RELACIÓ I DEFINICIÓ DE CONTROLS QUE S’HAN DE FER D’ACORD AMB EL DECRET 375/88 D’1 DE DESEMBRE DE1988
SUBSISTEMA CONNEXIONS 1 ELECTRICITAT
2.2 Telecomunicació per cable
In the previous section we have shown that verbs that show the inchoative-causative alternation are unaccusative, externally caused verbs. In this section we address the issue of the derivation of this alternation. As we have mentioned at the beginning of section 2.4, in the literature on the topic there is no agreement on the direction of the derivation of the inchoative-causative alternation.
L&RH (1995), Reinhart (2002) and Chierchia (2004 [1989]) agree in deriving the inchoative-causative alternation from a fundamentally transitive frame: according to this view, the intransitive form of a verb like melt is derived by an ‘arity’ operation, i.e. an operation that affects the syntactic valence of the verb, from the transitive basic entry of the verb, in a similar way as in the case of passives (cf. Reinhart & Siloni 2005). While in the passive form of the verb, the arity operation disables the syntactic realization of the argument corresponding to the Agent θ-role, but this role is not eliminated and still assigns the role in the semantics, in the unaccusative form the arity operation eliminates the theta-role all together (cf. example 20, from Reinhart & Siloni 2005)20, i.e. there is a decausativization mechanism:
(20) a. The icei was melted ti (with a candle). b. The icei melted (*with a candle).
L&RH (1995) agree with these authors in considering unaccusative verbs as basically causative; they assume a single lexical semantic representation (a causative one) for both the transitive and the unaccusative form of these verbs. According to them, these verbs have a dyadic lexical representation, but under certain circumstances they can express only the internal argument and are realized as
20 One of the ways to test the difference is by adding an Instrument. A passive can occur with an Instrument (20a), which is generally licensed only if an Agent role is available in the semantic representation. In contrast, with an unaccusative verb (20b), no Instrument can be licensed, since the external role of the transitive entry is completely eliminated (cf. Reinhart & Siloni 2005:399).
95 monadic predicates in the syntax (21a). In contrast, unergative verbs have an inherently monadic representation (21b):
(21) a. break: [[x DO-SOMETHING] CAUSE [y BECOME broken] b. walk: [x walk]
As it can be seen from the representation in (21a), for an alternating verb is assumed a complex lexical semantic representation formed by two subevents linked by a causal relation. This bi-eventive analysis presents a causing subevent and the subevent that specifies the change associated with the verb (cf. also Dowty 1979, Pustejovsky 1991). In contrast, unergative (non-alternating intransitive) verbs are basically monadic and do not have the predicate CAUSE in their representation (21b), which is why they do not have a causative variant21.
The first argument L&RH (1995) provide in favour of this analysis of alternating verbs comes from the observation that the selectional restriction on the object (in the transitive use) and on the subject (in the intransitive use) do not coincide for any verb; however, the set of possible subjects for the intransitive variant of a verb apparently is a subset of the set of the possible objects for the transitive use of the same verb22. See
the examples in (22) and (23) from L&RH (1995:85):
(22) a. Antonia broke the vase / the window / the bowl / the radio / the toaster. b. The vase / The window / The bowl / The radio / The toaster broke. c. *Antonia broke the cloth / the paper / the innocence.
d. *The cloth / The paper / The innocence broke. BUT:
e. He broke his promise / the contract / the world record. f. *His promise / The contract / The world record broke.
21 Recall that Hale & Keyser (1993, 1998, 2002) consider unergative and transitive verbs as having the same structure (cf. chapter 1, fn.61), i.e. an “lp-monadic structure” (in relation to the arguments which must appear internal to the lexical configuration associated with a lexical item, not in relation to syntactic adicity): the argument structure configuration projected by the head contains just one argument, i.e. the complement. In contrast, unaccusative verbs like break are characterized by a dyadic type configuration and have a specifier (i) (cf. Hale & Keyser 1998, 2002).
(i) Head tu Spec Head
tu Head Comp
22 This is due to the fact that with certain objects the eventuality described cannot come about without the intervention of an agent (detransivization is possible where an externally caused object can come about without the intervention of an agent) (cf. also Haspelmath 1993).
(23) a. Jean opened the door / the window. b. The door / The window opened. BUT:
c. This book will open your mind. d. *Your mind will open from this book.
L&RH (1995) assume that the basic variant of the verb is the one which imposes less selectional restrictions; if it were the other way around, they claim that it would be difficult to derive the variant with the looser restrictions in a plausible way. Therefore, according to this view, the causative variant is the basic form, i.e. the one with the looser selectional restrictions.
