Semantic constraints on the distribution of English higher adverbs

ES 25 (2003-04) - pp. 103-122 SEMANTIC CONSTRAINTS ON THE DISTRIBUTION OF ENGLISH HIGHER ADVERBS1 JOSÉ MARÍA GARCÍA NÚÑEZ Universidad de Cádiz The present study focuses on the different distribution of a group of current English adverbs and adverb occurrences, viz higher adverbs. Regarding distribution, these adverbs can be divided into monoperipheral and biperipheral higher adverbs. Delving into the semantics of these two classes of adverbs leads to a distinguishing semantic feature: eventivity. The work puts forward a number of data and arguments favouring the hypothesis that only biperipheral higher adverbs are eventive. Finally, the hypothesis is explored that focalization, an event-related phenomenon according to current studies, lies at the source of the different distribution of monoperipheral and biperipheral higher adverbs. 1. ENGLISH HIGHER ADVERBS Regarding their structural semantic properties, adverbs and adverb occurrences can be divided into those that can and those that cannot take scope over other sentence constituents. The adverb always, for example, can take a quantified NP in its scope. Contrariwise, it is impossible for an adverb like well to scope outside a quantifier. The contrast is observable in (1)-(2), respectively. (1) Mary always greets everyone. (2) Lewis treats everyone well. One interpretation of (1) has it that, on every relevant occasion, Mary greets the set of individuals denoted by the quantifier everyone. On the other hand, whatever the denotation of well in (2), it will be true as many times as individuals are picked out by the quantifier. On the premise that scope relations have a configurational basis (Chomsky 1977; May 1977; 1985), I will call always a higher adverb and well a lower adverb. Higher adverbs also interact with the scope of other categories. They can take scope over other adverbs, a possibility not open to lower adverbs. (3)-(4) (with the brackets providing a rough analysis of the relevant configurations) illustrate this point. (3) Mary always [greets everyone twice] (4) * Lewis [greets everyone twice] well 1 An earlier version of this paper was presented at the 2002 LAGB Spring Meeting. I am grateful to participants in this conference for interesting comments and criticisms. JOSÉ MARÍA GARCÍA NÚÑEZ 104 In most of the cases, it is not the adverb proper, but one of its structural occurrences that is either higher or lower. The different occurrences of luckily in (5)-(8) provide relevant examples. (5) Anthony luckily saved both goals. (6) Anthony saved both goals luckily. (7) Anthony luckily [saved a goal twice] (8) * Anthony [saved a goal twice] luckily. For the sake of brevity, I will henceforth use the term higher adverb to refer to both adverbs and adverb occurrences. Higher adverbs constitute the object of study of this work. More precisely, I will deal with the linguistic contrasts between two markedly different classes of higher adverbs. In the following section I will outline and exemplify these differences. 2. THE PROBLEM NEEDING ANALYSIS There is a very important distributional difference among the adverbs that come out higher by the test put forward above. Unlike lower adverbs, whose distribution seems to be highly idiosyncratic,2 higher adverbs exhibit a very regular distributional asymmetry. Some of them can occur both in the left and in the right periphery, while others are confined to the left periphery. Let us see what this means. The higher adverbs in (1)-(8) above are distributionally “well behaved”. Neither always nor luckily can occur to the right of the main verb (i.e. the right periphery) without rendering the sentence ungrammatical or taking on a lower reading, respectively. (9) * Mary greets everyone always. (10) Anthony saved both goals luckily. Unlike these adverbs, higher adverbs like reluctantly or twice can occur on both peripheries. (11) Ralph (reluctantly) greeted everyone twice (reluctantly). (12) Ralph (twice) greeted everyone reluctantly (twice). I will adopt the terminological convention of calling the adverbs in (9)-(10) monoperipheral higher adverbs (MHAs henceforth), and the ones in (11)-(12) biperipheral higher adverbs (BHAs henceforth). 