Table Of Content
- About the author(s)/editor(s)
- About the book
- This eBook can be cited
- Valency in verbs and verb-related structures – A theoretical background (Anna Malicka-Kleparska / Maria Bloch-Trojnar)
- Passives, unaccusativity, and nominalisation (Artemis Alexiadou)
- How much verb in an adjective? – A DM perspective on the structure of passive potential adjectives in Polish (Maria Bloch-Trojnar)
- The omission of internal arguments in verbal and nominal structures (Bożena Cetnarowska)
- When a verb has more than one valence frame (Lars Hellan)
- The interaction between event structure and argument structure in light of some Hungarian and English facts (Éva Kardos / Gergely Pethő)
- Synthetic compound adjectives in Polish – A case of the morphology-syntax interface (Joanna Kolbusz-Buda)
- The morpho-syntax of reflexive impersonals in Polish (Anna Malicka-Kleparska)
- The unbearable lightness of HAVE(ing) (Marijana Marelj)
- Antipassive-like features in selected English and Polish object drop constructions (Katarzyna Mroczyńska)
- The internal structure of the synthetic compounds based on the passive participle in English (Sebastian Wasak)
- On the properties of arguments in Polish na-prefixed constructions (Sławomir Zdziebko)
- Series index
The John Paul II Catholic University of Lublin
The papers contained in this collection are the offshoot of the fruitful discussions by the participants of the “Valency in clauses and verb related structures” workshop held during the 5th meeting of the Linguistics Beyond and Within Conference (LINGBAW) at the John Paul II Catholic University in Lublin on the 18th and 19th of October 2017. Valency or argument structure (AS) is traditionally defined as the list of obligatory participants in a situation which receive syntactic expression. The father of the notion of valency is Tesnière (1959), who compared the build-up of the core elements of a clause to the chemical valency of the atom (cf. Przepiórkowski 2018). In his view, clauses are like molecules whose structure largely depends on the valency value of the verb. The presence of arguments (‘actants’ in his terminology) is required by the verb and “completes” its meaning, while adjuncts (‘circumstants’) are optional and their number and type are not limited by the lexical properties of the verb. Tesnière’s (1959) publication aroused immense interest in the valency properties of verbs and verb-related structures. This area occupies a central position in syntax-oriented generative linguistics, which views all non-pragmatic, strictly grammatical aspects of language as being ultimately derived from the generative syntactic skeleton dressed in morphological and semantic content. The structure determines the appearance of arguments and satellites, and so, in order to discover the grammatical structure of a language, the valency properties of different head elements (ultimately responsible for apportioning valency values) have to be studied in detail.1 Issues relating to valency and argument structure continue to fuel intensive research at the Lexicon-Syntax and Syntax-Semantics interface, as is evidenced by the myriad of recently published miscellanies (e.g. Erteschik-Shir and Rapoport 2005; Everaert, Marelj ← 7 | 8 → and Siloni 2012; van Gelderen, Barðdal, and Cennamo 2013; Bachrach, Roy and Stockall 2014 to mention just a few) and special issues of journals (e.g. Lingua 141 (2014) edited by Borik and Mateu; Lingua 149 (2014) edited by Arche, Fábregas and Marín; Word Structure 10(1) (2017) edited by Fábregas and Marín). This is because argument structure realisation raises fundamental questions regarding the overall structure of grammar and the role of particular modules.
In the 1980s and 1990s lexicon was the unchallenged locus of argument structure and argument structure alternations (Jackendoff 1990; Pustejovsky 1991, 1995; Levin and Rappaport Hovav 1995; Rappaport Hovav and Levin 1998). The introduction of the Minimalist Program in syntax (Chomsky 1993) coincided with the rise of the Distributed Morphology approach (Halle and Marantz 1993), which stripped the verbal root of its semantic content and paved the way for a more syntax-oriented perspective on argument structure (Arad 2003; Harley 1995; Borer 2005; Ramchand 2008). Grimshaw’s (1990) seminal monograph drew attention to AS-realisation in deverbal nouns and instigated a heated debate on the issue of AS-preservation in category-changing morphology and found an expression in literature so robust that it can hardly be related in an introductory chapter such as this. Nominalisations have recently been discussed in monographs couched both in the constructionist (e.g. Borer 2003; Alexiadou 2001) and lexicalist tradition (e.g. Bloch-Trojnar 2013), as well as collections of papers (edited by Giannakidou and Rathert 2009; Alexiadou and Rathert 2010; Bloch-Trojnar and Malicka-Kleparska 2017). The status and representation of participles and their corresponding adjectives has been of interest since Wasow (1977), giving rise to competing constructionist (e.g. Anagnostopoulou 2003; Embick 2004; Alexiadou and Anagnostopoulou 2008) and lexicalist accounts (e.g. Levin and Rappaport Hovav 1986; Sleeman and Verheugd 2000). Relatively recently deverbal adjectival structures, both potential and dispositional, have come under scrutiny with regard to their AS-licensing potential and aspectual properties (e.g. Massuet-Oltra 2013; Martin, Piteroff and Pross (eds.) 2016; Fábregas and Marín 2017).
