Rappaport Hovav Research Articles

Unaccusative or unergative: The case of the English verb to die

The study examines the class status of the verb TO DIE in English. The verb under scrutiny (treated as a member of a semantically coherent class of disappearance verbs, together with disappear, expire, lapse, perish, vanish ) is tested against the six syntactic unaccusativity diagnostics valid for English. It is shown that three diagnostics do not work for the verb TO DIE , i.e., (1) auxiliary selection (inapplicable to all verbs in Modern English), (2) causative alternation, e.g., Philip died./ *The soldier died Philip, since the verb TO DIE belongs to a non-caused disappearance verb class, as argued by Levin (1993) and Levin and Rappaport Hovav (1995, 281–283), and (3) resultative constructions (although some collocations would be theoretically permissible, e.g., *He died stiff , no such sentences appear in the language corpus). On the other hand, the remaining three diagnostics yield positive results, i.e., (4) the occurrence of the adjectival participle, but with a reservation that it must be placed in a post-nominal position (*a DIED uncle / an uncle DIED in an accident), (5) there -insertion (the only existing example from the COCA Corpus: There DIED a myriad), and (6) the locative inversion diagnostics (the only instance found in the literature by Levin and Rappaport Hovav (1995, 303): (…) this year also DIED the possibility of turning the cup races (…)). Since only three unaccusativity tests out of the six mentioned above seem to work for the verb TO DIE , it might be problematic to treat it as a member of the unaccusative class. Additionally, the instances provided to illustrate the three diagnostics valid for the verb TO DIE rarely occur in the literature and the available corpora, and therefore they should rather be viewed as exceptions, which would cast serious doubt on the unaccusative status of the verb TO DIE . This would lead us to the conclusion that the verb TO DIE should be regarded as a real example of an Unaccusative Mismatch (Levin 1986), i.e., a clash between the results of two or more unaccusative diagnostics (discussed in Grimshaw 1987; Zaenen 1993; Levin and Rappaport Hovav 1995).

Forceful Contact in a Result Prominent Language

<p>Levin and Rappaport Hovav (1995) remind us that manner and result verbs often exhibit complementary distribution within a given language. They also note that when a main verb lexically specifies manner or result, the complementary component can be expressed outside the verb, in a satellite constituent of some sort. In Rappaport Hovav and Levin (2010), manner/result complementarity constrains verb root lexicalization. Building on this, Erteschik-Shir and Rapoport (2010) examine English verbs of contact, e.g. <em>smear</em>, <em>splash</em>, whose complements specify a result relation between moveable object and stationary locatum. Classically, these verbs show a locative alternation with holistic ~ partitive interpretations (Levin 1993).</p><p>For this paper I examine forceful contact expressions in Emai (West Benue Congo, Edoid in Williamson and Blench 2000). Relatively strict SVO, Emai manifests little inflectional morphology and few prepositions. Its motion predications express manner and result as one verb in series with another (<em>la</em> ‘run’ and <em>shan</em> ‘move through’ for ‘run through’) or as verb plus postverbal particle (<em>si<span style="text-decoration: underline;">o</span>n</em> ‘thread’ and <em><span style="text-decoration: underline;">o</span></em> ‘onto’ for ‘thread onto’).</p><p>Lacking verbs in series or postverbal particles, forceful contact in Emai reflects simple and complex predications. Predicates like transitive <em>hian</em> ‘strike’ or <em>so</em> ‘collide, bang into’ take a subject expressing moveable object and a direct object conveying locatum (<em>òjè só ùdék<span style="text-decoration: underline;">è</span>n</em> [Oje collide wall] ‘Oje collided with / banged into the wall’).</p><p>Related complex predications explicitly code contact means. English near equivalents are ‘punch,’ ‘kick,’ ‘peck,’ ‘bite’ and ‘pinch.’ Corresponding Emai predications with <em>hian</em> or <em>so</em> require means of contact, expressed as a body-part noun (<em>èkpà</em> ‘fist,’ <em>ízà</em> ‘heel,’ <em>úkpà </em>‘beak,’ <em>àk<span style="text-decoration: underline;">ò</span>n</em> ‘tooth,’ <em>éhì<span style="text-decoration: underline;">é</span>n</em> ‘fingernail’) covertly linked to a subject referent, followed by a preposition marked locatum. With a nonhuman locatum, body-part means (serving as moveable object) occurs in direct object position and a place noun locatum appears as preposition <em>vbi</em> object (<em>òhí só ékpá vbì ìtébù</em> [Ohi collide fist LOC table] ‘Ohi’s fist collided with the table/ Ohi’s fist banged on the table’). With a human locatum, the means body-part nominal is retained but a distinct body-part noun occurs as <em>vbi</em> object and its external possessor immediately follows the verb and precedes the moveable object (<em>òhí só ójé ékpá vbì <span style="text-decoration: underline;">è</span>ò</em> [Ohi collide Oje fist LOC face] ‘Ohi’s fist collided with Oje’s face / Ohi punched Oje in the face’). Human locatum predications also show an alternation lacking the <em>vbi</em> phrase (<em>òhí só ójé èkpà</em> [Ohi collide Oje fist] ‘Ohi’s fist collided with Oje / Ohi punched Oje’). There is however no simple transitive predication that combines as core arguments possessors, with or without their body parts (*<em>òhí só òjè</em> ‘Ohi punched Oje’), or subject possessor and locatum (*<em>òhí só ùdék<span style="text-decoration: underline;">è</span>n</em> ‘Ohi punched the door’). As a consequence, result is privileged in Emai expression of forceful contact, downgrading manner through use of nominal forms and their partitive relations. English, as manner prominent in this domain, reveals a contrary type, where manner is obligatorily verb expressed (<em>punch</em>) and result is optionally available via prepositional satellite (<em>on</em>). </p>

