UNIVERSITA’ DI PISA
Dipartimento di Filologia, Letteratura e Linguistica
Corso di Laurea Magistrale in Linguistica
TESI DI LAUREA
Towards an Italian VerbNet:
a descriptive analysis of argument alternations.
Candidata:
Veronica Viola
Relatore:
Prof. Alessandro Lenci
Correlatore:
Prof. essa Giovanna Marotta
Esperto Esterno:
Dott. Gianluca Lebani
ANNO ACCADEMICO 2012-2013
“In the world “out there,” there are no verbs, no speech events, and no adjacency pairs. There are particles of matter moving around in certain recurrent and yet not fully predictable patterns. We interpret such experiences as and through symbolic means, including linguistic expressions. That’s what it means to be human.” (Alessandro Duranti) To Emmet, for standing by my side while I face the daily challenges of being human. 2 Contents Contents ......................................................................................................................................... 3 Introduction .................................................................................................................................. 5 Chapter 1. Analysing verbs at the syntax-­‐semantics interface .................................... 7 1.1. The necessity of an integrated syntactic-­‐semantic framework ......................... 7 1.2. Lexical semantic representation ................................................................................ 10 1.2.1. The meaning-­‐based approach .............................................................................................. 12 1.2.2 Aktionsart: Vendler’s verb classes ....................................................................................... 13 1.2.3 Verb valency and argument structure ............................................................................... 16 1.2.4 Fillmore’s case frames and Thematic Roles ..................................................................... 20 1.2.5 From verb meaning to the underlying event structures ............................................ 26 Chapter 2. VerbNet: a verb classification based on argument alternations ......... 29 2.1. Argument Alternations: what they are and why they matter ........................... 29 2.1.2. Traditional argument alternations ..................................................................................... 31 2.1.3. Event Composition .................................................................................................................... 33 2.2. Verb classes ........................................................................................................................ 35 2.3. VerbNet: a computational verb lexicon for English .............................................. 36 Chapter 3. Towards a classification of argument alternations in Italian .............. 40 3.1. Subcategorisation frames for a sample of Italian verbs ..................................... 40 3.1.1. Resources ....................................................................................................................................... 40 3.1.2. Carving verbs subcategorisation frames ......................................................................... 43 3.2. Data extraction ................................................................................................................. 45 3.2.1. Automatic extraction of argument alternation candidates ...................................... 46 3.2.2. Manual identification of Italian frame alternations .................................................... 47 Chapter 4. A descriptive analysis of argument alternations in Italian .................. 48 4.1. Alternations that involve argumental sentences .................................................. 48 4.1.1. Alternations involving argumental sentences in a complement position ......... 49 4.1.2. Alternations involving argumental sentences in the subject position ................ 59 4.2. Alternations that involve noun or prepositional phrases (NP or PP) ............ 62 4.2.1. Alternations causing a change in the transitivity of the verbs ............................... 63 4.2.2. Alternations involving two intransitive variants ......................................................... 79 4.2.3. Alternations involving verbs with predicative complements ................................. 82 3 4.3. Alternations involving an argumental sentence and a phrase complement (NP or PP) ................................................................................................................................... 83 4.3.1. Alternations affecting a complement position .............................................................. 83 4.3.2. Alternations taking place in the subject position ......................................................... 94 4.3.3. Alternations requiring predicative complements ........................................................ 96 Conclusions ................................................................................................................................ 99 Acknowledgments ................................................................................................................. 105 Bibliography ............................................................................................................................ 107 APPENDIXES ............................................................................................................................. 116 APPENDIX A – List of Subcategorisation Frames in Il Sabatini & Coletti (2012)
...................................................................................................................................................... 116 APPENDIX B – List of the 1000 Top Frequent Italian Verbs Analysed in our Sample (see CD-­‐ROM) ........................................................................................................... 117 APPENDIX C – List of Subcategorisation Frames identified in the annotation procedure ................................................................................................................................. 118 APPENDIX D – Argumental Sentence Alternations ..................................................... 124 APPENDIX E – Alternations involving Noun or Prepositional Phrases (NP or PP)
...................................................................................................................................................... 127 APPENDIX F – Alternations involving an argumental sentence and a phrase complement (NP or NP) ....................................................................................................... 130 4 Introduction In this thesis we explore the linguistic phenomenon known as “argument alternations” (Levin & Rappaport 2005) in relation to Italian. Given its widespread diffusion in natural languages and the intellectual challenges it encourages, the study of this particular aspect has interested all researchers involved in both syntax and semantics, regardless of the theoretical framework in which they place their work. Unfortunately, as is the case with many other linguistic phenomena, the vast majority of the research conducted so far on argument alternation has concentrated almost exclusively on English, paying very little or no attention at all to other languages (DeLancey 1995; Guerssel et al. 1985). In relation to Italian, argument alternations have been the object of dedicated research only in the last decade and quite sporadically; also, most of the time, these works focus only on few specific different alternations (Cennamo (forthcoming); Jezek 2003; Lenci 2009; Lenci (in press)), while a complete and detailed classification of all the ones allowed in Italian, comparable to that available for English (Levin 1993), is still lacking. This thesis represents a first attempt to fill this gap, providing a descriptive analysis of argument alternations in Italian based on a sample of the 1000 top frequent Italian verbs, derived from a lexicographic resource and manually annotated, and semi-­‐
automatically extracted. In Chapter 1 we outline a brief history of the major theories of lexical semantic representation, highlighting their strong points, but also the issues that each of them inevitably raises and the problems they left unsolved. We also argue in favour of a theoretical framework that should embrace syntax and semantics at once, as demonstrated by the inherent incompleteness and poor predictive power of all those approaches that have focused on only one of these linguistic levels of investigation. In Chapter 2 we present the phenomenon investigated in this thesis, namely “argument alternations”, which we describe with a specific focus on the theoretical challenges they posit, particularly in relation to two of the most recent and debated contributions in the field, put forward by Roland & Jurafsky (2002) and Beavers (2006) respectively. In order to provide the most complete framework possible, we also briefly explore a similar and equally intellectually demanding phenomenon, commonly referred 5 to as “event composition”. Lastly, we describe VerbNet (Kipper et al. 2000; Kipper-­‐
Sculer 2005), the computational lexical resource available for English that not only has proven to be extremely valuable in both linguistics and Natural Language Processing (NLP) research, but that has also inspired this work. In Chapter 3 we show the methodology used to, on one hand, collect our data, illustrating in detail the lexical resources exploited (Il Sabatini & Coletti 2012; Lenci et al. 2012) and the annotation procedure, and, on the other, to semi-­‐automatically extract what we had set out to investigate, namely argument alternations in Italian. In Chapter 4 we provide a descriptive analysis of our findings, describing in depth each and every alternation we found: for most of them, we also attempt a first and necessarily partial semantic classification of Italian verbs, mainly following Levin’s (1993), but also proposing novel classes when it was felt that none of the ones available in her seminal work were compatible with our verbs. Moreover, it was thought best to divide our results into three major groups, according to the type of argument realisation allowed by each alternation, which could consist of: 1) two argumental sentences, 2) two noun or prepositional phrases, and 3) an argumental sentence and a noun or prepositional phrase. Hence, in order to build an Italian VerbNet, the necessary first step to undertake was the investigation and the consequent identification of argument alternations allowed by Italian verbs, and this is exactly what we tried to achieve with this thesis. Moreover, taking into consideration the complexity of the linguistic phenomenon we explored and the difficulties intrinsic to such an ambitious purpose, it is worth stressing that the semantic classification proposed is only a first attempt to organise Italian verbs and future research will most certainly have to focus on this aspect. These necessary premises notwithstanding, we hope that this work will encourage the Italian research community, and whosoever would like to conduct their work on Italian verbs, to carry out further research on argument alternations in Italian, not only extending the sample we used, but also basing the analysis on sources of information different from the ones we exploited. In fact, it is only with the collection of more corpus-­‐based evidence that the claims made in this thesis could be fully validated for Italian. 6 Chapter 1 Analysing verbs at the syntax-­‐semantics interface 1.1. The necessity of an integrated syntactic-­‐semantic framework The study of meaning in language, known as semantics, has interested the greatest thinkers and philosophers for a very long time, deeply fascinated and intrigued by the unique ability of mankind to interpret the world around them and express it through words. However, it was not until the second half of the last century that semantics truly became a central aspect in the description of the language faculty proposed not within philosophical traditions, but within proper linguistic frameworks (Berruto 1976; Cann, Kempson & Gregoromichelaki 2009). In modern linguistics, the interest in the way words represent meaning, or lexical semantics, can be traced back to Ferdinand de Saussure (Geeraerts 2010), whose intuitions have resulted in the so-­‐called “structural semantics”, a branch of the broader approach known as “structural linguistics”. In particular, De Saussure (1916) put forward a model according to which a linguistic sign or unit is a double entity consisting of a signifier (its phonic component realised as a sequence of either graphemes or phonemes) and a signified (the mental image evoked by the signifier), distinguishing thus between form and substance (De Saussure 1916, pp 97-­‐100). In an attempt to provide semantics with a formal and rigorous descriptive apparatus, Saussure’s model was first developed by Hjelmslev (1957), who not only renamed signifier and signified as expression plane and content plane respectively, but also proposed for the first time ever to adapt the analysis in distinctive features and phonemes used in phonetics to semantics, introducing the semantic features (Hjelmslev 1961). This theoretical framework was subsequently enriched by the work of Greimas (1966), who applied to semantic analysis the specular terms seme (the smallest unit of meaning recognisable) and sememe (a semantic unit made up by the combination of two or more semes). This interest for lexical semantics as a main topic of linguistic research was partly overshadowed by the advent of the transformational-­‐generative model, considered by many “the most radical and important change in direction in descriptive linguistics and in linguistic theory that has taken place in recent years” (Robins 1967, p. 226). This approach was inaugurated in 1957 with the publication of Chomsky’s Syntactic 7 Structures and became the dominant theoretical model in linguistics when his second main work, Aspects of the Theory of Syntax, was published in 1965. As the titles of Chomsky’s seminal works themselves would suggest, his original theory was built on the premise that syntax was the core component (or module: see Fodor 1983) of the language faculty and that the analysis of syntactic behaviour in different languages was the key to understanding it, thus giving very little prominence to other linguistic levels such as lexicon. As a partial reaction to the wide spread of Chomsky’s model, since the 1970s, over the past forty years we have witnessed a shift in the general focus of theoretical linguistics, from merely looking at how words can and do combine with each other (syntax) to the study of how actual meaning is represented in those words (lexical semantics), giving rise to many different theoretical frameworks which have since been put forward, such as Lexical Functional Grammar (Bresnan 1982), Functional Grammar (Dik 1978, 1989, 1997; Vossen 1995; Olbertz et al. 1998), Role and Reference Grammar (van Valin & Foley 1984) and the Generative Lexicon (Pustejovsky 1995). Moreover, notwithstanding the syntax-­‐centered nature of the generative school, generative linguists themselves begun to take lexicon more seriously in the light of the newly collected evidence, and Chomsky himself presented a revised version of his theory in which semantics was finally incorporated and its key role in understanding the nature of the language faculty recognised (Chomsky 1981). It is precisely in this renovated interest for semantics that one must place all the attempts made to study and understand verbal behaviour. The category of verbs has in fact drawn attention since ever and Plato himself viewed verbs as one of the “two kinds of vocal indication of being” (Plato 1921, 261.e), along with nouns, underlying how linguistic meaning does not truly emerge: Theaet. “In what way?” Str. “ For instance, “walks”, “runs”, “sleeps”, and the other verbs which denote action, even if you utter all there are of them in succession, do not make discourse for all that.” Theaet. “No, of course not.” Str. “And again, when “lion”, “stag”, “horse”, and all other names of those who perform these actions are uttered, such a succession of words not yet make discourse; for in neither case do the words uttered indicate action or inaction 8 or existence of anything that exists or does not exist, until the verbs are mingled with the nouns; then the words fit, and their first combination is a sentence, about the first and shortest form of discourse” [Plato 1921, 262b-­‐c] It is worth stressing how this passage of Plato’s Sophist seems to have anticipated the field of research known today as “linguistic interfaces”, by which we commonly refer “to the informational connections and communication among putative modules within the grammar” (Rachmand & Reiss 2007, p. 2): this new approach has been proving so fertile and successful for the understanding of the architecture of language that it does not come as a surprise that, in recent years, a significant part of the most intellectually engaging and challenging research has come precisely from the study of interfaces. Amongst the many possible interfaces, the syntax-­‐semantics one has received great attention, and for obvious reasons: not only it bridges a gap which for so long was considered unbridgeable, but it also deals with one of the most intriguing facts about language, namely the rising of meaning through the combination of lexical items, primarily that of verbs and nouns, as Plato’s aforementioned quote stresses. Hence, linguists have been facing the difficult task of elaborating models of lexical representation that would encompass both semantic and syntactic aspects at once, and, even though the theoretical frameworks so far proposed vary substantially in relation to one other, they all agree on the key role of verbs and of the mechanisms through which they project their lexical information onto the nouns they combine with: an apparently basic yet fundamental linguistic match, which has been defined –and rightly so – “the backbone of language and thought” (Moro 2012, p. 20). It has long been known that every single verb “evokes a certain number of elements in respect to which something is predicated” (Jezek 2003, p. 18): these elements commonly go under the name of arguments and they are syntactic realisations of the lexical-­‐semantic information that each and every verb carries. The operations through which this information is passed from the semantic level onto the syntactic one is an issue known in linguistics as mapping, and the analysis of the possible syntactic expressions of a verb’s arguments is called argument realisation: (1) a. Leslie swept b. *Kelly broke [Levin & Rappaport Hovav 1998, p. 102] 9 As the examples in (1) shows, a verb like to sweep is grammatically acceptable with only one argument (the subject), even in the absence of any context, whereas a verb like to break is not, unless we add a second argument (the object) that must be always syntactically realised: (2) Kelly broke the dishes/the glasses/her finger/the TV Because of its inherently twofold nature, this area of linguistics has appealed to many scholars and researchers, who turned it into the leading field for the study of the interaction between the syntactic level and the semantic one. In fact, as we briefly mentioned above, “since the 1980s, many theories of grammar have been built on the assumption that the syntactic realisation of arguments – their category type and their grammatical function – is largely predictable from the meaning of their verbs” (Levin & Rappaport Hovav 2005, p. 7), which means that every theoretical model of argument realisation has to feature at least a theory of lexical semantic representations of verbs and, moreover, a theory of how these representations are mapped onto the relevant syntactic representations. 1.2. Lexical semantic representation If there is a general agreement on interpreting the semantic representation of verbs as articulated in different levels (also called structures1), there is definitely less agreement on which are the levels, how many of them there are and how semantic information distributes among them. In fact, once we have established that “the semantic content of a lexical entry determines to a large extent its syntactic contexts of occurrence or subcategorization properties2” (Koening & Davis 2006, p. 71), and that this “semantic content” is organized at different levels, we still have to determine which facets of a verb meaning influence and, at the same time, limit its syntactic structure. Since verbs are, like any other lexical items, linguistic symbols that codify complex events happening in the world, it is easily understandable why they should, and indeed do, display an equally complex set of semantic properties. However, as 1 See 2.3, 2.4 and 2.5 for the description of the different structures. 2 See 2.3.1. and 2.3.2. 10 crosslinguistic differences show, not all of these properties play a role in determining the syntactic realisation of verbs and their arguments, therefore leading us to assume that “there is a set of semantic elements and relations that is much smaller than the set of cognitively available and culturally salient distinctions, and verb meanings are organised around them” (Pinker 1989, p. 166). According to this view then, it seems to be more accurate to interpret verbs as predicate of events, rather than as events themselves, namely as complex happenings in the world decomposable into the smaller and delimited phases of which they consist (subatomic semantic decomposition of events; see Parsons 1990, Arsenijević et al. 2013). Also, this assumption about verbs allows us to fully understand why languages vary greatly in relation to argument structure: in each and every one of them, verbs exhibit an idiosyncratic sensibility only to a selected subset of the wide range of semantic features associated with the event they describe, making every language a unique depiction of the world (Steiner 1975). Hence, following Levin & Rappaport (2005), we believe that only those aspects of meaning that determine argument realisation must be included in the lexical semantic representation of verbs, whereas all the others, being irrelevant for the set purpose, can be left out. Consequently, in providing verbs with a proper lexical semantic representation one should first analyse verbs and their syntactic realisations, trying to divide them into classes accordingly to their behaviour. Then, one should carefully look at the semantics of those verbs that fall into the same group, seeking out the underlying semantic properties that can be held responsible for their common argument realisation. The outcome of this thorough analysis should be the identification of verbs semantic classes present in a given language. However, as soon as many different -­‐ yet all seemingly plausible -­‐ semantic parameters arose as means to classify verbs, it became clear that “the isolation of meaning components appropriate to the characterisation of verbs in a particular semantic class presents a real challenge” (Levin & Rappaport Hovav 2005, p. 15), the failure of which can not only compromise the results, but also bring into question the reliability of the methodology itself employed to obtain them. 11 In the next paragraphs we briefly illustrate the main approaches so far put forward to organise verbs into different classes. Given the scope and purpose of this thesis we thought it best to give more space and pay more attention to those aspects related to the so-­‐called “argument structure” and, of course, to the models based on them: for the sake of completeness we will still present some of the other existing frameworks, which, however, will not be dealt with in depth but only touched on. 1.2.1. The meaning-­‐based approach The most basic and naïve verb classification is certainly the one based on the “analysis of the type of event expressed by a verb from a semantic point of view” (Jezek 2005, p. 107) namely its denotative meaning 3 , which is also responsible for the so-­‐called “prototypical meaning” of verbs (and also nouns). This is precisely the basic and broad meaning conveyed by any given lexical item and that, even in the case of polysemy or figurative language, can still be identified as the typical semantic content of that specific word. According to this parameter, verbs naturally fall into various categories, such as: -­‐