There is general agreement that the verb and its arguments contribute to the interpretation of the syntactic structure in which they occur. However, the nature of their interaction and the extent of that contribution is far from settled. Two major approaches have crystalised in the ongoing debate: the projectionist or “Lexicon-driven” approach, and the constructionist or “Syntax-driven” approach. Section 2 will provide a thumbnail sketch of the main assumptions and developments within these competing research programmes. Let us clarify at the outset that the use of the term “constructionist/syntax-driven” in this paper is equivalent to what some researchers define as “neo-constructional”, as opposed to the “traditional” ← 8 | 9 → constructional approach of Goldberg (1995, 2005), Jackendoff (1997) and Ruppenhofer and Michaelis (2010). In the latter approach, constructions are viewed as conventionalised pairings of meaning with syntactic functions, thus obliterating the boundary between lexicon and syntax and effectively placing syntax in the lexicon. Section 3 is an overview of the problematic research questions in the verbal domain, whereas Sections 4 and 5 are devoted to valency phenomena in nominals and adjectives respectively. Section 6 is a brief résumé of the specific problems addressed by the contributions to this volume.
2. The projectionist vs. constructionist perspective on argument structure
The projectionist approach, which may well be traced back to Pānini’s Sanskrit grammar, assumes that the verb is listed in the lexicon together with information on the number of arguments it requires, their configuration and possible alternations. To put it in more up-to-date generative terms, the verb is listed along with the specification of theta-roles, Argument Structure and Event Structure, and the syntactic structures merely mirror or project lexical information in accordance with the Projection Principle (Chomsky 1986: 29). This approach is epitomised in the collaborative work of Beth Levin and Malka Rappaport Hovav (Levin and Rappaport Hovav 1995, 2005; Rappaport Hovav and Levin 1998) and it envisages the existence of a cross-linguistically motivated set of semantic elements, namely theta roles,2 which the verb selects and which are matched with syntactic arguments by a set of universal Linking Rules and expressed linguistically by means of bound morphemes or light verbs (Kiparsky 2002). Alternations or multiple argument realisations, as in (1) below, are regarded as the result of the application of lexical rules to verbal lexical entries (and the existence of multiple lexical entries) or else can be ascribed to polysemy: ← 9 | 10 →
|b.||Pat ran to the beach.|
|c.||Pat ran herself ragged.|
|d.||Pat ran her shoes to shreds.|
|e.||Pat ran clear of the falling rocks.|
|d.||The coach ran the athletes around the track.|
(Rappaport Hovav and Levin 1998: 98)
The preponderance of such multiple realisation patterns resulting in the multiplication of lexical entries and overburdening them with information, which is, to a large extent, predictable, has led to the development of constructionist approaches. They assume that the verb/root does not determine the construction but the constructional elements license some of the complement structure. In various strains within the constructionist paradigm elements of the verb meaning, thematic and aspectual information are encoded in the syntactic structure by means of various functional and light verb projections. In some approaches, certain structures can be conceived of as arising in the lexicon (e.g. the l-syntax of Hale and Keyser 1993; Harley 1995), in others they are syntactic all way down (Borer 2003, 2005; Ramchand 2008; Alexiadou 2001, 2010ab; Alexiadou et al. 2015). In constructionist approaches, aspectual and argument relationships are inextricably connected since they are expressed by the same structure via particular functional projections and, as a consequence, render the specification of the number and types of arguments redundant. As Ramchand explains “once we decompose according to event predicational structures, we can make participant semantics (thematic roles in traditional terminology) parasitic on event structure positions. The participant relations are built up recursively from successively embedded event descriptions and ‘subject’ predications. The semantics of event structure and event participants is read directly off the structure and directly off information encoded by lexical items” (Ramchand 2007: 489).