2 A lower adverb like perfectly seems awkward in its left-peripheral occurrence in (i). (i) She perfectly learnt two languages However, the VP in (ii) seems to considerably improve its left-peripherality. (ii) She perfectly understood the question. The distributional differences that will be dealt with in relation to higher adverbs are more SEMANTIC CONSTRAINTS IN THE DISTRIBUTION OF ENGLISH HIGHER ADVERBS 105 (13) shows a possible distributional classification of higher adverbs. The notional classes as well as the general divisional lines are borrowed from Ernst (2000).3 (13) MONOPERIPHERAL HIGHER ADVERBS BIPERIPHERAL HIGHER ADVERBS Discourse-oriented (frankly, honestly) Evaluative (luckily, regrettably, unbelievably) Epistemic (probably, obviously, clearly) Evidential (reportedly, presumably) Agent-oriented (wisely, rudely, generously) Exocomparative (similarly, likewise) Domain (politically, historically, stylistically) Quantifier (always, usually) Aspectual (already, yet) Most duration (briefly, momentarily) Most location-time (immediately, now, then) Mental attitude (reluctantly, intentionally) Frequency (often, periodically, twice) Some questions arise given the above constellation of facts: (a) why are MHAs distributionally different from BHAs? (b) are distributional differences due to the existence of two adverb classes (MHAs and BHAs) maintaining deeper and more far-reaching grammatical differences? (c) if so, what are these differences and what is their nature? Questions (b) and (c) seem the right place to start our query, for it seems clear that a proper answer to these questions, if there turns out to be one, should suffice to provide an answer to question (a). 3. MHAS AND BHAS The literature on adverbs that will be reviewed in the next section is almost unanimous on the theoretical need to set up something like the partition in (13). However, given that, as will be shown below, some of the theoretical arguments put forward to support the partition in (13) are clearly objectionable, it seems convenient to resort to theory-neutral criteria to test the solidity of the MHA/ BHA 3 Unlike Ernst (2000), I make room among MHAs for quantifier adverbs such as always and usually, which clearly cannot occur in the right periphery. I also depart from Ernst’s classification in rating domain adverbs among left peripheral adverbs. Ernst’s decision is based on his judgement (Ernst 1985) that sentences like (i) and (ii) are synonymous. (i) Intellectually, they won him over. (ii) They won him over intellectually. I do not believe that the adverb in (ii) is identical to the one in (i). For example, unlike the former, the latter fails to take scope over negation. (iii) Intellectually, they didn’t win him over. They didn’t win him over intellectually. JOSÉ MARÍA GARCÍA NÚÑEZ 106 division. In the rest of this section I will turn to very simple facts about the classification of adverbs in order to test the validity of the division in (13). Adverb notional classes like the ones depicted in (13) (e.g. epistemic, agentoriented, evidential, etc) are settled on the basis of cooccurrence restrictions. In general terms, adverbs belonging in the same notional category do not cooccur. This is allegedly the reason why the sentences in (14)-(16) are ill-formed. (13) * Frankly, honestly, she kicked the ball first (14) * She obviously will probably kick the ball first (15) * She intentionally had reluctantly kicked the ball first If we delve further into the cooccurrence criterion, we see that adverb classes are not only occurrentially compatible or incompatible. When they are cooccurrent, they have to respect certain ordering restrictions. Some adverbs have to occupy more prominent structural positions than others in order for their cooccurrence to be grammatical. Assuming for reasons that will become clearer in section 6 below that in adverb clusters the adverb to the left is structurally more prominent than the one to the right, it follows from the different acceptability of the (a)/ (b) options in (17)- (19) that the adverb classes involved have different structural prominence. (16) a. Obviously, she fortunately kicked the ball first b. * Fortunately, she obviously kicked the ball first (17) a. She intelligently had always kicked the ball first b. * She always had intelligently kicked the ball first (18) a. She always