[Lingtyp] agent nominalization
Alex Francois
francois at vjf.cnrs.fr
Fri Jan 8 19:03:31 UTC 2016
dear Eitan,
In *Mwotlap*, an Oceanic language of northern Vanuatu, deverbal
nominalisation normally takes the form of reduplication.
1) *Event nominalisation*
Reduplication, for example, is what creates an "action noun" (*nomen
actionis*, referring to the event or State of affairs):
- *mat* 'die' => stem for deverbal noun *matmat*
--> citation form for this noun *na-matmat *'death'
(Note: Because the output stem is a noun, it takes the article prefix
*nV-* which forms part of a noun's citation form. Yet importantly, this
article *nV- *is not what carries out the derivation V>N, which is only
done by reduplication. *Matmat* behaves morphologically like any other
noun stem.)
- *dēm* 'think' => stem for deverbal noun
*dēmdēm *--> citation form for this noun *nē-dēmdēm *'thought, idea, mind'
If the verb is transitive and its patient is semantically relevant to the
meaning of the deverbal noun, then the patient noun is encoded as an N2
modifier to the deverbal N1 noun.
(a modifying noun takes the form of a bare noun, deprived of its article,
following the head noun — hence: <*nV*- N1 N2>):
- *galeg n-ēm̄* 'make a house'
--> *na-gagaleg ēm̄ *'the building of a house; house construction'
- *wēl na-lqōvēn* 'buy a woman'
--> *nē-wēlwēl lōqōvēn *'the purchase of a woman = marriage'
In these cases, the end result is both a case of derivation (through
reduplication) and of compounding (cf. the <N1 N2> structure).
2) *Agent nominalisation*
(*nomen actoris*, referring to the agent of an action)
a) *simple nominalisation*
One strategy for deriving actor nouns involves the same pattern as for
action nouns. Predictably, this sometimes results in nouns that are
polysemous between *nomen actionis* and *nomen actoris*:
- *vatgo* 'teach'
=> na-*vatvatgo* (1) '(art of) teaching, instruction'
(2) 'teacher'
- *tow n-eh* 'compose a song'
=> no-*towtow eh*
* <ART-NMLZR~compose song> *(1) 'song composition; poetry'; (2)
'song composer; poet'
- *tēy nē-bē* 'hold water'
=> nē-*tēytēy bē*
<ART-NMLZR~hold water>
(1) '(the art of) medicine'; (2) 'doctor'
- *tan̄ n-et* lit. 'touch a person' -> 'massage'
=> na-*tan̄tan̄ et*
* <ART-NMLZR~touch person> *(1) '(the art of) massage'; 'midwifery';
(2) 'masseuse; midwife'
b) *periphrastic nominalisation*
Another common strategy in producing *nomen actoris* is to use a head noun
*n-et* meaning 'person' followed by a *nomen actionis*:
{ *n-et* (*N1 N2*) }
{ *ART* person (*N1 N2*) }
- *vēgēl* 'bewitch'
=> *n-et vēgēpgēl*
<ART-person NMLZR~bewitch>
'sorcerer' [liter. 'person of bewitching']
- *tēy *nē-*tēnge* 'hold leaves'
=> *n-et tēytēy tēnge*
<ART-person NMLZR~hold leaf>
'healer'
Syntactically, the latter construction has 3 nouns in a modifying
relationship { N0 (N1 N2) }
'healer' is literally '(the) person of (the) holding of (the) leaves'
The same syntactic structure underlies nouns of instrument (where N0 is
simply 'thing' instead of 'person'):
- *lep *no-*totgal* 'take a picture'
=> *n-age leplep totgal*
<ART-thing NMLZR~take picture>
'camera'
Note that this structure is distinct from relative clauses or other forms
of subordination. It is a case of phrasal compounding — lexical compounds
larger than a phonological word.
The noun *n-et* is synchronically productive, being the hyperonym for human
referents. It is also used for equivalents of our indefinite pronouns (e.g.
'someone', 'nobody'…), or in relative clauses ('the *one* who'…,
'whoever'...), etc. I guess you could say that in Mwotlap, agent
nominalisation has a noun 'person' as its "historical origin" – but this is
in fact still the case in synchrony, with no particular trace of
morphologisation or phonological erosion.
c) *periphrasis with preposition*
Finally, the most productive strategy is also syntactically the most
elaborate.
As in (b) above, it involves a head noun (N0) modified by a *nomen actionis*;
except this time, the syntactic relation between N0 and (N1 N2) is encoded
by a preposition *bV-*, which can often be glossed "for". Thus the
structure is
{ *n-et* *bV-* (*N1 N2*) }
{ *ART* person *for* (*N1 N2*) }
- *lak* 'dance'
=> *n-et ba-laklak*
<ART-person for-NMLZR~dance>
'dancer'
- *lam̄ *nē-*vētōy* 'beat the drum'
=> *n-et ba-lam̄lam̄ vētōy*
<ART-person for-NMLZR~beat drum>
'drummer'
*N-et* 'person' is only singular; other numbers involve suppletion with a
small paradigm of morphemes labelled Human Collectives (dual *yoge*, trial
*tēlge*, plural* ige*):
- *lam̄ *nē-*vētōy* 'beat the drum'
=> *yoge ba-lam̄lam̄ vētōy*
<HUM:dual for-NMLZR~beat drum>
'the two drummers'
- => *ige ba-lam̄lam̄ vētōy*
<HUM:pl for-NMLZR~beat drum>
'(the) drummers'
Note that the structure with preposition has parallels with genuine
(non-derived) nouns:
- n-*ih* 'a bow'
= *n-et b*-*ih*
* <ART-person for-bow> * 'archer'
- nu-*skul* 'school'
= *ige bu*-*skul*
* <HUM:pl for-school> * 'schoolers'
Finally, my Mwotlap corpus sometimes shows alternation between two possible
structures for creating agent nouns (with semantic nuances I won't detail
here):
- *wos* 'to nail'
=> *n-et woswos*
* <ART-person NMLZR~nail> * 'builder, handyman, DIYer' (a habitual
property)
= *n-et bo*-*woswos*
* <ART-person for-NMLZR~nail> * 'the builder' (of a particular house)
- *tow n-eh* 'compose a song'
=> *no-**towtow eh*
* <ART-NMLZR~compose song> * (1) 'song composition; poetry';
(2) 'a song composer; a poet'
=> *n-et bo*-*towtow eh*
* <ART-person for-NMLZR~compose song> * 'the composer (of a
particular song)'
The latter contrast is reminiscent of the one that Benveniste reconstructed
for Proto Indo-European, in his 1948 study of nominal derivation in PIE (cf.
