Question on agentive nominalizations

[Bernhard Waelchli]

Dear Mark

Since you emphasize the parallelism between agent and action nominals, 
it might be worth looking into all the languages described as Sentential 
in Koptjevskaja-Tamm’s (1993) well-known typological study on 
nominalizations and her survey in WALS (2005), where the following 
languages are classified as having sentential action nominal 
constructions ("its arguments are signalled in the same way as 
verb-argument relations in finite clauses"): Akhvakh Apurinã Archi Avar 
Barasano Basque Bezhta Burushaski Canela-Krahô Chamalal Godoberi Ingush 
Kobon Korean Lak Lavukaleve Lezgian Nambikuára Nepali Nivkh Pirahã 
Quechua (Imbabura) Tamil Yagua Yupik (Sirenik). If there is a general 
parallelism between action nominals and actor nominals in gross syntax 
(nobody has really tested this cross-linguistically, as far as I know), 
these languages are expected to exhibit sentential actor nominals.

I am not sure whether I understand your question correctly, it would be 
easier to decide if the question was formulated in the form of a 
question or as a hypothetical universal. In case you are looking for 
languages where nomina agentis can occur in constructions with similar 
syntax as finite verb forms under certain conditions, here is a further 
example in addition to Sanskrit.

Livonian/Liv (Uralic, Finno-Ugric, Finnic), a moribund language of 
Latvia, uses nomina agentis as evidential present (for example, to 
express reported speech), which can occur both in independent clauses 
and in clauses introduced by a complementizer. The use of cases is the 
same as with finite verb forms.

läp:shi piet:i:sti, ku so:rlist tul’:id un i’edijid ma’g lo’igi un 
vut:a:jid vuoza: ul:z un pan’:it ko:ma si’z:el
children.PART:PL cheat.PST:3PL, that island.people.PL:NOM come:NA:PL and 
cut.NA:PL belly.NOM/ACC torn[ADV] and take.NA:PL meat.NOM/ACC out and 
put.NA:PL turbot.NOM/ACC in
'The (Livonian) children were lied to that the Saaremaa-people would 
come and cut their bellies and take the flesh out and put a turbot in' 
(Kettunen 1938: LXX).
(I am sorry for not being able to render the diacritics properly; 
Livonian has a much more complicated phonology than the closely related 
Finnish language. PART partitive, NA nomen agentis, nominative and 
accusative happen to have the same form for a large number of nouns)

However, the Livonian nomen agentis has also substituted the present 
active participle in some functions (Kettunen 1938: LXVIII). So, if you 
count this as an example of a participle it documents at least a form 
that has grammaticalized from an action nominal.

Anyway, the grammaticalization paths of actor nominals (Future in 
Sanskrit, Evidential in Livonian) might deserve more attention in 
grammaticalization studies. In Mordvin (Finno-Ugric) the 3rd person 
singular present has the same form as (the short and older form of) the 
actor nominal. According to Bubrix (1953: 98), the Mordvin second past 
form (intermediate between habitual past and imperfect in function) 
derives from the actor nominal plus the first past form of the verb 'to 
be'. There must be many more cases where some actor nominal forms 
grammaticalize to some finite verb constructions.

Kettunen, Lauri. 1938. Livisches Wörterbuch mit grammatischer 
Einleitung. Helsinki: Suomalais-ugrilainen seura.
Bubrix, D. V. 1953. Istoricheskaja grammatika èrzjanskogo jazyka. 
Saransk: Mordovskoe knizhnoe izdatel'stvo.

Best wishes,
Bernhard Waelchli