David's Hierarchy of Omissibility of Subjects

Charles Randriamasimanana C.H.Rand at MASSEY.AC.NZ
Tue Nov 24 23:34:14 UTC 1998

David's Hierarchy:

'verbal predicates >
nominal predicates with copula >
nominal predicates without copula

What this says is:

(1) if a language has all three constructions, and if it can omit
subjects at any point on the hierarchy (in a given context) then it can
omit subjects at all higher points.

(2) among "pure" languages (those with either obligatory copula in all
contexts or no copula at all), omission of subject will be more frequent
in languages with copula than in languages without.'

I agree with the general thrust of David' remark, but would like to go a
bit further than he does and look into the first part of his claim related
to 'verbal predicates'* and Null Subjects.

As I show in 'Evidence for Null Subjects in Malagasy' 1998a. Papers from
the Fifth Annual Meeting of the Southest Asian Linguistics Society.
Shobhana L. Chelliah and Willem J. de Reuse, eds. Arizona State University,
Tempe, Arizona. Pp 303-315, the distribution of Null Subjects in a language
like Malagasy is partially related to

(i) The absence of the element AGR(eement) under the Inflections for
Tense/Aspect --this being a simplification of the system of the language,
but will do as a first approximation:

(1)	Ireto m-ianatra ireto ny mpianatra.
	these pres-study these the student
	'The students are studying.'

(2)	Ity m-ianatra ity ny mpianatra.
	this pres-study this the student
	'The student is studying'

where the presence of AGR 'ireto' 'this-plural' or 'ity' 'this-singular'
imposes the relevant semantic interpretation on the subject NP 'ny
mpianatra'. If 'ireto' shows up, 'ny mpianatra' = 'the studentS', i.e.
plural obligatorily; but if 'ity' shows up, then 'ny mpianatra' = 'the
student' (singular) mandatorily.

The main point here is that you canNOT leave out the subject NP 'ny
mpianatra' in either (1) or (2) simply because AGR is present in the
sentence. The explanation I propose is that in Malagasy AGRs as well as a
number of tense particles (especially in the passive voice) have Case to
assign, hence the obligatory presence of the subject NP when AGR is there.
In a sense, this is a very welcome development as far as Chomsky's
Minimalist framework is concerned (specifically, his feature-checking

Now AGRs like 'ireto', 'ity', etc can only be strictly subcategorized by
Control predicates (verbs describing some kind of Deliberate Activity),
whereas Stative verbs usually cannot be accompanied by such AGRs at all (At
least, not in the intended interpretation). See the article above for
further detail and illustrative examples.

(ii) The above means that with a stative kind of predicate, it is usually
the case that the subject can be omitted --even in the first utterance of
an adjacent pair. For example, the following is fine:

(3)	Tsy mbola teraka hono.
	Neg yet  born   is-said
	'They say [ (s)he ] is not yet born who...'

(3) was published a few years ago in a Malagasy language daily [ The exact
reference is in the article noted above]. Note that 'teraka' is a root verb
with a stative kind of meaning and that the literal equivalent of [(s)he]
has been omitted in Malagasy.

Last but not least, the Malagasy phenomenon of Null Subjects does not
appear to be related to what is commonly referred to as 'pro-drop' in
Romance languages: For one thing, Malagasy does not show the so-called
'rich morphology', which would normally license the omission of the subject
NP in well-known languages such as Spanish or Italian.



*In the article mentioned above, I have examples illustrating the third
part of David's  statement, i.e. the possibility of omitting the subject NP
in an equative type of clause-- Malagasy being a language which does not
possess the literal equivalent of the English copula 'be'.

Charles Randriamasimanana, PhD in Linguistics
Linguistics, School of Language Studies
OMB 3.19
Massey University
Private Bag 11-222
Palmerston North
New Zealand

Telephone: (06)-356-9099, Extension 7059
Personal fax: (06) 359-3989

More information about the Lingtyp mailing list