reluctantly kicked the ball first. b. * She reluctantly always kicked the ball first. These structural constraints on cooccurrence have led linguists to analyse adverb classes as hierarchically arranged. The hierarchical arrangement of the adverb classes in (13) shows an interesting difference between MHAs and BHAs. The different classes of adverbs falling within the former group maintain a strict hierarchial relationship between one another (see (17)-(18) above) and to BHAs (see (19)), which are hierarchically less prominent than them. On the other hand, the hierarchical ordering of the adverb classes listed under the label BHAs is much less clear. The orderings in the (a)-(b) options of (20)-(21) do not seem to have consequences other than the different scope of the adverbs involved. (19) a. She soon kicked the ball reluctantly b. She reluctantly kicked the ball soon (20) a. She had already kicked the ball first b. She had often already kicked the ball The upshot is that BHAs are occurrentially different from MHAs. More specifically, the cooccurrence data suggests that BHAs constitute a more homogeneous adverb group than MHAs. As pointed out above, this allows us to seek for an explanation for the different distribution of higher adverbs at the boundary between BHAs and the rest of higher adverbs. Given the parallelism SEMANTIC CONSTRAINTS IN THE DISTRIBUTION OF ENGLISH HIGHER ADVERBS 107 between the cooccurrence and the distributional behaviour of higher adverbs, we can provisionally assume that distribution is but the grammatical manifestation of deeper differences between the adverb classes under study. It also seems clear that only by spelling out these underlying differences can we aim at fully understanding why higher adverbs are distributionally different. This is the programme that I will follow in the rest of this work. The BHA/ MHA opposition (or something close to it) has been the focus of attention of part of the current literature on adverbs. Before pursuing a new approach to the topic, I will assess the explanatory capacity of current theoretical accounts of the BHA/ MHA opposition. 4. PREVIOUS ACCOUNTS AND THEIR PROBLEMS 4.1. Syntactically centred approaches It is common place in current generative syntax that the left-peripheral occurrences of adverbs are derivationally more basic than the right-peripheral ones. Right-peripheral occurrences of higher adverbs are scope-ambiguous in a way that can be overtly expressed in the left periphery. (22), for example, can receive the interpretation of both (23) and (24). (21) Liz didn´t knock on the door twice reluctantly. (22) Liz reluctantly didn’t knock on the door twice. (23) Liz didn’t reluctantly knock on the door twice. The upshot is that these higher adverbs have different structural occurrences, the structural variation being overtly realised in the left-periphery but only covertly expressed in the right-periphery. These facts have led syntacticians to defend that higher adverbs are basegenerated in their left-peripheral positions and surface to the right as a result of movement.4 There are two major and competing approaches to the base-generation of adverbs, the so-called right-adjunction (Andrews 1983; Ernst 1984, 2002), and left-adjunction (Kayne 1993; Alexiadou 1997; Cinque 1999) accounts. Basically, right-adjunction and left-adjunction theorists differ in the way they have syntactic constituents surface to the right. The former allow for a certain amount of symmetry in the overall asymmetric design of X’ structure and defend that categories adjoining to syntactic projections are allowed to symmetrically branch to the right. The latter are conceptually opposed to right-adjunction. On their view, linearization demands that constituents be strictly ordered in terms of c-command, which allows only for structures that are radically asymmetric. In order to account for the right4 For an alternative view with BHAs base-generated as verb complements, see Larson (1990) and Alexiadou (1997). JOSÉ MARÍA GARCÍA NÚÑEZ 108 periphery facts reviewed, they posit a generalised constituent movement across adjuncts to higher projections. Right-adjunction advocates tend to favour an analysis of adverb distribution in terms of constituency. Jackendoff’s (1972) classical analysis makes the I(nflectional) P(hrase)/ V(erb) P(hrase) divide responsible for the different distributional behaviour of adverbs. Specifically, Jackendoff argues that VP adverbs are allowed to occur on both peripheries whereas IP adverbs are distributionally restricted to the left periphery. Empirically, the analysis fails to capture the division in (13), for some MHAs (e.g. agent-oriented adverbs) have been shown to be basegenerated in VP (Frey 2000). Theoretically, the account does not provide a motivated answer to question (a) above. Both the left-peripherality/ IPhood and the bi-peripheality/ VPhood associations follow from stipulation. Regarding left-adjunction accounts, the different distribution of the adverbs in (13) is held to follow from their different structural prominence. The higher in structure, the less likely an adverb will be to let its c-command domain be raised to projections high above. As Cinque (1999) points out, this is descriptively unobjectionable in view of the parallelism between adverb hierarchy and distribution. However, there is, to the best of my knowledge, no account of which is the precise functional category providing the structural divide behind the partition in (13). More importantly, there is no indication as to why this category should lead adverbs above it to block constituent movement. In failing to provide such an explanation left-adjunction accounts also leave question (a) above unanswered. 4.2. Modification approaches There is a group of studies (Williams 1994, 23-24; Peterson 1997; Shaer 2000) that, with more or less fortune, have tried to follow Heny’s (1973) early indication that the syntax and semantics of higher adverbs should be tackled in terms of the restrictive/ non restrictive opposition. Among them, Wiliams (1994) is the only one that provides an explanation for the distributional puzzle posed by higher adverbs. He observes that, as (26) shows, non restrictive modifiers generally cannot occur in the right periphery, unless separated from the rest of the sentence by a pause. (24) The Greeks as philosophical as Bob... (25) The Greeks, as philosophical as Bob... He argues that (27)-(28) depicts the adverbial counterpart of the pattern in (25)-(26), for, as (29) and (30) respectively show, the adverb in (27) is a restrictive modifier while the one in (28) is a non restrictive one, i.e. the latter can fall within the scope of negation while the former cannot. (26) John left reluctantly. (27) John left, probably. SEMANTIC CONSTRAINTS IN THE DISTRIBUTION OF ENGLISH HIGHER ADVERBS 109 (28) He didn´t reluctantly leave. (29) * He didn’t probably leave. The idea is that adverbs are inherently either restrictive or non restrictive. It hence follows that some adverbs can be used either restrictively or non restrictively. (30) He didn’t leave rudely. (31) * He didn’t rudely leave quickly. Williams’ proposal does not trivially link up to the fact that non restrictive adverbs are confined to the left-periphery. However, given the fact that the structural principle extensionally carries over to the adjectival domain, such an explanation may yet be thought to be worth pursuing. There is, nevertheless, strong empirical evidence that Williams’ restrictive/ non restrictive classification of adverbs does not really extensionally correlate with the distributional classification in (13). (33)-(34) clearly show that evidential adverbs are radically monoperipheral. (32) He reportedly left. (33) * He left reportedly. However, Williams’ negation test dubs these adverbs restrictive. (34) He didn’t reportedly leave. The data suggests that the distribution of adverbs does not correlate with their restrictive import, hence undermining the basis of a possible explanation for the distribution of adverbs along the lines suggested by Williams. The above revision shows that the currently available accounts of adverb syntax and semantics prove incapable of capturing the distributional pattern of higher adverbs and/or to provide a motivated account of it. In what follows I will pay attention to the structural semantic properties of the adverbs at issue in the hope that this will provide a more convincing answer to questions (a) and (c) above. 