movement verbs (to walk, to run, to stroll) / verbi di movimento (camminare, correre, passeggiare); -­‐
perception verbs (to hear, to see, to smell) / verbi di percezione (sentire, vedere, annusare); -­‐
throwing verbs (to throw, to toss, to hurl) / verbi di lancio (buttare, gettare, scagliare). Even though this way of classifying verbs is clear-­‐cut and theoretically plausible, as extensively discussed above, the semantics of verbs is more complex than this and, as further empirical research demonstrated, there is much more to verbs and verbal behaviour than just “knowledge of idiosyncratic word-­‐specific properties” (Levin 1993, p. 1). 3 “Denotation. The relation between a lexical unit and whatever objects, etc. it is used to refer to. E.g. bull denotes (in one account of its meaning) a class of animals, brown a property of individuals or objects, etc.” (Matthews 2007, p. 97). 12 1.2.2 Aktionsart: Vendler’s verb classes A different way of classifying verbs is based on a semantic parameter known as Aktionsart. Even though its first use dates back to the beginning of the XX century (Agrell 1908), it was not until the publication of Vendler’s famous article Verbs and Times (1957) that this level of verb meaning became widely known and used. The German term Aktionsart is commonly translated into English with the phrase “lexical aspect” and, in this framework, it refers to “different ways of viewing the internal temporal constituency of a situation” (Comrie 1976, p. 3). As we already argued, verbs are lexical items that describe structured events: since every happening in the world necessarily takes place over the course of a more or less defined length of time, looking at how the unfolding through time is represented by every verb can cast some light on its syntactic behaviour. Vendler bases his analysis of verbs behaviour on a parameter that he calls “time schemata”, indicating with it the internal temporal structure of a verb, regardless of any other time-­‐point that may refer to an external situation. According to this semantic property then, four different classes can be individuated, which are believed to refer to as many types of events: 1. States (or statives): to believe, to own, to exist, etc.; 2. Activities: to cook, to dance, to study, etc.; 3. Accomplishments: to build, to heal, to cut; 4. Achievements: to find, to become, to lose. Not only do they differ in relation to which tenses they allow and which temporal modifiers they can be paired with, but they also present different logical implications. Stative verbs typically go on for a certain period of time, during which the event described does not show any modification or any further decomposition into smaller sub-­‐phases. Activity verbs have certain duration, just like States, but contrary to them they can be analysed as a sequence of different stages, all of which are equally true. Accomplishment verbs have duration, just like the previous two classes, but they “have a climax, which has to be reached if the action is to be what it is claimed to be” (Vendler 1957, p. 145). 13 If verbs normally ascribed to the three classes presented so far all share the durativity element, those belonging to the fourth one do not: Achievement verbs typically describe so-­‐called punctual events, namely those events whose inherent temporal endpoint consist of just one short phase that corresponds to an instantaneous culmination, and before which there is no preparation phase. Vendler’s time schemata have thus been interpreted as an internal temporal structure articulated in binary aspectual features, the combination of which determines to which class a verb belongs (Fig. 1). Traditionally, there are three types: durativity, dynamicity and telicity4. The first one draws a fine line between events that happen overtime and events that are perceived as instantaneous; the second distinguishes dynamic events -­‐ typically processes -­‐ from states; the third is the least intuitive one and it describes events that tend towards a completion, the reaching of which is essential in order to consider the action completed (telic), in opposition to those that do not imply such an internal final point (atelic). DURATIVITY DYNAMICITY TELICITY States + -­‐ -­‐ Activities + + -­‐ Accomplishments + + + Achievements -­‐ + + Table 1: Combination of aspectual features in Vendler's classification As briefly mentioned above, the most striking fact about these features – which is also what has made them so widely used– is that they only allow for specific syntactic structures, the presence of which in a given class can then be inferred by looking at its characteristic features. Durative and dynamic verbs (Activities and Accomplishments) allow continuous tenses (“I am running”, “She is cutting the bread”), whereas the lack of just one of these two features (States and Achievements) is enough for a verb to not have such a structure (*“I am owing two houses”, *“He is finding the keys”). Moreover, atelic predicates (States and Activities) can be paired with the adverbial phrase “for x time” (“He was in college 4 These three are the most widely known and used, but not the only ones; some models include other aspects, such as: iterativity, ingressivity, egressivity, and so on. 14 for two years”, “ We cooked for an hour”), while telic ones (Accomplishments and Achievements) can’t (*“They built a bridge for three months”, *“We found the hotel for a day”). Lastly, only durative, dynamic and telic verbs (Accomplishments) can appear with the phrase “in x time” (“The wound healed in a week”), and only non-­‐durative, dynamic and telic verbs (Achievements) are compatible with the phrase “at time x” (“You arrived at 4PM”). Vice versa, the possibility for a verb to exhibit a certain syntactic behaviour is taken as a test to establish whether a verb belongs to a class or not (Fig. 1). Figure 1: Syntactic tests for Vendler's verb classes [ Jezek 2003, p. 33] Given the analysis we propose in this thesis, it is important to stress the value of actionality for the consequences it bares in relation to syntactic valency. In particular, the presence of the telicity feature seems to be responsible for the possibility of intransitive Activity verbs (3a) to become transitive Accomplishment verbs (3b), by means of introducing a direct object and, therefore, of effecting the syntactic valence of the verbs in question: (3) a. Chiara sings / Chiara canta b. Chiara sings a cake / Chiara canta una canzone Furthermore, it is worth mentioning that a significant application of Vendler’s classification is that conducted by Bertinetto for Italian (1986): even though it is limited to the indicative mood, it is still today an invaluable piece of work and, by far, the most complete and thorough actionality-­‐based analysis of the Italian verb system. After Vendler’s classification, the most renowned contribution in the description of verb behaviour based on aspectual properties was that of Dowty (1979), who proposed a representation of the four classes established by Vendler expressed through the aid of 15 logical terms. These “logical terms” consist of the following three semantic primitives, namely semantic basic concepts that are innately understood, but cannot be expressed in simpler terms: 1. BECOME: it indicates inchoativity, which is the instantiation of a change of state taken to be spontaneous and not attributable to a specific agentive entity; 2. DO: it reflects agentivity, namely the firm intention of a specific entity to act on a certain situation or other entity; 3. CAUSE: it establishes a causal relation between two otherwise separate events, which are put into correlation by either a spontaneous or an intentional event. When these primitives are applied to what Dowty considers “atom predicates” that cannot be further decomposed, namely Statives, they give rise to the other three classes of verbs and the processes through which they emerge are called by Dowty “logical structures” (Dowty 1979). So, for instance, a typical achievement verb such as to die, can be analysed as ‘(BECOME) dead (x)’, which means that “the entity (x) has undergone a change of state that has led it to now be dead”. 1.2.3 Verb valency and argument structure A third way of studying verbs and their syntactic behaviour is based on the semantic level now known as “argument structure”, which was first proposed by Tesnière (1959)5. According to this framework – and the ones which followed –, every sentence is built around a verb, which, just like an atom with its surrounding sub-­‐particles, is capable of exerting a force upon a variable number of phrasal elements, called first “actants”, then “arguments”, and defined as “les personnes ou choses qui participent à un degré quelconque au procès” (Tesnière 1959, p. 105). Also, the traditional concept of “government” or “rection” is replaced with that of “valence” or “verb valency”, clearly derived from the definition and use of this term in chemistry. 5 Other models based on this semantic level are Grimshaw (1990) and Lazard’s “actantial model” (1994). 16 Thus, depending on the number of arguments that a verb needs in order for its meaning to be fully realised, four classes of verbs have been identified6: 1. Avalent verbs (0 arguments): “It7 rains a lot”; 2. Monovalent verbs (1 argument): “She1 has aged”; 3. Divalent verbs (2 arguments): "The mother1 hugged the little boy2”; 4. Trivalent verbs (3 arguments): “He1 showed us2 the house3”. However, calculating the valence of a verb proved to be not as straightforward as it may seem at first sight. As Tesnière himself noted, a verb can also be accompanied by phrasal elements, known as adjuncts, that are not its arguments but still express “les circonstances de temps, lieu, manière, etc. dans lesquelles se déroule le procès” (Tesnière 1959, p. 102), as seen in the following examples: (3) a. My dad left on Tuesday b. I met my friend Julian the other day in a nice café in town Moreover, under particular circumstances certain verbs can leave one of their arguments syntactically unexpressed, which is therefore said to be optional: (4) a. “My boss takes ages to park [his car/motorbike/van/etc. > vehicles] b. “Amanda has finally come back [home/to work/etc. > places] As one can assume from these sentences, these verbs have two equally possible and acceptable syntactic structures, depending on the presence or not of one of their arguments (precisely those in italics). 6 A fifth class of tetravalent verbs (4 arguments: “They moved the couch from the sitting room to the bedroom”) has been proposed, but the debate is still quite far from being settled (Jezek 2005). 7 Here, “it” is what is normally called a “dummy subject”, namely a mere syntactic placeholder used in non-­‐pro-­‐drop languages, such as English, when the argument of a verb is semantically nonexistent and has no real reference, but it is still syntactically required. In pro-­‐drop languages, such as Italian, avalent verbs have in fact no subject at all (*“Il cielo piove”). 17 Another case is that of verbs like to cut, the meaning of which semantically and logically implies the use of a sharp manageable tool; however, the argument that should refer to it is actually never syntactically realised, as demonstrated by the sentences in (5): (5) a. “My brother always cuts the bread [with a knife]” b. “The hairdresser cut my hair nicely [with a pair of scissors]” Such arguments are referred to as default arguments (Pustejovsky 1995, p. 63). A similar yet slightly different case is that of verbs like to brush, whose meaning already incorporates one of its arguments: it differs from the previous scenario because, in order for this argument to be syntactically expressed, it needs to be further specified: (6) a. *“My mum always brushed my hair with the brush” b. “My mum always brushed my hair with my favourite Barbie’s brush” Such arguments commonly go under the name of shadow arguments (Pustejovsky 1995, p. 63). The examples discussed above show that “if it is true that the scene elicited by a verb evokes a certain number of elements in relation to which something is ‘predicated’, it is just as true that these such elements are not all equally and necessary realised at the syntax level” (Jezek 2003, p. 18). This striking evidence inevitably led to a much-­‐needed distinction between syntactic valence, which refers to the pattern of syntactically necessary arguments of a verb – else the sentence is ungrammatical – and semantic valence, which corresponds to the set of all the arguments implied by a verb at a logical level. Selectional restrictions Once the number of arguments that a verb needs to create well-­‐formed sentences has been established, it is necessary to validate their semantic content. If it is true that predicates select their arguments in terms of numbers, it is equally as true that, precisely because of their meaning, they also constrain them in terms of semantics. 18 Even though the first to systematically note that co-­‐occurring words always show some relations of semantic context was Porzig (1934), we owe the first extensive discussion about selectional restrictions to Chomsky, who talks about “strict subcategorisation rules” on one hand, and “selectional rules” on the other (Chomsky 1965). The latter is precisely what we commonly refer to as semantic-­‐selection (or s-­‐
selection: Chomsky 1986), and it is responsible for the restrictions between predicates and the semantic type of their complements. As for how to represent these restrictions and the semantic content of the arguments they allow, since the work of Hjelmslev (1961) within the structural theoretical framework, linguists have adapted the formalism of binary semantic features, which is briefly illustrated in the examples in (7): (7) a. to eat (2 arguments): [+ animate] _ [+edible]; b. to sleep (1 argument): [+ animate]. This aspect of argument structure is fundamental for whoever wishes to undertake a complete analysis of verbs because it shows how sentences can be syntactically well-­‐
formed, therefore grammatical, but still thoroughly unacceptable for any speaker of a given language, since they entail a violation of a semantic restriction. Once again, Chomsky gave a brilliant example of this when he drew a distinction between the two following sentences: (8) a. “Colourless green ideas sleep furiously”’ b. *“Furiously sleep ideas green colourless” As a matter of fact, he simply noticed that (8a) “though nonsensical, is grammatical, while (8b) is not grammatical” (Chomsky 1957, p. 15), which demonstrates how the two levels in question, namely syntax and semantics, must be kept separated in the study of language faculty (see also Moro 2006). 19 Subcategorisation restrictions Lastly, it is important to note that verbs influence their arguments in one more way, namely through another set of restrictions that work together with the semantic ones discussed in the previous paragraph. This property of verbs is called subcategorisation (or c-­‐selection: Chomsky 1986) and it “expresses restrictions between predicates and the syntactic category of their complements” (Grimshaw 1979, p. 279). Verbs can in fact allow the same number of arguments, but restrict their syntactic realisation in a different way. For instance, the verb “to rent” is divalent, just like the verb “to live”: however, while the former one allows only a direct object as second argument (NP), the latter one requires the same second argument to be expressed as an indirect object, realised as a prepositional phrase (PP). Such evidence is clearly what prompted Chomsky to recognise the necessity for a verb (V) to be categorised “in terms of a certain set of frames in which V occurs” (Chomsky 1965, p. 96): this set of arguments has been thus called subcategorisation frame ever since and, along with the set of semantic selectional restrictions, it is believed by many linguists to be part of the lexical entry of a verb (Grimshaw 1990). 1.2.4 Fillmore’s case frames and Thematic Roles A considerably different way of looking at arguments and their structure was first introduced by Charles Fillmore in his seminal paper “The case for case” (1968), which can indeed be considered the origin of the modern interest in semantic role lists. Moving from a syntactical perspective, Fillmore’s focus is on the transformational mechanisms that make it possible for the deep structure (DS) to turn into the surface structure (SS) of a language. Even though he obviously borrows the word case from the tradition of studies based on such “case languages” as Latin, Greek or Sanskrit, he does not use it to refer to grammatical categories, but to semantic functions. Given sentences like the following ones: (9) i. He hit the ball ii. He received a blow iii. He received a gift 20 iv. He loves her v. He has black hair [Fillmore 1968, p. 6] Fillmore notes how the same subject “He”, thus realised in the same syntactic position, takes part in the actions described by the verbs above in very different ways. Moving from this observation, he believes that such differences have to be ascribed to the semantic relations that each argument establishes with the verb. On this account, Fillmore posits the existence of a limited and presumably universal set of what he calls deep cases 8 , which aim to represent precisely all the possible semantic relations responsible for the heterogeneity of argument realisation. The interesting aspect of Fillmore’s model (later called case grammar) is that, even though it emerged from a strongly syntactic tradition, it is based on the assumption that the argument structure of a verb – namely its syntactic realisation – depends on the deep cases associated with it – namely a semantic relation –, and not the other way around. Every verb is then identified with a so-­‐called case frame (Fillmore 1968), which is construed as a configuration of its deep cases from which one can derive the argument realisation. Hence, according to this framework, “the full combinatory description of a verb would consist […] of a pairing of a case frame with the manner in which the phrases representing the individual cases are realized in the syntax” (Fillmore 2008, pp. 5-­‐6). In the following years, Fillmore’s case grammar received so much attention – along with mixed reviews – that even Chomsky, after including the subsystem “lexicon” as a base of the Universal Grammar (UG), postulated what he called Θ-­‐theory, namely a set of principles “concerned with the assignment of thematic roles such as agent-­‐of-­‐
action” (Chomsky 1981, p. 5). The principle that bears particular relevance in this context is the θ-­‐criterion, according to which every verb is bound to each of its arguments by a specific semantic relation, called θ-­‐role, assigned on the basis of the θ-­‐
position it occupies in the LF (“Logical Form”). Moreover, the information concerning this projection mechanism is coded in what Chomsky calls θ-­‐grid, which is thought to be part of the lexical entry of every verb (Chomsky 1981, pp. 34-­‐36). So, for instance, in the 8 For similar concepts, see Gruber (1965) and Jackendoff’s (1983) thematic relations, and Davidson’s (1984) primitive notions. 21 following sentences, Chomsky identified the θ-­‐positions in brackets and, depending on the specific θ-­‐grid of the verb in question, they will be assigned a certain θ-­‐role: (10) a. [They] persuaded [John] [that [he] should leave] b. [We] hold [that [these truths] are [self-­‐evident]] c. [We] hold [these truths] to be [self-­‐evident]] [Chomsky 1981, p. 36] Semantic Role lists Over the past decades, many linguists have tried to describe the semantic relations between verbs and arguments, usually formalising them in what has been called a semantic role list 9 , namely “a predetermined set of labels that identify arguments according to the semantic relation they bear to their verb” (Levin & Rappaport Hovav 2005, p. 35). The following one is a typical example of what they look like: a. Agent (A): the instigator of the event; b. Counter-­‐Agent (C): the force or resistance against which the action is carried out; c. Object (O)10: the entity that moves or changes or whose position or existence is in consideration; d. Result (R): the entity that comes into existence as a result of the action; e. Instrument (I): the stimulus or immediate physical cause of an event; f. Source (S): the place from which something moves; g. Goal (G): the place to which something moves; h. Experiencer (E): the entity which receives or accepts or experiences or undergoes the effect of an action. [Fillmore 1971b, p. 376] In spite of the fact that many semantic role lists have been so far proposed (see also Gruber 1965; Jackendoff 1972, 1976), as a matter of fact they all begin from the same 9 Semantic roles refers here to what early accounts (such as the ones described in the previous paragraph) call case frame (Fillmore 1968), thematic roles, θ-­‐roles or theta-­‐grid (Stowell 1981). 10 The role “Object” in Fillmore includes both the semantic roles of “Patient” and “Theme”, distinguished in the majority of semantic role lists proposed afterwards. 22 basic assumptions: first, they all see semantic roles as entities that cannot be semantically analysable; moreover, being predetermined, they are defined independently of verb meaning and, lastly, even though the lists vary in terms of both composition and size, they are all considerably small. Thus, the aim of a theory of semantic roles is to identify a universally acceptable canon of semantic roles that can be applied to any argument of any verb: the consequence of such a goal is that assigning semantic roles is taken as a means to bring out similarities and differences in verb meaning that are reflected in argument realisation. Therefore, it is possible to identify, for each role, a natural class of arguments, whose members not only establish the same relation with their verbs, but also allow the same morphosyntactic expression. For instance, the break class and the hit class examined by Fillmore in The Grammar of Hitting and Breaking (1970) show specific morphosyntactic behaviour, systematically shared by all the members of the respective class (see Chapter 2, §1). In order to explain the differences between the two types of class, Fillmore proposes distinct semantic role lists for the break verbs and the hit verbs, taken to be responsible for the superficial syntactic variances: (11) a. break: Agent, Instrument, Object b. hit: Agent, Instrument, Place [Fillmore 1970, p. 131] Since Fillmore’s breakthrough, approaches based on semantic role lists have received ample and, perhaps, predictable criticisms. The first issue related to such approaches concerns their basically syntactic nature: as noted by Ravin (1990), the fact that only the semantic aspects that have a syntactic realisation are taken into account is very limiting because, as discussed above (2.3), part of the semantic content of a verb can remain syntactically unrealised, but undoubtedly crucial for understanding and determining its actual syntactic behaviour when put in the context of a sentence. A solution to this issue was proposed by Dowty (1991), who maintained that semantic roles are not universal, but strictly defined in relation to the verbs they are selected by: every verb, in fact, influences the semantic roles that it assigns to its arguments by virtue of what Dowty calls lexical entailments, thus claiming that “the implication follows from the meaning of the predicate in question 23 alone” (Dowty 1991, p. 552), originating then L-­‐thematic roles (Lexical-­‐thematic roles) of that particular verb. According to this view, “establishing what semantic role an argument of a verb bears, then, requires a careful examination of the meaning of the verb and, in particular, the identification of the lexical entailments which the verb specifies for that argument” (Levin & Rappaport Hovav 2005, p. 39). However, another problem that seems to stem directly from this revised approach is the possibility for more semantic roles to emerge at a deeper analysis of verbs and their argument realisation, an issue known as role fragmentation, the most famous case of which is that of the role Agency: if Jackendoff (1983, p. 179-­‐183) divides it into Agent and Actor, and D. A. Cruse (1973, pp. 18-­‐21) split it in four ways (Volitive, Effective, Initiative, Agentive), Lakoff (1977, p. 244) goes as far as to propose fourteen different Agent roles, depending on their supposedly distinctive characteristics. Dowty fears that the adoption of “finer categorization of roles to achieve certain distinction” might lead to the inability to appreciate “generalizations by not being able to refer to the grosser Agent category” (Dowty 1991, p. 554). At the same time, extending the same semantic role to clearly different arguments is equally as undesirable, since they would prove to have little predictive power, consequently failing to appreciate differences in the behaviour of arguments11 of various verbs. One more issue concerning the interaction between the thematic level and the syntactic level is related to the possibility shown by verbs to realise, on one hand, the same thematic role in different syntactic fashions (argument alternation), and, on the other, different thematic roles with the exact same argument, as respectively illustrated in (12) and (13): (12) a. “HeAg gave the moneyTh to MaryRec” b. “HeAg gave MaryRec the moneyTh” [Jezek 2003, p. 25] (13) a. “TracyAg washed the carTh with an enormous spongeInst” b. “TracyAg washed the carTh with StacyCom” [Levin & Rappaport Hovav 2005, p. 41] 11 As argued in Levin & Rappaport Hovav (2005), objects are generally more prone to potential fragmentation then subjects. 