Origins / development of the Welsh verbnoun?

[Jonlang]

I was unsure where to put this, so I apologise if it would be better suited elsewhere.

I am still working on my Brythonic-ish conlang, but I don't know how to go about evolving a Welsh-like verbnoun. So, I figured I may as well try to find out how the Welsh VN evolved so I can learn something while I conlang. There seems to be almost nothing freely available online. I keep getting pointed to the paper The historical morphology of the Welsh verbal noun (Schumacher, 2000) but I don't have academic access to read it and it seems to be out of print everywhere.

[Linguoboy]

According to Jeffers ("Remarks on Indo-European Infinitives." Language 51, no. 1 (1975): 133-48), Celtic is something of a free-for-fall compared to other Indo-European languages:

In Celtic...there is virtually no limitation on what stem types may occur. The other IE languages show a systematic extension of some (or some small group of) nominal suffix(es) in the creation of a new morphological category, the infinitive; but Celtic at best appears merely to have developed a set of verb/noun associations, which need not even incorporate an element of formal identity (e.g., serc is the verbal substantive to caraid 'loves').

My impression of Welsh is that, as with plural formations, there's a strong tendency toward analogical extension. But whereas with plural endings the extensions generally follow semantic classifications (e.g. -od for animals), with verb-nouns the motivations are generally phonological (e.g. -u for verbs with a, ae, e, or y in the last syllable of the stem). The main exception is -io, which seems to be the default for verbs borrowed from English regardless of their phonological shape.

[Jonlang]

According to Jeffers ("Remarks on Indo-European Infinitives." Language 51, no. 1 (1975): 133-48), Celtic is something of a free-for-fall compared to other Indo-European languages:

In Celtic...there is virtually no limitation on what stem types may occur. The other IE languages show a systematic extension of some (or some small group of) nominal suffix(es) in the creation of a new morphological category, the infinitive; but Celtic at best appears merely to have developed a set of verb/noun associations, which need not even incorporate an element of formal identity (e.g., serc is the verbal substantive to caraid 'loves').

My impression of Welsh is that, as with plural formations, there's a strong tendency toward analogical extension. But whereas with plural endings the extensions generally follow semantic classifications (e.g. -od for animals), with verb-nouns the motivations are generally phonological (e.g. -u for verbs with a, ae, e, or y in the last syllable of the stem). The main exception is -io, which seems to be the default for verbs borrowed from English regardless of their phonological shape.

Though I very much appreciate the reply, I'm looking more for where the VNs came from, as a thing. They do jobs of nouns, verbs, and participles without changing form. Did different forms all collapse in and create VNs? I mean, I know how to use them (mostly) but where along the way from PIE to Welsh did the VN pop up?

[vegfarandi]

Never heard of this, sounds fascinating.

Have you tried finding an email address for the writer of the paper and asking for a copy? Most scholars in my experience are willing to share their papers if you ask nicely.

[Linguoboy]

Though I very much appreciate the reply, I'm looking more for where the VNs came from, as a thing. They do jobs of nouns, verbs, and participles without changing form. Did different forms all collapse in and create VNs? I mean, I know how to use them (mostly) but where along the way from PIE to Welsh did the VN pop up?

They're reconstructed for Proto-Celtic (all other living Celtic languages use them too) so they go back quite a ways. I read the quote I shared as saying that they originally developed from PIE infinitives but were subject to a broader range of formations and usages than infinitives in most other branches of PIE. In any case, with the origins that far back, any evolutionary account would be almost entirely speculative anyway.

Although the usage of Welsh verb-nouns has some interesting peculiarities, I don't know that it's all that exceptional. After all, in English, short infinitives can also behave nominally, verbally, and adjectivially without changing form. And other Indo-European languages (e.g. German, Cajun French) use an infinitive rather than a participle to form the progressive so it's not hard to see how this construction could have developed in the Celtic languages.

[Linguoboy]

Although I wasn't able to access Schumacher's article, I did find a dissertation by Randall Clark Gordon called "Derivational Morphology of the Early Irish Verbal Noun" which might answer some of your questions. He only mentions the Welsh verb-noun in passing, but given the clear developmental parallels in both languages, much of what he says is probably valid for Welsh as well. He chose to look at Irish because it's attested earlier so he thought that it would yield more clues to how this particular word class originated.

He starts off by establishing a typology of verbal noun formations. In order of frequency among the verbal roots in his corpus, they are:

Type 1. Deradical Formations. (55%) [most strong verbs]
Type 2. Deverbative Formations. (31%) [most weak verbs]
Type 3. Source Nouns. (9%) [denominative verbs]
Type 4. Deadjectival/Denominative Formations. (3%) [deadjectival verbs]
Type 5. Suppletive Verbal Nouns. (2%).