5. MHAS AND BHAS: SEMANTIC DIFFERENCES The hypothesis that I will explore in the rest of this work is semantic in nature and is rooted in the traditional view (Thomason and Stalnaker 1973; Bellert 1977) that the adverb hierarchy is based on differences in the semantic type selected by adverbs. The fact that BHAs are hierarchically subordinate to MHAs but not to one another suggests that the grammatical homogeneity of the former class might be due to a coincidence in the semantic type required by these adverbs. It is interesting to observe that the adverb classes included in the group of MHAs have frequently been related to different types. Thus, for example, agentoriented adverbs can be related to the expression of facts, or speaker-oriented adverbs to the illocutionary content of the utterance. This is the reason why the former adverbs are infelicitous in non factive contexts (see the different JOSÉ MARÍA GARCÍA NÚÑEZ 110 acceptability of (36) and (37)) or why the latter are awkward in illocutionary opaque contexts (compare (38) to (39)). (35) ? Mary thinks that John foolishly knocked on the door. (36) Mary knows that John foolishly knocked on the door. (37) * Mary said that John, frankly, knocked on the door. (38) Frankly, Mary said that John knocked on the door. BHAs do not demand such higher types, i.e. they can replace the adverbs in (36)-(39) without giving rise to unacceptability. The upshot is that whichever the type selected by BHAs, it must be lower than these two. A possible candidate is the proposition expressed by the utterance. Unlike agent-oriented and speaker-oriented adverbs, BHAs are part of the propositional content of utterances. This can be shown on the basis of Infantidou-Trouki’s (1993) test for propositionality. Infantidou-Trouki (1993) suggests that an interesting test to determine which adverbs contribute to the propositional content of the utterance is to embed them in a conditional clause and to check whether or not they scope within the conditional operator. If they do, they are part of the proposition expressed. According to this test, the BHAs in (40) and (41) are fully propositional, i.e. the conditions under which Chris will be scorned include his intentionally or often kissing Sandy passionately, respectively. (39) If Chris intentionally kisses Sandy passionately, he will be scorned. (40) If Chris often kisses Sandy passionately, he will be scorned. Contrariwise, agent-oriented and speaker-oriented adverbs are not propositional according to the conditional test, i.e. they are not part of the conditions that will lead to Chris’s being scorned in (42) and (43). (41) If Chris foolishly kisses Sandy passionately, he will be scorned. (42) If Chris, frankly, kisses Sandy passionately, he will be scorned However, the proposition cannot be the minimal type required by BHAs, for some MHAs are also propositional according to the conditional test. As pointed out by Infantidou-Trouki (1993), evidential adverbs are fully propositional, i.e. they scope within the conditional operator (e.g. in (44), Chris will be scorned if he is reported to kiss Sandy passionately). (43) If Chris reportedly kisses Sandy passionately, he will be scorned. If some MHAs are part of the proposition, propositionality cannot be the factor distinguishing BHAs from MHAs. Therefore the type modified by BHAs must be pursued below the propositional level. According to the current literature on event semantics (Davidson 1980; Parsons 1990) the semantic type closer to and lower than the proposition is the event. The point to be made in the rest of this paper is that the event5 is the semantic type required by BHAs. Fortunately, the empirical evidence on the linguistic 5 My use of the term event will be as broad as in the Neo-Davidsonian tradition (Parsons 1990), i.e. it will cover instances of genuine events as well as states. SEMANTIC CONSTRAINTS IN THE DISTRIBUTION OF ENGLISH HIGHER ADVERBS 111 realisation of the semantic type event is large enough to allow us to test the validity of this claim. In the sections that follow, I will check the behaviour of MHAs and BHAs in relation to some grammatical phenomena generally assumed to be related to the expression of events. 