24 This issue is made even more puzzling by the fact that not all the verbs which could potentially present the same argument alternation of “to give”, do in fact allow it, as demonstrated by the example given in (14): (14) a. HeAg returned the moneyTh to MaryRec b. *HeAg returned MaryRec the moneyTh [Jezek 2003, p. 25] Moreover, only specific thematic roles can share the same morphological expression in a given language12, and they do it so systematically that it is impossible to dismiss it as a random and accidental occurrence. Many scholars have in fact argued in favour of some sort of internal structure of semantic roles, convinced by the evidence discussed above that “the small set of unanalysed roles that characterises an ideal semantic role approach, then, is incompatible with linguistic reality” (Levin & Rappaport Hovav 2005, p. 42). One last problem relating to thematic roles rises from the claim made by many semantic role list approaches – particularly the earlier ones – that there may be at most one instance of each semantic role per sentence and that each argument can bear one and one only role (Fillmore 1968; Chomsky 1981). Unfortunately, even in this case linguistic evidence contradicts such assumption, as shown in Gruber (1965) and Jackendoff (1972, 1976, 1983), since many verbs present dual semantic role assignment not only when an argument bears two semantic roles at once – (12) –, but also when two arguments bear the same semantic role – (13) – and “with no apparent asymmetry in what is predicated of the two arguments on which to pin a distinction in role type” (Dowty 1991, p. 556): (12) a. Kelly rolled down the hill / Kelly è rotolata giù dalla collina = AGENT (Kelly rolled intentionally) b. Kelly rolled down the hill / Kelly è rotolata giù dalla collina = THEME (Kelly underwent a change of position over which she had no control) 12 For example, in English patients and recipients have the same syntactic realization in double object constructions, just like the preposition “with” usually indicates only instruments and comitatives. 25 (13) a. My brother resembles my grandfather / Mio fratello assomiglia a mio nonno = THEME -­‐ THEME After more than forty years since Fillmore’s influential work, it is now believed that “thematic roles do not represent a strong explicative instrument, they certainly aren’t a semantic primitive and, in more that one occasions, even their existence has been called into question” (Jezek 2003, p. 26). Thematic roles do not in fact feature in the majority of current approaches to lexical semantic representation and the only attempts to reintegrate them were made by purely semantic frameworks, such as Role and Reference Grammar (Foley & van Valin 1984; van Valin & La Polla 1997), in which they are referred to as macro-­‐roles, and Dowty’s proto-­‐roles (Dowty 1991). 1.2.5 From verb meaning to the underlying event structures A yet different attempt to overcome the problems and limitations of semantic roles entails a shift of focus from verbs and their syntactic behaviour to the underlying event structure they actually represent, clearly stemming from a strongly semantic perspective. These new theoretical proposals assume that argument realisation derives directly from verb meaning, reason why what needs to be decomposed into more basic elements is, in fact, verb meaning itself, and not semantic roles. Such approaches are commonly known as predicate decomposition approaches and, even though they have been elaborated in the work of many semanticists in various contexts, they are all based on the idea that the meaning of a verb can be “formulated in terms of one or more primitive predicates chosen to represent components of meaning that recur across significant sets of verbs” (Levin & Rappaport Hovav 2005, p. 69) and that stands for what is usually called event structure. One example of predicate decomposition is that put forward by Dowty (1979) and briefly examined above (see 2.2): the primitives found in this account (DO, CAUSE, BECOME) are widely used and can indeed be considered the fundamental ones in all predicate decompositions. Since verbs denote many and very different types of events, it is intuitive that the sole use of primitives could not do justice to the huge variety of happenings in the world and their linguistic codification, potentially leading to a misrepresentation of both. In 26 order to avoid such a risk then, these accounts maintain that verb meaning is composed not only of primitives, but also of a second kind of basic information responsible for the idiosyncratic element of verb’s meaning, commonly referred to as root (Pesetsky 1995). Given three deadjectival verbs belonging to the same class (causative change-­‐of-­‐state verbs), the standard representation of their meanings is illustrated in the following example: a. dry: [[ x ACT ] CAUSE [ y BECOME <DRY> ] ] b. open: [[ x ACT ] CAUSE [ y BECOME <OPEN> ] ] c. shorten: [[ x ACT ] CAUSE [ y BECOME <SHORT> ] ] [Levin & Rappaport Hovav 2005, p. 71] Hence, the event structure of verbs stems from the combination of its root and one or more primitives, allowing predicates to be part of a certain class while, at the same time, retaining their semantic distinctiveness. From the example above, it is also clear that such representations allow verbs belonging to the same semantic class (change-­‐of-­‐state) to have decompositions with common substructures (x ACT > CAUSE > y BECOME <…>), with roots of the same ontological type (dry, open, short all denote a transformation from state A to state B), recurring in the exact same positions in these substructures (<…>). The two most salient consequences of predicate decompositions are, on the one hand, the possibility of identifying every semantic class of verbs with a given decomposition that represents the basic event structure of its members; on the other, if a verb is assigned a proper decomposition, one could predict which semantic class it will belong to. It does not come as a surprise then that predicate decomposition approaches have received much praise: not only do they feature an immediate and economic apparatus of representation, but, drawing a distinction between two equally necessary semantic building blocks (primitive and roots), they also allow us to appreciate variety within generalisations. Moreover, by their own nature, predicate decompositions entail relations between arguments, providing a good explanation as to why only certain co-­‐
occurrences of arguments are systematically observed, while others seem to never be possible (see 2.4.1). Lastly, the association of every primitive with a specific argument 27 reveals the substructures of the more general event structure of every given verb, enabling us to look at their properties and, consequently, understand the reasons behind argument realisation. 28 Chapter 2 VerbNet: a verb classification based on argument alternations 2.1. Argument Alternations: what they are and why they matter In the previous chapter, while introducing argument structure and how a semantic role list approach deals with it (see 1.2.4.1), we briefly mentioned the linguistic phenomenon known in the literature as argument alternation: as the discussion about give clearly demonstrated, a verb can realise its arguments in more than just one syntactic way, thus giving rise to a noticeable variation in how they can be syntactically expressed. As the investigation of such linguistic aspects deepened, it became apparent that argument alternations are in fact the rule, rather than the exception. In fact, the majority of verbs, in English as well as in all the other languages studied so far, (DeLancey 1995; Guerssel et al. 1985) seem to display at least one form of the broader set of phenomena called multiple argument alternations, which mainly consist of “pairs of sentences with the same verb, related by paraphrase or subsumption” (Levin & Rappaport Hovav 2005, p. 186). A prime example of such alternations is the causative-­‐inchoative alternation (also known as anticausative alternation or, simply, causative alternation; see Haspelmath 1993), possibly the most widely examined of all of them and illustrated in the following sentences: 1) a. Rebecca broke the pencil. b. The pencil broke. 2) a. Maria opened the door. b. The door opened. 3) a. Thomas dried the clothes. b. The clothes dried. [Piñón 2001, p. 346] As is apparent, this alternation concerns verbs that have “an intransitive as well as a transitive use, where the intransitive use typically denotes a change-­‐of-­‐state event undergone by some entity and the transitive use denotes that this change-­‐of-­‐state event 29 has been brought about or caused by some different entity” (Schäfer 2009, p. 641). Therefore, it is understandable that the transitive variants (a) have also been called causative, and they can be paraphrased as “‘cause to V-­‐intransitive’”, whereas the intransitive ones (b) have been labeled inchoative. Further research has also demonstrated that a superficial and general semantic similarity between two or more verbs cannot be taken as a valid criterion to establish or predict what alternations they might present and, consequently, share, as illustrated very well in the famous case of break and hit. In “The Grammar of hitting and Breaking” (1970), Fillmore focuses precisely on this pair because, given their shared basic meaning, one would be prone to assume that they behave in a similar syntactic fashion. However, as the following sentences show, this is not the case: 4) The stick broke. 5) *The tree hit. [Fillmore 1970, p. 126 – 128] Even though both verbs can indeed be “characterized as agent-­‐act-­‐on-­‐patient verbs” (Levin & Rappaport Hovav, 2005, p. 1), break and hit appear nonetheless to act in a dissimilar way, and, more importantly, not only in relation to the causative-­‐inchoative alternation described above. One more difference that Fillmore notices in his aforementioned work is that related to the possible readings of sentences that feature “stative adjectives” derived from the two verbs in question, presented below: 6) The window was {broken}. 7) The window was {hit}. [Fillmore 1970, p. 131] As one may intuitively notice, while the sentence in 1) can be “understood either as passive or as description of states13”, the sentence in 2) “can be understood only as passive” (Fillmore 1970, p. 131). This also means that sentence like 1) carry an idiosyncratic ambiguity unknown to sentences like 2). In a later article, Fillmore (1977, 13 Called also eventive reading in Levin & Rappaport Hovav (2005, p. 2). 30 pp. 74-­‐79) also stressed that sentences like those in (8) cannot be construed as simple paraphrases, whereas those in (9) can: (8) a. John broke the fence with the stick b. John broke the stick against the fence (9) a. John hit the fence with the stick b. John hit the stick against the fence The flexibility of syntactic behaviour exhibited by so many verbs has, of course, challenged all those frameworks based on the assumption that the argument structure of a given verb is encoded in its lexical entry, which alone determines the projection of arguments onto syntax. However, if this were the case, it would be difficult to explain why a verb should display more than one way to syntactically realise the exact same arguments, and why verbs with similar meaning do not allow the same argument alternations (see I.2.4.1.). This theoretical issue has led linguists to reconsider the status of all the instances of multiple argument realisations, drawing a distinction between at least two major types: traditional argument alternations and so-­‐called “event composition”. 2.1.2. Traditional argument alternations These alternations are by far the most studied ones: they consist of an alternate realisation of a single set of arguments, such as those illustrated above in (8) and (9). Traditionally, the possibility for a verb to realise its arguments in two different ways was explained by structural approaches by assuming a single lexical entry for both forms. These are believed to have a common underlying syntactic structure: however, one realisation is construed as the basic one, from which the other one is derived by means of transformational rules. This is the case of the active-­‐passive alternation illustrated in (10) and described by Chomsky in these terms (Chomsky 1957): (10) a. The mouse ate the cheese b. The cheese was eaten by the mouse [Levin & Rappaport Hovav 2005, p. 196] 31 Explanations of this kind were also proposed for other alternations, such as the dative alternation and the locative alternation (Hall 1965). However, it quite soon became apparent that not all the instances of argument alternations could actually be explained in this way, as the discussion above regarding hit and break illustrated. Recently, many linguists and researchers have in fact stressed that, even though alternations are traditionally construed as twofold syntactic realisations of the exact same argument, “the definition of alternation does not exclude the fact that each syntactic variant is accompanied by specific semantic and pragmatic inferences that determine a change in the interpretation of a sentence depending on the way the same argument is realised at the syntactic level” (Lenci 2008, p. 12). It is within this new approach to argument alternations that positions like Beavers’ (Beavers 2006) and Roland and Jurafsky’s (Roland & Jurafsky 2002) have to be placed. In relation to the object-­‐oblique alternation, Beavers proposed “The Principle of Contrast for Alternations”, according to which, if a verb allows two or more different argument realisations it is because they necessarily express some sort of contrast, analysable in terms of different parameters (affectedness of the participants, telicity of the event described, control over the event, etc.). For instance, Beavers notes how, in a particular type of direct/oblique alternation in English, the direct realisation is specified for some effect that may or may not be total in any sense (11a) while the corresponding oblique realisation is not specified at all (11b): (11) a. John cut the bread b. John cut at the bread [Beavers 2006, p. 116] This is demonstrated by the fact that only the second one can occur with an explicit context in which no affectedness occurs at all (12b), while the first one is not acceptable (12a): (12) a. John cut at the pie, but in his drunken state managed to miss it entirely b. *John cut the pie, but in his drunken state managed to miss it entirely [Beavers 2006, p. 116] 32 Along the same lines, Roland and Jurafsky (2002) compared five different and variously derived sources of subcategorisation information in order to investigate whether it is actually true that different frames allowed by the same verb convey the same meaning. They examined the effect of verb sense on frame probability by focusing on instances of ambiguous verbs and their findings demonstrated that different senses of the same verb differ in frame probability, just as different verbs differ in frame probability. This evidence prompted them to propose the Lemma Argument Probability hypothesis (Roland & Jurafsky 2002, p. 336), according to which it is not possible to make claims about the probabilistic subcategorisation frames of a given verb, but only about the probabilistic frame of each and every sense of any given verb. So, for instance, the probability for a certain frame of the verb to charge varies greatly if the sense we are investigating is that of “to accuse” as in (13a), or is that of “to bill”, as shown in (13b): (13) a. Separately, a Campeau shareholder filed suit, charging Campeau, Chairman Robert Campeau and other officers with violating securities law b. Currently, the Government charges nothing for such filings [Roland & Jurafsky 2002, p. 337] The current status of classic argument alternations is still hotly debated and a clearer view on the matter could be achieved only with further research, possibly on languages other than English. 2.1.3. Event Composition Another kind of phenomenon belonging to the broad class of multiple argument realisation is the one we defined “event composition”, and which has been widely recognised as very different from the traditional argument alternations discussed above, as illustrated by the sentences in (11): (11) a. Pat ran b. Pat ran to the beach c. Pat ran herself ragged d. Pat ran her shoes to shreds 33 e. Pat ran clear of the falling rocks f. The coach ran the athletes around the track [Rappaport Hovav & Levin 1998, p. 98] As one can see, the sentences in (11) follow a progression of complexity of the event described by means of adding either a novel argument, an adjunct or both. A traditional model that views argument realisations as encoded in the lexical semantic representation of the verb that allows them, and syntactically expressed through linking rules (Chomsky 1965, Hall 1965), would simply assume that every verb has as many lexical semantic representations as the argument alternations it allows for. However, as we have already argued, cases like the ones illustrated in (11) are the rule, rather than the exception, and it seems to be computationally implausible and counterintuitive at best to assume six different entries for the verb to run, each for every possible argument realisation. However, as already discussed in I.2.5, verbs denote many and very different types of events that take place in the extra-­‐linguistic reality and, since some of those happenings are more complex than others, it is not surprising that even their linguistic representations would follow this dynamic. This fact can be illustrated by the examples in (11), where there is a crescendo of complexity in the events described by each sentence, progressing from the most basic one (11a), with only one participant, to (11f), with two participants and the information about the direction of the “running”, with intermediate possibilities in (11b-­‐11e). This evidence has prompted some linguists to develop approaches commonly referred to as “constructional approaches” (Carrier & Randall 1993, Goldberg & Jackendoff 2004, Rappaport Hovav & Levin 2002), according to which part of the meaning of a complex sentence (e.g. 11d, 11e, 11f) is not conveyed only by lexical elements, but also by the phrasal patterns according to which those lexical items are built upon. Moreover, it is worth stressing that these regular patterns are greatly found in English (Goldberg & Jackendoff 2004, pp. 533-­‐535), and most likely in every other language, hence confirming the idea put forward by these approaches that the structuring itself of any given sentence is somehow responsible for the semantic content it conveys. 34 Thus, sentences like those in (11) can be analysed and decomposed in terms of a basic event (11a), variously incremented through the adding of an extra argument-­‐
taking predicate (11c), an additional argument (11b) or both (11d, 11e, 11f), giving rise in all cases to complex events derived by a basic simple one (11a). 2.2. Verb classes Regardless of the approach taken to tackle the issues related to multiple argument realisations, one thing all researchers agree upon is the possibility for certain verbs that allow specific argument realisations (whether it is argument alternation or event composition) to be clustered together in distinct semantic classes, in English and well as in other languages (Talmy 1985, 1991; Green 1973; Choi & Bowerman 1991). In fact, it has long been recognised that verbs present regular syntactic patterns of argument realisation that are not idiosyncratic, but they are shared by other verbs, whose semantic coherency is far from being random or casual. Going back to the examples in (8) and (9), Fillmore also noticed that the verbs to break and to hit are indeed representative of a larger set of verbs, such as those listed in (12): (12) a. Break verbs: bend, fold, shatter, crack b. Hit verbs: slap, strike, bump, stroke [Fillmore 1970, p. 130] This distinction allowed for a finer classification of those verbs that, just like break and hit, were thought to be very similar, yet, which presented quite different syntactic realisations difficult to account for. Based on the possibility of verbs to take part in certain given realisations, many different classes were established, and, considering the inherent semantic coherency of all of them, it became clear that their characteristic argument realisation patterns could be ascribed to the semantic properties of each class (Guersell et al. 1985; Hale & Keyser 1986, 1987). However, it also became apparent that the identification of the facets of verb meaning that were responsible for the syntactic expression of a given class was not a small task: on the one hand, verbs that were expected to fall into a certain class failed to (see the example of the verb to return in I.2.4.1), while, on the other hand, the same 35 argument realisation can be shared by verbs belonging to different classes, as shown in (13) by to scream and to say, a manner of speaking verb and a content of speaking one respectively: (13) a. Evelyn screamed (to Marilyn) to go b. Evelyn said (to Marilyn) to go [Levin & Rappaport Hovav 2005, p. 16] Even though the classification of verbs based on their syntactic patterns can be problematic and controversial, verb classes have proven to be a great tool for investigating verb behaviour with noticeable predictive ability. The only existing verb classification of this kind is the one proposed by Levin (Levin 1993) for English and fully exploited in the construction of VerbNet, a computational resource for verbal behaviour that we will present in the next paragraph. 2.3. VerbNet: a computational verb lexicon for English With the aim of providing a reliable computational verb lexicon that could be used in as many Natural Language Processing (NLP) tasks as possible, Kipper and colleagues built VerbNet (VN) (Kipper et al. 2000; Kipper-­‐Sculer 2005), which is an extensive computational verb lexicon for English. The starting point of VN was Levin’s verb classification (Levin 1993), which is still today the largest verb classification for English. What makes this classification such a useful tool for investigating verb behaviour is the fact it features both semantic and syntactic verb profiles. In particular, Levin based her work on the assumption that verbs which display the same argument alternation (or sets of alternations) also share certain facets of meaning that make it possible to group verbs into semantically coherent classes. Levin worked on a sample of 3,024 verbs and, after a thorough examination of their syntactic behaviour, she identified 79 argument alternations involving NN and PP complements, according to which verbs were classified into 49 broad semantic classes, further subdivided into 192 fine-­‐grained semantic classes. In order to create a more solid and reliable lexical resource, at different stages Kipper and colleagues expanded Levin’s original verb classification, enlarging the 36 sample of verbs taken into consideration, and, consequently, introducing novel classes and subclasses at once: the last available version of VN (VerbNet 3.2) features 8537 verbs represented, 273 main classes and 214 subclasses14. Also, for the purpose of enriching each entry with a great variety of semantic information, they mapped VN with other important on-­‐line lexical resources, such as WordNet (Miller 1990, Fellbaum 1998), Xtag (XTAG Research group 2001), FrameNet (Baker et al. 1998), and OntoNotes Sense Groupings (Hovy et al. 2006), therefore providing over 90% token coverage of the Proposition Bank data (Palmer et al. 2005). Moreover, they provided each entry with a complete description of its properties, so that, in VN, each verb is not only ascribed to a specific semantic class and described by a set of co-­‐members, but it is also defined in terms of more traditional semantic information, such as thematic roles, semantic predicates, selectional restrictions and, of course, syntactic frames (Fig.2). Figure 2: Lexical entry for the enforce class in VN A typical entry in VN is illustrated in Fig.1, which represent the enforce class. The complete description of such a class is articulated in three distinct sections: the first provides information about the members of the class, which, in this case, are control, 14 Freely consultable at: http://verbs.colorado.edu/verb-­‐index/. 37 enforce and impose. For each of them there is also the corresponding reference in the databases mentioned above (e.g. WN: WordNet, G: Groupings). The second section gives the information related to thematic roles and their selectional preferences: for instance, for the verbs listed in this class, there has to be an Agent, which needs to be [+ANIMATE & +ORGANIZATION], and a Theme, for which there is no indication of specific selectional preferences. The third and last part indicates the remaining properties of the verbs included in this class, namely a set of frames consisting of: I. A brief description of the syntactic frames allowed by the verbs of the class (“NP V NP” and “NP V THAT S”); II. An example of a sentence for each frames; III. A syntactic description in which each thematic role is associated to the corresponding position in the sentence given; IV. A set of semantic predicates that includes a temporal function indicating in which phase of an event the predicate is true: in this case, it is the preparatory stage (during(E)). Over the past decade, VN has been received with great acclaim by the research community, proving to be a reliable computational tool in a great number of NLP tasks, such as: automatic verb acquisition (Swift 2005), semantic role labelling (Swier & Stevenson 2004; Yi et al. 2007), robust semantic parsing (Shi & Mihalcea 2005), word sense disambiguation (Dang 2004), building conceptual graphs (Hensman & Dunnion 2004) and creating a unified lexical resource for knowledge extraction (Croch & King 2005). Given its unparalleled success, one would expect to find a VN – or at least a comparable lexical resource – for many other languages; however, this is not the case. In fact, thus far, no attempt has been made to build such a precious resource by non-­‐
