Andrew Radford - Linguistics An Introduction [Second Edition]

On the basis of our discussion so far, we can see that any theory of the interpretation of sentences is going to have at least two prerequisites: an account of the semantic contributions of the words appearing in sentences and an account of the syntactic structure of sentences. In section 12, we examined some of the issues which arise in the study of word meaning, and previous sections of this part of the book have developed a syntactic theory to a point where we can associate

semantically appropriate syntactic structures with examples such as (401) and

(403). However, consideration of a wider range of examples indicates the need to further extend our resources.

Sentence meanings and Logical Form

333

Thematic roles

Consider first the simple sentences in (405):

(405) a.

The Dark Destroyer cracked the nut

b.

The hammer cracked the nut

In both cases, we have a simple declarative clause, and section 20 has provided considerable detail on how such clauses should be analysed syntactically.

Nothing has been proposed there to suggest that there is any syntactic distinction between (405a) and (405b), and it would appear to follow by the Principle of Compositionality that the only semantic distinction between the clauses is due to the presence of the DP the Dark Destroyer in (405a) as opposed to the hammer in (405b). But is this correct? Some evidence to suggest that it might not be arises if we embellish the examples in (405) with an adverb such as deliberately.

(406) a.

The Dark Destroyer cracked the nut deliberately

b.

?The hammer cracked the nut deliberately

Here, the question mark preceding 406b corresponds to the judgement that there is something odd about this sentence – we are not claiming that it is ungrammatical. The oddness can be identified with the fact that hammers are not the sort of things that act deliberately, intentionally, etc. This, in turn, is linked to the fact that if we consider the events portrayed by (406a, b), the individuals designated by the DPs the Dark Destroyer and the hammer play different roles in those events. For (406a), the Dark Destroyer is an Agent, who, by virtue of his (or her) own volition, acts in such a way as to crack the nut. This may involve crushing it in the hand, stamping on it, throwing it against a wall or, indeed, hitting it with a hammer. By contrast, in (406b), the hammer, unless personified in a science fiction context, is an Instrument lacking independent volition, used by an Agent to achieve the

cracking. And, of course, we can easily construct a sentence in which both the Agent and Instrument roles are explicitly expressed:

(407)

The Dark Destroyer cracked the nut with a hammer

There is a further observation that lends support to the claim that the Dark Destroyer in (406a) is importantly different to the hammer in (406b). It will be recalled from section 19 (p. 263) that the co-ordination test is a useful means for checking the status of sequences of words as constituents in a syntactic representation. This test referred to constituents ‘of the same type’ and, in this context, we can consider (408):

(408)

?The Dark Destroyer and the hammer cracked the nut

Once more, we are not claiming that (408) is ungrammatical, and we are certainly not suggesting that the Dark Destroyer and the hammer are anything other than DPs, but we are claiming that there is something odd about it. It should be readily apparent now where we are locating this oddness: (408) involves the conjunction

334

senten ces

of two DPs, but whereas one of these DPs fulfils the role of Agent, the other is an Instrument, and this difference in roles is sufficient to induce the oddness of the co-ordinate structure in (408). Agent and Instrument are referred to as thematic roles, usually abbreviated to θ-roles (θ being the Greek letter ‘theta’).

Having introduced two θ-roles in the context of some simple examples, we can now ask whether there are additional θ-roles that play a part in the semantic representation of sentences. In fact, there have been a number of proposals for such an inventory, and here we shall simply mention and illustrate some of the most common members of such inventories.

Alongside Agent and Instrument, it has been customary to propose a role of

Affected Object or Patient, illustrated by the the nut in (405a, b) and (407), as well as the italicised DPs in (409):

(409) a.

The paediatrician examined the baby

b.

The postman delivered the letter

Often, those affected objects undergoing a change of state involving location or movement, as in (409b), are seen as instantiating a distinct thematic role, designated Theme, and if we take this step, the baby will be a Patient in (409a), but the letter will be a Theme in (409b). Obviously, given such a characterisation of Themes, some DPs in subject position, such as that in (410), might also be seen as expressing this role:

(410)

The letter arrived

It is customary to recognise a set of θ-roles linked to spatial notions, these being expressed by DPs occurring as the complements of different prepositions in

English. Thus, we see the roles of Location, Source and Goal, expressed by the italicised DPs in (411a, b, c), respectively: (411) a.

The train is in the station

b.

The train came from St Pancras

c.

The train travelled to Lille

And there are more, but it would be of dubious value to try to offer a complete review of possibilities, along with uncertainties, in this introductory context.

Rather, we shall conclude this brief discussion of θ-roles by raising an issue of major importance if we take seriously the matter of determining a semantic

representation for a sentence.

Supposing that we are persuaded that a semantic representation for a sentence must include an indication of (a) the thematic roles expressed by the sentence and (b) how those roles are expressed. Specifically, if we take, say, (409a), how can we ensure that our semantic representation of this sentence includes the information that Agent and Affected Object are expressed, that Agent is expressed by the paediatrician, and that Affected Object is expressed by the baby (for present purposes, we are not distinguishing between Patient and Theme).

Let’s take the presence of the two θ-roles first. Clearly, these can’t be associated with the nominals themselves as inherent properties, since the fact that the

Sentence meanings and Logical Form

335

paediatrician is an Agent and the baby is an Affected Object in (409a) is a fact about that sentence: in (412), the thematic roles of the two DPs are reversed.

(412)

The baby examined the paediatrician

However, it does seem plausible to suggest that the two θ-roles are due to the presence of the verb examine. Any examining event will necessarily involve some entity doing the examining and another entity, not necessarily distinct (the paediatrician examined himself), being examined. Accordingly, we can suggest that it is a lexical property of the verb examine that it requires an Agent and an Affected Object, this information being part of the lexical entry for the verb (see 115 in

section 10).

So how do the two θ-roles that the lexical representation of examine brings to the sentence get properly linked to the DPs, the paediatrician and the baby? One possibility that has been explored is that this linking could be straightforwardly determined by syntactic structure. Thus, for (409a), on the basis of the system that has been developed in previous sections, we have the structure in (413 ) (we do not represent the internal structure of the DPs, as these are not relevant to the issue being considered):

(413)

CP

C TP