5.1. Perceptual reports The exact syntactic import of events is the focus of much current empirical research. An eventive denotation has been traced in nominal (Parsons 1990, 132135; Grimshaw 1990, 47-63), functional (Travis 1988; Kratzer 1994) and clausal projections (Higginbotham 1983; Vlach 1983). Especially convincing are the arguments in favour of the event denotation of the so-called naked infinitive complements to perception verbs. As Kearns (2000: 193-195) notes, the complements to the verb see in (45) and (46) are not of the same semantic type. Specifically, the latter denotes a proposition, while the former clearly does not: a proposition cannot be physically perceived. (44) Maurice saw Lydia talk to Ralph twice. (45) Maurice saw that Lydia talked to Ralph twice. Postulating events as the semantic denotation of naked infinitive complements to perception verbs is not only intuitively appealing but also empirically adequate. It has, for example, been noticed that the complement clauses in (45) and (46) are not equally transparent to reference. Referentially identical definite descriptions bring about different truth conditions in the complement clause of (46), but not in that of (45). This can be seen by comparing (45) and (46) with (47) and (46), respectively. In a context where Ralph is the Dean, (45) and (47) are synonymous, while (46) neither entails nor is entailed by (48). (46) Maurice saw Lydia talk to the Dean. (47) Maurice saw that Lydia talked to the Dean. Higginbotham (1983), Vlach (1983) and Parsons (1990: 15-17) argue that the different referential transparency of the complement clauses of see in (45) and (46) is due to the fact that the former denotes an event while the latter denotes a proposition. Turning to higher adverbs, it is interesting to observe that infinitival complements to perception verbs discriminate between the two classes in (13). As (49) and (50) show, only BHAs are acceptable in these contexts. (48) Maurice saw Lydia immediately/ intentionally/ briefly talk to the Dean twice. (49) * Maurice saw Lydia obviously/ reportedly/ intelligently talk to the Dean twice. It hence seems that, unlike MHAs, BHAs are fully compatible with, and indeed demand, the expression of an event. The possibility arises of relating BHAs JOSÉ MARÍA GARCÍA NÚÑEZ 118 6.2. Spanish –mente adverbs As already pointed out, right displacement of adverbs is an optional syntactic operation in English. This allows the adverb to be superficially placed in two different positions, one of them marked for focus, the other a topicalised position. A different constellation of facts is observed in the case of Spanish -mente adverbs. These adverbs are like prepositional phrases in requiring right-displacement.10 At the same time, it follows from the above picture of facts that monoperipheral higher –mente adverbs will block right-displacement for semantic reasons (viz they cannot be focused). The combined outcome of this morphosyntactic and semantic constraint on linearisation is that Spanish MHAs will be confined to explicitly topicalised positions in the left periphery. This is born out by the acceptability pattern in (96)-(101). (95) * Carmen descuidadamente olvidó las llaves a la vista de todos Carmen carelessly forgot the keys within everyone’s sight (96) * Carmen felizmente había depositado las llaves en un lugar seguro. Carmen happily had left the keys in a safe place (97) * Carmen olvidó las llaves a la vista de todos descuidadamente Carmen forgot the keys within everyone’s sight carelessly (98) * Carmen había depositado las llaves en un lugar seguro felizmente Carmen had left the keys in a safe place happily (99) Descuidadamente, Carmen olvidó las llaves a la vista de todos Carelessly, Carmen forgot the keys within everyone’s sight (100) Felizmente, Carmen había depositado las llaves en un lugar seguro Happily, Carmen had left the keys in a safe place On the basis of the above analysis, we expect Spanish higher –mente adverbs to pattern with English higher –ly adverbs in contexts where their different morphosyntax does not bring about a difference in distribution. One such context is provided by passive constructions. As is well known, Spanish –mente adverbs are allowed to freely occur in the left periphery of passive sentences. (101) Las preguntas habían sido (cuidadosamente) contestadas (cuidadosamente). The questions had been (carefully) answered (carefully). Under the circumstances, the theory put forward above predicts that the semantic constraints