A Constructional-Cognitive Analysis of Chinese Directionals

This paper investigates the syntactic and semantic behaviour of directional items in Mandarin Chinese, proposing a constructional-cognitive model. Simultaneously, it argues against the traditional analysis of directionals as part of a compound. Instead, they are treated as proper larger constructions signaling subordination, based upon structural and functional evidence. Particularly, the verbs of motion are presented as catenative verbs (Aarts 2011, Huddleston and Pullum et al 2002) similar to those found in English, showing event complexity (eg Jackendoff 2002, Levin and Rappaport Hovav 2005). The two events then constitute mental spaces at the conceptual structure (Fauconnier 1994, 1997) and project into the entire motion-directional construction, which reflects the mappings of language and thought.

The Lexicon–Syntax Interface: Root Semantics As An Indirect Determinant Of Intransitive Verb Syntax In Cree

This paper presents a study of the (Northern East) Cree intransitive verb final (verbalizing suffix) -piyi ‘move’, which occurs in eventive unaccusatives and in two classes of unergative: vehicular motion verbs and verbs of emission. Generally, Algonquian intransitive finals systematically correlate with an external argument, or an internal argument, but not with both. We argue that -piyi is no exception, correlating with an internal (theme) argument. A null verb final, which correlates with an external argument (an internal causer), derives the unergatives, in which cases -piyi is lexicalized into a complex root. Root semantics are an indirect predictor of verb type in that they reflect patterns of legitimate pairing with either -piyi or the null final; e.g., -piyi pairs with root[STATE] to yield an unaccusative, while the null final pairs with the lexicalized root [root[PATH]+piyi] to yield an unergative. This analysis is compatible with a little-v analysis of finals (among others, Ritter and Rosen 2010). The unaccusatives have a spontaneous occurrence reading which, following Davis and Demirdache (2000), we attribute to -piyi suppressing the “causing process” sub-event in a causative representation. Our analysis is thus compatible with the view that some, if not all, unaccusatives are semantically causative (among others, Levin and Rappaport Hovav 1995).

The syntax of argument structure by Leonard H. Babby (review)