English speaking research groups, Italian included. Moreover, a verb classification based on argument alternations akin to the one proposed by Levin (Levin 1993), and used by Kipper and colleagues to build VN, was not available for Italian. Taking this into consideration, with this work we tried precisely to compensate for this shortage – although only partially –, carrying out a descriptive 38 analysis of those argument alternations we found using an annotated sample of Italian verbs, which we present in detail in the following chapters. Hence, this thesis represents the very first step towards the construction of an Italian VerbNet, with the aim of providing the Italian research community, and whosoever would like to conduct their work on Italian verbs, with a computational lexical resource comparable to that available for English. 39 Chapter 3 Towards a classification of argument alternations in Italian The development of a classification of argument alternations for Italian verbs has been carried out in a two-­‐stage process. In the first stage (see section 3.1), the subcategorisation frames for a sample of the most frequent Italian verbs have been manually extracted from an Italian monolingual dictionary. In the second phase (see section 3.2), we semi-­‐automatically identified the most significant alternations in our annotated sample. 3.1. Subcategorisation frames for a sample of Italian verbs 3.1.1. Resources In order to identify the subcategorisation frames for Italian verbs, we used the most updated online version of the only Italian dictionary that indicates the verb valency in the lexical entry, namely Il Sabatini Coletti 2012 (S&C: Sabatini and Coletti, 2012)15. Overall, this dictionary registers 185,000 headwords and definitions, 80,000 examples, 83,000 etymologies with 82,500 dates of first attestations, the conjugation of all verbs and division into syllables and pronunciation for all entries. In relation to the verb class, the dictionary contains 9467 lexical entries, of which 1746 are marked as high frequency verbs. Fig. 1 represents the typical structure of a verbal lexical entry. The first information given is the division in syllables, followed by the indication of the word class (“v.”) and few information about the conjugation (“irr.: coniug. come porre”). Then, for each verb there is a list of its senses, clustered together according to the subcategorisation frame they share16. For instance, the verb imporre illustrated in Fig.1 has nine possible senses, four of which occur with transitive frames ([sogg-­‐v-­‐arg-­‐prep.arg], [sogg-­‐v-­‐arg]), while the remaining five occur with pronominal frames ([sogg-­‐v], [sogg-­‐v-­‐arg], [sogg-­‐v-­‐prep. arg]. However, the formalism used by the authors of S&C neglects a piece of information that is crucial for the purpose of our work, that is the specification of the 15 http://www.elexico.com/en/Products/all/1642812-­‐1930385-­‐1970536-­‐1926039.html 16 See Appendix A for a complete list of the subcategorisation frames available in S&C. 40 preposition introducing the prepositional phrase (PP). For instance, in the first subcategorisation frame of imporre (Fig.3), it should be specified that the preposition introducing the indirect argument is a, a piece of information that gets lost in marking the slot in question as a generic “prep. arg.”. Figure 3: The lexical entry of the verb imporre in S&C We overcame this shortcoming by consulting the online version of LexIt, an automatically built corpus-­‐based lexical resource on Italian argument structure (Lenci et al. 2012) 17 . The data were extracted from two automatically parsed corpora, La Repubblica (Baroni et al. 2004: approximately 380M tokens), and a dump of the Italian Wikipedia database (approximately 152M tokens). In LexIt, each lemma is associated with a syntactic profile and a semantic profile. The latter describes the set of possible slot fillers and their selectional preferences. The former generalises over the range of syntactic environments in which the lemma occurs by representing them as a set of frames and a set of slots. Crucially, for each slot the preposition that introduces every prepositional argument is specified (see Fig. 4) and, overall, there are 3873 verbs from La Repubblica and 2831 verbs from the Italian Wikipedia corpus. 17 http://sesia.humnet.unipi.it/lexit/ 41 Figure 4: Syntactic profile of the verb imporre in LexIt If we look at Fig.2, what we see is the set of subcategorisation frames allowed by the verb imporre and sorted by frequency. In total, there are ten frames, four of which occur with transitive frames (subj#obj, subj#obj#comp-­‐a, subj#obj#comp-­‐in, subj#compl-­‐a#inf-­‐di), two of which occur with intransitive frames (subj#0, subj#comp-­‐
a), and the remaining four occur with intransitive pronominal frames (subj#si#0, subj#si#compl-­‐in, subj#si#compl-­‐in, subj#si#obj). As one can see, every time there is a prepositional argument (marked as “compl-­‐“) within a given frame, there is also the precise indication of what preposition introduces that complement (e.g. “compl-­‐a”). These two resources have complementary advantages and limitations. While S&C is able to discriminate between the different verb senses, the minimal lexical unit in LexIt is the verbal lemma. S&C, however, lacks the information about the prepositions introducing prepositional arguments, whereas, as we discussed above, LexIt features this important information. At the same time, the subcategorisation frames found in S&C are more reliable than the automatically extracted ones found in LexIt, due to the inherent difficulty of automatically carving syntactic frames from corpora (Schulte im 42 Walde 2009). Lastly, since the data in LexIt are grounded on linguistic productions attested in two corpora, it is possible to associate every verb with its subcategorisation frames frequency in the two corpora, therefore obtaining a precise figure for each and every lemma. On the contrary, it is not possible to carry out this operation with the frames registered in S&C: although the most frequent lemmas are marked, there is no further indication of how frequent their occurrence actually is, both in terms of the overall frequency and in terms of relative frequency. For all these reasons, it was thought best to integrate the two resources described above, supplementing the shortcomings of one with the strengths of the other and vice versa. 3.1.2. Carving verbs subcategorisation frames In order to have a sample of manageable size, we first took the 1746 verbs marked as high frequency verbs in S&C and then we matched them with the corresponding verbs in La Repubblica corpus: this way we assigned each verb a precise frequency value, thereby ordering all of our verbs accordingly. Once the frequency list was drawn up, we narrowed our sample down to the 1000 top frequent verbs. We then manually identified for each verb the corresponding frames registered in S&C, filtering out the technical, archaic and literary uses, integrating the information about the prepositions available in LexIt. We obtained the Excel table in Appendix B in which, for each of the 4450 verb-­‐frame pairs, the following properties were encoded: I. Frequency in the La Repubblica Corpus: a number that indicates the exact frequency value of each verb in the aforementioned corpus; II. Subcategorisation frame: the formalism adopted consisted of a series of abbreviations, which indicated the arguments of the predicate in question, separated by hash tags. Noun phrases (NP) were further divided according to their syntactic position (subj = subject / obj = object), whereas prepositional phrases (PP) were accompanied by the preposition that introduces them curly brackets ({a}PP). As for verbs, we indicated the pronominal form by adding a suffix “–si” to the abbreviation for “verb” (V-­‐si). Since we also took into account 43 argumental sentences, we used curly brackets to mark the complementiser introducing them, and the abbreviations “inf” and “fin” to indicate an infinitive clause and a finite clause respectively ({di}inf / {che}fin). For instance, one of the frames of the verb imporre in S&C is [sogg-­‐v-­‐prep.arg]: this frame was enriched by integrating the information relative to the preposition found in the corresponding frame in LexIt, namely [subj#obj#comp-­‐a], and the resulting frame was registered in our file as [NP_subj#V#NP_obj#{a}PP] (Table 2): S&C: [sogg-­‐v-­‐prep.arg] [NP_subj#V#NP_obj#{a}PP] LexIt: [subj#obj#comp-­‐a] Table 2: Combination of the information in S&C with that found in LexIt We registered a total of 136 subcategorisation frames, which we list and briefly explain in Appendix C. III. Thematic Roles: for each argument we indicated the relevant thematic role, basing our classification on the inventory of 36 roles proposed in the context of the VerbNet project (VerbNet Annotation Guidelines, pp. 19-­‐22). So, for the frame of the verb imporre in Fig.3, we identified the following thematic roles (Table 3): [NP_subj#V#NP_obj#{a}PP] NP_subj: AGENT NP_obj: THEME {a}PP: RECIPIENT Table 3: Thematic roles of the frame [NP_subj#V#NP_obj#{a}PP]of imporre IV. Reference in S&C: for each entry we also specified the frame and, if present, the sense, in which all the information listed was valid (first and second digit respectively). For instance, for the verb imporre in Fig.1, there will be five numbers indicating which one of the five frames allowed by it we are referring to; 44 moreover, for those frames that presented a subdivision in senses, we added a second digit specifying to which one of these senses it pertains. V. Selectional Preferences: for each slot of each frame, we also highlighted its selectional preferences, namely the semantic constraints that every verb impose on its argument in order for the clause to be acceptable. Also in this case, we used the information available in LexIt, listing for every slot of every verb its possible semantic categories, taken from the 24 WordNet super-­‐senses (Miller, 1990; Fellbaum 1998): ANIMAL, ARTIFACT, ACT, ATTRIBUTE, FOOD, COMMUNICATION, KNOWLEDGE, BODY PART, EVENT, NATURAL PHENOMENON, SHAPE, GROUP, LOCATION, MOTIVATION, NATURAL OBJECT, PERSON, PLANT, POSSESSION, PROCESS, QUANTITY, FEELING, SUBSTANCE, STATE, TIME. So, for the frame we discussed above of the verb imporre, we identified the following selectional preferences (Table 4): [NP_subj#V#NP_obj#{a}PP] NP_subj: Person/Group NP_obj: Act/Event/Process/State {a}PP: Person/Group Table 4: Selectional preferences of the frame [NP_subj#V#NP_obj#{a}PP] for imporre 3.2. Data extraction In the first part of this project, we wanted to investigate which frames were associated with the verbs in our sample. Once the annotation procedure described in 3.1.2 was completed, we moved on to second phase, namely the identification of: a) the argument alternations allowed by the verbs in our sample; b) how frequent they are and for which verbs they are valid. Point a) has been accomplished by exploiting a semiautomatic procedure, consisting of an automatic identification of the top correlated frames, followed by a 45 manual inspection. On the other hand, point b) was carried out manually, based on the filtered list of plausible alternations obtained in phase a). 3.2.1. Automatic extraction of argument alternation candidates Given the very high number of possible frame combinations (10,011), as well as the high number of verb-­‐frame pairs identified (4,450), the identification of possible argument alternations could not have been conducted manually. Therefore, we decided to automatically identify the possible candidates comparing the distribution of frames over our target verbs. More specifically, we represented each frame as a vector of binary values whose dimensions indicate whether a verb allows that frame (1) or not (0), and whose dimensionality equals the total number of our target verbs (1000). Thus, the frames [NP_subj#V#NP_obj], [NP_subj#V], [NP_subj#V#NP_obj#{a}PP], [NP_subj#V#{di}inf] are associated to vectors in a manner similar to that shown in Table 5. This illustrative toy space encodes, for instance, the fact that the [NP_subj#V#NP_obj] frame is allowed by imporre, consegnare and dire, but not for arrivare and piovere. On the contrary, the [NP_subj#V] frame is allowed only by arrivare. imporre arrivare piovere consegnare dire NP_subj#V#NP_obj 1 0 0 1 1 NP_subj#V 0 1 0 0 0 NP_subj#V#NP_obj#{a}PP 1 0 0 1 1 NP_subj#V#{di}inf 1 0 0 0 1 Table 5: Examples of vectoral representation of subcategorisation frames In order to identify which of the 10,011 possible frame pairs may constitute a real alternation for at least a subset of our verbs, we calculated the correlation between the corresponding vector pairs, filtering out all those pairings whose correlation failed to reach the 0.2 threshold. The outcome of this procedure was a set of 174 potential argument alternations, for each of which we recorded the list of verbs allowing the candidate alternation. 46 3.2.2. Manual identification of Italian frame alternations The list of potential argument alternations, however, is not fully reliable. Part of the regularity identified by means of correlating frame vectors, indeed, is the by-­‐product of well-­‐known linguistic phenomena, of which the most influential is probably verb polysemy. Even though we based our work on the assumption that a verb sense can be expressed by different subcategorisation frames (Levin 1993), it is true that, in some cases, different frames can also be associated with different senses of the same lemma (Roland & Jurafsky 2002). For instance, the verb sentire may be realised within the following frames: [NP_subj#V#{che}fin] / [NP_subj#V#{di}inf]; however, these two argument structures are paired with two different senses of the verb in question, the first of which corresponds to the English verb to hear, while the second corresponds to the English verb to feel. A minor consequence of this phenomenon is that an argument alternation valid for many verbs can also feature verbs for which it is not so, as illustrated by the example of sentire. Regular patterns of verbal polysemy, however, may interest the whole set of verbs that our automatic procedure associated to a given potential argument alternation. In these cases, it is the frame alternation itself that needs to be marked as incorrect and, eventually, removed from our dataset. For instance, this is what happened with the alternation [NP_subj#V#{da}PP] / [NP_subj#V#{per}inf] registered for the verbs ripartire and venire: for both of these verbs, in fact, the two different frames simply refer to two different semantic patterns, the first one meaning “to leave a place again” and “to come from a place” respectively, and the second meaning “to head towards a destination” and “to go somewhere to achieve something” respectively. Given the impossibility to handle these issues automatically, we decided to manually inspect our plausible frame pairs, verifying whether each verb associated with a given alternation was a proper case of argument alternation or not, thus filtering out those data that turned out to be inconsistent. In this phase, alternations involving only one verb were ruled out as well. In the next chapter we will present an in-­‐depth analysis of our final results, individually discussing each alternation and its peculiarity. 47 Chapter 4 A descriptive analysis of argument alternations in Italian 4.1. Alternations that involve argumental sentences If it is true that most of the research conducted so far on argument realisation has focused on arguments as phrases – either noun phrases (NP) or prepositional phrases (PP) –, it is equally as true that arguments can be realised as subordinate clauses, giving rise to what is known in the literature as a “complex sentence” (see Renzi, Salvi & Cardinaletti 1988, vol. I, I.1.6). Depending on the verb considered and the participants involved in the action it describes, the clause in question is introduced by different types of conjunctions or pronouns (complementisers), as the following examples illustrate: (1) E’ possibile che Gianni parta domani “It is possible that Gianni leave tomorrow” (2) Gianni ha detto che partirà domani “Gianni said that he will leave tomorrow” (3) Gianni non sa chi partirà domani “Gianni does not know who will leave tomorrow” [Graffi 1994, p. 115] If we tried to omit the dependant clauses, we would be left with three agrammatical sentences, respectively: *È possibile, *Gianni ha detto, *Gianni non sa18. This fact shows clearly that the clauses in italics exhibit a syntactic behaviour comparable to that typically attributed to arguments, therefore saturating the valence of the verbs. Given this evidence, Graffi suggested that they should be considered as a different class of dependant clauses altogether, calling them “argumental sentences” (Graffi 1994, p. 116) and further dividing them into nominal clauses – when they act like subjects (1) –, completive clauses – when they function as complements, either direct ones (2) or indirect ones –, and indirect interrogative clause – when they correspond to an interrogative sentence (3) –. The first and third sentences could be somehow acceptable, but only in particular circumstances, such as that of elliptical constructions. 18
48 In this section we will therefore list and describe precisely those alternations that involve a set of arguments both expressed by such clauses, the analysis of which represents an original contribution in more than one way: not only have they never been described in relation to Italian, but not even Levin (1993) took them into account in her seminal work that has inspired the vast majority of contemporary research on argument alternation, including this thesis. 4.1.1. Alternations involving argumental sentences in a complement position In the following paragraphs we will focus our attention on those alternations we found involving “argumental sentences” taking place within the VP, which we referred to as completive clauses. These clauses can have the verb in the infinitive or finite form; the selection of one of these two possibilities influences, on the one hand, the overt realisation of the subject within the argumental sentence in question (which is possible only within the latter option) and, on the other, the complementiser chosen to introduce the dependant clause. It is also important to note that, with the exception of 1.1.6 and 1.1.7., these alternations do not bring about a change in the transitivity of the verb and, consequently, that they do not entail a change in the number of arguments that a verb displays. 4.1.1.1. NP_subj#V#{che}fin#{a}PP / NP_subj#V#{di}inf#{a}PP. In the first alternation discussed here, verbs present three arguments: an external one (the subject), a direct internal one (the object) and an indirect internal one (the dative). They also realise their direct internal argument as what we defined as an “argumental sentence”; however, they allow two different ways of expressing the sentence in question: (4) a. Ti Auguro che tutto vada bene “I wish (to) you that everything go well” b. Ti auguro di fare un buon viaggio “I wish (to) you to have a pleasant journey” 49 In the possibility exemplified in (4a), the argument is realised as a subordinate clause introduced by the complementiser che (“that”) and with its verb in a finite form; on the other hand, in the possibility illustrated in (4b) the argument is expressed as an infinitive sentence introduced by the complementiser di (“of”) (Renzi, Salvi & Cardinaletti 1991, XIII.1.1.2; Giovanardi 1986, pp. 107-­‐110) and, of course, with the verb in the infinitive form. Within our annotated sample, the alternating verbs are: augurare, garantire, ricordare, assicurare, promettere, gridare, confidare, annunciare, raccontare, chiedere, comandare, raccomandare, dire, dichiarare, scrivere, giurare, confessare, comunicare, riferire19. Moreover, following Levin’s verb classification (1993), these verbs can be grouped into different semantic sub-­‐classes that show a high internal semantic coherency and which we present below: a. VERBS OF FUTURE HAVING: garantire, assicurare, promettere, giurare; b. VERBS OF TRANSFER OF A MESSAGE: raccontare, chiedere, scrivere; c. VERBS OF MANNER OF SPEAKING: gridare; d. SAY VERBS: confidare, annunciare, dire, comunicare, riferire; e. GET VERBS: comandare; f. DECLARE VERBS: dichiarare, confessare; g. CHARACHTERISE VERBS: ricordare, raccomandare; h. LONG VERBS: augurare. However, these verbs could be classified in a slightly different way if we took as a relevant parameter the possibility of a verb to have, in the realisation introduced by che, the verb in the subjunctive form rather than in the indicative form. In fact, this distinction is strictly related to the modality implied by each verb, namely the commitment of the speaker towards his or her own utterances (Lyons 1977): the indicative mood is used in so-­‐called “factual utterances”, namely sentences that the speaker considers to be either true or untrue. On the contrary, the subjunctive mood is used in so-­‐called “nonfactual utterances”, namely those sentences in which the speaker suspends his or her judgement in relation to the truth of the sentence. 19 To wish, to guarantee, to remind, to assure, to promise, to shout, to confide, to announce, to tell, to ask, to order, to recommend, to say, to declare, to permit, to write, to swear, to confess, to communicate, to report. 