Empirical Issues in Syntax and Semantics 9 (EISS 9 ... - CSSP - CNRS
Empirical Issues in Syntax and Semantics 9 (EISS 9 ... - CSSP - CNRS
Empirical Issues in Syntax and Semantics 9 (EISS 9 ... - CSSP - CNRS
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
. N ⇔ [ vP v INCH [ AktP Akt ]]<br />
c. TYR ⇔ [ vP v TR [ AktP Akt CAUSE ]], where CAUSE ∈ {I-CAUSE, G-CAUSE}<br />
S<strong>in</strong>ce out of the three items <strong>in</strong> (37), only LA is specified for the VP node, it is the only option<br />
for spell<strong>in</strong>g out VP. In effect, LA always surfaces <strong>in</strong> denom<strong>in</strong>al verbs regardless of what features<br />
v <strong>and</strong> Akt bear. The spell-out of other components depends on their featural content.<br />
Class 1 verbs are based on the I-CAUSE relation. Depend<strong>in</strong>g on the TR/INCH feature on v,<br />
two configurations are theoretically available. In (48), the whole structure is spelled out by LA.<br />
Other competitors (N for Akt I-CAUSE <strong>and</strong> TYR for v TR <strong>and</strong> Akt C-CAUSE ) lose to LA accord<strong>in</strong>g to (34),<br />
s<strong>in</strong>ce LA is the biggest match. In (39), N is the only option for spell<strong>in</strong>g out v INCH . It competes<br />
with LA for Akt CAUSE , but loses the competition due to (35): unlike N, LA bears the I-CAUSE<br />
feature <strong>and</strong> is thus more ‘specific’.<br />
⏟<br />
[ vP<br />
… v TR<br />
[ AktP<br />
… Akt I-CAUSE<br />
[ VP<br />
V [ XP<br />
… X … ]]]]<br />
(38)<br />
LA<br />
Class 1; transitive<br />
(39) [ v P<br />
… v INCH<br />
[ AktP<br />
… Akt I-CAUSE<br />
[ VP<br />
V [ XP<br />
… X … ]]]] Class 1; <strong>in</strong>choative<br />
⏟<br />
N<br />
⏟<br />
LA<br />
Class 2 verbs differ crucially <strong>in</strong> that they are based on the INCR relation. This prevents LA,<br />
specified as Akt I-CAUSE , from realiz<strong>in</strong>g Akt INCR <strong>and</strong>, due to monotonicity of spell-out, from<br />
realiz<strong>in</strong>g v as well. The role of LA is thus restricted to spell<strong>in</strong>g out VP. Akt INCR is lexicalized by<br />
N <strong>in</strong> both causative <strong>and</strong> <strong>in</strong>choative configurations, as <strong>in</strong> (40) <strong>and</strong> (41), respectively. In addition,<br />
N spells out v INCH <strong>in</strong> (41), where it is the only c<strong>and</strong>idate. It cannot lexicalize v TR , however, due to<br />
the feature mismatch, <strong>and</strong> this is where TYR takes over, (40).<br />
⏟<br />
TYR<br />
⏟<br />
[<br />
⏟ VP<br />
V<br />
(40) [ vP<br />
v TR<br />
[ AktP<br />
Akt INCR<br />
N<br />
⏟<br />
N<br />
(41) [ v P<br />
v INCH<br />
[ AktP<br />
Akt INCR<br />
LA<br />
⏟<br />
[ XP<br />
… X … ]]]] Class 2; transitive<br />
[ VP<br />
V [ XP<br />
… X … ]]]] Class 2; <strong>in</strong>choative<br />
LA<br />
We believe that the analysis just outl<strong>in</strong>ed has a number of attractive properties. First, it<br />
suggests that class 1 <strong>and</strong> class 2 transitives are reduced to the same syntactic configuration<br />
consist<strong>in</strong>g of projections of v, Akt <strong>and</strong> V. Transitives that belong to class 1 <strong>and</strong> class 2 are then<br />
correctly predicted to be identical <strong>in</strong> all relevant respects, <strong>and</strong> their similarities exemplified <strong>in</strong><br />
(22)–(25) follow naturally. Secondly, <strong>and</strong> crucially, represent<strong>in</strong>g the relation between subevents<br />
with<strong>in</strong> AktP opens a way of expla<strong>in</strong><strong>in</strong>g why class 1 <strong>and</strong> class 2 transitives have different<br />
morphological shapes: the analysis allows to relate this difference to the semantic opposition<br />
between I-CAUSE <strong>and</strong> INCR, hence to account for the observations <strong>in</strong> §3.2. If the Akt head<br />
<strong>in</strong>troduces I-CAUSE, it is spelled out by the LA morpheme, as well as V <strong>and</strong> v TR . If, on the other<br />
h<strong>and</strong>, Akt is specified as [INCR], LA is no longer available, <strong>and</strong> N is called for; v is then realized<br />
by TYR, ‘the causative morpheme’.<br />
We are <strong>in</strong> a position of summariz<strong>in</strong>g the argument for RPD based on this material. There<br />
are two classes of denom<strong>in</strong>al transitives <strong>in</strong> Tatar, which only differ as to the properties of the<br />
relation between subevents (INCR versus I-CAUSE) <strong>and</strong> morphological makeup, be<strong>in</strong>g identical<br />
<strong>in</strong> all other respects. This suggests that both classes are associated with the same hierarchical<br />
structure, <strong>and</strong> the difference has to do with the way this structure is phonologically realized.<br />
Assum<strong>in</strong>g RPD with the Akt head <strong>in</strong> between v <strong>and</strong> V enables us to account for two facts. First,<br />
a transitive verb under relevant circumstances is realized by four, not three, pieces of<br />
morphology, whereby the Akt head receives a unique spell-out, dist<strong>in</strong>ct from the exponents of<br />
218