1974 [1967]:155):
- *dH3-*tér*-
(>Gr. δοτήρ) ‘a giver [a habitual property]’
- *déH3-*tor*-
(>Gr. δώτωρ) ‘the giver [in a particular situation]’
Reference:
- Benveniste, Emile. 1948. *Noms d'agent et noms d'action en
indo-européen*. Paris: Librairie d'Amérique et d'Orient.
- —— 1974 [1967]. Fondements syntaxiques de la composition nominale.
In *Problèmes de Linguistique Générale*, vol.2. Collection Tel.
Paris: Gallimard, 145–162.
For Mwotlap, you can find more examples and explanations in my thesis:
- François, Alexandre. 2001. *Contraintes de structures et liberté dans
l'organisation du discours: Une description du mwotlap, langue océanienne
du Vanuatu*. Doctoral thesis in Linguistics. Paris: Université Paris-IV
Sorbonne. [open access
<http://alex.francois.free.fr/AFpub_books_e.htm#02>]
=> see pp.*227-236*, and also 181-183.
with best wishes,
Alex
_________
Alex François
Directeur, LACITO-CNRS <http://lacito.vjf.cnrs.fr/membres/francois.htm>,
France
Australian National University
<https://researchers.anu.edu.au/researchers/francois-a>, Canberra
Academia page <https://cnrs.academia.edu/AlexFran%C3%A7ois>
Personal homepage <http://alex.francois.free.fr/>
__________________
2016-01-06 11:07 GMT+01:00 Eitan Grossman <eitan.grossman at mail.huji.ac.il>:
> Dear all,
>
> I am writing to ask a question about 'agent'* nominalizations across
> languages. I am interested in agent nominalizers that do or don't have
> known diachronic sources, in the attempt to understand which diachronic
> pathways are attested (and hopefully, their relative frequency/rarity). For
> example, some languages have:
>
> (a) bound morphemes whose diachronic source is clearly identifiable,
> whether lexical (Japanese -nin or -sya 'person; Khwe and Meskwaki are
> similar, or Japanese -te 'hand') or grammatical (Serbo-Croatian -l(o) from
> an original instrumental meaning, perhaps similarly for Afroasiatic m-).
> (b) bound morphemes whose diachronic source may be mysterious or
> reconstructible as such to the proto-language (Quechuan -q?,
> Malay-Indonesian peng-/pe-?).
> (c) free morphemes whose diachronic source is clearly identifiable
> (Ponoapean olen ''man of')
> (d) more complex constructions involving the reduction of modifier clauses
> of some sort (Coptic ref- < ultimately from 'person who verbs')
> (e) rarer morphosyntactic alternations, like reduplication of the initial
> syllable (Hadze, Serer), vowel length (Akan), vowel raising (+breathiness)
> (Nuer)
> (f) no such nominalizer mentioned, or explicitly mentioned that there is
> no dedicated agent noun construction. In some languages, ad hoc formation
> via relatives is the only (Tlapanec), main, or a supplementary strategy
> (e.g., Indonesian relativizer yang).
> (g) zero conversion
>
> There is nice paper by Luschuetzky & Rainer in STUF 2011, but it deals
> almost exclusively with affixes and only rarely mentions diachronic
> information.
>
> From a *very* preliminary survey of grammars, it looks like the origin of
> agent nominalizers is often pretty obscure, and the shortest and most bound
> morphemes look to be very old, quite expectedly. Identifiable lexical
> sources seem to converge around 'person, thing' or body parts. Reduction of
> complex constructions to an affix seems to be rare but attested.
>
> *So, here's the question: in your languages, is the diachronic source of
> agent nominalizers identifiable? * I'd be grateful for any information
> you might be willing to share!
>
> Best,
> Eitan
>
> *Disclaimer: even though this is a common term, most languages I've seen
> don't single out the semantic role of agent, and this is often noted in
> theoretical discussions. Also, such nominalizations don't have to be
> derivational or even 'morphological.'
>
>
>
> Eitan Grossman
> Lecturer, Department of Linguistics/School of Language Sciences
> Hebrew University of Jerusalem
> Tel: +972 2 588 3809
> Fax: +972 2 588 1224
>
> _______________________________________________
> Lingtyp mailing list
> Lingtyp at listserv.linguistlist.org
> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20160108/831732d2/attachment.htm>
More information about the Lingtyp
mailing list