50 4.1.1.2. NP_subj#V#{che}fin / NP_subj#V#{di}inf The second alternation can be considered a variation of the one described above, with the only difference that verbs that present this alternation have only two arguments: an external one (the subject) and a direct internal one (the object). However, as for the one in 1.1.1., the alternation involves a twofold way of expressing the internal argument, realised in both cases as an argumental sentence: (5) a. La situazione impone che tutti partecipino “The situation requires that everyone get involved” b. La situazione impone di essere uniti “The situation requires to stick together” The two argumental sentences allowed are akin to those analysed in 4.1.1.1.: a dependant clause introduced by che and with the verb in the subjunctive form in (5a), and an infinitive clause introduced by the complementiser di in (5b). According to our results, this alternation is allowed by the following verbs: ignorare, disporre, scoprire, tacere, imporre, badare, escludere, vedere, riscoprire, protestare, stabilire, intuire, gridare, temere, confidare, sopportare, constatare, supporre, tollerare, annunciare, pensare, ammettere, deliberare, immaginare, dubitare, fingere, sapere, convenire, ipotizzare, nascondere, smentire, esigere, ritenere, sperare, pretendere, dire, aspettare, riconoscere, dichiarare, credere, dimenticare, sostenere, aggiungere, prescrivere, ottenere, decidere, negare, accettare, dimostrare, sognare, testimoniare, rivelare20. A number of observations can be made: a. There is a group of verbs that allow also the alternation in 4.1.1.1., all belonging to either the VERBS OF MANNER OF SPEAKING class or to the SAY VERBS class: gridare, confidare, annunciare, dire, dichiarare. Obviously, since they all refer to the action of speaking, they semantically imply an interlocutor, somebody to whom something is said: 20 To ignore, to dispose, to discover, to omit, to require, to care for, to count out, to see, to rediscover, to protest, to establish, to grasp, to shout, to fear, to confide, to bear, to ascertain, to presume, to tolerate, to announce, to think, to acknowledge, to deliberate, to imagine, to doubt, to pretend, to know, to agree, to suppose, to hide, to retract, to demand, to consider, to hope, to expect, to say, to wait, to recognize, to declare, to believe, to forget, to maintain, to add, prescribe, to obtain, to decide, to deny, to accept, to demonstrate, to dream, to testify, to reveal. 51 however, depending on the context, this argument can be expressed (see 4.1.1.1.) or not (see above). b. A second group is composed of those verbs that, just like in (5a) and (5b), allow this alternation without implying a necessary co-­‐reference between the subject (grammatical and logical) of the main verb and that of the dependant clause, since the verb does not exert any necessary control over the latter. These verbs are disporre, imporre, stabilire, impedire, prescrivere, decidere, and, in the infinitive option, they all entail a generic reference, which, in fact, does not need to be syntactically realised: (6) a. Il codice prescrive di essere prudenti “The code prescribes to be cautious” b. Il codice prescrive che la domanda sia fatta in carta bollata “The code prescribes that the request be sent on stamped paper” c. A third group arises in opposition to the second one and it consists of all the other verbs left out until now: ignorare, scoprire, tacere, attendere, badare, escludere, vedere, riscoprire, protestare, intuire, temere, sopportare, constatare, supporre, tollerare, pensare, ammettere, deliberare, immaginare, dubitare, fingere, sapere, ipotizzare, nascondere, smentire, esigere, ritenere, sperare, pretendere, ipotizzare, nascondere, smentire, esigere, ritenere, sperare, pretendere, aspettare, riconoscere, credere, dimenticare, sostenere, aggiungere, ottenere, negare, accettare, dimostrare, sognare, testimoniare, rivelare. Despite the heterogeneity of this list, all these verbs have one fundamental semantic aspect in common that explains why they can be clustered together: the actions described by them cannot be transferred onto other referents maintaining the same truth value conditions, since the verb controls the embedded subjects, which, consequently, has to coincide with that of the infinitive clause (See Serianni 1989, XIV.35-­‐36): (7) Mia sorella sostiene di cucinare molto bene “My sister claims to cook very well” 52 A sentence like the one in (7) can only be interpreted as “My sister thinks that she cooks well” and, in Italian, no other referent could be indicated or suggested as the controlled subject of the infinitive dependant clause. Moreover, another peculiarity of the verbs listed in b. and c. is that, in the infinitive realisation, some of them allow the presence of a PP introduced by {a}, as shown in (8): (8) Il rumore ci impedisce di dormire “The noise impedes (to) us to sleep” If we integrated this property as a valid parameter for a finer-­‐grained sub-­‐classification of the verbs presented in b. and c., we would have the following distribution (Fig. 1): Parameters {a}PP Co-­‐reference of the subjects in Tacere, the infinitive form *{a}PP ammettere, Ignorare, scoprire, attendere, nascondere, negare, badare, dimostrare, rivelare escludere, vedere, riscoprire, protestare, intuire, temere, sopportare, constatare, supporre, tollerare, pensare, deliberare, immaginare, dubitare, fingere, sapere, ipotizzare, smentire, esigere, ritenere, sperare, pretendere, aspettare, riconoscere, credere, dimenticare, sostenere, aggiungere, ottenere, accettare, sognare, testimoniare, rivelare Unrequired co-­‐reference of the Imporre, subjects in the infinitive form impedire, Disporre, stabilire, decidere prescrivere Table 6: A finer-­‐grained classification of verbs in b. and c. in the infinitive realisation 53 4.1.1.3. NP_subj#V#{che}fin / NP_subj#V#{come}fin A similar alternation to the one discussed above is the one presented in this section, which is illustrated by the following sentences: (9) a. La maestra ha notato che gli studenti erano stanchi “The teacher noticed that the students were-­‐indic tired” b. La maestra ha notato come gli studenti fossero stanchi “The teacher noticed how the students were-­‐subj tired” With regard to the first realisation, the only difference with the corresponding clause examined in 1.1.3. is that, here, the mood required is the indicative (erano). As for the second possibility allowed by this alternation, namely the one in (9b), we find another completive clause, not an infinitive one: this clause is introduced by typically explicative conjunctions and adverbs, such as come, and it requires the subjunctive mood (fossero). The fact that this alternation is, at a deeper analysis, quite different from the one presented in 1.1 is confirmed also by the verbs that present the one we are discussing here and that are not found in the classes of verbs that allow the one examined in the first paragraph. Based on our results, these verbs are: provare, sottolineare, notare, concepire, ricordare, vedere, giudicare, imparare, sapere, intendere, stabilire, prevedere21. Once again, there is a striking semantic consistency among these verbs and, still Following Levin (1993), we believe that they can be grouped in the classes listed below: a. DECLARE VERBS: provare, sottolineare, giudicare; b. CONJECTURE VERBS: notare, concepire, vedere, sapere, intendere, prevedere; c. CHARACTERISE VERBS: ricordare, stabilire; d. LEARN VERBS: imparare. 21 To prove, to stress, to note, to conceive, to remember, to see (not to be construed as a perception verb though, but, metaphorically, as “to understand”), to judge, to learn, to know, to understand, to establish, to foresee. 54 4.1.1.4. NP_subj#V#{che}fin#{a}PP / NP_subj#V#{come}fin#{a}PP The fourth alternation we present is similar to the one in 4.1.1.1. with respect to the valence of the verbs that allow it (three arguments: an external one, a direct internal one and an indirect internal one). However, this is pretty much all they have in common, as the examples below show: (10) a. Il poliziotto ha spiegato agli studenti che l’autocontrollo è vitale “The policeman has explained that self-­‐control is vital to the students” b. Il poliziotto ha spiegato agli studenti come l’autocontrollo sia vitale agli “The policeman has explained how self-­‐control be-­‐subj vital to the students” The verbs that, in our sample, present this alternation are: spiegare and mostrare22. As the examples above suggest, the only difference with the one discussed in 4.1.1.3. is the presence here of a third core argument, realised as {a}PP and typically referring to the person or group to whom something is explained or shown. 4.1.1.5. NP_subj#V-­‐si#{che}fin / NP_subj#V-­‐si#{di}inf This alternation differs from the ones discussed so far in respect to the morphological form of the verbs that allow it: they are all reflexive verbs, namely verbal forms in which an unstressed clitic pronoun accompanies the verb, becoming a whole with it (see Serianni 1989, XI.18-­‐29). (11) a. Il calciatore si dispiace che la squadra abbia perso “The footballer is sorry-­‐pron that the team lost-­‐subj” b. Il calciatore si dispiace-­‐pron di aver perso “The footballer is sorry to have lost” As the examples in (11b) shows, the relation between the subject (both logical and grammatical) of the main clause and the one in the infinitive clause is the same described in 4.1.1.2., namely an implied co-­‐reference. Moreover, also the sentence in 22 To explain, to show. 55 (11a) confirms what we said above in relation to the possibility of the two clauses to have two different subjects (once again, both grammatical and logical) only if the dependant one is an explicit subordinate. The verbs that, according to our analysis, appeared to allow this alternation are: convincersi, augurarsi, assicurarsi, illudersi, ricordarsi, sorprendersi, immaginarsi, dispiacersi, attendersi, aspettarsi, accorgersi, sognarsi23. It is worth mentioning the fact that, even though all these verbs exhibit the same behaviour with regard to the alternation discussed here, further distinctions can be made between them. With the exceptions of accorgersi, for which the pronoun does not carry any specific value but is simply a formal component of the verb, and dispiacersi, which has the form dispiacere but used only as an impersonal verb (see 4.1.2.3.), all these verbs present also the standard form without the unstressed pronoun. The most interesting thing about this twofold realisation is the different function that the pronoun, depending on the meaning of the verb, carries out in the pronominal variant, which can have noteworthy consequences in relation to the value of the argumental sentence. Based on the argument structure of the verbs listed above, the following sub-­‐
classes can in fact be individuated: a. NP_subj#V#NP_obj: immaginare, attendere, aspettarsi, sognare. Verbs like these take the pronominal form which, however, does not affect the argument structure, since they all remain transitive, as shown in (12): (12) a. Stava attendendo il figlio “He was waiting his son” b. Si attendeva di essere premiato / che lo premiassero “He expected-­‐pron to be awarded / that they awarded-­‐subj him” As for the semantic content, the pronoun si does not entail any major shift in the meaning conveyed; however, it does indicate a greater emotional involvement of the subject (intensive si); 23 To convince, to wish, to assure, to delude, to remind, to surprise (oneself), to imagine, to expect, to notice, to dream. 56 b. NP_subj#V#NP_obj: sorprendere. This case is different from the ones in a. because the pronominal form prompts a drastic change in the structure of the verb, which not only becomes intransitive, but also requires a new argument realised as a PP or, in this case, as an argumental sentence introduced by the complementiser {di}: (13) a. La sua risposta ha sorpreso tutti “His answer surprised everyone” b. Ci sorprendemmo di vederlo ritornare / che finisse così “We surprised-­‐pron to see him coming back / that it would end like this” c. NP_subj#V#NP_obj#{a}PP: augurare, assicurare, ricordare,. In these cases, the pronoun in the pronominal form takes the role of the {a}PP in the non pronominal one, and the argumental sentence that of NP_obj (indirect reflexive verbs): (14) a. Ti ricordo la promessa fatta “I remind (to) you the promise you made” b. Mi ricordai che avevo un appuntamento / Si ricorda sempre di farmi gli auguri “I remembered-­‐pron that I had an appointment” / “She always remembers to wish me happy birthday” d. NP_subj#V#NP_obj#{di}PP: illudere, convincere. In relation to such verbs, the pronoun of the reflexive form indicates the NP_obj of the standard one, and the argumental sentence works as the {di}PP (direct reflexive verbs): (15) a. Convincerò i colleghi dell’opportunità del provvedimento “I will convince my workmates of the convenience of the measure b. Mi convinsi di aver sbagliato / Ci siamo convinti che l’imputato è innocente “I convinced-­‐pron to have done something wrong / We convinced-­‐pron that the accused is innocent 57 4.1.1.6. NP_subj#V#{come}fin#{a}PP / NP_subj#V-­‐si#{come}fin As we briefly anticipated at the start of this section, this alternation implies a change in the syntactic valence of the verb: (16) a. Ti spiego come si fa “I will explain to you how it is done” b. Ora mi spiego come mai non mi saluta “Now I explain-­‐pron how he doesn’t say hi to me” The only two verbs in our annotated sample that allow this alternation are: spiegare and domandare 24 , two illocutionary verbs 25 belonging to the broad class of VERBS OF COMMUNICATION (Levin 1993). As the examples above illustrate, the realisation in (16a) presents three arguments: a subject, namely the person or group who initiates the communicative act, a direct object, namely what is communicated, and an indirect object, which refers to the interlocutor to whom something is communicated. Furthermore, the direct object is expressed as a dependant clause introduced by an interrogative pronoun or conjunction, such as come (“how”), whereas the indirect one is realised as a PP introduced by the preposition a (“to”). In the realisation in (16b), the verb takes the pronominal form and has only two arguments: a subject and a direct object, which is still realised as a subordinate. The indirect object, on the other hand, is represented by the pronoun si, which, in this case, acts precisely as the interlocutor otherwise expressed by the PP, as the paraphrase of (16b) in (17) demonstrates: (17) Ora spiego a me stesso come mai non mi saluta “Now I explained to myself how he doesn’t say hi to me” Verbs that show this particular syntactic and semantic construction are called indirect reflexives (see Serianni 1989, XI. 21). 24 To explain and to ask. 25 “Illocutionary. Applied in the theory of speech acts to the force that an expression of some specific form will have when it is uttered.” (Matthews 1997 (2007), p. 184). 58 4.1.1.7 NP_subj#V#{di}inf#{a}PP / NP_subj#V-­‐si#{di}inf This alternation is very similar to the one discussed in 4.1.1.6 in that it implies a change in the syntactic verb valency: (18) a. Ha permesso al figlio di uscire “He let his son to go out” b. Si permette di andare in vacanza dove vuole “She affords-­‐pron to go on holiday wherever she wants” The verbs in our sample that allowed this alternation are: proporre, permettere, imporre, augurare, assicurare, offrire, ricordare, impedire, rimproverare, risparmiare 26 . Even though these verbs could be ascribed to different classes, they all refer to actions that necessarily entail an affected participant expressed as a PP introduced by the preposition {a}, as illustrated in (18a). In the pronominal form, this argument is not present syntactically, but, as for 4.1.1.6, it is still semantically available precisely in the pronoun si, which co-­‐refers to the subject, as the paraphrase of (18b) in (19) shows: (19) Permette a se stessa di andare in vacanza dove vuole “She affords to herself to go on holiday wherever she wants” Even in this case, the verbs that allow for this alternation are indirect reflexives. 4.1.2. Alternations involving argumental sentences in the subject position In the following paragraphs we will deal with those alternations we found involving argumental sentences taking place in the subject position. In all the cases presented, an argumental clause occupies the subject position (usually filled with a noun), and all the verbs that allow the syntactic structures discussed below are unaccusative intransitive verbs. They also act as impersonal verbs27, thus taking only a third person singular inflection. 26 To propose, to permit, to impose, wish, to assure, to offer, to remind, to impede, to scold, to spare. 27 “[…] constructions lacking a referential subject” (Malchukov & Ogawa 2011, p. 20). 59 Moreover, all the alternations we found allow for an infinitive sentence (which is either introduced by the complementiser di or by the 0 complementiser) and a finite one (which is always introduced by the complementiser che). The selection of one or other of the options depends, of course, on the possibility “to refer the semantic content of the nominal clause to an entity, which, in this case, is syntactically expressed as a noun phrase” (Rensi, Salvi & Cardinaletti 1991, p. 665), and which, of course, takes the position of the subject within the argumental sentence. 4.1.2.1. NP_subj{che}fin#V / NP_subj{di}inf#V In this alternation, the argument that can be expressed in two ways is the one that functions as the subject of the main verb: (20) a. Pare di non ottenere nessun risultato “Seems-­‐impers to not achieve any results” b. Pare che nessun risultato sia ottenibile “Seems-­‐impers that no result is achievable” As these examples suggest, the argumental sentences in (20a) and (20b) act as syntactic subjects of the verbs in the main clause, and they always come after the main verb28. One last thing to note is that, within the argumental sentences in question, in order to maintain the impersonality value expressed by (20a), the NP-­‐object in (20a) moves to the NP-­‐subject position in (20b). Furthermore, it is worth underlining the fact that the mood required in (20b) is always the subjunctive (see Renzi, Salvi & Cardinaletti 1991, XIII.1.2.4.6). There are only three verbs in our survey that present this alternation: sembrare, parere, accadere29. Levin (1993) does not deal with these verbs in respect to their impersonal senses30, thus she did not establish any semantic class for this type of verbs. 28 See the agrammaticality of sentences such as “*[Di non ottenere nessun risultato] pare” (see Renzi, Salvi & Cardinaletti 1988, vol. II, XIII.1.2.4.4). 29 To appear, to seem, to happen. 30 To appear is classified among the APPEAR VERBS, which “describe the appearance of an entity on the scene” (Levin 1993, p. 258), whereas to happen is ascribed to the VERBS OF OCCURRENCE class, which “describe the occurrence of an event” (Levin 1993, p. 261). The verb to seem is not even included in her survey. 60 However, we feel that, in relation to Italian, one is much needed and we propose to call it HAPPEN VERBS. 4.1.2.2. Alternation NP_subj{0}inf#V / NP_subj{che}fin#V A very similar, almost identical, alternation to the one described in the previous paragraph is the one shown in the following examples: (21) a. Bisogna raccogliere i soldi “Needs-­‐impers to collect the money” b. Bisogna che i soldi siano raccolti “Needs-­‐impers that the money be collected” This alternation differs from the one examined in 4.1.2.1. with respect to the infinitive clause in (21a): here, we have a zero infinitive clause (or bare infinitive clause), namely a sentence whose verb is an infinitive not preceded by any complementiser. As for the variation in (20b), the impersonality aspect is once again preserved by moving the NP_obj of the argument clause in (21a) to the subject position of the same clause in (21b), as we discussed in relation to 4.1.2.1. The verbs in our sample that present this alternation are only two: occorrere e bisognare31 . The semantic coherency of these verbs is so blatantly strong that we propose to establish a different class for them, called NEEDS IMPERSONAL VERBS. 4.1.2.3. Alternation NP_subj{0}inf#V#{a}PP / NP_subj{che}fin#V#{a}PP Even though the alternation analysed here might seem very similar to the one discussed above, this is actually not the case. If anything, it is closer to the one we presented in 4.1.2.1., even if, here, the presence of the indirect internal argument realised as a PP introduced by the preposition a adds the experiencer of the action expressed by the main verb: 31 To need, to be necessary. 61 (22) a. A noi conviene partire subito “To us mattersimpers to leave immediately” b. A noi conviene che tu parta subito “To us mattersimpers that you leavesubj immediately” From the analysis of our data, it emerged that the verbs allowing this alternation are: dispiacere, convenire, risultare32. Given the fact that, in English, these verbs act in a very different way, it does not come as a surprise that they are not dealt with by Levin (1993); however, as for the previous cases, we believe that they deserve to be ascribed to a semantic class and we propose HAPPEN VERBS WITH A PP. 4.2. Alternations that involve noun or prepositional phrases (NP or PP) In this second section we present the alternations we found that, contrary to the ones presented in the first part of this chapter, involve noun phrases (NN) or prepositional phrases (PP), as illustrated by the examples below: (23) Gianni ha visto la sua amica con un binocolo (NP) “Gianni has seen his friend with a binoculars” (NP) (24) Gianni ha visto la sua amica con un binocolo (PP) “Gianni has seen his friend with a binoculars” (PP) [Graffi 1994, p. 89] This kind of alternation is by far the most studied cross-­‐linguistically and “they have been a mainstay of syntactic research since the beginning of generative grammar” (Dowty 2000, p. 111). These alternations cause either a different syntactic realisation of the participants involved in the event described by the predicate (as in (25), where the subject position is occupied by the Agent in a. and by the Location in b.), or a change in the valence of the verb (as in (26) where the predicate is transitive in a. and intransitive in b.): 32 To be sorry, to suit, to result. However, it is important to stress that, in Italian, these verbs can only be impersonal, with the experiencer always realised as dative. 62 (25) a. Bees are swarming in the garden b. The garden is swarming with bees [Salkoff 1983, p. 288] (26) a. Janet broke the vase b. Crystal vases break easily [Dowty 2001, p. 179] In this section we present the results of our analysis in relation to these types of so-­‐