Reviewed by: The syntax of argument structure by Leonard H. Babby Peter Arkadiev Leonard H. Babby. The syntax of argument structure. Cambridge: Cambridge University Press, 2009, pp. xviii + 307. [Cambridge Studies in Linguistics, 120.] In the book under review, Leonard Babby,1 an eminent linguist renowned for his influential work on many aspects of the syntax of Russian, presents an ingenious and sophisticated theory of morphosyntax stemming from his previous studies and in many important ways diverging from the current mainstream of generative grammar. While for most proponents of Government and Binding and Minimalist Program “argument structure” is either read off the semantic representation of predicates (as in, e.g., Levin and Rappaport Hovav 1995as in, e.g., Levin and Rappaport Hovav 1998) or reduced to syntactic configurations (e.g., Borer 2005, Ramchand 2008), Babby argues for a view that argument structure is an autonomous module of grammar, and in fact the core one, from which syntactic representations are built, and which cannot be fully reduced to semantics. For Babby, argument structure is not just a static representation, but a dynamic component of language where various operations such as passivization, causativization, etc., apply, and which is crucial for the determination of the initial syntactic representation of clauses. Despite its explicit theory-constructing goals, the book is thoroughly empirically-oriented, presenting a number of in-depth case studies of Russian morphosyntax, such as the structure and syntax of “short” (SF) and “long” (LF) forms of adjectives, control of infinitives, participles and gerunds, and the predicate instrumental construction, each elucidating particular aspects of the theory. A reader, accustomed to associating the term “argument structure” with such issues as “unaccusativity,” “dative shift,” or “spray/load alternations,” might be surprised by Babby’s choice of subject matter. However, the book shows, and quite convincingly, [End Page 259] that the phenomena Babby investigates can be elegantly and revealingly accounted for from the chosen perspective. The book consists of an introduction, five chapters, endnotes, references and indices. In the introduction (pp. 1–10) Babby provides the reader with a very useful nutshell presentation of the main theoretical claims of the book and locates his conception of argument structure against the background of current generative theorizing. Importantly, Babby acknowledges from the outset that his theoretical thinking has been largely motivated by his empirical work on Russian, whose morphosyntax, being different from English in many important respects, cannot necessarily be adequately accounted for by “Anglocentric” approaches reducing all grammatical complexity to basic syntactic operations such as Merge and Move. In chapter 1 “The structure of argument structure” (pp. 11–73) Babby presents the conceptual and technical basics of his theory of argument structure and syntactic projection. The theory rests on the notion of diathesis consisting of two “autonomous but related tiers” (p. 13) of theta selection and category selection, in part going back to the corresponding notion developed in the Leningrad school of typology (Xolodovič 1974). Two assumptions about diathesis underlying the whole theory are the following: (i) all predicates and productive predicate affixes in all languages have the same universal diathesis structure (see below); (ii) syntactic operations such as movement cannot alter the predicate’s diathesis and, consequently, the basic grammatical relations in the clause. From (ii) it follows that many argument-structure-related operations often assumed to be syntactic, e.g., passivization or subject suppression with infinitives, occur at the level of argument structure and not in syntax proper. Therefore, argument structure is a dynamic morphosyntactic component of grammar where the lexically determined diathesis of the verbal stem is successively combined (“amalgamated”) with the diatheses of whatever productive affixes2 the verb attaches, and this interaction between the diatheses of the predicate stem and affixes determines the potential syntactic configurations in which the predicate occurs. Technically, the diathesis is a 2 × 4 grid where the upper tier represents up to three thematic roles and the lower tier, in turn, the syntactic categories corresponding to them; see (1) below for a tabular notation [End Page 260] and (2) for the corresponding linear notation of the diathesis of a three-argument ditransitive predicate. The order of theta positions in the upper tier and their mapping to phrase structure...

On the internal eventive properties of -er nominals

Deverbal -er nominals (driver, lifesaver, blender) may have eventive properties as they may or may not entail the occurrence of an actual event. For Rappaport Hovav and Levin (1992), eventivity is linked to animacy. More recent studies have concluded that eventivity is always present but comes in different aspectual flavors (e.g. episodic/dispositional, as in Alexiadou and Schäfer, 2010). This paper argues, on the basis of French data, that both views are partially correct in the sense that instruments, which are not animate, are never eventive, and that eventive nominals (always animate) fall into two groups depending on whether they involve an episodic event or a generic one. The eventive properties are demonstrated by the interpretation of -er Ns in combination with modifying adjectives, and specifically the event-related meanings of attributive adjectives (cf. Larson, 1998), a context often neglected by previous studies. For eventive nominals, the episodic/generic sense is contributed by the argument structure under the assumption that eventive nominals are built on complex verbal structure lacking with non-eventive Ns. We argue that -er nominals can be integrated in a larger view of deverbal nouns in a syntactic approach to complex word formation that distinguishes nouns derived from roots and argument structure nominals derived from full verbal phrases. Thus, we contribute to a view in which the correlation between event reading and argument structure (as commonly assumed for event-denoting deverbal nominals since Grimshaw, 1990) is uniformly maintained.

