Ludmila Veselovská
5. Countability Constraint on Oblique mnoha “a lot OBL ” + [NP]
As demonstrated above in the contrast between (4a) and (4b), the Q mnoh-a “a lot” is not unacceptable in every oblique context. It appears to be ungrammatical only in those contexts where the following qN is uncountable.9
The next table shows the number of examples of the Q mnoh-a “a lotOBL” in a variety of oblique case contexts following prepositions as they were found in the Syn2015 corpus.10
8 The corpus is a part of the Czech national corpus, and it has 121,667,413 positions—it is therefore a quite reliable source of modern Czech data. I am indebted to Monika Pitnerová for her help with the corpus search. Without her the data would not be complete.
9 In other words, while in NOM/ ACC form the Czech mnoho “a lot” may be translated like both “much/many,” in oblique contexts (i.e., with the inflection mnoh-a) the reading “much” is not available.
10 The search was done concentrating on the prepositional obliques, but similar examples could be formed with verbs selecting oblique noun phrases, as the example below shows.
A NUMBER CONSTRAINT OF CZECH QUANTIFIED NOMINALS
44
(7) The variety of Oblique contexts for the Q mnoh-a “a lotOBL” after P (b) Stěžoval si na mnoh-o / mál-o / *pět-Ø žlutého oleje
complained REFL about many / fewNOM? / fiveNOM? [yellow oil]MS.GEN
“He complained about a lot / little / *five yellow oil.”
(7) The variety of Oblique contexts for the Q mnoh-a “a lotOBL” after P
Case Tokens ... Most frequent P Tokens
GENITIVE total 1,729 from/of z + Q + NGEN 1,174
DATIVE total 155 to k + Q + NDAT 128
LOCATIVE total 4,277 in v(e) + Q + NLOC 2,798
INSTRUMENTAL total 1,154 with s(e) + Q + NINS 601
total 7,315
(10)
PP P0 DP
[uQ]
D0 NumP
D0 [+Q]
[uQ] Num NP [+Q]
AP nP
NBR + √...
(11) when [+COUNT]
(a) s mnoh-a / ?mál-o/*-a (žlutými) brouky (b) s *mnoh-a / *mál-o/*-a (žlutým) olejem
with a lot of / few-little yellow bugs/oil
All the above 7,315 examples were checked manually and all following nominals were both countable and [+PL].
The examples in (8) show some of the thousands of oblique forms of the Qs found in the corpus: a DAT assigned by a verb in (a), a LOC assigned by a preposition in (b) and also a post-nominal GEN in (c). In line with the generalization formulated here, all the examples found with the Q mnoh-a “a lotOBL” were countable and [+PL].
(8) (a) za války pomohl mnoha Čechům
during war helped [+DAT] many CzechsPL.DAT
“during the war he helped many Czechs”
(b) v mnoha případech stát reagoval opožděně in [+LOC] many casesPL.LOC state reacted late
“in many cases the state reacted late”
(c) absorbuje vlivy mnoha kultur
absorbs influences [+GEN] many culturesPL.GEN
“it absorbs the influences of many cultures”
The corpus search thus confirms the existence of a constraint which in Czech requires that the quantified nominal complex following the two Qs in oblique contexts is count-able [+PLURAL].
I propose that this constraint can be explained using a small modification of the deriva-tion of the structures containing Qs as in Pesetsky (2013). The steps are described in (9) in the paragraphs (a)–(i) and illustrated for scheme (10), which illustrates examples (11) = (4).
(i) To pomohlo mnoh-a našim lidem / *naší energii
it helped REFL[+DAT] manyOBL our peoplePL.DAT / *our energySG.DAT
“It helped many of our people.”
LUDMILA VESELOVSKÁ
(9) Derivation of the structures with Qs (cf. Pesetsky2013)
(a) Each nominal category is assigned a “primeval” GEN by some nominal categorial head. I assume the relevant categorial head is related to the Number head (NBR in [10]), which in Czech is merged very low and forms a part of the noun stem.
(b) After the merge of NBR, the projected N complex carries the primeval GEN which gets spread to its modifiers via NP-internal agreement.
(c) To project a structure with a Q, some Q (e.g., mnoho/málo/pět “a lot / little-few/
five” [5 & up]) merges in the high Num head position.