called “classical argument alternations”, providing new data for Italian, which, up until now, still consist only of the works of Jezek (2003), Lenci (2009), Lenci (in press) and Cennamo (forthcoming). 4.2.1. Alternations causing a change in the transitivity of the verbs Regardless of the idiosyncratic peculiarities of the alternations presented in this first part, there is a common thread that makes it possible to cluster them all together: in fact, in all of them the verbs, originally transitive, turn into syntactically unaccusative ones taking the pronominal form33. In the next paragraphs we will therefore go through each and every one of these alternations, focusing on their specific characteristics. 4.2.1.1. NP_subj#V#NP_obj#{da}PP / NP_subj#V-­‐si#{da}PP In the first option of this alternation (27a), the verb has three arguments (an NP subject, an NP object, and a {da}PP); in the second option (27b), the verb actually becomes unaccusative, taking the pronominal form indicated by the presence of the pronoun si: (27) a. Hai buttato i sassi dalla finestra “You threw stones from the window” b. Lui si è buttato dal trampolino “He jumped-­‐pron from the trampoline” 33 For a discussion on the relation between pronominal verbs in Italian and unaccusativity see Jezek (2003), III.1.2. 63 The verbs in our sample that allow for this alternation are: rilanciare, buttare, escludere, difendere, gettare, dividere, staccare, spostare, lanciare, sciogliere, allontanare, sollevare, distrarre, trarre, ritirare, levare, sfilare, separare, liberare, ritrarre, riparare, salvare34. Given the semantic coherency of these verbs, it is not surprising that they can be grouped into well definite sub-­‐classes, once again based on those found in Levin (1993): a. THROW VERBS: rilanciare, buttare, gettare, lanciare; b. REMOVE VERBS: escludere, difendere, dividere, staccare, spostare, sciogliere, allontanare, sollevare, distrarre, trarre, ritirare, levare, sfilare, separare, liberare, ritrarre, riparare, salvare. The verbs in the first class describe situations in which an entity is suddenly set in motion and then moves without the help of the agent that originated the action. This entity is realised as an object in (27a) and as a subject in (27b). In both cases, the PP introduced by the preposition da (“from”) indicates the starting point of the action, the SOURCE35. The verbs belonging to the second class, on the other hand, describe actions in which an entity is removed from either a physical place, as in (28a), or a situation, as in (28b) (again realised as a PP introduced by da): (28) a. Allontana la sedia dal muro! “(You) remove the chair from the wall!” b. Allontanati dai pericoli! “(You) remove-­‐pron from dangers!” One last point worth stressing is that, in Italian, all the verbs listed are, in the pronominal variant and only with an agentive subject, direct reflexive verbs, namely verbs that describe actions “which the subject carries out on themself, treating himself or herself as object” (Lepschy 1986, p. 12). This means that a finer-­‐graded distinction has to be made between those verbs that allow only for an agentive subject, and those that do not, allowing for both an agentive and a non-­‐agentive subject. In fact, in the 34 To throw again, to throw out, to exclude, to defend, to toss, to divide, to unplug, to move, to throw, to untie, to remove, to lift, to distract, to pull, to pull back, to take away, to pull out, to separate, to free, to withdraw, to shelter, to save. 35 See VerbNet Annotation Guidelines, p. 21 (downloadable at http://verbs.colorado.edu/verb-­‐
index/). 64 former case, verbs are and can only be direct reflexives, whereas, in the latter, two options are given: a direct reflexive one when the subject is agentive, and an intransitive pronominal one when the subject is an inanimate non-­‐agentive entity. Considering this new parameter, the following classification is also possible and valid: I. DIRECT REFLEXIVE VERBS: rilanciare, buttare, gettare, lanciare, escludere, difendere, distrarre, trarre, ritirare, levare, separare, liberare, ritrarre, riparare, salvare. These verbs always entail an animate agentive subject and, in the pronominal form, even though they syntactically belong to the class of unaccusative verbs, their semantic valency is still somehow transitive, with the pronoun si acting as the object and co-­‐referring to the same entity of the subject. It follows that a sentence like (29) can be paraphrased as shown in (30): (29) Il suicida si è gettato dalla finestra “The suicide victim threw-­‐pron from the window (30) Il suicida ha gettato se stesso dalla finestra “The suicide victim threw himself from the window” II. DIRECT REFLEXIVE/INTRANSITIVE PRONOMINAL VERBS: dividere, staccare, spostare, sciogliere, allontanare, sollevare, sfilare. Depending on the agentivity of the noun filler in the subject slot, in the pronominal realisation these verbs can be direct reflexive, in which case the pronoun si does act as a semantic object co-­‐referring with the subject (28b), but they can also be intransitive pronominal verbs, giving rise to a causative-­‐inchoative alternation in which a sentence like (31) cannot be paraphrased as (32): (31) La nave si è allontanata dalla costa “The ship removed-­‐pron from the shore (32) *La nave ha allontanato se stessa dalla costa *“The ship has removed itself from the shore 65 4.2.1.2. NP_subj#V#NP_obj#{di}PP / NP_subj#V-­‐si#{di}PP This alternation presents the same structure as the one discussed in 4.2.1.1., with the only syntactic difference being the preposition introducing the PP, which, in this case, is di (“of”). (33) a. Ho convinto i colleghi della validità della teoria “I have convinced the colleagues of the validity of the theory” b. Mi sono convinto del mio errore “I convinced-­‐pron of my mistake” From the analysis of our sample, it emerged that this alternation is allowed by the following verbs: convincere, privare, fornire, ricoprire, riempire, caricare, svuotare, dotare, investire, colmare, incaricare, circondare, coprire36. They can be grouped in the following sub-­‐classes: a. FILL VERBS: ricoprire, riempire, caricare, svuotare, colmare, circondare, coprire; b. VERBS OF FULFILLING: fornire, dotare, investire, incaricare37; c. VERBS OF POSSESSIONAL DEPRIVATION: privare; d. VERBS OF SOCIAL INTERACTION: convincere. These verbs are, like the ones in 4.2.1.1., direct reflexive and, therefore, what we said in relation to the semantic value of the pronoun si is valid also for the verbs listed above; this means that (33b) can be reformulated as (34): (34) Ho convinto me stesso del mio errore As for the semantic role of the PP introduced by di, in the three first classes described above it refers to an entity that is either moved (in a.), given (in b.) or taken away (in c.). 36 To convince, to deprive, to supply, to lavish, to fill up, to load, to empty, to provide, to assign, to fill to the brim, appoint, to overload, to cover. 37 In Levin (1993), to assign and to appoint are classified as a VERB OF FUTURE HAVING and as an APPOINT VERB respectively. However, given their meaning in Italian and their syntactic behaviour, we believe that they fit the VERBS OF FULFILLING class better. 66 Also, the NP in the object position can describe either a physical place, as in (35), or an animate entity, as in (36) and (37): (35) Coprire le pareti di quadri “To cover the walls with paintings” (36) Incaricare un amico della commissione “To appoint a friend to an errand” (37) Privare i dipendenti della paga “To deprive workers of the salary” As for the verb convincere in d., the NP refers always to a person or a group, while the PP introduced by di can be construed as a STIMULUS, namely a “cause in an event that elicits an emotional or psychological response” (VerbNet Annotation Guidelines, p. 21). 4.2.1.3. NP_subj#V#NP_obj#{a}PP / NP_subj#V-­‐si#{a}PP This alternation shares the basic structure we have dealt with until now, but contrary to what we have seen so far, it features a PP introduce by the preposition a (“to”, “beside”, “with”): (38) a. Ha abituato i figli allo studio “She accustomed the children to studying” b. Si sono abituati a una nuova vita “They have accustomed-­‐pron to a new life” Within our annotated sample, the following verbs allow for this alternation: disporre, affidare, donare, mostrare, alternare, opporre, adattare, abituare, mescolare, affiancare, costringere, agganciare, esporre, consacrare, adeguare, accordare, unire, iscrivere, preparare, sottrarre, paragonare, attaccare, raccomandare, presentare, votare, appassionare, associare, allineare, vendere, avvicinare, indirizzare, dichiarare, dare, predisporre, sommare, sottoporre, accostare, concedere, rivolgere, consegnare, interessare, 67 dedicare38. What all these verbs have in common, in spite of their obvious semantic differences, is the fact that they all describe actions whose accomplishment necessarily implies either a final point to reach, or a general entity without the participation of which the action is not completed: in both cases, it is expressed as a PP introduced by the preposition a. This implied ending point, however, can refer to various entities, depending on which the following sub-­‐classes are possible: I. {a}PP = physical place: affiancare, attaccare, iscrivere, avvicinare, accostare (VERBS OF MOTION); (39) a. Affiancheremo il mobile alla scala “We will move the wardrobe beside the stair” b. Ti sei affiancato a un auto “You moved-­‐pron beside a car” II. {a}PP = person/group: affidare, donare, promettere, raccomandare, presentare, vendere, indirizzare, dare, concedere, consegnare (VERBS OF CHANGE OF POSSESSION), dichiarare, rivolgere (VERBS OF COMMUNICATION); (40) a. Ti presento a mia madre “I introduce you to my mother” b. Presentarsi ai nuovi colleghi” “To introduce-­‐pron to the new colleagues” III. {a}PP = situation or state of things: disporre, adattare, abituare, costringere, esporre, consacrare, adeguare, preparare, sottrarre, allineare, votare, appassionare, predisporre, obbligare, sottoporre, interessare, dedicare (mainly VERBS OF SOCIAL INTERACTION) ; 38 To prepare, to entrust, to donate, to show, to alternate, to oppose, to adjust, to accustom, to mix, to place side by side, to force, to hook, to expose, to consecrate, to adapt, to reconcile, to join, to enrol, to get ready, to devote, to thrill, to associate, to adjust, to sell, to approach, to address, to declare, to give, to predispose, to sum up, to undergo, to pull close, to grant, to address, to give, to interest, to dedicate. 68 (41) a. Sottrassero il ragazzo agli sguardi indiscreti dei presenti “They took the boy away from the indiscrete gazes of the people” b. Ti sei sottratto a una responsabilità “You took away-­‐pron from a responsibility” IV. {a}PP = third necessary entity: opporre, alternare, mescolare, agganciare, accordare, paragonare, associare, sommare (VERBS OF COMBINING AND ATTACHING). (42) a. Accordo i tendaggi all’arredamento “I match the curtains with the furniture” b. Il colore delle pareti si accorda a quello del divano “The colour of the walls matches-­‐pron with that of the couch” 4.2.1.4. NP_subj#V#NP_obj#{in}PP / NP_subj#V-­‐si#{in}PP The peculiarity of this alternation lies in the preposition introducing the PP, namely in (“in, into”): (43) a. Il provvedimento ha integrato i portatori di handicap nella società “The measurement has integrated disabled people within society” b. Mi sono integrato nel nuovo ambiente di lavoro “I have integrated-­‐pron in the new workplace” The verbs in our sample that allow this alternation are: proiettare, integrare, immergere, situare, rinchiudere, specializzare, inserire, inquadrare, trasformare39. Even though these verbs are semantically heterogeneous, in all of them the preposition in refers to a situation, a place or a physical entity that can be construed as the final point of the action described by the verb. We propose now a sub-­‐classification, partly derived from Levin (1993) – a., b., c. –, partly proposed here for the first time – d., e. –: 39 To project, to integrate, to put effort in, to immerse, to situate, to close in, to specialise, to insert, to contextualise, to transform. 69 a. PUT VERBS: immergere, rinchiudere, inserire; b. AMALGAMATE VERBS: integrare; c. TURN VERBS: trasformare; d. VERBS OF CONTEXTUALISATION: situare, inquadrare; e. VERBS OF DEDICATION: proiettare, specializzare. In a manner similar to that discussed in relation to the verbs in 4.2.1.1., we believe that a further distinction can be made depending on the semantic value of the pronoun si in the pronominal realisation, an aspect that is strictly related to the possibility for these verbs to have only agentive subjects or not: I. DIRECT REFLEXIVE VERBS: rinchiudere, integrare. These verbs can only have an animate agentive subject and, in the pronominal form, even though they are unaccusative verbs, their semantic valency is still transitive, with the pronoun si acting as the object of the verb and co-­‐referring to the same entity of the subject, as demonstrated by the possibility to paraphrase (43b) as (44): (44) Ho integrato me stesso nel nuovo ambiente di lavoro “I integrated myself in the new workplace” II. DIRECT REFLEXIVE/INTRANSITIVE PRONOMINAL VERBS: immergere, inserire, trasformare, proiettare, specializzare. These verbs allow for both an agentive subject, in which case the pronoun si does act as a semantic object co-­‐referring with the subject (45), and a non-­‐agentive subject, therefore becoming intransitive pronominal verbs (46), which, unsurprisingly, cannot be paraphrased as (47): (45) Il fidanzato di mia sorella si è inserito bene nella compagnia “My sister’s boyfriend integrated-­‐pron well in the group” (46) Il piano si inserisce nella struttura portante “The flat surface integrates-­‐pron into the carrier structure” (47) *Il piano inserisce se stesso nella struttura portante *“The flat surface integrates itself into the carrier structure” 70 III. INTRANSITIVE PRONOMINAL VERBS: situare, inquadrare. These verbs can only have non-­‐
agentive subjects and, in the pronominal realisation, the pronoun si indicates the inchoativity of the action described (48) and it could never be paraphrased as (49): (48) Il provvedimento si inquadra in un vasto piano di interventi “The measure places-­‐pron within a broad intervention program” (49) *Il provvedimento ha inquadrato se stesso in un vasto piano di interventi *”The measure has placed itself within a broad intervention program” This particularity is due to the fact that, while the verbs in I describe actions that the agent carries out on themselves or on one of their qualities, the verbs in III indicate actions that can only be transferred onto external entities, as the agrammaticality of the sentence in (50) shows: (50) *Mi sono situato in un momento economico sfavorevole *”I placed myself in an unfavourable economic time” On the other hand, as we have discussed above, the verbs in II lay something in between those in I and III, allowing for both types of subject and, consequently, allowing for both interpretations of the pronoun si. 4.2.1.5. NP_subj#V#NP_obj#{su}PP / NP_subj#V-­‐si#{su}PP This alternation has the same basic structure as the ones presented so far, but with the PP introduced by the preposition su (“on”): (51) a. Hanno basato i calcoli sui dati ufficiali “They based the evaluations on official data” b. L’accusa si basa sulle testimonianze “The prosecution bases-­‐pron on the depositions” 71 Within our sample of verbs, those that present this alternation are: fondare, proiettare, basare, concentrare. Since none of these verbs is taken into consideration in Levin (1993), we propose our own sub-­‐classification: a. BASE VERBS: fondare, basare; b. VERBS OF FOCUS: proiettare, concentrare. The interesting fact about these verbs is that, even though it is still true that they are direct reflexives, in the pronominal form the entity indicated by the pronoun si does not coincide with the one denoted by the subject, but rather with one of its faculties, as demonstrated by the fact that (52) cannot be paraphrased as (53), but, in fact, by something along the lines of (54): (52) Mi sono concentrato sul libro “I concentrated-­‐pron on the book” (53) ?Ho concentrato me stesso sul libro ?“I concentrated myself on the book” (54) Ho concentrato l’attenzione/lo sguardo sul libro “I concentrated my attention/my mind on the book” 4.2.1.6. NP_subj#V#NP_obj#{con}PP / NP_subj#V-­‐si#{con}PP This alternation presents, once again, the same general structure discussed so far, but with a PP introduced by the preposition con (“with”): (55) a. I contadini scambiano grano con petrolio “Farmers swap wheat with petrol” b. Mi sono scambiato con il compagno di banco “I swapped-­‐pron with my deskmate” 72 From the analysis of our sample, it appears that this alternation is allowed by these verbs: conciliare, combinare, alternare, scambiare, confrontare, mescolare40. Accordingly to Levin (1993), the following sub-­‐classes are possible: a. AMALGAMATE VERBS: conciliare, alternare, scambiare, confrontare; b. MIX VERBS: combinare, mescolare. As the names of the classes suggests, all these verbs describe actions in which an entity is mixed up or amalgamated with another one. Moreover, the striking inherent semantic coherency of these verbs is confirmed by the fact that, in Italian, the preposition that indicates the conjunction of two or more elements is precisely the one found in both realisations, namely con (see Serianni 1989, VIII.89-­‐91). With regard to the pronominal form, all these verbs allow for both an agentive and a non-­‐agentive subject, the selection of which is responsible for a direct reflexive interpretation (56) or an intransitive pronominal one (58), strictly followed by the possibility or not of a paraphrase, as demonstrated in (57) and (59) respectively: (56) A lavoro, mi alterno con la mia amica “At work, I alternate-­‐pron with my friend” (57) A lavoro, alterno me stessa con la mia amica At work, I alternate myself with my friend (58) In un musical, i dialoghi si alternano con le canzoni “In a musical, dialogues alternate-­‐pron with songs” (59) *In un musical, I dialoghi alternano se stessi con le canzoni *”In a musical, dialogues alternate themselves with songs” 4.2.1.7. NP_subj#V#NP_obj / NP_subj#V-­‐si The alternation we discuss now is by far the most common one within our sample, including approximately 20% of the verbs: given this substantial inventory and for reasons of space, we thought it best to list all the verbs only in the appendix, discussing 40 To reconcile, to cover again, to alternate, to swap, to confront, to mix. 73 here only their syntactic and semantic properties. Moreover, we chose to attempt a classification not based on Levin’s classes, but on the particular semantic value that the pronominal form carries: this allowed us to focus on the most salient features of the verbs and the different entailments that the alternation has. Therefore, the classification we propose does not claim to be either thorough or the only one possible: on the contrary, we believe that a deeper semantic classification of these verbs, preferably based on a larger sample, is indeed needed and desirable. We will present here the different sub-­‐classes we found, with the indication of the most representative verbs for each class, which, however, we fully list in the Appendix: I. DIRECT REFLEXIVES: isolare, lavare, rinnovare, liberare, negare, allontanare, scoprire, schierare, esprimere, giustificare, assicurare, uccidere, ammazzare, allenare, licenziare, accettare, consolare, contraddire, tormentare, interrogare, valorizzare, ferire, escludere, spiegare, umiliare, nascondere41. The semantic peculiarity of these verbs is the fact that, in the pronominal form, subject and object co-­‐refer to the same entity, as illustrated by the semantic equivalence of (60a) and (60b) (Serianni, XI. 18): (60) a. Spesso si giustifica “Often he justifies-­‐pron b. Spesso giustifica se stesso “Often he justifies himself” From a syntactic perspective, however, they are intransitive verbs that, in fact could never take a direct object, as shown in (61a), but, if necessary, could be accompanied by an indirect object (61b): (61) a. *Vi contraddite voi *“You contradict yourself you” b. Vi contraddite (con le vostre azioni) “You contradict-­‐pron (with your actions)” 41 To isolate, to wash, to renovate, to liberate, to deny, to push, to discover, to deploy, to express, to justify, to assure, to kill, to murder, to train, to fire, to accept, to consulate, to contradict, to torment, to interrogate, to value, to injure, to exclude, to explain, to humiliate, to hide. 74 II. RECIPROCAL REFLEXIVES: conoscere, sospettare, rispettare, combattere, fronteggiare, controllare, attirare, ritrovare, rivedere, stimare, sfidare, abbracciare, rincorrere, inseguire, sfiorare, scegliere, dividere, stringere, picchiare, disturbare, odiare, respingere, frequentare, temere, incontrare, sposare, vedere, trovare, baciare, soccorrere 42 . What makes these verbs interesting is the fact that they “express an action that two or more subjects carry out and undergo at once” (Battista Moretti & Orvieto 1983, vol. III, p. 83). This semantic property shows its influence on the syntactic construction allowed by these verbs in that they can only have a plural subject (62) and would not be grammatically acceptable with a singular one (63): (62) Ci siamo incontrati tanto tempo fa “We met-­‐pron a long time ago” (63) *Luca si è incontrato ieri *”Luca met himself yesterday” I. CAUSATIVE-­‐INCHOATIVE ALTERNATION: chiudere, spaventare, confondere, emozionare, abbassare, restringere, staccare, ridurre, intrecciare, sbloccare, scatenare, rovinare, turbare, piegare, conservare, spezzare, spaccare, rompere43. These verbs allow the argument alternation extensively examined in 2.1., namely the causative-­‐inchoative alternation. As we discussed in the second chapter, this alternation consists of a pair of sentences whose verb expresses the same situation, generally a change of state, but starting from different semantic perspectives, which are reflected in the two different realisations. In the causative one, the verb meaning includes an agentive participant who causes the situation (64), whereas in the inchoative variant the verb meaning excludes such agent and presents the situation as occurring spontaneously (65): (64) Lucia ha spezzato un biscotto “Lucia broke a biscuit” 42 To know, to suspect, to respect, to fight, to face, to control, to attract, to rediscover, to see again, to esteem, to challenge, to hug, to chase, to run after, to brush against, to choose, to split, to cling, to beat up, to disturb, to hate, to repel, to hang out, to fear, to meet, to marry, to see, to come across, to kiss, to help. 43 To close, to frighten, to confuse, to move, to lower, to shrink, to remove, to reduce, to intertwine, to unblock, to instigate, to ruin, to upset, to bend, to keep, to crack, to smash, to break. 75 (65) La corda si è spezzata “The rope has broken-­‐pron” The verbs in our sample that allow this alternation are mainly CHANGE OF STATE VERBS (Levin 1993), such as spezzare, rovinare, piegare and rompere. However, there are numerous PSYCH-­‐VERBS (Levin 1993), such as spaventare, emozionare and turbare: even in this case, the inchoative realisation does not feature an agentive participant responsible for prompting the action, which is still presented as a spontaneous happening (66): (66) Non mi emoziono facilmente “I am not moved-­‐pron easily” 4.2.1.8. NP_subj#V#NP_obj#{...}PP / NP_subj#V-­‐si#{...}PP This alternation is, to a certain extent, syntactically similar to the one discussed in 4.2.1.7, but with some notable semantic differences. The first thing to notice is that, contrary to the previous alternation, the argument structure of this one is made more complex by the introduction of a PP; moreover, this PP is not introduced by a definite preposition, but, depending on the verb meaning and on the noun filler of the slot in question, it can be preceded, in both realisations, by different prepositions, as shown in (67): (67) a. Marco ha buttato il riso agli sposi / la maglia in lavatrice / acqua sul fuoco “Marco threw rice at the newlyweds / the jumper in the washing machine / water on the fire” b. Marco si è buttato in acqua / sul letto / contro qualcuno “Marco threw-­‐pron in the water / on the bed / against someone” The verb in our sample that allow for this alternation are: infilare, girare, disporre, informare, calare, orientare, buttare, muovere, trascinare, collocare, gettare, voltare, fissare, versare, rilanciare, avviare, aggiungere, aprire, spingere, insinuare, accordare, avvolgere, stabilire, spostare, rovesciare, lanciare, imprimere, spargere, dirigere, piazzare, 76 sistemare, ambientare, stampare, stendere, mettere, rivolgere, abbandonare, porre44. As for the verbs in 4.1.7., we propose a classification of these verbs based on the particular semantic value that the pronominal form carries: I. DIRECT REFLEXIVES: girare, disporre, calare, orientare, buttare, muovere, trascinare, gettare, voltare, rilanciare, avviare, spingere, stabilire, spostare, lanciare, dirigere, piazzare, sistemare, ambientare, stendere, mettere, rivolgere, abbandonare, porre (VERBS OF MOTION), informare, accordare45 (VERBS OF COMMUNICATION). When all these verbs take the pronominal form, the pronoun si does act as a semantic object, co-­‐