Improvised Causativization

AbstractIn this paper, I argue a special kind of causative alternation that arises from a speaker's immediate need to express the ‘forced control’ semantics of an imposed causer. This type of causativization, termed improvised causativization, is not limited by the argument structure of an intransitive base verb, whether it is unergative or unaccusative. Based on this observation, I propose that causativization be syntactically represented through a cause phrase that is added after a vP is completed; specifically, at the step of derivation when an intransitive profile is completed (cf. Levin & Rappaport Hovav 1995). Along with a claim in lexical semantics (Pylkkänen 2002, cf. Neeleman & van de Koot 2010) that cause relates a caused to a causing event, I argue that the mediating role of the cause is based on a set of felicity conditions. Under this analysis, variations at different levels, i.e., individual, dialectal, diachronic, or language acquisitional, can be accounted for without further stipulation.

Processing correlates of verb typologies: Investigating internal structure and argument realization

The paper investigates the processing correlates of core verb features used to generate verb typologies. The aim was to contrast the effects of verb internal structure (as in semantic/conceptual features e.g., Jackendoff [1991]; Levin and Rappaport Hovav [2005]) with argument structure and argument realization in sentence processing. To this end, we designed a self-paced reading task with internal structure and argument realization as variables in verb processing. Results showed that absence of a prototypical Agent or mismatch between thematic hierarchies and argument realization yields longer processing times possibly due to some form of thematic reanalysis or as a reaction to a “surprise” effect by the human processor while detecting a non-expected thematic assignment. No effects of verb internal structure were found. We take this result as an indication that argument structure and argument realization play an important role in verb recognition during sentence processing. We further propose that this reflects their prominent role in verb representation and we suggest that this finding could give additional support to verb classifications based on verb argument structure features compared to those based on internal structure.

Anticausatives in Sinhala: involitivity and causer suppression

Many recent theories of causative/inchoative alternations adopt an anticausativization analysis, wherein the inchoative is derived from the causative via some operation that eliminates the causer argument from a verb’s argument structure, provided the causer is semantically unspecified for agentivity (Levin and Rappaport Hovav 1995; Chierchia 2004; Koontz-Garboden 2009). We explore the properties of such an analysis for causative/inchoative alternations in Colloquial Sinhala, which are overtly indicated via a volitive/involitive stem contrast on the verb. We argue that the alternation arises from a causer suppression operation that deletes the causer syntactically but preserves it semantically, albeit formally marking it as unresolvable for agentivity. This prevents the verb from occurring in certain contexts requiring agentivity of a syntactically active argument, including the volitive stem, although such a reading may be derived pragmatically. The patient in turn shows a case alternation that we argue reflects two ways the suppressed causer can be interpreted—via reflexivization or existential binding. These data, we argue, support an anticausative analysis as elimination of causers unspecified for agentivity. They also expand the typology of ways agentivity is encoded and left unspecified, how causer elimination can occur, and what types of overt morphology can indicate the alternation.

L’influence de la volition sur la distribution des verbes à alternance causative dans les résultatives de l’anglais

This article is a detailed examination of the distribution of verbs undergoing the causative/inchoative alternation in the English resultative. More precisely, the purpose of this paper is to show, on the basis of an extensive corpus investigation of naturally occurring data, that already-existing analyses of the resultative structure – Simpson 1983, Levin & Rappaport Hovav (L&RH) 1998, 1999, 2001, Wechsler 1997 – are insufficient to account for the full range of data collected. On the one hand, Simpson’s and L&RH’s approaches fail to make accurate predictions as to the distribution of alternating change of state verbs in the resultative frame. On the other hand, although Wechsler’s analysis is much more promising, I show that it nonetheless needs to be refined in light of a previously unnoticed fact. Indeed, the distribution of verbs undergoing the causative/inchoative alternation in the COCA indicates that their intransitive variants are much more likely to occur in resultatives than their transitive counterparts. In order to account for this phenomenon, I propose that a new explanatory factor, namely volitionality, influences the acceptability of alternating verbs in the resultative frame.

