Adverb orientation: semantics and pragmatics

ELIA 3, 2002, 299-315 ADVERB ORIENTATION: SEMANTICS AND PRAGMATICS José María García Núñez Universidad de Cádiz Orientation is a well known property of some adverbs in English. Early approaches to the topic simply offered a systematisation of paraphrase relations between these adverbs and the corresponding adjectives. Current cognitive (Nakamura, 1997) and Event Logic (García Núñez, 1999) approaches have discovered simpler semantic components in adverb orientation. This study presents cases of Subject Oriented Adverbs pragmatically oriented to non explicit participants, and demonstrates that the proposed analyses in terms of simpler meaning components provide a good basis for a predictive and explanatory account of adverb orientation to both explicit and implicit participants. Key words: thematically dependent adverbs, event adverbs, adverb scope, thematic structure, deontic modality 1. Adverb orientation Since Jackendoff (1972), a number of adverbs in English have been cited as exhibiting the property of orientation. In very general terms, orientation can be described as the process by which the adjectival root of an adverb can be ascribed to some participant in the event modified by the adverb; furthermore, orientation is assumed to target one particular argument: sentence subject. To illustrate this introductory definition, consider sentences (1)-(4) and their respective paraphrases in the corresponding (a) sentences. Estudios de lingüística inglesa aplicada 300 J.M. García Núñez ELIA 3, 2002, pp. 299-315 (1) Martha willingly sent the secret files to John. (1a) Martha was willing to send the secret files to John (2) John was willingly sent the secret files by Martha. (2a) John was willing to be sent the secret files by Martha. (3) Martha wisely sent the secret files to John. (3a) It was wise of Martha/ Martha was wise to send the secret files to John. (4) John was wisely sent the secret files by Martha. (4a) It was wise of John/ John was wise to be sent the secret files by Martha. As can be seen in the examples, both active and passive subjects can be targets for orientation1. Some authors (McConnell-Ginet 1982, Ernst 1984, Wyner 1994) have noted that orientation is a thematically rather than a syntactically based 1 Here we omit cases of orientation to passive deep subjects like the one in (i) (paraphrased in (ii)). (i) John was sent the secret files wisely by Martha. (ii) John was sent the secret files in a wise manner by Martha. In other words, we will focus on cases of what can be very generally classified as disjunct and preverbal adjunct occurrences of adverbs. The preverbal adjunct occurrence of wisely in active sentences like (iii) is ambiguous between a subject-oriented and a manner reading. Here we are interested only in the former. (iii) Martha wisely sent the secret files to John. Leaving out postverbal adjunct occurrences of oriented adverbs will not affect the discussion below. Adverb orientation: Semantics and… 301 ELIA 3, 2002, pp. 299-315 process2. Sentences where adverb orientation fails to take place in the presence of a subject are taken to constitute good evidence that thematic conditions (more specifically, agentivity) rather than syntactic conditions are the key feature for orientation. (5) * Martha reluctantly received the letter. (6) * John was wisely seen by Martha. 2. Orientation: a semantic/ pragmatic problem A less observed fact about these subject-oriented adverbs is that some of them can at times be oriented to participants outside the argument structure of the sentence, which is tantamount to saying outside the explicit content of the utterance. Quirk et al. (1981: 625-626) puts forward and discuss examples like (6) and (7), where wisdom is ascribed to some participant inferable from the context of utterance. (7) Wisely, the meeting ended early today. (8) Wisely, Bill’s car is in a garage overnight throughout the winter. In these examples the participants whose wisdom is highlighted by the adverb, though implicit and inferable from context, are still identifiable as the individuals who brought about the events described in the sentences, i.e., are agentive participants. Another important thing to notice is that orientation to an implicit agent involves sentence modification, while orientation to a thematic agent involves VP modification. That is, in (7) and (8) the whole sentence is judged wise and the person who caused the event remains implicit. On the other hand, in (9) we have two possible interpretations: either the sinking of 2 There is nevertheless variation among these authors as to how strong the correspondence between syntactic and thematic structure is. They are not hence equally likely to keep to the subject-orientation hypothesis. 