that's really /b/" in a language where voicing is not phonemic...

But the word for "she who is a mother to me" breaks down morphologically as be.mother.to-attributive-3>1 and parallels the predicative form be.mother.to-indicative-3>1 for "she is a mother to me".

First off, the ordinary English for "to be mother to" is just "to mother". You're just phrasing it oddly to make it sound more special. Likewise, what you call "attributive" seems to just be "noun" - obligatory marking of all nouns (which is frankly a little redundant, as you point out yourself - what do you actually gain from this marking? English gets on fine without it). Your "3>1" affix is a little more interesting - basically you seem to be saying that your nouns take possessives, and that you have different possessives for possessed items that are also 1st or 2nd person referents. I've never heard of this happening in a language, but it wouldn't surprise me if it did. It's actually quite an interesting idea. I guess you could just treat it as person marking combined with a possessive, in which case it would be like normal nominal person marking, which is rare but not that weird - Sumerian, for instance, famously marks nouns for person.

I'm not sure why you're calling "noun" and "verb" "attributive" and "predicative" - particularly because this doesn't really match the normal sense of an attributive.
[In English: "she mothers me" - verb; "my mother ate the cat" - noun; "to the mother country" - attributive (instinctively I'd say adjective, but grammarians may have reasons for preferring to call that a sort of appositive noun construction).]

It's also best to be careful where you seem to imply a contrast between "attributive" (i.e. nominaliser) and "indicative". Because of course you can have non-indicative nouns, so there's no necessary reason why the nominalising suffix should pattern alongside moods. Although it's not impossible, of course.

Given that it derives transparently from the verb root and differs in syntactic role but not semantics, does it really make sense to classify it as a distinct lexical class?

going back to my earlier points:
a) it may make sense, yes, because in practice there will be roots used almost always as nouns and others used almost always as verbs, and it might be helpful for dictionaries to make note of the difference;
b) does it matter? That's just an issue of dictionary notation, not anything more significant. Why worry about "how to do it" questions if all you're doing is making a usage frequency note, essentially saying that "the word for 'mongoose' is used as often as a verb as as a noun" and "the word for 'abhor' is used as often as a noun as as a verb"?
c) again, if you're only talking about lexical classifications, then why mess around with strange glosses for things? Glosses are about syntax, showing how the sentence works, rather than about lexicographical organisation.

First, that assumes that they are "verbal stems" to begin with. And secondly: English has plenty of affixes that apply productively and (barring irregularities) transparently to any verbal (or nominal) stem. But we don't say that "duckise" ('cause to be or be infested with a duck') is a inflection...

Well ok, what would you say constitutes an inflection as opposed to derivation?

Salmoneus wrote:a) it's not clear that languages can actually work like that, because it seems inevitable that some concepts will never in practice be found in verbal or nominal forms without the need for exceptional marking;

I should point out that this conlang is polysynthetic with considerable influence from languages like Inuktitut and Mohawk, heavy use of incorporating morphology, and so forth. Hence even basic concepts like "water" would frequently occur in situations other than free-standing nouns, in stems like water-there.is and water-consume.

Your "3>1" affix is a little more interesting - basically you seem to be saying that your nouns take possessives, and that you have different possessives for possessed items that are also 1st or 2nd person referents.

The language has polypersonal agreement and the notation "3>1" indicates a third person subject and first person object. The root for mother is a transitive verb requiring both a subject (for the mother herself) and an object (for her child). Or if you prefer, it is a noun which obligatorily inflects for its own person and that of its possessor in ways paralleling transitive verbs.

I'm not sure why you're calling "noun" and "verb" "attributive" and "predicative" - particularly because this doesn't really match the normal sense of an attributive.
[In English: "she mothers me" - verb; "my mother ate the cat" - noun; "to the mother country" - attributive (instinctively I'd say adjective, but grammarians may have reasons for preferring to call that a sort of appositive noun construction).]

I was using it in the sense of an attributive verb, as described here.

I'm all for merging or rethinking syntactic categories, but the noun-verb one is pretty hard to get rid of, for many of the reasons Sal points out.

Salmoneus wrote:c) one reason why it's not that interesting is that lacking a lexical distinction doesn't mean there is not a syntactic distinction (though it's not unusual for that distinction to be blurred in some cases).

I'd like to expand on this, because I think conlangers probably still underestimate syntax. If you find yourself writing "A verb in the indicative or subjunctive appears with several arguments, all verbs in the attributive"... well, it's just kind of a weirdly jargonized way of saying "verbs have nouns as their subjects and objects". Conventions in language are there for convenience and understanding; don't burden your readers.