In Hale and Keyser (1993) some features (e.g. the full set of phonological features, a particular R complement, the manner feature) are regarded as inherent features of roots, which can be linked to internal or external arguments and licensed in their presence. In more radical approaches, couched in the neo-Davidsonian tradition (Ramchand 2007), meaning components directly project syntactic-structure-yielding aspectual information, argument number and thematic role information. The structure in (2), taken from Ramchand (2007: 492), is a representation of the transitive telic event of farmers drying cocoa beans. There are three subevental components in its event structure: causal/initiational, process ← 10 | 11 → and result state. Each of these subevents forms part of a hierarchical-embedded structure with its own subject, i.e. the initiator, undergoer and resultee:
Recognising the existence of subevents in the structure of a macro event makes it possible to circumvent the polyadicity problems besetting projectionist approaches. The representation in (2) captures the fact that the verb dry in English can be a pure process (3a), can have a resultative interpretation (3b), can be conceived of as a caused process (3c) or both simultaneously (3d) (Ramchand 2007: 490):
|a.||The cocoa beans dried on the racks (for three days).|
|b.||The cocoa beans dried bone dry (in just two days).|
|c.||The farmers dried the cocoa beans (for three days).|
|d.||The farmers dried the cocoa beans bone dry (in just two days).|
By severing the external argument from the verb and keeping the internal one as part of the AS of the verb, Kratzer (1996, 2000) and her followers (e.g. Alexiadou et al. 2015) can be placed in the semi-Davidsonian camp. They assume that external arguments are not arguments of verbs but are introduced by the Voice head. The presence of this functional projection can, among others, be diagnosed empirically by the possibility of agent-oriented adverbial modification, by-phrases and instrumental phrases. The internal argument and event implications are associated with the categorising head v.
3. Valency in verbs
As valency was initially linked with the verb alone, it is not surprising that the first in-depth studies dealing with valency were devoted to the clausal properties that pivoted on the verb. A monograph written by Beth Levin in 1985, and only published in 1993, entitled English verb classes and alternations was of crucial importance for the generative approach,3 since it supplied the standard for recognising various valency types (Levin uses the term ‘diathesis’) and their alternations. Soon afterwards other linguists followed suit, applying this canon to the valency systems of other Indo-European languages and refining it at the same time (see e.g. Alexiadou and Anagnostopoulou 2004; Reinhart and Siloni 2005; Pylkkänen 2008; Koontz-Garboden 2009; Alexiadou and Doron 2012; Doron and Labelle 2010, to name just a few).
Levin (1993) observes that certain verbs may appear in transitive and intransitive valency frames. The same verb may also take a given number of arguments but spell them out in distinct orders and forms (see especially Rappaport Hovav and Levin 1998; Segal and Landau 2012; Levin and Rappaport 2010). The arguments in transitive and intransitive frames may also be distinguished by different semantic characteristics and morphological build-up (see e.g. Reinhart and Siloni 2005). Depending on these characteristics, particular verb subclasses enter different frames of valency alternations.
In particular, the object of a transitive verb (which is treated as the basic verbal variant) may appear as the subject of an intransitive clause containing the same verb form. When the agentive participant in the predication is logically implied in the intransitive variant, this type of valency rearrangement is named the middle alternation (see e.g. Dixon 1991; Fellbaum 1986; Fagan 1992; Levin 1993) and the intransitive clause contains the middle verb. This is illustrated in (3) below, where the first sentence shows the regular transitive valency pattern (3a), while the second sentence is middle: the subject undergoes a change of state, while the agentive cause of the change is merely implied (3b):
A different type of relationship between a transitive and an intransitive sentence occurs where a transitive verb loses its direct object, or rather demotes it to the function of an adjunct. The resulting valency frame is called conative (see e.g. Dixon 1991; Levin 1993), as can be observed in (4) below. Here, the subject of the clause is unaffected by the valency rearrangement:
|(4)||Remember, though, that the sound will also cut at each shot-change, so it’s wise to make your in-camera picture cuts at times when no one is actually speaking.|
The event named by such a clause is characterised by the lack of completion of the activity (Levin 1993).