Manner and Result in the Roots of Verbal Meaning

Rappaport Hovav and Levin (2010) argue that verbs fall into (at least) two classes: result verbs (e.g., break) and manner verbs (e.g., run). No verb encodes both manner and result simultaneously, a truth-conditional fact that Rappaport Hovav and Levin argue follows from how verb meanings are composed at the level of event structure. However, a key issue in verifying this claim is isolating truth-conditional diagnostics for manner and result. We develop and review a number of such diagnostics and show that there are verbs that encode both meanings together, counterexemplifying their truth-conditional complementarity. However, using evidence from scopal adverbs, we argue that when the meanings occur together, they are encoded in a single, undecomposable manner+result root at event structure. This fact validates complementarity as a fact about how many and what types of roots may occur in an event structure, though it also argues for a richer typology of roots than is typically assumed, including those encoding manner and result simultaneously.

PPs without Disguises: Reply to Bruening

Bresnan and Nikitina (2009) and Rappaport Hovav and Levin (2008) show that, contrary to standard assumptions, fixed-theme idioms may appear in to-constructions under certain pragmatic circumstances. Bruening (2010a) contends that the cases they present are in fact R(ightward)-dative shifts, double object constructions with the object projected to the right. In this article, we argue that Bruening’s proposed theoretical apparatus is unnecessarily complex and ad hoc and falls short of explaining the main facts it is supposed to deal with, massively overgenerating. A regular PP structure is argued to be empirically more adequate and conceptually simpler, avoiding the main problems of the R-dative shift analysis. New empirical evidence concerning pairlist readings and scope freezing also suggests that the empirical facts about idioms should be reconsidered in completely different terms.

The Role of Topicality in Subject Selection

The problem of subject selection is one of the frequently discussed topics concerning the lexical semantics-syntax interface. In the present article we have critically reviewed three representative approaches: the thematic hierarchy approach, Randall`s (2010) lexical decomposition approach, and Dowty`s proto-role approach. Dowty`s (1991) analysis seems to have greater explanatory power than the other two, but it is far from satisfactory in that it cannot account for subject selection in sentences containing symmetrical predicates or location subject predicates. Building upon the observations by Levin and Rappaport Hovav (2005), Wunderlich (2006), and Dowty (2000), we have claimed that topicality of an argument plays a crucial role in subject selection when the two coarguments have (almost) the same number of proto-role entailments and that the semantic idiosyncrasies of the location subject sentence are attributable to the topical nature of its subject, not to the polysemous nature of the predicate, as Dowty (2000) claims.

Mandarin Secondary Predicates

In the first part, I provide some data of secondary predicates in English and Mandarin on consequence-depictives (SUBJ-oriented) and resultatives (OBJ-oriented), which adopt an intransitive verb/adjective for their secondary predicate. In the second half, I present an account of the linking issue on ”resultative” compound predicates in Mandarin Chinese, building on the LFG/LMT work of Her (2007), who assumed that the argument structures of each predicate merge to give a composite structure, which determines whether a resultative sentence is semantically causative or not, and from which the arguments link to grammatical functions. I argue here that the facts require a more articulated semantics, for unlike Her's analysis, the determination of causativity and the linking of the arguments of the two predicates is fully an issue of semantics; specifically, I argue that there are two types of secondary predicates in terms of their semantics, namely those with internally- and externally-caused changes of state (see Levin and Rappaport Hovav: 1995, McKoon and Macfarland: 2000), which are respectively ”indirect-causative” and ”direct-causative”; causativity should be categorised into three types, non-causative, indirect-causative, and direct causative. I further argue that the argument undergoing internally-caused change always links to Actor and that the one undergoing externally-caused change (a truly ”affected” argument) always links to Undergoer.

Dative verbs

In a recent paper, M. Rappaport Hovav and B. Levin (2008) challenge the predominant view of the English dative alternation, which takes all alternating verbs to have two meanings and, concomitantly, associates each meaning with a particular syntactic realization (e.g. S. Beck and K. Johnson 2004, G. Green 1974, K. Hale and S.J. Keyser 2002, H. Harley 2003, M. Krifka 1999, 2001, R. Oehrle 1976, S. Pinker 1989). On this accepted view, the first meaning, a caused possession meaning, schematized in (4a), is said to be realized by the double object variant (the (a) sentences in (1)-(3)), while the second meaning, a caused motion meaning, schematized in (4b), is said to be realized by the to variant (the (b) sentences).

Possession of a Controlled Substantive: Light ‘have’ and Other Verbs of Possession

No abstract.