Syntactic categories are defined, naturally enough, by their syntactic behavior. If something can occur as the argument of a verb, then it's a noun. Its morphological derivation is irrelevant, interesting though it is in the morphology section.

At the same time, syntactic categories are fuzzy. There isn't a strict line between nouns and verbs; the line is already fudged by things like gerunds, participles, infinitives, and nominalizations. So rather than asking "is it a noun or a verb?" the better question may be "how nouny or verby is it?" For your 'attributives', can you do nouny things like pluralization, possessive, diminutive, replacement by personal pronouns, adding descriptive adjectives, topicalization? Do they resist verby things like tense, aspect, irrealis? Note that a particular form may have both nounlike and verblike features.

You should also, I think, worry about the language development view. How does a child learn these words? Does it learn the verb 'to mother' first, and only later encounter 'a mother'? That seems very unlikely! It's going to learn a word X which refers to its mother. The natural supposition is that X means 'mother'. To insist that the word is really a verb is like insisting that 'goodbye' is really a wish that God be with you.

A counter-argument would be that children learn words like 'gimme' as single morphemes, and later realize that these are two words 'give me'. But such reanalyses apply only to a few words. If your reanalysis applies to thousands of words, you have to consider whether you're describing the language or forcing a theory onto it.

zompist wrote:You should also, I think, worry about the language development view. How does a child learn these words? Does it learn the verb 'to mother' first, and only later encounter 'a mother'? That seems very unlikely! It's going to learn a word X which refers to its mother. The natural supposition is that X means 'mother'. To insist that the word is really a verb is like insisting that 'goodbye' is really a wish that God be with you.

A counter-argument would be that children learn words like 'gimme' as single morphemes, and later realize that these are two words 'give me'. But such reanalyses apply only to a few words. If your reanalysis applies to thousands of words, you have to consider whether you're describing the language or forcing a theory onto it.

I'm sure there's a language somewhere which does handle it's kinship system with a set of verbs like that described above, but I need to look further into that.

Just as another test case, is "forget-me-not" a noun? My dictionary thinks so (and I agree). Is it "really" an imperative VP?

Frislander wrote:

zompist wrote:You should also, I think, worry about the language development view. How does a child learn these words? Does it learn the verb 'to mother' first, and only later encounter 'a mother'? That seems very unlikely! It's going to learn a word X which refers to its mother. The natural supposition is that X means 'mother'. To insist that the word is really a verb is like insisting that 'goodbye' is really a wish that God be with you.

A counter-argument would be that children learn words like 'gimme' as single morphemes, and later realize that these are two words 'give me'. But such reanalyses apply only to a few words. If your reanalysis applies to thousands of words, you have to consider whether you're describing the language or forcing a theory onto it.

I'm sure there's a language somewhere which does handle it's kinship system with a set of verbs like that described above, but I need to look further into that.

Certain languages of Arnhem Land have "kinship verbs" - "X is mother to me;" "I am brother to Y" etc. They are interesting for several reasons, one of which is that they have transparent lexical derivations (e.g. Gurr-goni a-rrirrmi-rri-pu "he held him", ie. "he is his father")

zompist wrote:If something can occur as the argument of a verb, then it's a noun.

What if you can have a finite verb as the argument of a verb? One of my conlangs can do just that (as well as conjugate another verb form for both person/number of its subject and animacy/number of its head)

zompist wrote:I'd like to expand on this, because I think conlangers probably still underestimate syntax. If you find yourself writing "A verb in the indicative or subjunctive appears with several arguments, all verbs in the attributive"... well, it's just kind of a weirdly jargonized way of saying "verbs have nouns as their subjects and objects". Conventions in language are there for convenience and understanding; don't burden your readers.

So you would consider it a case of the duck test: if this verbal form serves the same function as nouns, inflects for nominal attributes like possession, and so forth, it makes more sense simply to call it a noun.

Syntactic categories are defined, naturally enough, by their syntactic behavior. If something can occur as the argument of a verb, then it's a noun. Its morphological derivation is irrelevant, interesting though it is in the morphology section.

But what if the same lexeme can occur in both verbal and nominal roles depending on its inflection? Consider the case system of Finnish, with its extensive collection of locative cases that serve adverbial functions. Would you consider nouns in those cases to have become adverbs?