referring with the subject itself, as illustrated by the possible reformulation of (68b) as (69): (68) a. I fondatori hanno stabilito la sede dell’azienda a Roma / nella capitale “The founders have established the headquarters of the business in Rome / in the capital” b. Ho deciso di stabilirmi a Milano “I have decided to establish-­‐pron in Milan” (69) Ho deciso di stabilire me stesso a Milano46 “I have decided to establish-­‐pron myself in Milan” II. CAUSATIVE-­‐INCOHATIVE ALTERNATION: infilare, collocare, versare, aggiungere, aprire, insinuare, avvolgere, rovesciare, spargere, imprimere, stampare. As seen in 4.2.1.7., these verbs allow for two different realisations whose meaning varies substantially: the first one (70a) is a causative sentence in which the subject refers to either an agent or a cause responsible for the instantiation of the action; in the second one (70b), the subject corresponds to a semantic object that undergoes the action described by the verb, which is presented as random and in no way attributable to an agentive participant: 44 To pierce through, to turn, to put in order, to inform, to lower, to orientate, to throw, to move, to carry, to place, to toss, to turn, to cast again, to address, to add, to open, to push, to insinuate/to creep in, to agree, to wind up, to establish, to transfer, to spill, to cast to impress, to spread, to direct, to position, to set, to locate, to print, to spread out, to put, to turn again, to abandon, to put. 45 These two verbs of communication are quite problematic since the pronominal form allows for an interpretation in between the typical direct reflexive one and the intensive reflexive one, without clearly falling into either of them. 46 This sentence is perfectly formed in Italian, but it could be used only in a marked context, such as that allowed by topicalisation mechanisms. 77 (70) a. Ho rovesciato la minestra sul tappeto “I spilt the soup on the carpet” b. Il latte si è rovesciato per terra “The milk spilt-­‐pron onto the floor” As for the PP, it refers to either the ending point of the action, as shown in (70), or simply a physical entity on which the action is imposed, as in (71): (71) a. Ha impresso le proprie orme sulla sabbia “She impressed her footsteps on the sand” b. Le orme si imprimono sulla sabbia “Footsteps impress-­‐pron onto the sand” III. INTENSIVE REFLEXIVES: fissare. This class consists of only one verb, but it is representative of a type of verbs that allow a pronominal form, in which, however, the pronoun does not act a semantic object co-­‐referring to the subject, but rather with one of the subject’s qualities (72): (72) Fissa sempre lo sguardo sulla mia ragazza “He always fixes his gaze on my girlfriend” This fact is also confirmed by the impossibility of a sentence like (73) to be paraphrased as (74), since the pronoun in the former sentence simply indicates a greater involvement of the subject, and not the subject itself: (73) Si è fissato con quella ragazza / in un progetto “He fixed-­‐pron with that girl / in a project” (74) *Ha fissato se stesso con quella ragazza / in un progetto *“He fixed himself with that girl / in a project” 78 4.2.2. Alternations involving two intransitive variants The alternation we discuss in this section differs from the ones presented so far with respect to transitivity: in the case examined below, in fact, verbs are always intransitive in both the realisations allowed. 4.2.2.1. NP_subj#V / NP_subj#V#{...}PP This alternation involves a change in the number of arguments required by the verb and, as we will see, it is connected to a more specified use of the verb: (75) a. Nel dopoguerra, molti furono costretti ad emigrare “During the post-­‐war period many people were forced to emigrate” b. I miei nonni emigrarono in Germania “My grandparents emigrated to Germany” The verbs within our sample that allow this alternation are: tornare, girare, correre, giocare, finire, accorrere, uscire, battere, emigrare, terminare, perdere, bastare, combattere, risalire, arrivare, durare, oscillare, riuscire, volare, nascere, giungere, ricadere, cadere, scivolare, avanzare, votare, venire, crescere, piombare, sorgere, comparire, reagire, saltare, slittare, lavorare, picchiare, salire, vagare, precipitare, rientrare47. They can be grouped into the following sub-­‐classes (Levin 1993), the last three of which (p., q., r.) we propose here for the first time: a. VERBS OF INHERENTLY DIRECTED MOTION: tornare, uscire, risalire, arrivare, giungere, ricadere, cadere, avanzare, venire, piombare, salire, precipitare, rientrare; b. ROLL VERBS: girare, oscillare, scivolare, slittare; c. RUN VERBS: correre, accorrere, volare, saltare; d. LEAVE VERBS: emigrare; e. PERFORMANCE VERBS: giocare; 47 To come back, to spin, to run, to play, to end, to hasten, to exit, to hit, to emigrate, to finish, to loose, to be enough, to fight, to go back up, to arrive, to last, to swing, to succeed, to fly, to be born, to reach, to fall again, to fall, to slide, to advance, to vote, to come, to grow, to precipitate, to rise, to appear, to react, to jump, to slid, to work, to beat, to go up, to wander, to plunge, to come back in. 79 f. BEGIN VERBS: finire, terminare; g. HIT VERBS: battere, picchiare; h. APPEAR VERBS: nascere, sorgere, comparire; i. VERBS OF CALIBRATABLE CHANGE OF STATE: crescere; j. MEANDER VERBS: vagare; k.. OTHER ALTERNATING VERBS OF CHANGE OF STATE: perdere; l. KNEAD VERBS: lavorare; m. MEET VERBS: combattere; n. MARVEL VERBS: reagire; o. VERBS OF FULFILLING: votare; p. NEED VERBS: bastare; q. VERBS OF DURATION: durare; r. VERBS OF SUCCESS: riuscire. Even though this group includes semantically heterogeneous verbs, they are all verbs whose meaning can be further specified through the addition of a PP, the preposition of which is greatly determined by the specifying complement selected by each verb, as the few examples in (76) and (77) demonstrate: (76) Sta venendo una macchina A car is coming La trottola non gira più The spinning top does not spin anymore Ai bambini piace saltare Kids like to jump Possiamo andare a giocare? “Can we go to play?” La partita è terminata “The match has ended” Il sole è sorto “The sun has risen” La ditta sta perdendo “The business is loosing” 80 L’esercito combatte valorosamente “The army is fighting bravely” I soldi non bastano mai “Money is never enough” Sono tessuti che durano “These are fabrics that last” L’impresa è riuscita “The endeavour succeeded” (77) Vieni subito da me [DIRECTION] La Terra gira intorno al Sole [DIRECTION] [MANNER] Il gatto gioca con il gomitolo [INSTRUMENT] [PLACE] [PLACE] [QUALITY] [PURPOSE] [BENEFICIARY] “Come here to me” “The Earth spins around the Sun” Saltare come un cerbiatto “To jump like a fawn” “The cat is playing with the ball of wool” Il sentiero termina nel bosco “The path ends in the wood” La città sorge su una rupe “The city rises on a cliff” Perdere di autorità “To loose in authority” Combattere per la liberta “To fight for freedom” Ti basta un foglio? “Is one paper sheet enough for you?” Il bel tempo è durato per una settimana “The good weather lasted for a week” Riuscire nell’esame [TIME] [ACTIVITY] “To succeed in the exam” 81 This alternation can be interpreted as an instantiation of the phenomenon known as “argument drop” or “null arguments” (Cole 1987; Huang 1995), according to which core arguments of a given verb can be syntactically unexpressed under particular circumstances. The most widely known type of “argument drop” is the “object drop” (Keller & Lapata 1998), whereas, in this case, we have a case of “indirect object drop”, clearly related to the possibility shown by these verbs to feature in a more general context or, on the contrary, in a more specified one. 4.2.3. Alternations involving verbs with predicative complements A predicative complement usually “consists of a noun or adjective that, referred to the subject or the direct object, is used to determine and complete the meaning of the verb” (Serianni 1989, p. 95). Given its peculiar nature, we thought it best to keep it separate from the other alternations examined in the two previous sections. 4.2.3.1 NP_subj#V#Cpred / NP_subj#V#Cpred#{a}PP This alternation implies a change of valence in the verbs that allow it, with the addition of a PP introduced by the preposition a (“to”) in one of the two possible realisations (78b): (78) a. Tutti gli sforzi sono apparsi vani “All the efforts seemed vain” b. L’esito dell’incontro apparve scontato a tutti i presenti “The outcome of the meeting seemed predicted to all the present people” The verbs in our sample that allow this alternation are: riuscire, apparire, risultare48. They all belong to the class of copulative verbs, which are verbs that, similarly to verb to be, have a generic semantic content and therefore need to be further specified by a predicative complement. The realisation in (78b) differs from the one in (78a) for the presence of a PP that simply indicates the EXPERIENCER (see VerbNet Annotation Guidelines, p. 21), namely 48 To come across as, to seem, to turn out to be. 82 the person or group aware of the action described by the combination of the copulative verb and the predicative complement. 4.3. Alternations involving an argumental sentence and a phrase complement (NP or PP) The third type of alternations we found can be considered to be somewhere between the two presented so far: in all the cases discussed below in fact, verbs allow for alternations in which one realisation involves either a noun phrase (NP) or a prepositional phrase (PP), replaced in the other case by an argumental sentence. Some of them entail a change in the syntactic valency of the verbs, while some others don’t. Moreover, we classified them depending on the syntactic position affected by the alternation, namely the complement position or the subject one; however, we have decided to keep separate those alternations that involve predicative complements, since, as we will see more in detail, they can be considered a unique class on their own. 4.3.1. Alternations affecting a complement position All the alternations examined in this section concern the complement position, which can be a direct complement (NP_obj) or an indirect complement (PP). We will first discuss those alternations that do not involve a change in the syntactic valency of the verb, followed by those that do so. 4.3.1.1. NP_subj#V#NP_obj#{a}PP/NP_subj#V#{di}inf#{a}PP This alternation is allowed by trivalent verbs and it involves the direct complement, namely the object: in the first realisation (79a), this slot is occupied by a noun filler (NP_obj), whereas in the second one (79b) there is an argument clause introduced by the complementiser di and with the verb in the infinitive form. On the other hand, the indirect complement introduced by the preposition a (“to”) and indicating a somehow affected participant remains unchanged in both options: 83 (79) a. Ordinò una marcia forzata ai soldati “He ordered a forced march to the soldiers” b. Ordinò agli studenti di preparare il testo entro una settimana “She ordered to prepare the text within a week to the students” The verbs in our sample that allowed for this alternation are: rimproverare, imporre, risparmiare, garantire, ricordare, giurare, promettere, gridare, proporre, ordinare, confidare, proibire, annunciare, raccontare, chiedere, augurare, comandare, offrire, raccomandare, sussurrare, dire, predicare, vietare, dichiarare, permettere, suggerire, denunciare, scrivere, assicurare, confessare, concedere, comunicare, consigliare, riferire, impedire49. Following Levin’s verb classes (Levin 1993), these verbs could be divided in two broad classes: a. VERBS OF COMMUNICATION: rimproverare, ricordare, giurare, promettere, gridare, proporre, confidare, annunciare, raccontare, chiedere, augurare, raccomandare, sussurrare, dire, predicare, dichiarare, suggerire, denunciare, scrivere, assicurare, confessare, comunicare, consigliare, riferire; b. VERBS OF SOCIAL INTERACTION: imporre, risparmiare, garantire, ordinare, proibire, comandare, offrire, vietare, permettere, concedere, impedire. The most interesting thing about this alternation is that it overlaps with the one discussed in 4.1.1.1.: the majority of the verbs that allow for it, in fact, display a threefold way of expressing the direct complement, which can be realised as a noun, as an infinitive argumental sentence introduced by di and as an argumental sentence introduced by che and with the verb in a finite form. Taking the same verb we used in the examples in 4.1.1.1., the three following possibilities are therefore given: (80) a. Augurai buone vacanze al mio amico “I wished happy holiday to my friend” 49 To scold, to impose, to spare, to guarantee, to remind, to swear, to promise, to scream, to propose, to order, to confide, to forbid, to announce, to tell, to ask, to wish, to command, to offer, to recommend, to whisper, to say, to preach, to prohibit, to declare, to allow, to suggest, to denounce, to write, to assure, to confess, to grant, to communicate, to advise, to report, to impede. 84 b. Ti auguro di fare un buon viaggio “I wish (to) you to have a pleasant journey” c. Ti Auguro che tutto vada bene “I wish (to) you that everything go well” As a confirmation of the validity of our findings, we also found the other possible alternation, namely the one between (80a) and (80c), which we will discuss in the following paragraph. 4.3.1.2. NP_subj#V#NP_obj#{a}PP/NP_subj#V#{che}fin#{a}PP This alternation complements the one examined in the previous paragraph in that it presents, on the one hand, a realisation with a noun filler (81a) and, on the other, a realisation with an argumental sentence, introduced in this case by the complementiser che (“that”) and with the verb at a finite form (81b). Furthermore, even in this alternation, the indirect complement does not present any variation: (81) a. Promisero aiuto a un compagno “They promised help to a friend” b. Ti prometto che arriverò in orario “I promise (to) you that I will arrive on time” The verbs that, within our annotated sample, allowed this alternation are: augurare, mostrare, garantire, ricordare, assicurare, promettere, gridare, confidare, ripetere, annunciare, raccontare, chiedere, comandare, raccomandare, dire, dichiarare, segnalare, spiegare, permettere, denunciare, proporre, scrivere, giurare, confessare, concedere, insegnare, comunicare, riferire50. As with the previous alternation, the following sub-­‐
classification is possible: 50 To wish, to show, to guarantee, to remind, to assure, to promise, to scream, to confide, to repeat, to announce, to tell, to ask, to command, to recommend, to say, to declare, to signal, to explain, to allow, to denounce, to propose, to write, to swear, to confess, to grant, to teach, to communicate, to report. 85 a. VERBS OF COMMUNICATION: augurare, ricordare, assicurare, promettere, gridare, confidare, ripetere, annunciare, raccontare, chiedere, raccomandare, dire, dichiarare, segnalare, spiegare, denunciare, proporre, scrivere, giurare, confessare, comunicare, riferire; b. VERBS OF SOCIAL INTERACTIONS: mostrare, garantire, comandare, permettere, concedere, insegnare. As we briefly mentioned above, the alternation presented here and the one presented in 4.3.1.1. are indeed interchangeable for the majority of the verbs listed in these two sections, but not for all of them. In fact, there are five verbs (mostrare, ripetere, segnalare, spiegare, insegnare) that appear in this alternation, but that do not seem to allow the one presented in 4.3.1.1.. At the same time, there are twelve verbs that present the reverse situation (rimproverare, imporre, risparmiare, ordinare, proibire, offrire, sussurrare, predicare, vietare, suggerire, consigliare, impedire). In order to fully understand such fine-­‐graded distinctions, a larger sample of verbs is certainly necessary and it was thought best to not hazard hasty interpretations that could easily be proven wrong. Also, a cross-­‐linguistic investigation could cast some more light on the alternation we have just illustrated, providing data from various languages and, therefore, allowing for more plausible claims. 4.3.1.3. NP_subj#V#NP_obj#{a}PP / NP_subj#V#NP_obj#{a}inf The alternation presented here involves the indirect complement, which, even though it is always introduced by the preposition a (“to”) can be expressed as a noun (PP) or as an argumental sentence with the verb in the infinitive form, as illustrated in (82): (82) a. Educammo i giovani al rispetto degli altri “We educated young people to the respect for others” b. Educate vostro figlio a essere tollerante! “(You) educate your son to be tolerant!” The verbs in our sample that allow this alternation are: disporre, autorizzare, invitare, stimolare, abituare, indurre, motivare, incoraggiare, costringere, condannare, esercitare, 86 educare, ammettere, trattenere, spedire, sollecitare, delegare, forzare, ridurre, obbligare, mettere, destinare51. The following semantic classes can be identified: a. PSYCH-­‐VERBS: stimolare, abituare, indurre, motivare, incoraggiare, esercitare, sollecitare, ridurre, obbligare; b. VERBS OF SOCIAL INTERACTION: disporre, autorizzare, invitare, costringere, condannare, educare, ammettere, trattenere, spedire, delegare, forzare, mettere, destinare. One thing to notice about these verbs is that, in spite of their different meanings, they all entail a second animate participant expressed as the direct object that is affected by the action described the verb, whose meaning is further specified by the indirect object. 4.3.1.4. NP_subj#V#NP_obj#{che}fin / NP_subj#V#NP_obj#{di}PP As for the previous alternation, even in this case the complement involved is the indirect object, which can be realised either as a noun introduced by the preposition di (83a), or as an argumental sentence introduced by the complementiser che (“that”) and with the verb in a finite form (83b): (83) a. Avviserà i parenti della partenza “She will warn the relatives about the departure” b. Ho avvisato tutti che la lezione è sospesa “I have warned everybody that the lesson is cancelled” There are only two verbs in our sample that allow for this alternation: avvisare and convincere
52
. Both verbs can be ascribed to the broad class of VERBS OF COMMUNICATION, even though, contrary to the majority of the verbs in this class, they both entail a communication aimed at obtaining a reaction from the interlocutor, an aspect that makes these two verbs less prototypical than the other verbs of communication. 51 To prepare, to authorise, to invite, to stimulate, to accustom, to induce, to motivate, encourage, to compel, to condemn, to accustom, to educate, to admit, to hold, to send, to urge, to delegate, to force, to reduce, to obligate, to place, to destine. 52 To warn, to convince. 87 4.3.1.5. NP_subj#V#{a}PP / NP_subj#V#{a}inf This alternation is allowed by intransitive verbs and it involves the indirect object, which is always introduced by the preposition a (“to”), but can be realized as a noun (84a) or as an argumental sentence (84b): (84) a. Rinunciai a un progetto “I gave up on a project” b. Rinunciò a partire “She gave up on going away” Within our annotated sample, the verbs that allow for this alternation are: contribuire, arrivare, rinunciare, scappare, pervenire, provvedere, giocare, venire, badare, pensare, tenere, ritornare, mirare, aspirare, concorrere, tendere53. These verbs can be divided in the following broad semantic classes: a. VERBS OF MOTION: arrivare, scappare, pervenire, giocare, venire, ritornare; b. VERBS OF PERSONAL INVOLVEMENT: contribuire, rinunciare, provvedere, badare, pensare, tenere, mirare, aspirare, concorrere, tendere. If the first class is taken from Levin (1993), the second one is not: given the underlying meaning of all the verbs in b., it was thought best to put them into a new class that would encompass the common thread, namely the emotional and personal involvement of the subject in the achievement of the action described by the verb. 4.3.1.6. NP_subj#V#NP_obj#{a}PP / NP_subj#V-­‐si#{a}inf This alternation is one of the two alternations in this section that brings about a change in the syntactic valency of the verbs that allow for it: in one of the possible realisations, in fact, they have a direct object and an indirect one realised as a noun (85a), while in the other one they take the pronominal form and thus display only the indirect 53 To contribute, to arrive, to give up on, to run off, to reach, to provide, to play, to come, to be careful, to think, to value, to come back, to aim at, to strive, to combine to, to tend. 88 argument (85b), which, however, is realised as an argumental sentence. In spite of the very different syntactic structures involved, the indirect argument is introduced by the preposition a (“to”) in both possibilities: (85) a. Obbligammo il nemico alla resa “We forced the enemy to defeat” b. Mi sono obbligato a finire il libro entro l’anno “I have forced-­‐pron to finish the book within the year” The verbs in our sample that allow this alternation are: disporre, convincere, costringere, obbligare, esercitare, indurre, abituare, ridurre, trattenere54, and they all belong to the VERBS OF SOCIAL INTERACTION class. Moreover, all of these verbs are direct reflexive verbs, which means that, in the pronominal form, the pronoun acts as a direct object that co-­‐refers to the subject, as the paraphrase of (85b) in (86) illustrates: (86) Ho obbligato me stesso a finire il libro entro l’anno “I have forced myself to finish the book within the year” 4.3.1.7. NP_subj#V#NP_obj#{per}PP / NP_subj#V-­‐si#{per}inf This alternation is similar to the one examined in 4.3.1.6., in more than one respect: not only does it involve a change in the syntactic valence of verbs, but it also allows for the indirect object to be realised as a noun (87a) or as an argumental sentence (87b), introduced in both cases by the preposition per (“for, to”): (87) a. Sto preparando i bambini per la scuola “I’m preparing the kids for school” b. Ci stiamo preparando per andare a cena “We are preparing-­‐pron to go for dinner” 54 To prepare, to convince, to force, to obligate, to accustom, to induce, to accustom, to reduce, to hold. 89 The verbs in our sample that allow for this alternation are: preparare and sacrificare55. Both verbs refer to actions that implicitly tend towards a final purpose, which is realised as an indirect object introduced precisely by the preposition that, in Italian, has a “destinative-­‐final value” (Serianni 1989, p. 350), namely per. Moreover, we are once again presented with direct reflexive verbs since the pronoun in (87b) acts as a semantic direct object co-­‐referring to the subject, as its possible paraphrase in (88) demonstrates: (88) Stiamo preparando noi stessi per andare a cena “We are preparing ourselves to go for dinner” 4.3.1.8. NP_subj#V-­‐si#{per}PP/ NP_subj#V-­‐si#{per}inf This alternation is very much like the one we discussed in 4.3.1.7: they both involve the indirect object, which is introduced by the preposition per (“for, to”) and alternatively realised as a noun (89a) or as an argumental sentence (89b). However, contrary to the previous one, this alternation presents the verb in the pronominal form in both possibilities: (89) a. Ci organizzammo per un’escursione “We organised-­‐pron for a field trip” b. Ci organizzammo per uscire presto il mattino “We organised-­‐pron to leave early in the morning” Within our annotated sample, the following verbs allowed for this alternation: organizzare, preparare, sacrificare56. As expected, they overlap with those found in the previous alternation, with the exception of organizzare57, which, according to S&C, does not allow for the frame illustrated in (87a), therefore not taking part in the alternation in 4.3.1.7. Lastly, these verbs are once again direct reflexives. 55 To prepare, to sacrifice. 56 To organise, to prepare, to sacrifice. 57 The verb to organise is by far the least prototypical of all, allowing for an interpretation which is to be placed somewhere in between the properly direct reflexive one and the properly intensive reflexive one. 90 4.3.1.9. NP_subj#V-­‐si#{di}PP / NP_subj#V-­‐si#{di}inf In this alternation, verbs take the pronominal form in both realisations and the argument affected is the indirect one, which can be expressed as a noun in (90a), or as an argumental sentence in (90b), but introduced by the preposition di (“of”) in both: (90) a. Mi sono dimenticato del tuo compleanno “I have forgotten-­‐pron (of) your birthday” b. Mi dimenticavo di farti gli auguri “I forgot-­‐pron to wish you happy birthday” The verbs in our annotated sample that allowed for this alternation are: curare, convincere, accusare, assicurare, pentirsi, sorprendere, occupare, stupire, ricordare, accontentare, dimenticare, incaricare, accorgersi, vantare58. On the basis of Levin (1993), the possible semantic classes can be identified: a. PSYCH-­‐VERBS: convincere, accusare, pentirsi, sorprendere, stupire, ricordare, accontentare, dimenticare, accorgersi, vantare; b. VERBS OF SOCIAL INTERACTION: curare, assicurare, occupare, incaricare. However, a second classification is possible if we select as a valid parameter the semantic value of the pronoun si, which, in fact, varies greatly: I. DIRECT REFLEXIVE VERBS: convincere, accusare, incaricare. In these cases, the pronoun si does act as a semantic direct object, co-­‐referring to the subject, as the paraphrase of (91) as (92) shows: (91) Mi accusai del delitto “I accused-­‐pron of the crime” (92) Accusai me stesso del delitto “I accused myself of the crime” 58 To take care of, to convince, to accuse, to make sure, to repent, to surprise, to look after, to astonish, to remember, to please, to forget, to appoint, to notice, to boast. 91 II. INTENSIVE PRONOMINAL VERBS: curare, assicurare, pentirsi, sorprendere, occupare, stupire, ricordare, accontentare, dimenticare, accorgersi, vantare. In relation to these verbs, the pronoun si does not function as a semantic argument, but it simply indicates a greater involvement of the subject in the action described by the verbs, as the agrammaticality of the paraphrase of (93) as (94) demonstrates: (93) Non mi stupisco più delle sue affermazioni “I am not surprised-­‐pron anymore by his statements” (94) *Non stupisco più me stesso delle sue affermazioni *“I do not surprise myself anymore of his statements” One last thing worth mentioning in relation to this class is that not all of these verbs show the same pattern in the corresponding standard form, namely the one with a direct object and the verb in a non-­‐pronominal form. First of all, there are two verbs that exist only in the pronominal form and do not have a non-­‐pronominal realisation (see Serianni 1989, XI.24): pentirsi and accorgersi. Moreover, while some verbs still imply a degree of control exerted by the subject (curare, assicurare, occupare, vantare), some others do not, actually conveying a sense of utter lack of control over the action on behalf of the subject (sorprendere, stupire, ricordare, dimenticare). Lastly, the verb accontentare is somehow different from all the others: in the non-­‐pronominal realisation it describes an action totally controlled by the subject and which has direct consequences only on the entity represented by the object, without the possibility of a further specification (95a); on the other hand, in the pronominal variant it describes an action that, in a certain way, the subject undergoes, being left with no other choice than that expressed by the PP (95b): (95) a. I nonni accontentano spesso i nipoti “Grandparents often please grandchildren” b. Mi accontenterei del terzo posto “I would please-­‐reflex of a third place” 92 4.3.1.10. NP_subj#V-­‐si#{che}fin / NP_subj#V-­‐si#{di}PP This alternation partly overlaps the one in 4.3.1.9.: not only do they both have verbs in the pronominal form, but they also involve the indirect object, which, in this case, can be realised as a noun introduced by the preposition di (“of”) (96a), or as an argumental sentence introduced by the complementiser che (“that”) and with the verb in a finite form (96b): (96) a. Mi assicurai della verità delle affermazioni “I verified-­‐pron the truth of the statements” b. Mi assicurai che tutto sia in ordine “I verified-­‐pron that everything is in order” As one would expect, the verbs in our sample that allow for this alternation partially coincide with those found in 4.3.1.9.: convincere, stupire, assicurare, ricordare, sorprendere, accorgersi, and they are all PSYCH-­‐VERBS. Moreover, also in this case it is possible to put forward a second different sub-­‐