The other argument in favour of this analysis comes from Chierchia (2004 [1989]), who points out that unaccusative verbs that lack a paired transitive use, e.g. come, are exceptional and would be expected to have such a use, since they are basically dyadic. He suggests that a verb like come is related to a causative verb meaning something like ‘bring’, but this causative verb is either not lexicalized or lexicalized by a verb not morphologically related to the intransitive use23. Chierchia (2004 [1989]) assumes that these verbs, diachronically and across dialects, tend to oscillate in valency from transitive to intransitive and vice versa; he cites an Italian unaccusative verb, crescere ‘grow’, which, according to him, in standard Italian lacks the causative variant, stressing that in some dialects it has a causative use, with the meaning ‘raise (children)’24. Examples of this type can be easily found across dialects. In some Italian dialects (e.g. Sicilian), for example, unaccusative verbs like entrare ‘enter’,
uscire ‘exit’, scendere ‘descend; go down’, which in standard Italian are only
intransitive, can be used transitively (24):
(24) a. Mamma ha uscito la carne dal frigo per il pranzo. mother exited the meat from fridge for the lunch
‘Mom took out the meat from the fridge for lunch’
23 In other languages this alternation (‘come’-‘bring’) can be expressed by the same verbal root, i.e. by a lexical causative. For instance, in Chinese the verb 来 lái means both ‘come’ and ‘cause to come, send (here), bring’, apparently acting as an alternating verb (cf. Lü 1980:252): 客人来了 kèren lái le ‘guest come ASP = The guests are here’ vs. 他来过两封信 tā lái guo liǎng fēng xìn ‘he come ASP two CL letter = He sent (here) two letters’, 请再来一瓶啤酒 qǐng zài lái yī píng píjiǔ ‘please again come one CL beer = bring me another bottle of beer, please’ (cf. also chapter 4, exx. 20a-c).
24 In standard Italian, the verb crescere ‘grow’ actually has a causative variant. De Mauro – Il dizionario della lingua italiana per il terzo millennio (DM 2000) registers three transitive uses of crescere: 1) allevare, educare ‘raise (children)’; 2) far diventare più grande; accrescere, aumentare ‘make grow; increase’ (not frequent); 3) nei lavori a maglia: aumentare ‘add, increase (in knitting)’ (cf. also Sabatini Coletti: Dizionario della lingua italiana, SC 2008).
97 b. Ho entrato la macchina nel garage.
entered-1SG the car in garage ‘I put the car in the garage.’
c. Mi scendi le chiavi per favore? me descend the keys please ‘Can you please bring me the keys (downstairs)?’
L&RH (1995:87) point out a similar example in English, where the verb
deteriorate, usually intransitive, is used transitively: The pine needles were deteriorating the roof. Chierchia (2004 [1989]) points out that unergative verbs, in
contrast, are “stable” and are not expected to show the alternation, since they are inherently monadic.
Moreover, L&RH (1995) highlight the fact that Nedjalkov (1969), in a survey on sixty languages on the behaviour of the verbs break and laugh, found out that in most languages the transitive form of the verb break is unmarked, while the intransitive form is identical to the transitive form (i.e. labile verbs) or derived from the causative form. The fact that the intransitive form is morphologically marked apparently confirms that these verbs are basically causative, whereas their intransitive use is derived: the morphological marking indicates the non-expression of the external cause. In contrast, in the majority of languages observed by Nedjalkov (1969), the causative form of laugh is morphologically more complex than the non-causative form (cf. also Haspelmath 1993). According to L&RH, this is due to the fact that these verbs are inherently monadic verbs, whose lexical semantics lack a causative predicate.
Finally, the causative analysis is supported by Chierchia’s (2004 [1989]) remark about Italian: according to Chierchia, the use of the adverbial da sé ‘by itself’ (i.e. ‘without outside help’)25 reflects the presence of a cause argument in the lexical representation of the verb (cf. ex. 25, from Chierchia 2004 [1989]:43):
(25) a. La porta si è aperta da sé.
The door opened by itself
‘The door opened by itself.’
Apparently this adverbial modifies a cause, which identifies itself as the theme argument. The intransitive verbs that do not participate regularly in the alternation do not appear with this adverbial (cf. L&RH 1995). Actually, unergative verbs in English
25 In English by itself can mean both ‘without help’ and ‘alone’. With the unaccusative form of alternating verbs it means ‘without help’.
can appear with by itself adverbial, but in the sense of ‘alone’ (cf. fn. 25), e.g. She
walked by herself (i.e. unaccompanied).