posited above will be in force, yielding the distributional pattern identical to the one found in English, i.e. allowing higher readings of –mente MHAs when these adverbs occur to the left of the lexical verb and blocking them when they occur in the right periphery. The prediction is born out by facts. Only in 10 This may be due to –mente being like a preposition in morphosyntactically triggering rightdisplacement. For an antisymmetric comparison of –ly and –mente along these lines cf. Déchaine and Tremblay (1996) and references therein. SEMANTIC CONSTRAINTS IN THE DISTRIBUTION OF ENGLISH HIGHER ADVERBS 119 (103) and (104) is it possible to impose a higher (non manner) reading on descuidadamente and felizmente, respectively. (102) Las llaves habían sido descuidadamente olvidadas a la vista de todos The keys had been carelessly forgotten within everyone’s sight. (103) Las llaves habían sido felizmente depositadas en un lugar seguro The keys had been happily left in a safe place (104) ?? Las llaves habían sido olvidadas a la vista de todos descuidadamente. The keys had been forgotten within everyone’s sight carelessly. (105) ?? Las llaves habían sido depositadas en un lugar seguro felizmente. The keys had been left in a safe place happily. 7. CONCLUDING REMARKS This study has argued for the grammatical plausibility of the class of BHAs. I have put forward evidence concerning perceptual reports, focusability and quantifier adverbs that what distinguishes BHAs from the rest of higher adverbs is the fact that they predicate of the event variable. It has also been shown that, once the connections between eventivity and focusability are properly spelt out, the special distribution of BHAs ceases to be a mystery. The biperipheral occurrence of BHAs follows straightforwardly from their eventivity and from the economy of right-displacement. Likewise, it has been shown that the fact that the rest of higher adverbs are not related to the semantic type event is the reason for their not being right-displaceable. Looked at from a theoretical perspective, the proposed semantic theory has the merit of providing a motivated account of facts that remain obscure under current syntactically or modification oriented approaches. Obviously, the possibility exists that the semantic type event syntactically correlates with some or other functional projection. This issue is the focus of much current research (Kratzer 1994, Pustejovsky and Tenny 2000). However, even though this turned out to be the case, a strictly syntactic account of the distribution of higher adverbs would still be expected to offer the kind of trivial connection between adverb types and linearisation which the analyses presented above provides. In a broader theoretical perspective, the work bears out some ideas and proposals in current syntactic and semantic literature. On the one hand, the analysis clearly supports the widespread position that right-peripheral occurrences of adverbs are derived from left-peripheral ones and that, in the absence of morphosyntactic triggers for right-displacement, this movement operation is semantically rather than syntactically constrained. On the other hand, the empirical results of this work support Davidson’s claim that events are a fundamental constituent of the semantics of natural language. More specifically, the data and arguments handled above support Herburger’s (2000) position that the semantic JOSÉ MARÍA GARCÍA NÚÑEZ 120 import of focus is to arrange the internal structure of the event quantification expressed by the sentence. REFERENCES Alexiadou, Artemis. Adverb Placement: A Case Study in Antisymmetric Syntax. Amsterdam: John Benjamins, 1997. Andrews, Alexander. “A Note on the Constituent Structure of Modifiers.” Linguistic Inquiry 14 (1983): 695-697. Atlas, Jay. “Topic-comment, Presupposition, Logical Form, and Focus Stress Implicatures: the Case of Focal Particles Only and Also.” Journal of Semantics 8 (1991): 127-147. Bellert, Irena. “On the Semantic and Distributional Properties of Sentential Adverbs.” Linguistic Inquiry 7 (1977): 337-351. Chomsky, Noam. “Deep Structure, Surface Structure and Semantic Interpretation.” Semantics: an Interdisciplinary Reader in Philosophy, Linguistics and Psychology. Eds. Danny Steinberg and Leon Jakobovits. Cambridge: Cambridge University Press, 1971. 