result, the LF-representation of the sentence has two thematic interpretation, namely the external argument of kiss and the internal argument of kiss. Since the verb kiss requires two arguments and there is no other predicate in the derivation, the condition shown in (16) is met.
Now, let us turn to M-Case and the PF-representation. After the introduction of T into the derivation, the operation Move is applied to Mary, and M-Case of Mary takes [Move to [Spec, T]] as an argument. Since null argument does not have M-Case, it does not require an overt movement. Since the nominal phrase Mary, which requires a morpho-phonological interpretation, has an argument of its M-Case and the M-Case gives us back a morpho-phonological interpretation as its value, we can say that all and the only necessary morpho-phonological interpretation is in the PF-representation of the sentence. Therefore, in the derivation shown in (83), the condition in (15) is met. Since the both condition (15) and (16) is met by the structure shown i n (83), the condition in (14) is met and the derivation converges.
In the next section, I will focus on the property of null argument, and show that by adopting the variation of the property between languages, we can analyze the anti-passive.
consider what will happen if the null argument is introduced as an internal argument into the derivation. This will lead us to give an explanation on the linguistic variation on the acceptance of anti-passive.
Anti-passive is a construction whose internal argument is syntactically demoted. The phenomenon can be exemplified by the following sentences.
(86) a. Active in Dyirbal
Yabu-Ø ŋuma-ŋgu bura-n.
mother-ABS father-ERG see-PAST
‘Father saw mother,’
b. Anti-Passive in Dyirbal
ŋuma-Ø buɽal-ŋa-ɲu (yabu-gu).
father-ABS see-ANT-PAST mother DAT (Palmer 1994: 18) (87) a. Active in Chukchee
ətləg-e keyŋ-ən penrə-nen.
father-ERG bear-ABS attack-3SG.3SG.AOR
‘Father attacked the bear.’
b. Anti-Passive in Cukchee
ətləg-en penrə-tko-gʔe (keyŋ-etə).
father-ABS attack-ANT-3SG. AOR bear-DAT (Palmer 1994: 177)53
(88) a. Active in Greenlandic Inuit Juuna-p Anna-Ø kunip-p-a-a.
Junna-ERG Anna-ABS kiss-IND-TRANS-3SG.3SG
‘Juuna kissed Anna.’
b. Anti-Passive in Greenlandic Inuit Juuna-Ø (Anna-mik) kunis-si-v-u-q.
Juuna-ABS Anna-INST kiss-ANT-IND-INTR-3SG
(Bittener and Hale 1996: 36) (89) a. Active in Chamorro
Un-patek i ga’lagu 2SG(ERG)-kick the dog
‘You kicked the dog.’
b. Anti-Passive in Chamorro
Mam-atek hao gi ga’lagu.
ANT-kick 2SG(ABS) LOC dog
‘You kicked at the dog.’ (Cooreman 1988: 578) As shown in (86), (87), (88) and (89), in some languages such as Dyirbal, Chukchee, West Greenlandic Inuit, and Chamorro, anti-passive is permitted.
Since the languages shown above are all ergative languages, the external argument of the transitive verb is marked as ergative, and the internal argument of the transitive verb is marked as absolutive in active sentences. In anti-passive sentences, the external argument of the transitive verb is marked as
sentence. Since I have not built up the analysis for the ergative Case system, I will focus only on the thematic relation in this thesis.54 The active sentences have the following structure, and the D-Cases in the derivation will be as follows.
(90) Active55 vP
NP1 v’
V1 VP
V2 NP2
(91) a. D-Case of NP1 in (90)
⨍D: Merge V1 → V, internal argument b. D-Case of NP2 in (90)
⨍D: Merge V2 → V, internal argument
As shown in (90), active transitive sentences have the two nominal phrases.
The D-Case of the nominal phrase which is Merged to V1 has a value as an external argument of the verb as shown in (91a). The D-Case of the nominal phrase which is Merged to V2 has a value as an internal argument of the verb as shown in (91b). Since the predicate in transitive sentences requires two thematic interpretation in the LF-representation, the LF-representation of (90) succeeds to have all and the only necessary interpretation, the condition in (16) is met.
Now, let us turn to anti-passive sentences. Likewise passive sentences, the demotion of the internal argument in anti-passive sentences will be explained by using the null-argument as an internal argument of the predicate.
The anti-passive sentence has the following structure, and the D-Cases in the derivation will be as follows.