(d) Then an abstract of overt D head merges, which has the [unQ] feature.
(e) To satisfy the [unQ] feature of D, the closest [Q] (i.e., the Num) moves to D.
It “undermerges” (i.e., right adjoins) to D and becomes its complement.
(f) The D category assigns a D-case (NOM) to its complement, i.e., to [Q]/Num.
In standard situations the D complement is constituted by the following (quantified) NP.
In these structures with the undermerged Q, however, the complement of D is the right adjoined Q itself and therefore it is the Q that receives the D-case.
(b) Stěžoval si na mnoh-o / mál-o / *pět-Ø žlutého oleje complained REFL about many / fewNOM? / fiveNOM? [yellow oil]MS.GEN
“He complained about a lot / little / *five yellow oil.”
(7) The variety of Oblique contexts for the Q mnoh-a “a lotOBL” after P
Case Tokens ... Most frequent P Tokens
GENITIVE total 1,729 from/of z + Q + NGEN 1,174
DATIVE total 155 to k + Q + NDAT 128
LOCATIVE total 4,277 in v(e) + Q + NLOC 2,798
INSTRUMENTAL total 1,154 with s(e) + Q + NINS 601
total 7,315
(10)
PP P0 DP
[uQ]
D0 NumP
D0 [+Q]
[uQ] Num NP [+Q]
AP nP
NBR + √...
(11) when [+COUNT]
(a) s mnoh-a / ?mál-o/*-a (žlutými) brouky (b) s *mnoh-a / *mál-o/*-a (žlutým) olejem
with a lot of / few-little yellow bugs/oil
A NUMBER CONSTRAINT OF CZECH QUANTIFIED NOMINALS
46
In a NOM context the NP retains Pesetsky’s “primeval GEN.” Although the whole nominal complex can later on appear in some other (later) case assigning contexts, the phase border makes the lower NP domain inaccessible for any other case marking.
In ACC contexts I accept Pesetsky’s (2013) proposal that the V case (=ACC) is assigned (in Czech) only to a disjunction of [+FEM/+ANIM/+PRON/not -SG]. Because the Q is none of the above, the NP retains the primeval GEN which gets freezed by a phase border as in (g) above.
In Oblique contexts:
• P merges with DP/QP, which checks the [Q] feature,
• P assigns P-case to the DP/QP,
• Pesetsky’s “One suffix rule” allows only the last Case (=P case) to be realized.
The above derivation (taken from Pesetsky [2013]) explains several points:
• Why the partitive (primeval) GEN disappears in Oblique contexts. It does so because P-Case overrides it.
• Why Qs cannot appear without the GEN cl.: This is because the [+Q] of D requires a complement.
• Why Subject-Verb agreement can only be default (3SN) with QP subjects:
This is because the φ of D are deficient, given that they make the features of the lower domain inaccessible (they are blocked by the undermerge of Q).
However, the description of the derivation above, which follows Pesetsky’s proposals, does not explain why the Qs mnoho/málo “a lot / few-little” require [+COUNT/+PL] in Oblique contexts. At the same time the above description has not pointed out sufficiently one aspect of Pesetskyʼs (2013) analysis. The author assumes that Russian paucals (the low cardinals 2–4) represent primitive features of NBR (a lower Number head) which move to the higher Num head (and then to D). In Czech, the paucals exhibit patterns, as in (2) above, and therefore there is no reason to assume the position of NBR in Czech is separate from the nominal stem. NBR forms a part of the φ features of the NP and as evidenced by the default agreement, those features are not accessible on the DP level when the QP is in the subject position.
On the other hand, based on the constraint on countability, I propose that the merge of P, which requires a [Q] feature, triggers the movement of the NBR features to the level of the DP. Crucially, this movement happens only when those NBR features are marked, i.e., countable and [-SG].
If the above proposals are on the right track, I conclude that the so-called “oblique”
morphology on the Czech Qs, i.e., -a/-i mentioned in the discussion of (5), does not
LUDMILA VESELOVSKÁ
realise any case at all, but it instead realises PLURAL, the marked feature (i.e., [-SG]) of NBR present on D/Num.
6. Paradigmatic Gap with the Q