Salmoneus wrote:Consider, for example, the English sentence, "let's science the shit out of the problem". Look at the subtantive words: let, science, shit, and problem. "Let" is both a noun and a verb, "science" is normally a noun but here is being used as a verb and "shit" is both a noun and a verb"; "problem" isn't normally used as a verb, though it's easy to see it being used that way, and in any case the root can easily be turned into a verb with affixes. And, for good measure, the proposition "out", if it is a proposition here and not an adverb or a part of a verb, is also both a noun and a verb.

Except in English there actually is a degree of strong lexical distinction between verbs and nouns. Why? Because there's no pattern of how you can use either as the other, it depends only on the word itself.

Compare these three:

“I phoned a little boy”
“I parented a little boy”
“I scapegoated a little boy”

In the first one, “phoned” is short for “used a phone [to communicate] with”. The source noun denotes the instrument of the action.
In the second one, “parented” is short for “was a parent to”. The source noun denotes the subject of the action.
In the third one, “scapegoated” is short for “made a scapegoat out of”. The source noun denotes the object of the action.

The very same process has been used three times with three different results. The semantics of the source noun has influenced the formation of the resulting verb. Ergo, it's rather a new lexical entity than a morphological one; rather zero-derivation than zero-morpheme affixation.

malloc wrote:

zompist wrote:I'd like to expand on this, because I think conlangers probably still underestimate syntax. If you find yourself writing "A verb in the indicative or subjunctive appears with several arguments, all verbs in the attributive"... well, it's just kind of a weirdly jargonized way of saying "verbs have nouns as their subjects and objects". Conventions in language are there for convenience and understanding; don't burden your readers.

So you would consider it a case of the duck test: if this verbal form serves the same function as nouns, inflects for nominal attributes like possession, and so forth, it makes more sense simply to call it a noun.

Mostly, yeah. You should use the closest simple word unless it causes extra confusion.

Syntactic categories are defined, naturally enough, by their syntactic behavior. If something can occur as the argument of a verb, then it's a noun. Its morphological derivation is irrelevant, interesting though it is in the morphology section.

But what if the same lexeme can occur in both verbal and nominal roles depending on its inflection?

What of it? We're so used to this in English that English speakers may forget that null derivation isn't a process in all languages. Yet if you have good syntactic tests, they will provide a good answer for the forms in a given utterance.

(More interesting is the case where both inflections apply... e.g. a form has both aspect and plural morphemes. Again, it's not strange if a form is part-verb, part-noun. But there might still be useful rules that tell us which predominates in a particular instance.)

Consider the case system of Finnish, with its extensive collection of locative cases that serve adverbial functions. Would you consider nouns in those cases to have become adverbs?

As a syntax fan, I tend to think we should ban the word "adverb". So far as I know, it's easiest to understand Finnish cases as cases, but then I'm familiar with cases. If someone wasn't, I'd probably explain them by analogy with prepositions, not adverbs.

Pole, the wrote:“I phoned a little boy”
“I parented a little boy”
“I scapegoated a little boy”

In the first one, “phoned” is short for “used a phone [to communicate] with”. The source noun denotes the instrument of the action.
In the second one, “parented” is short for “was a parent to”. The source noun denotes the subject of the action.
In the third one, “scapegoated” is short for “made a scapegoat out of”. The source noun denotes the object of the action.

The very same process has been used three times with three different results. The semantics of the source noun has influenced the formation of the resulting verb. Ergo, it's rather a new lexical entity than a morphological one; rather zero-derivation than zero-morpheme affixation.

It's true that the verbalization can have various meanings, but that can be true of any derivation-- e.g. "a retiree" is someone who retires, an "employee" is someone who is employed. Or consider how -ize works in visualize, socialize, strategize, criticize.

So Haleza Grise wrote:

Frislander wrote: I'm sure there's a language somewhere which does handle it's kinship system with a set of verbs like that described above, but I need to look further into that.

Certain languages of Arnhem Land have "kinship verbs" - "X is mother to me;" "I am brother to Y" etc. They are interesting for several reasons, one of which is that they have transparent lexical derivations (e.g. Gurr-goni a-rrirrmi-rri-pu "he held him", ie. "he is his father")

Also the Iroquoian language Oneida (see page 99 of this PDF), closely related to Mohawk, which malloc named as an influence. It's likely that Mohawk itself also has kinship verbs, but I don't have a source at hand right now. The Iroquoian languages are very verby in general, but they clearly don't go as far as eliminating the N-V distinction.

Zompist wrote:At the same time, syntactic categories are fuzzy. There isn't a strict line between nouns and verbs; the line is already fudged by things like gerunds, participles, infinitives, and nominalizations. So rather than asking "is it a noun or a verb?" the better question may be "how nouny or verby is it?" For your 'attributives', can you do nouny things like pluralization, possessive, diminutive, replacement by personal pronouns, adding descriptive adjectives, topicalization? Do they resist verby things like tense, aspect, irrealis? Note that a particular form may have both nounlike and verblike features.