(92) Anti-Passive56 vP
NP v’
V1 VP
V2 null-arg
(93) a. D-Case of NP in (92)
⨍D: Merge V1 → V, external argument b. D-Case of null-arg in (92)
⨍D: Merge V2 → V, internal argument
As shown in (92), anti-passive sentences have one overt nominal phrase and one null argument. The D-Case of the overt nominal phrase has [Merge V1] as its argument, and gives us back [V, external argument] as its value. The D-Case of the null-argument has [Merge V2] as its argument, and give us back [V, internal argument] as its value. Since there are two necessary thematic interpretation in the LF-representation, the condition in (16) is met.
According to our analysis, the difference between passive sentences and
anti-passive sentences is the position of a null-argument. Passive sentences have the following structure, and the D-Cases in the derivation will be as follows.
(94) Passive vP
null-arg v’
V1 VP
V2 NP
(95) a. D-Case of null-arg in (94)
⨍D: Merge V1 → V, external argument b. D-Case of NP1 in (94)
⨍D: Merge V2 → V, internal argument
As shown in (94), a passive sentence has one overt nominal phrase and one null argument. The D-Case of the overt nominal phrase has [Merge V2] as its argument, and gives us back [V, internal argument] as its value. The D-Case of the null-argument has [Merge V1] as its argument, and give us back [V, external argument] as its value. Since there are two necessary thematic interpretation in the LF-representation, the condition in (16) is met.
If the language has a null-argument which takes only [Merge v] as its argument, the language permits passive alone, and not anti-passive. If the language has a null argument which allows [Merge V] alone as its argument, the
null-argument which can take both [Merge v] and [Merge V] as its argument, the language permits both passive and anti-passive. The parametric difference of passivization can be deduced to the property of the null-argument.
In order to make our argumentation clearer, let us consider the following sentence in English, which permits passive alone.
(96) *Johni was ti kissed null-arg.
As shown in (96), the sentence which has an overt external argument and a null internal argument is ungrammatical in English. The nominal phrase John cannot be interpreted as an external argument of the verb kiss. If there is a sentence “John was kissed,” than John in the sentence will be interpreted as an internal argument of the verb. Adopting the assumption that in English a null argument cannot be used as an internal argument of the verb, we can explain the reason of the ungrammaticality. It is difficult for us to explain the ungrammaticality without the assumption. Let us consider the sentence in (96) in detail. The structure and the Cases of the nominal phrases in (96) is as follows.
(97) Structure of (96) TP
Johni T’
T vP
ti v’
kiss1-PASS VP
kiss2 null-arg (98) Cases of John in (97)
a. ⨍D: Merge kiss1 → kiss, external argument
b. ⨍M: Move to [Spec, T] → morphological relation with T (99) Case of null argument in (97)
⨍D: no eligible argument → undefined
According to (97), the structure of the sentence in (96) seems to have no problem. As shown in (98b), the M-Case of John has [Move to [Spec, T]] as its argument, and give us back [morphological relation with T] as its value. Since there is no other nominal phrase which requires a morpho-phonological interpretation in the derivation, the condition in (15) is met by the PF-representation of the sentence.
Now let us turn to the LF-representation of the sentence. As shown in (98a), the D-Case of John has [Merge kiss1] as an argument and gives us back
arguments, the LF-representation of the sentence will meet the condition in (15) and the derivation will converge, if the D-Case of the null argument has [kiss, internal argument] as its value as a result of taking [Merge kiss2] as its argument.
However, as shown in (99), the D-Case of null-argument does not have an eligible argument. This is because [Merge kiss2] is not an eligible argument for the D-Case of the null argument. Therefore, the LF-representation of the sentence fails to have all the necessary thematic interpretation, and as a result of the shortage, the derivation crashes.
In order to provide the abovementioned explanation to the phenomena, the assumption that the null argument in English can only be used as an external argument is crucial. This may sounds rather ad hoc, but if we adopt this assumption, we can easily account for the language which accepts anti -passive:
the parametric difference is caused by the difference of the null -argument.
To sum up, in English, Accusative Case is not available for the internal argument of the verb in passive sentences because PASS absorbs the ability of v to be an argument of M-Case. The M-Case of the internal argument takes [Move to [Spec, T]] as a domain, and since the null argument does not require the domain of M-Case, condition in (15) is met. Because the verb has a null argument as its external argument, the sentence does not have an overt external argument.
If a null argument is able to be used as an internal argument of the sentence, the sentence will be an anti-passive sentence. The (in)possibility of (anti-)passive sentences can be explained by assuming that the D-Case of the
deduced to the difference in property of the null argument in the language.