Having assumed that alternating verbs have a dyadic lexical semantic representation (LSR), the unaccusative form is assumed to be derived from an operation of detransitivization, as described in (26), from L&RH (1995:108)26:
(26) a. Intransitive break
LSR [[x DO-SOMETHING] CAUSE [y BECOME broken]]
Lexical binding Ø Linking rules
Argument structure <y>
b. Transitive break
LSR [[x DO-SOMETHING] CAUSE [y BECOME broken]] Linking rules
Argument structure x <y>
The condition for the verb to decausativize is that the verb must not impose any lexical specification on the causing subevent; in this way, when it is used intransitively, the external cause argument of the verb is understood as not being lexically specified. Therefore, the external argument can be left unexpressed. According to L&RH (1995), this is possible because the alternating verbs express only the resultant state, leaving the causing event unspecified: for example, in the sentence Mark broke the door, only the change of state of the door is specified, while the causing event remains unspecified; a lot of different activities could have caused the change of state. The decausativizing process is possible when the event can be conceived as occurring spontaneously.
L&RH’s analysis (1995) seems to be confirmed by data of ‘anticausativizing’ languages (cf. 2.2), where the unaccusative form of the alternation is the marked one. However, as we have seen in 2.2, the morphology of the causative alternation varies greatly across languages and many languages prefer causativizing to anticausativizing
26 L&RH (1995:108) suggest that the binding of the external cause takes place in the mapping from the lexical semantic representation to the argument structure. The evidence that the binding of the external cause takes place before argument structure comes from comparison with passive verbs. Grimshaw (1990) shows that the operation that derives passive verbs involves binding a position in the lexical syntactic representation of a verb (argument structure), preventing the expression of that argument in the syntax.
99 morphology. As we have already mentioned, Haspelmath (1993) points out that languages that prefer anticausatives are spoken in Europe, while languages that prefer causatives are spoken elsewhere: he states that the absence of causative morphology and anticausative derivations seems to be an European areal feature.
According to Haspelmath (1993), the anticausative type is favoured by the probability of an outside force to bring about the event. In contrast, the causative type is favoured if the event is quite likely to happen without outside force: “The more typical the change of state is, the more likely the causative type will be unmarked” (cf. Haspelmath 1993:103). Verbs like freezing, drying, sinking and melting do not need an agentive instigator, thus they are more likely to appear in alternations when the causative variant is marked (causative type). In contrast, verbs like splitting,
breaking, closing, opening, gathering express things that human beings do, therefore
they are more likely to appear in alternations when the intransitive variant is marked (anticausative type). However, according to Haspelmath (1993:103) the correlation is only typical, not necessary: people may sink, dry, melt, freeze things; things may split, close, break spontaneously.
Haspelmath (1993) argues that if the semantic properties of a word are only the objective semantic features discovered by semantic decomposition, then causatives are always semantically more complex than inchoatives, thus the causative type follows the direction of the semantic derivation, and the existence of the anticausative type is odd. However, events that are more likely to occur spontaneously will be associated with a conceptual prototype of a spontaneous event, and this will be expressed in a structurally unmarked way. In contrast, events that are more likely to be brought about by an external agent will be associated with a stereotype of a caused event, so the caused event will be expressed in a structurally unmarked way (anticausative type) (cf. Haspelmath 1993:107). Nevertheless, Haspelmath stresses the fact that this form-meaning correlation is only a tendency.
Given the great cross-linguistic variation, with languages that prefer anticausativizing morphology vs. languages that prefer causativizing morphology, Ramchand (2008) argues that the morphological argument to show the direction of the derivation of alternating verbs is equivocal. According to Ramchand (2008), if we were to consider the morphological behaviour, we should assume that the derivation can potentially go in either direction.
Moreover, Ramchand (2008) revises L&RH’s (1995) argument about selectional restrictions on arguments (cf. exx. 22 and 23). As we have seen, L&RH (1995) state that the fact that the set of possible internal arguments are looser in the intransitive version than in the transitive one speaks in favour of a derivation of transitive to intransitive: alternating verbs are basically causatives. However, Ramchand (2008) notices that in some cases the selectional restrictions are looser in the transitive version, as in the examples in (27), from Ramchand (2008:84), where the idiomatic interpretation is available only in the intransitive use of the verb collapse:
(27) a. The tent collapsed. b. Mary collapsed. c. Sue collapsed the tent. e. *Sue collapsed Mary.
Ramchand (2008) is not convinced that L&RH’s (1995) argument on argument selection actually works, even if their generalization represents the dominant pattern. She argues that the assumption of the superiority of the transitive to intransitive derivation is due to a particular conception of the lexicon and of its role in expressing selectional restrictions.