To me, this is the really interesting aspect of this topic. In my own conlang Hkətl’ohnim, which also has kinship verbs but also a separate lexical category of true nouns, the kinship verbs (and many other verbs that are used idiomatically with a nominal meaning) can do all of those nouny things listed by Zompist, although not always in the same way as other nominals; for instance pluralization is done by using a 3PL>1SG verbal prefix instead of a 3SG>1SG one, while true nouns [which incidentally also inflect for person] have a different set of number prefixes. But they can also do all of the usual verby things. They are morphologically identical to other finite verbs (except possibly where it doesn't make sense semantically), just used syntactically as nouns.

Here's a short overview how kinship verbs work in Hkətl’ohnim, using -pīkə- ‘be a brother of’:

Hepīkə.
həy-pīkə
3SG>1SG.NPST-brother
He is my brother.

Həłpīkə.
həł-pīkə
3SG>2.NPST-brother
He is your brother.

Həhpīkə.
həh-pīkə
1SG>3:DEICTIC.NPST-brother
I am his brother.

Nipīkə.
ni-pīkə
3PL>1SG.NPST-brother
They are my brothers. (pluralization)

tuči qa-hepīkə
Ø-tuči qa=həy-pīkə
3SG-dog of=3SG>1SG.NPST-brother
my brother's dog (possessive)

Hesērupīkə.
həy-<sēru>-pīkə
3SG>1SG.NPST-<baby>-brother
He is my little brother. (diminutive - idiomatically formed via noun incorporation for all kinship terms, using ‘baby’ even when the brother is already an adult, but distinct from the diminutive of true nouns, which use a suffix -li)

Hemətloh (tūn) hepīkə.
həy-mətloh (tu-ːn) həy-pīkə
3SG>1SG.NPST-talk.to (1SG-ACC) 3SG>1SG.NPST-brother
My brother is talking to me. (kinship verb used as subject of another verb; overt object pronoun unusual but grammatical)

Həhmətloh hepīkə (tu).
həh-mətloh həy-pīkə (tu)
1SG>3:DEICTIC.NPST-talk.to 3SG>1SG.NPST-brother (1SG)
I'm talking to my brother. (kinship verb used as object of another verb; overt subject pronoun unusual but grammatical - but note that kinship verbs cannot take case suffixes like pronouns and true nouns do!)

Həhmətloh (ūn) (tu).
həh-mətloh (u-ːn) (tu)
1SG>3:DEICTIC.NPST-talk.to (3SG-ACC) (1SG)
I'm talking to him. (replacement by personal pronouns)

Həhmətloh hepīkə hrusku.
həh-mətloh həy-pīkə h-rusku
1SG>3:DEICTIC.NPST-talk.to 3SG>1SG.NPST-brother 3SG.NPST-strong
I'm talking to my strong brother. (descriptive adjective added)

Nya hepīkə, həhmətloh.
nya həy-pīkə, həh-mətloh
TOP 3SG>1SG.NPST-brother, 1SG>3:DEICTIC.NPST-talk.to
As for my brother, I'm talking to him. (topicalization)

For the more verby things I'm using a less permanent relationship, -əłučiň- ‘be a spouse of’ because it makes more sense semantically, but it would be grammatical with ‘brother’ too:

Wīłučin.
wi-əłučiň
3SG>1SG.PST-spouse
She was my wife. (past tense)

Hēłučēnihk.
həy-əłučiň-ihk
3SG>1SG.NPST-spouse-PROSP
She will be my wife soon. (prospective aspect)

Hēłučēnim.
həy-əłučiň-īmə
3SG>1SG.NPST-spouse-RES
She has just become my wife. (resultative aspect)

Hēłučēma.
həy-əłučiň-ːma
3SG>1SG.NPST-spouse-POT
She might become my wife. (potential mood, irrealis)

Həłəłučēmu.
həł-əłučiň-ːmōr
3SG>2.NPST-spouse-obviously
She is obviously your wife. (inferred evidentiality)

These variants can of course be used in a nominal function too:

Hēmətloh hēłučēnihk.
həy-mətloh həy-əłučiň-ihk
3SG>1SG.NPST-talk.to 3SG>1SG.NPST-spouse-PROSP
My future wife is talking to me.