Remarks on causative verbs and object deletion in English

Rappaport Hovav and Levin [Rappaport Hovav, M., Levin, B., 1998. Building verb meanings. In: Butt, M., Geuder, W. (Eds.), The Projection of Arguments: Lexical and Compositional Factors. CSLI Publications, Stanford, pp. 97–134] contend that result verbs disallow object deletion because of their lexical semantic properties. Their point is that the distinction between result verbs and manner verbs with their different event structure representation constitutes the important factor, which dictates the possibility of the variation of argument realization, of which object deletion represents one instance. Responding to their claim, Goldberg [Goldberg, A.E., 2001. Patient arguments of causative verbs can be omitted: the role of information structure in argument distribution. Language Sciences 23, 503–524] presents the evidence which mainly concerns the object deletion of causative verbs (which correspond to result verbs) in English in order to show that the distinction is not substantial. The purpose of this paper is twofold. One is to present several pieces of evidence for Goldberg’s contention based on the behavior of causative verbs with respect to object deletion. It is made evident that at least some causative verbs in fact behave even freer than Goldberg’s principle predicts. The other purpose is to examine some aspects of Goldberg’s principle and demonstrate that it has a wider applicability than originally intended, which eventually indicates that the principle comprises a part of a general set of conditions on the object deletion in English. The conclusion is that object deletion is not so much sensitive to the distinction between causative verbs and non-causative verbs as Rappaport Hovav and Levin claim it to be, and therefore object deletion fails to be a good diagnostic tool for the differentiation of the two verb classes. In other words, causativity is not a good parameter for the possibility of object deletion.

Arguing our way to the Direct Object Restriction on English resultatives

Drawing on Hoekstra’s (1988f) work on so-called ‘small clause results’ and Marantz’s (1992) work on the way-construction and its relation to resultative constructions, in this article I argue my way to the conclusion that the so-called ‘Direct Object Restriction’ (DOR) on English resultatives must be reinstated, despite Rappaport Hovav and Levin’s (2001) claims to the contrary. First, I review some of the main properties of resultative constructions that appear to motivate the syntactic approach, whose main descriptive tenet is the DOR. In particular, I show that the present analysis of the conflation process involved in the formation of complex resultatives allows us to offer an adequate explanation of their syntactic properties. Second, I put forward a relational syntactic analysis of the so-called ‘way-construction’. In particular, I show that the present analysis helps us understand why the DOR holds for this idiomatic resultative-like construction as well. Finally, I deal with some exceptional cases put forward by Verspoor (1997) and Wechsler (1997), reviewed by Rappaport Hovav and Levin (2001), which appear to contradict the DOR.

Ne-cliticisation and split intransitivity

I consider a number of constructions with ne-cliticisation, which at first sight would seem to be problematic vis-à-vis the hypothesis that the Italian partitive clitic ne is a diagnostic of unaccusativity. Structures with ne-cliticisation can receive an existential interpretation in sentence focus. I argue that, in the putatively non-canonical domains, ne realises the argument of a stage-level existential predicate (see Carlson 1977; Diesing 1992; Pustejovsky 1995), which is not spelled out in syntax, but only figures in the semantic representation of the sentence. My findings highlight the role of focus structure in unaccusativity phenomena (see Van Valin 1993a; Levin & Rappaport Hovav 1995; Lambrecht 2000, among others), and support the analysis of split intransitivity in terms of non-deterministic correspondence between discourse, semantics and syntax.

Morphology, argument structure, and lexical semantics: the case of Spanish auto- and co- prefixation to verbal bases

This paper deals with two Spanish prefixation processes, namely auto- and co- prefixation, which affect the participants in the event expressed by the verbal bases to which they apply. From our point of view, these two prefixation processes provide new evidence for the claim that it is necessary to distinguish two levels of lexical representation in the study of the relation between lexical semantics, morphology, and syntax, in the line of Booij (1992), Levin and Rappaport Hovav (1995, 1998), and Sadler and Spencer (1998). We will show that only verbs belonging to a specific lexical class can undergo co- prefixation. Therefore, it seems that this process has to be accounted for at the level of lexical semantic structure. As for auto- prefixation, it will be shown that this morphological process can apply to verbs belonging to very different lexical classes. This seems to indicate that auto-prefixation has to be accounted for at the level of argument structure. However, we will show that the existence of an additional semantic restriction governing auto- prefixation makes it necessary to have reference both to a lexical syntactic level of representation like argument structure and to a semantic level of representation like lexical semantic structure.