Ramchand (2008) observes that in a lexicalist system, where verbal meaning is completely stored in the lexical item and any non-predictable meaning is stored in the lexicon together with the lexical item, idiosyncratic transitive versions cannot be explained by additive rules: adding semantic content cannot provide a simple predictive system, unless the added content is the same for every alternation. However, in a constructivist approach, idiom formation can be associated to larger structures: thus, it is possible to associate an idiomatic content to a transitive verb built up from an intransitive root via a causational head (cf. Ramchand 2008:84). Furthermore, Ramchand points out that in a constructivist approach the intransitive to transitive derivation does not have to be specified as a rule: the derivation is predicted as the outcome of structure building during the course of the derivation. According to Ramchand, under this view, transitivization is more regular and transparent than detransitivization. Ramchand’s aim is not to provide arguments for the superiority of transitivization, but rather to deny the assumption that detransitivization is the only possible derivation.
Ramchand (2008) tries to demonstrate that in English transitivization is more economic than detransitivization. She considers transitive verbs specified for [init,
101 proc] features, like melt and hammer, and verbs specified for [init, proc, res] features, like break and throw. Some verbs in both classes have an intransitive variant, while some others do not: e.g. melt and break are alternating verbs, while hammer and
throw are not (cf. exx. 28 and 29, from Ramchand 2008:85).
(28) [INIT, PROC] VERBS
a. Karena melted the butter. b. The butter melted.
c. Karena hammered the metal. d. *The metal hammered. (29) [INIT, PROC, RES] VERBS
a. Alex broke the stick. b. The stick broke. c. Ariel threw the ball. d. *The ball threw.
Looking at these verbs from a detransitivizing perspective, Ramchand (2008) points out that the derivation of these alternations could be possibly realized in two ways: either by conflation of the Initiator and the Undergoer roles or by a kind of role suppression, i.e. the Initiator role is completely missing27. In any case, it would be necessary to add to the lexical entries some specifications that clearly state that detransitivization is possible. Therefore, a verb like break [init, proc, res], besides its features, should specify that it is distinct from verbs like throw [init, proc, res], which have the same category features (and the Aktionsarten properties derived from them) and can undergo detransitivization.
In contrast, Ramchand (2008) assumes that causativization is more economic and provides a simpler system. She observes that intransitive verbs with an initiation component (where the same DP is both the Initiator and the Undergoer, i.e. represent a composite role) cannot causativize: e.g. *Michael ran Karena (cf. Ramchand 2008:86). Therefore she assumes that English has a process of causativization, as a result of automatic structure building, which forms transitive verbs from verbs that do not contain a [init] specification and which is allowed for the presence of a default null init head. Alternating verbs like melt and break, thus, are listed as [proc] and [proc, res], respectively, rather than as [init, proc] and [init, proc, res] (cf. exx. 28 and
27 In Ramchand’s system, the mechanism of argument identification can be equated to the creation of composite roles (each specifier position does not need to be filled with a distinct DP). Moreover, argument suppression can be equated to the possibility of non-projecting category features of the root (cf. Ramchand 2008:85).
29). The transitive version would then be built by introducing a layer on top of their structure due to the null init head, which has the semantics of general causation28. The representation of alternating verbs like melt and break would be as illustrated in (30), representing (28a) and (28b), and (31), representing (29a) and (29b) (cf. Ramchand 2008:86-87):
(30) a. Intransitive melt [proc]
procP
tu
the buttertu
<melt> XP b. Transitive melt initPtu
Karenatu
init Ø procPtu
the buttertu
<melt> XP28 Hale & Keyser (1998) too assume a derivation from intransitive to transitive for English verbs through structure building. According to Hale & Keyser, unergative verbs cannot transitivize (e.g. *laugh the child) because the structural type of their lexical argument structure (monadic type) lacks a specifier (cf. chapter 1, fn.61), and thus there is no place in the lexical structure for the surface object of an hypothetical transitive clause (cf. Hale & Keyser 2002:15). In contrast, unaccusative verbs are characterized by a dyadic type configuration and have a specifier (cf. fn.21). These verbs may have a transitive counterpart, which is derived by insertion into the complement position of a verbal structure (ii). (i) V (ii) V tu tu DP V V V tu tu V X DP V tu V X
As far as the derivation of the alternation is concerned, Hale & Keyser (1998) suppose that in English, where the transitivity alternation is not itself associated with any overt morphology, alternating verbs have the simplest structure and, consequently, the simplest derivation, i.e. the intransitive one (i). Therefore, they assume that, in the absence of overt morphology, the direction is always from the simpler structure (intransitive or inchoative) to the more complex (transitive). For evidence in favour of this hypothesis, see Hale & Keyser (1998).
103 (31) a. Intransitive break [proc, res]
procP