Həhmətloh həłəłučēmu.
həh-mətloh həł-əłučiň-ːmōr
1SG>3:DEICTIC.NPST-talk.to 3SG>2.NPST-spouse-obviously
I'm talking to the woman who is obviously your wife.

zompist wrote:

Pole, the wrote:“I phoned a little boy”
“I parented a little boy”
“I scapegoated a little boy”

In the first one, “phoned” is short for “used a phone [to communicate] with”. The source noun denotes the instrument of the action.
In the second one, “parented” is short for “was a parent to”. The source noun denotes the subject of the action.
In the third one, “scapegoated” is short for “made a scapegoat out of”. The source noun denotes the object of the action.

The very same process has been used three times with three different results. The semantics of the source noun has influenced the formation of the resulting verb. Ergo, it's rather a new lexical entity than a morphological one; rather zero-derivation than zero-morpheme affixation.

It's true that the verbalization can have various meanings, but that can be true of any derivation-- e.g. "a retiree" is someone who retires, an "employee" is someone who is employed. Or consider how -ize works in visualize, socialize, strategize, criticize.

That's why I am calling verbalization a derivational process in English, as opposed to an inflectional one. (If verbalization were inflectional then you could say: yeah, English has no verbs as a separate lexical class, which is kind of the point of every “verbless” language.)

Pole, the wrote:That's why I am calling verbalization a derivational process in English, as opposed to an inflectional one.

Ah, OK, I'm fine with that.

zompist wrote:(More interesting is the case where both inflections apply... e.g. a form has both aspect and plural morphemes. Again, it's not strange if a form is part-verb, part-noun. But there might still be useful rules that tell us which predominates in a particular instance.)

Well, the words marked as attributive verbs can still inflect for verbal categories, including polypersonal agreement even (which seems decidedly contrary to nouns). The "attributive" marking simply indicates that they serve as arguments rather than predicates. Since both ordinary verbs and "nouns" take pronominal suffixes indicating their subjects (or their person, animacy, and number in the case of "nouns"), they would both indicate plurality by using a plural subject pronoun.

As a syntax fan, I tend to think we should ban the word "adverb". So far as I know, it's easiest to understand Finnish cases as cases, but then I'm familiar with cases. If someone wasn't, I'd probably explain them by analogy with prepositions, not adverbs.

My point is that inflections do not generally change of the word to which they apply.

zompist wrote:Just as another test case, is "forget-me-not" a noun? My dictionary thinks so (and I agree). Is it "really" an imperative VP?

Not really an argument in the discussion, but that example reminded me of the dozen-or-so nouns in Bulgarian formed out of imperative + object, e.g.:

въртиопашка
rotate-2.SG.IMP-tail
flighty, thoughtless, frivolous woman

прескочикобила
jump_over-2.SG.IMP-mare
leapfrog

препъникамък
trip_over-2.SG.IMP-stone
tricky obstacle

Any other languages with such or similar feature?
And yes, the last one does literally translate to 'trip a stone (over sth.)!' instead of 'trip over a stone!'

Zju wrote:Not really an argument in the discussion, but that example reminded me of the dozen-or-so nouns in Bulgarian formed out of imperative + object, e.g.:

[…]

Any other languages with such or similar feature?

Do Polish ad hoc constructions like „szukajka” (search-IMP-NOMZ, “a thing that can be used to search for something”), „sprawdzajka” (check-IMP-NOMZ, “a thing that can be used to check something”) count?

(Google notes that even zero-nominalization can be used here, so an imperative verb form „szukaj” (search-IMP) can be turned to a noun „szukaj” (the “search” functionality).)

It's an interesting thing, but not quite the same.

Cedh wrote:Also the Iroquoian language Oneida (see page 99 of this PDF), closely related to Mohawk, which malloc named as an influence. It's likely that Mohawk itself also has kinship verbs, but I don't have a source at hand right now. The Iroquoian languages are very verby in general, but they clearly don't go as far as eliminating the N-V distinction.

Exactly. English treats relationships like "mother" and "friend" as nouns whereas my conlang treats them as transitive verbs whose arguments refer to the participants in those relationships. One goal of the "attributive" (or "nominal" if you consider that more accurate) form is ensuring symmetry between stems when they occur in verbal forms (like the indicative and imperative) and when they function as arguments of verbs.

Now it seems that many dispute whether the nominal and verbal forms of a given stem are really the same lexeme. I maintain that they are, since changing the stem from verbal to nominal changes the syntactic role of the word without changing its semantics or even eliminating its arguments (the nominal form of "she befriends me" would still have pronominal suffixes for 3rd person subject and 1st person object). Treating them as separate lexemes seems like treating nouns in different cases as separate lexemes.