In Types 3 and 4, the verb is a secondary derivation from a noun or adjective and the corresponding noun or adjective is then used as a verbal noun.

Gordon considers Type 1 to be the oldest type. Among the suffixes attached to PIE roots to derive verbal nouns are: *-o- [n.], *-io- [n.], *-to- [n.,m.], *-tlo- [n.], *-ā- [f.], *-iā- [f.], *-ma- [f.], *-na- [f.], *-ta-[f.], *-agla- [f.], *-axta- [f.], *-i- [n., also f.?], *-ni- [f.], *-ti- [f.], *-tion- [f.], *-u- [m.], *-mu- [m.], *-tu- [m.], and *-(s)men- [n.], so a pretty good range of PIE nominal/adjectival suffixes is represented. Some of these are also attached to verb stems to form Type 2 derivations, primarily *-tu-. In fact, more than 80% of Type 2 verbal nouns are derived by means of this suffix, which becomes ever more common over time and displaces earlier Type 1 formations.

This was what I was getting at earlier when I talked about the power of analogy: once certain distinctive endings become primarily associated with a particular morphosyntactic category (in this case, verbal noun stems), they tend to spread to other stems in the same category. Obviously, the details differ for Welsh--it's not a descendant of *-tu- that spread widely but but descendants of vocalic suffixes like *-o-, *-io-, *-i-, *-iā-, *-ā-, etc.--but the phenomenon is the same.

You can actually observe this process in all branches of PIE. Vedic Sanskrit, for instance, has over a dozen different ways of forming infinitives, but Classical Sanskrit has only one ending, -tum. Gordon argues that the process of infinitive formation seen in other branches (by which certain inflected forms of verbal nouns or adjectives became fossilised and generalised) was never really completed in Celtic. It got as far as associating a particular abstract verbal noun with each verb stem and systematically using these in certain constructions with verbal characteristics but these formations were never fully integrated into the verbal system in the way we see happen elsewhere.

[Hallow XIII]

According to Jeffers ("Remarks on Indo-European Infinitives." Language 51, no. 1 (1975): 133-48), Celtic is something of a free-for-fall compared to other Indo-European languages:

In Celtic...there is virtually no limitation on what stem types may occur. The other IE languages show a systematic extension of some (or some small group of) nominal suffix(es) in the creation of a new morphological category, the infinitive; but Celtic at best appears merely to have developed a set of verb/noun associations, which need not even incorporate an element of formal identity (e.g., serc is the verbal substantive to caraid 'loves').

My impression of Welsh is that, as with plural formations, there's a strong tendency toward analogical extension. But whereas with plural endings the extensions generally follow semantic classifications (e.g. -od for animals), with verb-nouns the motivations are generally phonological (e.g. -u for verbs with a, ae, e, or y in the last syllable of the stem). The main exception is -io, which seems to be the default for verbs borrowed from English regardless of their phonological shape.

Though I very much appreciate the reply, I'm looking more for where the VNs came from, as a thing. They do jobs of nouns, verbs, and participles without changing form. Did different forms all collapse in and create VNs? I mean, I know how to use them (mostly) but where along the way from PIE to Welsh did the VN pop up?

If you're just interested in the ways to generate such polyvalent forms, the case of English -ing may prove instructive.

[hwhatting]

Though I very much appreciate the reply, I'm looking more for where the VNs came from, as a thing. They do jobs of nouns, verbs, and participles without changing form. Did different forms all collapse in and create VNs? I mean, I know how to use them (mostly) but where along the way from PIE to Welsh did the VN pop up?

They're reconstructed for Proto-Celtic (all other living Celtic languages use them too) so they go back quite a ways. I read the quote I shared as saying that they originally developed from PIE infinitives but were subject to a broader range of formations and usages than infinitives in most other branches of PIE. In any case, with the origins that far back, any evolutionary account would be almost entirely speculative anyway.

You can actually observe this process in all branches of PIE. Vedic Sanskrit, for instance, has over a dozen different ways of forming infinitives, but Classical Sanskrit has only one ending, -tum. Gordon argues that the process of infinitive formation seen in other branches (by which certain inflected forms of verbal nouns or adjectives became fossilised and generalised) was never really completed in Celtic. It got as far as associating a particular abstract verbal noun with each verb stem and systematically using these in certain constructions with verbal characteristics but these formations were never fully integrated into the verbal system in the way we see happen elsewhere.

The second account is what I know. PIE didn't have infinitives, it had a large number of different verbal noun formations, and Celtic is conservative both in not having infinitives and in not reducing the formations of verbal nouns to one or a small number of prevalent types.