302 J.M. García Núñez ELIA 3, 2002, pp. 299-315 the ship is judged as wise and Joan is the agent that caused the event, or the whole event (including the agent) is judged as wise; on the latter interpretation the agent responsible for the bringing about of the event must again be an implicit agent, someone who had Joan sink the ship. (9) Wisely, Joan sank the ship. It is important to notice that (7) and (8) are not suitable contexts for all types of oriented adverbs. Willingly, which was classified as an oriented adverb on a par with wisely, cannot orient to an implicit agent, as the unacceptability of (10) makes manifest. (10) * Willingly, the meeting ended early today. Any solution to the problem of orientation must be consistent with the differences and similarities between oriented adverbs. Failure to do so will necessarily mean missing the facts about the grammar/ pragmatic distinction, or those about the semantic homogeneity of adverb orientation, respectively. On this sketchy first approximation to (7)-(10), we are, on the one hand, dealing with the thematic concepts (less so with the syntactic ones) advocated by previous theories of adverb orientation, and, on the other, resorting to a pragmatic inferential process which certainly goes beyond the limits of syntax or semantics. In so doing, we have come across the sort of general theoretical problem we want to address in this work: how can pragmatics be made continuous with grammar?, how can grammar bias pragmatic inferential processes? It is at this point possible to set the goals of this work. First of all, it seems necessary to establish what kind of linguistic phenomenon orientation is. Is it a grammatically or a pragmatically constrained phenomenon? Should we analyse adverb orientation as a fully grammatically constrained Adverb orientation: Semantics and… 303 ELIA 3, 2002, pp. 299-315 phenomenon, or should we let our pragmatic abilities do part of the work? Secondly, do oriented adverbs form a homogeneous class, or are there differences between them?, do these differences impinge on or determine the kind of orientation they realise? In the next section, we assess the empirical benefits of three theoretical proposals that might in principle be able to account for the revised facts. The first two rely on a fully grammatical treatment of adverb orientation; the last one, the one we will argue for here, allows for a pragmatic treatment constrained by semantic factors. 3. Grammatically-based accounts of adverb orientation The first theoretical possibility we will explore consists in analysing adverb orientation together with other thematic phenomena as lexico-grammatically specified. In this perspective, subject oriented adverbs (SOAs henceforth) would assign thematic properties to a number of arguments in the structure in which they appear. This possibility has been explored in relation to adjectival modification and the thematic structure of NPs by Higginbotham (1985). He argues quite convincingly that modifiers bear a thematic grid or empty argument structure of their own which, as it were, is subordinated to the thematic grid of the lexical category they modify and borrows arguments from it. To illustrate this position, consider two sentences like (11) and (12). (11) Mary sank the ship. (12)Mary willingly sank the ship. The proposal under revision would claim that the transitive verb sink in (11) contains the grammatical information that its external argument is an Agent and its internal argument a Theme/ Patient. Being a lexico- 304 J.M. García Núñez ELIA 3, 2002, pp. 299-315 grammatical specification, failure to meet this requirement yields an ungrammatical expression. We can now, on the basis of this sketchy presentation, hypothesise what would be the rough outlines of a lexico-grammatical treatment of adverb orientation: the thematic grid of willingly in (12) would contain an empty slot to be filled by a thematic agent in the argument structure defined by the predicate sink. The proposal would work acceptably well for willingly, which, according to the reviewed facts, systematically orients to the Agent. However, it would fail to account for the orientation of sentence-modifying carelessly to an implicit agent. On this basis, we can provisionally conclude that adverb orientation to a thematic Agent is encoded in the grammatical content of reluctantly but not in