Another transitive/intransitive valency alternation with the object taking over as the subject of the intransitive verb is the causative/inchoative alternation. Here, however, no logical agent is implied by the clause (see e.g. Haspelmath 1987, 1993; Alexiadou and Anagnostoupoulou 2004; Koontz-Garboden 2009). (5a) below illustrates the causative use and (5b) its inchoative counterpart:
|a.||Baseball has had four black managers since Jackie Robinson broke the sport’s racial barrier in 1947.|
|b.||I can’t help it if Pitt has the kind of bones that break easily, can I?|
The distinction between middle and inchoative uses of intransitive verbs is a subtle one. It really depends on the meaning of the verb involved in forming a clause and on pragmatic considerations concerning how much agency is logically implied by a sentence, and consequently whether an intransitive clause has a middle interpretation. The problem is aggravated by the fact that in many languages (e.g. Slavic, Romance), like in English, middles and inchoatives are not differentiated by clear morphological exponents. However, the presence of adverbial agent-oriented elements in clauses may be treated as an indicator of their middle, rather than inchoative, character.
Another valency rearrangement can be identified where a verb loses its object but the object remains logically present, though non-specific. This is what Levin (1993) calls the unspecified object alternation but it is also referred to as the absolute use of verbal forms. The appearance of absolute uses of verbs has been discussed in detail by Ruppenhofer and Michaelis (2014) (see also Fillmore 1986; Resnik 1993, 1996; Goldberg 2005, 2006). An example of this valency alternation is given in (6) below:
|a.||Was she actually there and, if so, did she actually eat any of the food?|
|b.||His hands were manacled behind his back except when he ate or slept. ← 13 | 14 →|
Such “incomplete” valency is attributed, for the most part, to the indefinite properties of the object or to the semantics of the verb itself which prompts the identity of the object. Reflexive and reciprocal clauses are also characterised by this kind of valency (e.g. dress, hug). In the example above it is not important what is being eaten and the non-specific object food is self-evident. Among other potentially objectless verbs Levin (1993: 34) mentions verbs which point to body parts and, consequently, infallibly prompt the object of the clause, e.g. blink (eye), clap (hands), nod (head), etc. Other circumstances which are considered favourable for deleting logical objects may stem from the context which clarifies the structure of the event. For instance, the deleted argument can be an anaphor of an argument mentioned elsewhere:
|(7)||Zrobiła mi na drutach bezrękawnik-namiot. Był za długi, więc przywiozłam go do skrócenia […]ona pruła (zero object), ja zwijałam wełnę w kłębki. (Polish)
‘She had knitted me a tent-like sleaveless sweater. It was too long, so I brought it with me to be shortened […] she unpicked (the sweater), and I wound wool into skeins.’
Object deletion phenomena actually constitute one of the more obscure aspects of valency theory, perhaps because they are very closely linked to pragmatics and are notoriously difficult to handle by syntax alone.
We can note yet another interesting valency rearrangement effect if we consider dative alternation, which was probably the valency phenomenon granted the most attention in early generative studies (see e.g. Jackendoff 1990; Dowty 1979; Marantz 1984). The reason for this intense interest was the fact that this alternation seemed to embody a syntactic alternation relationship obtaining between two realisations of a single “deep” structure (see e.g. Akmajian and Heny 1975). Certain verbs alternate between the valency frames with two “bare” objects (8a), and one “bare” object and a PP object (8b):
|a.||His unusual topic gave Fry trouble with the title of his lecture.|
|b.||Both were brought in from the garden – home-grown – and never left the kitchen until Cook gave them to Edith for the table.|
In recent years such valency frames have been analysed as applied structures (Pylkkänen 2008).
The locative alternation also reveals interesting properties in that it is characterised by the exchange of functions between a bare and a prepositional argument. While in the dative alternation one of the arguments specifies in whose favour (or disfavour) the event is carried out, in the locative alternation the objects specify ← 14 | 15 → the substance and its location in two possible orders: either the location is spelled out by an NP and the substance by a PP (9a), or vice versa (9b):
|a.||Teams of workmen yesterday cleared oil-covered shingle from Calshot Spit (above), a nature conservation area, as part of a big clean-up operation after a tanker spilled 6,000 gallons of heavy Middle East crude into Southampton Water.|
|b.||Have they cleared that chicken off the roundabout?|
This valency rearrangement has attracted much attention as it seems to shed light on the amount of information which has to be entered into the lexicon and that which has to be supplied by the derivational component (see e.g. Levin and Rappaport Hovav 2010; Rappaport Hovav and Levin 2010; Alexiadou and Anagnostopoulou 2011; Koontz-Garboden and Beavers 2011; Beavers and Koontz-Garboden 2012; Segal and Landau 2012).