183-216. Chomsky, Noam. Essays on Form and Interpretation. New York: North-Holland, 1977. Cinque, Guglielmo. “A Null Theory of Phrase and Compound Stress.” Linguistic Inquiry 24 (1993): 239-297. Cinque, Guglielmo. Adverbs and Functional Heads. New York: OUP, 1999. Davidson, Donald. Essays on Actions and Events. Oxford: Clarendom, 1980. Déchaine Rose-Marie, and Mireille Tremblay “Adverbial PPs and Prepositional Adverbs in French and English.” University of Calgary Working Papers in Linguistics 7 (1996). Emonds, Joseph. A Unified Theory of Syntactic Categories. Dordrecht: Foris, 1985. Ernst, Thomas. Towards and Integrated Theory of Adverb Position in English. Bloomington: Indiana University Linguistic Club, 1984. Ernst, Thomas. “On the Odd Syntax of Domain Adverbs.” Language Research 21/ 2 (1985): 167-177. Ernst, Thomas. “M-command and Precedence.” Linguistic Inquiry 25 (1994): 327-334. Ernst, Thomas. “Semantic Features and the Distribution of Adverbs.” ZAS Papers in Linguistics 17 (2000): 79-99. Ernst, Thomas. The Syntax of Adjuncts. Cambridge: Cambridge University Press, 2002 Frey, Werner. “Syntactic Requirements on Adjuncts.” ZAS Papers in Linguistics 17 (2000): 107-135. Grimshaw, Jane. Argument Structure. Cambridge, Mass.: MIT Press, 1990 Heim, Irene. The Semantics of Definite and Indefinite Noun Phrases. PhD thesis, University of Massachusetts at Amherst, 1982 Heny, Frank. “Sentence and Predicate Modifiers.” Syntax and Semantics Vol. 2. Ed. John Kimbal, New York and London: Seminar Press, 1973. 217-245. Herburger, Elena. What Counts: Focus and Quantification. Cambridge, Mass.: MIT Press, 2000. Higginbotham, James. “The Logic of Perceptual Reports: an Extensional Alternative to Situation Semantics.” Journal of Philosophy 80 (1983): 100-127. Infantidou-Trouki, Ellie. “Sentential Adverbs and Relevance.” Lingua 90 (1993): 69-90. Jackendoff, Ray. Semantic Interpretation in Generative Grammar. Cambridge, Mass.: MIT Press, 1972. Kayne, Richard. The Antisymmetry of Syntax. Cambridge, Mass.: MIT Press, 1993. SEMANTIC CONSTRAINTS IN THE DISTRIBUTION OF ENGLISH HIGHER ADVERBS 121 Karttunen, Laurie, and Stanley Peters “Conventional Implicature.” Syntax and Semantics 11. Eds. Choon-Kyu Oh, and David Dinneen, New York: Academic Press, 1979. 1-56. Kearns, Kate. Semantics. New York: St. Martin’s Press, 2000. Kratzer, Angelika. “On External Arguments.” University of Massachusetts Occasional Papers in Linguistics 17 (1994): 103-130. Larson, Richard. “Double Objects Revisited: Reply to Jackendoff.” Linguistic Inquiry 21 (1990): 589632 Lewis, David. “Adverbs of Quantification.” Formal Semantics of Natural Language. Ed. Edward Keenan, Cambridge: Cambridge University Press, 1975. 3-15. May, Robert. The Grammar of Quantification. MIT, PhD thesis, 1977. May, Robert. Logical Form: its Structure and Derivation. Cambridge, Mass.: MIT Press, 1985. Parsons, Terence. Events in the Semantics of English. Cambridge, Mass: MIT Press, 1990. Peterson, Paul. Facts, Propositions, Events. Dordrecht: Kluwer, 1997. Pustejovsky, James, and Carol Tenny, eds. Events as Grammatical Objects. Oxford and New York: OUP, 2000. Quirk, Randolph, Jan Svartvik, Geoffrey Leech, and Sidney Greenbaum. A Comprehensive Grammar of the English Language. London: Longman, 1985. Reinhart, Tanya. “Interface Strategies.” OTS, Utrecht University, 1995. Rizzi, Luigi. “The Fine structure of the Left Periphery.” Elements of Grammar. Ed. Liliane Haegeman Dordrecht: Kluwer, 1997. 281-337. Rooth, Mats. “Focus.” The Handbook of Contemporary Semantic Theory. Ed. Shalom Lappin, Oxford: Blackwell, 1994. 271-298. Shaer, Benjamin. “Syntactic Position and the Readings of Manner Adverbs.” ZAS Papers in Linguistics 17 (2000): 265-287. Thomason Richmond, and Robert Stalnaker. “A Semantic Analysis of Adverbs.” Linguistic Inquiry 4 (1973): 195-220. Travis, Lisa. “The Syntax of Adverbs.” McGill Working Papers in Linguistics. Special Issue on Comparative Germanic Syntax (1988): 280-310. Vlach, Frank. “On Situation Semantics for Perception.” Synthese 54 (1983): 129-152. Williams, Edwin. Thematic Structure in Syntax. Cambridge, Mass.: MIT Press, 1994. Wyner, Adam. Boolean Event Lattices and Thematic Roles in the Syntax and Semantics of Adverbial Modification. Ph.D. thesis, Cornell University, 1994.