that of carelessly. An alternative theoretical approach would rest on an identification of orientation possibilities and structural locations. As we have seen (cf. foot note 1) the two relevant structural locations are preverbal position and sentence position. This approach is the one adopted by Jackendoff (1972), Thomason and Stalnaker (1973), Travis (1988) or Bowers (1993). All of them have tried to account for the orientation facts on the assumption that syntactic position suffices to constrain interpretation. Omitting details about the precise nature of the theories launched, we can see that these proposals work well in the case of preverbal adverbs (which orient to agent arguments), but can hardly provide a stable target for orientation in the case of sentence SOAs (the agent argument in the case of reluctantly, the agent argument or an implicit agent in the case of sentence carelessly). Grammatically-based accounts of SOAs do not seem to guarantee the empirical coverage of the facts pursued in this work. In the following sections, we present a theory of SOAs which both urges an interactive view of the semantics/ pragmatics interface and yields empirical benefits in relation to the revised facts. Adverb orientation: Semantics and… 305 ELIA 3, 2002, pp. 299-315 4. Subject-oriented adverbs and deontic modality Nakamura (1997) from a Cognitive Grammar perspective, and García Núñez (1999) within an Event Logic framework provide accounts of adverb orientation which make use of two intimately tied concepts: thematic structure and deontic modality3. The idea is that SOAs express some kind of deontic evaluation of the events which they modify4. Roughly, the adverbs in (13) and (14) somehow refer to some person (the speaker in (13), the agent in (14)5) who judges the reported event as undesirable for some reason or other, and who is responsible for the bringing about of the relevant event. (12) Martha foolishly put the secret files out on the kerb. (13) Martha reluctantly put the secret files out on the kerb. Close consideration of these constituents of adverb orientation and their grammatical realisations will lead, we believe, to a fully explanatory account of the facts and problems sketched out in sections 1 and 2 above. 3 The idea appears more clearly in García Núñez (1999) than in Nakamura (1997), for this latter author, though also identifying subject oriented adverbs with modal semantics, fails to distinguish between modal deontic target (the person who evaluates) and thematic target (the person responsible for the action). 4 As we are not dealing here with modalisation of non factual propositions (as is the case with modal deontic verbs), deontic modality should be understood more in its desiderative (right/ wrong evaluating) than in its instrumental sense. After all, as Lyons states in relation to deontic modality, “...the so-be-it component (...) is intended to have an instrumental meaning, but it may be assumed to have developed out of, and to include, an ontogenetically prior desiderative” (Lyons, 1977: 826). 5 As this distinction proves crucial both for distinguishing SOAs and for the revised facts, we will, following Wyner’s (1994) terminology, refer to wisely-type SOAs as Event Adverbs (EAs henceforth), and to reluctantly-type SOAs as Thematically-Dependent Adverbs (TDAs henceforth). The former adverbs are also known as agent-oriented adverbs (Frey 2000), the latter as mental attitude adverbs (Ernst 2000). 306 J.M. García Núñez ELIA 3, 2002, pp. 299-315 Once we accept that oriented adverbs are like deontic readings of modal verbs, we simply have to relocate concepts and categories which are well-established in modal semantics. For example, it is fairly uncontroversial that deontic modal verbs require an agent, i.e., can only operate on a proposition whose thematic structure includes an Agent, or whose processing in a context involves the presence of an agent. This can be seen in (15) (with an explicit agent) and (16) (with an implicit agent). (14) Martha must put the secret files away. (15) Martha must receive the secret files tomorrow (you/someone must see to it that she does). Likewise, modal deontic verbs also imply an evaluator (a deontic source in Lyons’s (1977: 824) terms). In both (15) and (16), this evaluator coincides with the speaker,a s the paraphrases in (15a) and (16a), respectively, show. (15a) It is desirable from the speaker’s point of view that Martha should put the secret files away. (16a) It is desirable from the speaker’s point of view that Martha should receive the secret files tomorrow. The fact