Some valency alternations are more language specific than those illustrated above. One such example is the locative preposition drop alternation, i.e. a valency rearrangement allowing the verb to appear with or without a characteristic preposition, as illustrated below in (10a) and (10b) respectively (Levin 1993: 43):
|a.||We climbed up the steps to intihuatana, the hitching post of the sun.|
|b.||I climbed the stairs to Charles’s office.|
In Slavic languages the preposition of the prepositional phrase is replaced by a prefixal element on the verb (see e.g. Svenonius 2004). Below, we give an example of this rearrangement taken from Polish.
|‘Julian ran across a dark yard.’|
|‘Julian crossed a dark yard.’|
Similar semantic effects seem to accompany this alternation both in English and in Slavic. Namely, the variant without the preposition is described by Levin (1993) as holistic, while that with the preposition as partitive. For the Slavic data, the distinction between perfective vs. imperfective seems to be a more accurate description. ← 15 | 16 →
Many verbal roots appear in an extensive spectrum of valency frames (see e.g. Alexiadou and Lohndal 2017: 91–92). We may illustrate this with some (shortened) examples taken from the BNC:
|a.||Not everyone who reads the book will be able to sympathise with Justin and Ursula. (transitive structure)|
|b.||It was true of Eliot, as it was true of Dr Johnson, that he needed to read himself into a writing disposition. (causative reflexive)|
|c.||In the original version, now lost, of the chapter in which she reads to Raskolnikov the gospel story of the raising of Lazarus. (applied: PP, NP)|
|d.||She reads him the story of the Gadarene swine, the novel’s Epigraph. (applied: NP, NP)|
|e.||The caption material under two images reads as follows: … (middle)|
|f.||Writing is wonderfully clear: once the book is opened it just reads itself out. (middle reflexive)|
|g.||He closes his book at seven and goes walking and then reads. (absolute)|
The spectrum of valency frames available to particular verbs depends on the verb and the language which is being examined.
To make this richness of form and meaning even more amazing, verbal valency is pliable, depending on the pragmatic context. For example, a transitive verbal structure can be used in an intransitive way or vice versa, as exemplified in (13) below. The sentences (taken from the BNC) show that the basically transitive verb fish can be used in a middle (intransitive) structure (13a) or the basically intransitive verb march can be used in a transitive causative frame (13b):
|a.||During heavy conditions, Arbroath Breakwater fished well.|
|b.||Each pupil had to detail a movement in front of the squad pretending to be a drill sergeant and my detail was to get my squad from the stand easy position, slope arms and march them some 50 yards, then bring them back to the original position order arms and stand at ease.|
Such linguistic creativity constitutes a problem for approaches that link verbal valency with root semantics. For instance, Levin (1993) expresses the opinion that valency types and alternations are not so much inherent and unpredictable properties of verbs but rather result from identifiable elements in their semantics. She claims that the middle valency frame can be found with verbs that signify the uncaused change of state, and the causative/inchoative alternation when the change of state is caused. Similar views on verbal valency properties are voiced in other lexicalist writings (see e.g. Levin and Rappaport Hovav 1995; Rappaport Hovav and Levin 1998). ← 16 | 17 →
However, later generative studies show very clearly that the relationship of the valency frame with the verb meaning can be quite illusory. Borer’s (2005: 13) approach represents the most radical form of the constructivist attitude to valency. According to her, not only the meanings of verbs but also the category of lexical items result from the construction in which particular morphemes are placed. Consequently, it is the valency properties of a construction that decide about the semantics of a clause, and not vice versa. In other words, a morpheme may be moulded into a specific valency frame if the need arises. Her views are shared by e.g. Marantz (2005) and Ramchand (2008). They find support in examples such as those given in (13) above. However, the fact that such uses are felt to be creative may argue against this line of reasoning.
Many questions await definite answers and verbal valency phenomena continue to open up new areas for further inquiry.
4. Valency in deverbal nominals
Argument Structure preservation/realisation or, to be more precise, the obligatory presence of verbal arguments in deverbal nominalisations remains a matter of debate and a major bone of contention between the adherents to constructionist theories (e.g. Borer 2003, 2005; Alexiadou 2001, 2009, 2010ab; Harley 2009; Roeper 2005) and lexicalist models (e.g. Newmeyer 2009; Bierwisch 2009).
- ISBN (PDF)
- ISBN (ePUB)
- ISBN (MOBI)
- ISBN (Hardcover)
- Publication date
- 2019 (April)
- valency morpho-syntax syntax-semantics interface argument structure event structure
- Berlin, Bern, Bruxelles, New York, Oxford, Warszawa, Wien, 2019. 263 pp., 1 fig. b/w, 7 tables