that evaluators can coincide with sentence agents is not a peculiarity of oriented adverbs either. A deontic modal verb like would rather also implies an agentive evaluator. (16) Martha would rather keep the secret files in a safe place. (17a) It is desirable from Martha’s point of view that she should keep the secret files. A partial definition of adverb orientation can be offered at this point: it is the process through which a subject-oriented adverb selects an agent. This definition is partial because it does not account for the perceived syntactic differences in orientation. In other words, the theory will not be complete Adverb orientation: Semantics and… 307 ELIA 3, 2002, pp. 299-315 until this new conception of orientation is shown to bear on the above reviewed syntactic constraints on orientation. In order to explain the syntactic realisation of orientation, it is necessary to take into account the semantic type modified by the relevant adverbs. In the light of a deontic classification of oriented adverbs, the question of what state of affairs is modified by the adverb becomes a central one. As is well known, deontic modal verbs operate on propositions expressing potential or possible states of affairs (states or events). Roberts (1986: 212-213) or García Núñez (1999: 439-442) claim that TDAs and EAs modify predicates denoting actual events. A sort of complementary relation seems to obtain: deontic modality as expressed by modal verbs denotes the desirability of a state of affairs which is to be brought about by some agent; deontic modality as expressed by an oriented adverb denotes the desirability of an actual event brought about by some agent. To put it graphically, the modal force of two sentences like (18) and (19) would differ in the way shown by (18a) and (19a). (17) Martha must burn out the secret files (18a) Deontic pole: desirable Evaluator: speaker Potential state of affairs: Martha burns the secret files. (18) Martha wisely burnt out the secret files (19a) Deontic pole: desirable Evaluator: speaker Actual event: Martha burnt the secret files. Now that the three essential components of deontic assertions have been identified, a natural explanation can be offered for the syntactic constraints on orientation. Remember the facts: both adjunct and disjunct occurrences of TDAs force thematic orientation of these adverbs to sentence agents; adjunctive EAs force the same interpretation, but disjunctive EAs 314 J.M. García Núñez ELIA 3, 2002, pp. 299-315 References Bowers, J. 1993 “The Syntax of Predication”, Linguistic Inquiry 24: 591656. Dillons, G.L. 1973. “Some Postulates characterising Volitive NPs, Journal of Linguistics 10: 221-233. Ernst, T.B. 1984. Towards an Integrated Theory of Adverb Position in English, Bloomington: Indiana University Linguistic Club. Ernst, T. 2000. “Semantic Features and the Distribution of Adverbs” in C. Fabricius-Hansen, E. Lang and C. Maienborn (eds.) Approaching the Grammar of Adjuncts. ZAS Papers in Linguistics 17: 79-99. Frey, W. 2000. “Syntactic Requirements on Adjuncts” in C. FabriciusHansen, E. Lang and C. Maienborn (eds.): Approaching the Grammar of Adjuncts. ZAS Papers in Linguistics 17: 107-135. García Núñez, J.M. 1999. Semántica de los Adverbios Temo-Dependientes del Inglés Actual. Cádiz: Servicio de Publicaciones de la Universidad de Cádiz. García Núñez, J.M. 2002. “Focusability, Quantification and Distribution in English Higher Adverbs”, paper delivered to the Spring Meeting of the LAGB, Liverpool. Higginbotham, J. 1985. “On Semantics”, Linguistic Inquiry 16: 547-593. Jackendoff, R. 1972. Semantic Interpretation in Generative Grammar. Cambridge, Massachusetts: MIT Press. Lyons, J. 1977. Semantics, Cambridge: Cambridge University Press. McConnell-Ginet, S. 1982. “Adverbs and Logical Form”, Language 58: 144184. Nakamura, W. 1997. “A Cognitive Approach to English Adverbs”, Linguistics 35: 247-287. Adverb orientation: Semantics and… 315 ELIA 3, 2002, pp. 299-315 Quirk, R. et al. 1985. A Comprehensive Grammar of the English Language. London: Longman. Roberts, I. 1986. The Representation of Implicit and Dethematized Subjects. Dordrecht: Foris. Thomaon R. and R. Stalnaker 1973. “A Semantic Theory of Adverbs”, Linguistic Inquiry 4: 195-220. Travis, L. 1988. “The Syntax of Adverbs”, McGill Working Papers in Linguistics, 20: 280-310. Wyner, A.Z. 1994. Boolean Event Lattices and Thematic Roles in the Syntax and Semantics of Adverbial Modification. PhD Thesis, Cornell University. Zubizarreta, M.L. 1982. Levels of Representationin the Lexicon and in the Syntax. Dordrecht: Foris.