classification based on the semantic value of the pronoun si: I. DIRECT REFLEXIVE VERBS: convincere. II. INTENSIVE PRONOMINAL VERBS: stupire, assicurare, ricordare, sorprendere, accorgersi. 4.3.1.11. NP_subj#V-­‐si#{a}PP / NP_subj#V-­‐si#{a}inf This alternation is to be paired with the one in 4.3.1.3., with the only difference being that, in this case, both realisations require the verb to be in the pronominal form. In both alternations, however, the involved argument is the indirect object, which is introduced by the preposition a (“to”) and expressed as a noun (97a), or as an argumental sentence (97b): (97) a. Non mi presterei mai a queste cose “I would never lend-­‐pron to these things” 93 b. Ti presteresti a sostenere le spese? “Would you lend-­‐pron to support the expenses?” The verbs in our sample that allow for this alternation are: preparare, rilanciare, costringere, determinare, obbligare, prestare, rassegnare, abbassare, rimettere, disporre, adattare, abituare, ridurre, indurre59 and they can all be classified as VERBS OF SOCIAL INTERACTION. Moreover, with the exception of rassegnare, which, in the standard form has a completely different meaning (“to resign”), all the verbs listed above are direct reflexive verbs, which means that, once again, the pronoun si acts as the semantic direct object that co-­‐refers to the subject, as the possibility of paraphrasing (97b) as (98) confirms: (98) Presteresti te stesso a sostenere le spese? “Would you lend yourself to support the expenses?” 4.3.2. Alternations taking place in the subject position In all the cases we present in this section, verbs allow their subject to be realised either as an argumental clause or as a noun; moreover, all the verbs that allow the syntactic structures examined below are unaccusative intransitive verbs and, in the realisation with the argumental sentence, they act as impersonal verbs, thus taking only a third person singular inflection. 4.3.2.1. NP_subj#V#{da}PP / NP_subj{che}fin#V#{da}PP This alternation takes place in the subject position, which can be realized as a noun (99a) or as an argumental sentence introduced by the complementiser che (“that”) and with the verb in a finite form (99b): (99) a. Dalle tue parole non consegue niente di buono “From your words nothing good follows” 59 To prepare, to restart, to force, to induce, to force, to lend, to put up with, to lower, to start again, to get ready, to adjust, to accustom, to reduce, to persuade. 94 b. Da ciò consegue che non è possibile partire “From this follows-­‐impers that leaving is not possible” There were only two verbs in our sample that allowed for this alternation: conseguire and risultare60. As for the verbs in 4.1.2.1., they can be classified as IMPERSONAL VERBS REQUIRING NOUN CLAUSES. 4.3.2.2. NP_subj#V#{a}PP / NP_subj{che}fin#V#{a}PP This alternation parallels the one we discussed in 4.1.2.3., with the only difference being that, in this case, the subject can be realised as a noun (100a) or as an argumental sentence introduced by the complementiser che (“that”) and with the verb in a finite form (100b): (100) a. A nessuno importano queste sciocchezze! “To nobody matter these silly things!” b. Ai ragazzi importa che tu sia qui con noi “To the guys matters-­‐impers that you are here with us” The verbs in our annotated sample that allowed for this alternation are: risultare, capitare, importare, dispiacere, sfuggire, convenire61. As for the verbs in 4.1.2.3., we propose the novel class HAPPEN VERBS. However, the two alternations need to be kept apart, as demonstrated by the fact that not all the verbs that allow for one, also allow for the other one: capitare, importare and sfuggire, in fact, do not allow for the realisation with the zero infinitive clause: (101) *A tutti capita perdere la pazienza *“To everybody happens-­‐impers lose their patience” 60 To result, to follow. 61 To result, to happen, to matter, to be sorry, to elude, to be convenient. 95 4.3.2.3. NP_subj#V#{a}PP / NP_subj{0}inf#V#{a}PP This alternation presents the reverse situation of the one discussed in the previous paragraph, allowing the subject to be realised as either a noun (102a) or as a zero infinitive clause (102b): (102) a. Mi preme solo la tua felicità “To me matters-­‐impers only your happiness” b. “A noi tutti preme essere presenti” “To all of us matters-­‐impers be-­‐inf present” Within our sample, the following verbs allowed for this alternation: stare, spettare, risultare, capitare, dispiacere, premere, convenire62. Once again, we propose to establish an original semantic class, which we call 0-­‐INFINITIVE HAPPEN VERBS. 4.3.3. Alternations requiring predicative complements This final section is dedicated to those alternations we found that involve a predicative complement: in all the cases discussed below, the alternation concerns the direct object, which can be realised as a noun or as an argumental sentence. As we will see more in depth in each specific case, some alternations are also responsible for a change in the syntactic valency of the verbs that allow for it, while some others are not. However, regardless of their idiosyncratic peculiarities, they all require a predicative complement. Moreover, the verbs in our annotated sample that allowed for the following alternations are always the same: credere, considerare and giudicare63. These verbs belong to the class traditionally known as “copulative verbs”, which encompasses all those verbs that express a state of being, rather than an action, and that, in order to do so, require a predicative complement. More in depth, the three verbs in question belong to the subclass of “estimative verbs” (Serianni 1989, XI.6-­‐d). 62 To be up to, to concern, to result, to happen, to be sorry, to matter, to be convenient. 63 To believe, to consider, to judge. 96 4.3.3.1. NP_subj#V#NP_obj#Cpred / NP_subj#V#{0}inf#Cpred The first alternation takes place in the direct object position, which can be syntactically expressed as a noun (103a) or as an argumental clause introduced by the zero complementiser (103b): (103) a. Vi credevo sinceri “I believed you sincere” b. Ho creduto opportuno agire in questo modo “I believed right act-­‐inf in this way” 4.3.3.2. NP_subj#V#NP_obj#Cpred / NP_subj#V#{che}fin#Cpred This alternation once again concerns the direct object, which, as seen in 4.3.3.1., can be realised as a noun (104a), but, contrary to the previous alternation, also as an argumental sentence introduced by the complementiser che (“that”) and with the verb in a finite form (104b): (104) a. Lo considero il miglior giocatore del mondo “I consider him the best player in the world” b. Considero sconveniente che tu rimanga qui “I consider inappropriate that you remain here” 4.3.3.3. NP_subj#V#{0}inf#Cpred / NP_subj#V-­‐si#Cpred The third alternation still involves the direct object, but, contrary to the ones discussed in the previous paragraphs, it also brings about a change in the syntactic valence of the verbs that allow for it. In fact, in the first realisation allowed, the direct object is expressed as a zero infinitive clause (105a), while, in the second one, the verb takes the prenominal form, acting as the semantic object (105b): (105) a. Ho giudicato opportuno agire cosi “I judged adequate act-­‐inf like this” 97 b. Si giudica furbo “He judges-­‐pron smart” Given the semantic value of the pronoun si in the pronominal realisation, the three verbs that allow this alternation are direct reflexives, and, as in all the cases illustrated so far, the following paraphrase of (105b) is possible: (106) Giudica se stesso furbo “He judges himself smart” 4.3.3.4. NP_subj#V#{che}fin#Cpred / NP_subj#V-­‐si#Cpred As the alternation presented din the previous paragraph, also the one we discuss here allows for a pronominal realisation in which the pronoun si acts as a semantic direct object (107b). however, contrary to the alternation in 4.3.3.3. and very much like the one in 4.3.3.2., it also allows for a realisation in which the direct object is expressed as an argumental clause introduced by the complementiser che (“that”) and with the verb in a finite form (107a): (107) a. Credo giusto che tu vada in persona “I believe right that you go in person” b. Mario si crede il più bravo di tutti “Mario believes himself the best” One last thing to note is that, even in this case, they are direct reflexive verbs, as the paraphrase of (107b) shows: (108) Mario crede se stesso il più bravo di tutti “Mario believes himself the best” 98 Conclusions This thesis investigated argument alternations in Italian, with the aim of providing new insights on what is unanimously considered a complex linguistic phenomenon, to the extent that its theoretical status is still today hotly discussed by all the different frameworks involved in the ongoing debate. After a manual annotation of a sample consisting of the 1000 most frequent Italian verbs derived from a lexicographic resource, and a semi-­‐automatic extraction of the argument alternations they allow, we were finally able to single out those of them that could actually be construed as proper instantiations of this widespread aspect of natural languages. Our findings provided us with a total of 37 valid argument alternations for Italian (Fig.6, Fig. 7, Fig. 8), some of which could even be further sub-­‐divided into finer-­‐graded clusters. The broader classification we proposed is based on the syntactic properties of the arguments involved in the alternations we found, and it is made up of three different classes, divided as follows (Fig.5): 1) Argumental sentence alternations: these alternation involve a set of arguments that are both expressed by so-­‐called “argumental sentences”, namely sentences that act as arguments saturating the syntactic and semantic valence of verbs; 2) Alternations involving a noun or prepositional phrase (NP or PP): in these alternations, verbs often undergo a change in their syntactic valency and the arguments affected are either noun phrases (NN) or prepositional phrases (PP); 3) Alternations involving an argumental sentence and a phrase complement (NP or PP): in these alternations, verbs allow two realisations of one of their arguments, which, in one case, is expressed as a noun phrase (NP) or a prepositional phrase (PP), replaced in the other one by an argumental sentence. As we have already argued, these results are only partial and could not be further from being considered definitive: the first thing to do would be to enlarge the sample of verbs examined, so that more subcategorisation frames could be identified and, consequently, more potential argument alternations would emerge and, moreover, new senses for the alternations already singled out would be available. 99 The analysis reported in this thesis moved from data extracted from a lexical resource, namely a manually built dictionary. However, it would be important to acquire information about subcategorisation frames and their frequency, as well as about argument alternations, also from large corpora: in this way, we would be able to integrate the precise, but somehow limited information derived from a lexicographic resource, born out of the intuitions of the lexicographer, with that obtained from more objective but also noisier broad-­‐coverage corpora. Hence, an interesting extension of this work would be the application of our method to data extracted from different kinds of collections of “real world” texts, such as spontaneous sentence production corpora, technical language corpora, child speech corpora and, possibly, second language learners production corpora. Such analysis could cast more light on this phenomenon, providing at once a larger quantity of verbs and a more variegated use of their senses in case of polysemy. The exploitation of psycholinguistic-­‐derived data, moreover, could reveal interesting patterns of development in the acquisition and, possibly, loss of argument alternations, which, in turn, could provide valuable insights on the nature and/or the organisation of the language faculty. Finally, one more aspect that future research will have to focus on is the semantic classification of Italian verbs based on the argument alternations they allow, in a manner similar to that followed by Levin (1993). Every time it was possible, we did attempt a classification of verbs, mainly exploiting the semantic classes available in Levin, but also putting forward new ones when strictly Italian phenomena did not find a suitable match in English. However, a complete semantic classification of verbs is no small task and many issues have to be taken into account, from the identification of the facets of meaning responsible for the syntactic realisations examined, to the inevitable theoretical challenges posited by finer-­‐graded divisions. We will leave to future investigations the challenge of developing a coherent and extensive semantic classification of Italian verbs based on their syntactic behaviour. Notwithstanding the inevitable limitations of our thesis, our work is intended as a first step that will encourage further research on the behaviour of Italian verbs at the syntactic-­‐semantic interface, which would ultimately benefit at once different areas of current linguistic research, such as theoretical linguistics, computational linguistics and psycholinguistics. 100 101 Figure 5: Argument alternations in Italian Figure 6: Argumental sentence alternations 102 103 Figure 7: Alternations involving a noun or a prepositional phrase (NP or PP) 104 Figure 8: Alternations involving an argumental sentence and a phrase complement (NP or PP) Acknowledgments This thesis is the outcome of three long years that, to a large extent, I wish had gone differently. However, there are people who deserve to be thanked and who are worth all my respect and gratitude. First of all, I want to thank my supervisor Prof. Alessandro Lenci for showing me that everything can be interesting and challenging – even Computational Linguistics –, as long as it is taught and approached with love, passion and curiosity. His sincere and thorough support, along with an impeccable work ethics, encouraged me to become a better linguist, accepting that I will never know enough, and that it is ok not to in order to become a good researcher. I am also thankful to my co-­‐supervisor Prof. Giovanna Marotta, who agreed to take part in this work at a short notice, nonetheless showing great enthusiasm and genuine interest for the subject of this thesis. I would like to thank my external examiner Doc. Gianluca Lebani, whose generous help throughout the writing of this thesis was constant, meticulous and simply invaluable. Also, during my time at the Laboratory of Computational Linguistics, he patiently supervised my unsuccessful attempts to develop any sort of informatical skills, kindly waiting until I eventually gave up altogether (which happened regularly, much to his amusement). Moreover, a huge thank goes to Alice, Francesca and Irene, three young women who, each in a different way, will be forever associated with this journey: they are responsible for filling these years with endless coffee breaks and random gatherings during which doubts, laughter and hopes were evenly shared. You made bearable what otherwise would have not been so. This thesis would have never existed if it had not been for my unconventionally extraordinary family. I owe my parents a debt I could never, ever repay: not only did they allow me to go to University for a second time, but they never stopped having faith in me, not even when I gave them good reasons to. They offered help and support, but never once imposed any decision of theirs on me, and they are still allowing me, day by day, to turn into the person I choose to be. Also, I was blessed with the most extravagant and generous sister and the smartest most loving brother: they managed to put up with me for 26 and 14 years respectively, at my best and my worst, getting on my nerves 105 every now and then, but still rooting for me every step of the way, no matter what, the way only siblings can do. Last, but definitely not least, I wish to thank Emmet, whose endless patience and unconditional love literally made it possible for me to get here today: he kept things in perspective, always offering a positive view -­‐ even when I did not think it possible -­‐ and a safe shelter every time a new storm hit. His kind Irish nature and his genuine and devoted passion for everything he does have been inspiring me to not just become a better linguist, but a better human being and, for that, I could never thank him enough. Go raibh mile maith agaibh. 106 Bibliography Agrell, S. (1908), Aspektänderung und Aktionsartbildung beim polnischen Zeitworte: Ein Beitrag zum Studium derindogermanischen Präverbia und ihrer Bedeutungsfunktionen, Lunds Universitets Arsskrift, new series, I, iv.2. Arsenijević, B., Gehrke, B., Marín, R. (2013), “The (De)composition of Event Predicates”, in Studies in the Composition and Decomposition of Event Predicates, B. Arsenijević, B. Gehrke and R. Marín, R. (eds.), Springer, Dordrecht, pp. 1-­‐26. Baker, C F., Fillmore, C. J. & Lowe, J. B. (1998), “The Berkerley FrameNet Project, in Proceedings of the 17th International Conference on Computational Linguistics, Montreal, pp. 86-­‐90. Baroni, M., Bernardini, S., Comastri, F., Piccioni, L., Volpi, A., Aston G. and M. Mazzoleni (2004), “Introducing the la Repubblica corpus: A large, annotated, TEI(XML)-­‐
compliant corpus of newspaper Italian”, in Proceedings of the Fourth International Conference on Language Resources and Evaluation (LREC’04), Lisbon, 24-­‐30 May 2004, pp. 1771-­‐1774. Beavers, J. T. (2006), “Argument/Oblique Alternations and the Structure of Lexical Meaning”, Doctoral Thesis, Stanford University. Berruto, G. (1976), La Semantica, Zanichelli, Bologna. Bresnan, J. (ed.) (1982), The Mental Representations of Grammatical Relations, The MIT Press, Cambridge Mass. Cann, R., Kempson, R. & Gregoromichelaki, E. (2009), Semantics. An Introduction to Meaning in Language, Cambridge University Press, Cambridge. Carrier, J. & Randall, J. H. (1992), “”The Argument Structure and Syntactic Structure of Resultatives”, Linguistic Theories 23, pp. 173-­‐234. Cennamo, M. (forthcoming), “Valency Patterns in Italian”. Chomsky, N. (1957), Syntactic Structures, Mouton, The Hague. Chomsky, N. (1965), Aspects of the Theory of Syntax, The MIT Press, Cambridge MA. Chomsky, N. (1981), Lectures on Government and Binding, Foris, Dordrecht. 107 Chomsky, N. (1986), Knowledge of Language: Its Nature, Origin and Use, Praeger, New York. Choi, S. & Bowerman, M. (1991), “Learning to express motion events in English and Korean: The influence of language-­‐specific lexicalization patterns”, Cognition, 41 (1-­‐3), pp. 83–121. Cole, P. (1987), “Null Objects in Universal Grammar”, Linguistic Inquiry 18, pp. 597–612. Comrie, B. (1976), Aspect, Cambridge University Press, Cambridge. Croch, D. & King, T. H. (2005), “Unifying Lexical Resources”, in Proceedings of Interdisciplinary Workshop on the Identification and Representation of Verb Features and Verb Classes, Saarbrücken, Germany. Dang, H. T. (2004), “Investigations into the Role of Lexical Semantics in Word Sense Disambiguation, Doctoral Thesis, CIS, University of Pennsylvania. Davidson, D. (1980), “The logical form of action and sentences”, in D. Davidson, Essays on Actions and Events, Clarendon Press, Oxford, pp. 105-­‐148. De Saussure, F. (1916 (1955), Cours de Linguistique Générale, Payot, Paris. DeLancey, S. (1995), “Verbal Case Frames in English and Tibetan”, unpublished ms. Department of Linguistics, University of Oregon, Eugene OR. Dik, S. C. (1978), Functional Grammar, North-­‐Holland, Amsterdam. Dik, S. C. (1989 (1997)), The Theory of Functional Grammar. The Structure of the Clause, ed. by Kees Hengeveld, Mouton de Gruyter, Berlin. Dik, S. C. (1997), The Theory of Functional Grammar: Complex and Derived Structures, ed. by Kees Hengeveld, Mouton de Gruyter, Berlin. Dowty, D. (1979), Word Meaning and Montague Grammar, Reidel, Dordrecht. Dowty, D. (1991), “Thematic Proto-­‐Roles and Argument Selection”, Language 67, pp. 547-­‐619. Dowty, D. (2000), “’The garden swarms with bees’ and the fallacy of ‘argument alternation’”, in Polysemy: theoretical and computational approaches, Y. Ravin and C. Leacock (eds.), Oxford University Press, Oxford, pp. 111-­‐128. 108 Dowty, D. (2001), “The semantic asymmetry of argument alternations (and why it matters)”, in Making Sense: From Lexeme to Discourse, in honor of Werner Abraham at the occasion of his retirement (Groninger Arbeiten zur germanistischen Linguistik Nr. 44), G. van der Meer and A. G. B. ter Meulen (eds.), Center for Language and Cognition Groningen, Groningen, pp. 171-­‐86. Fellbaum, C (ed.) (1998), “WordNet: An electronic lexical database”, Language, speech and communications, MIT Press, Cambridge MA. Fillmore, C. J. (1968), “The case for case”, in Universals in Linguistic Theory, E. Bach & R. T. Harms (eds.), Holt, Rinehart and Winston, New York, pp. 1-­‐88. Fillmore, C. J. (1970), “The Grammar of Hitting and Breaking”, in Readings in English Transformational Grammar, R. A. Jacobs and P. S. Rosenbaum (eds.), Ginn and Co., Waltham MA, Toronto, London, pp. 120-­‐133. Fillmore, C. J. (1971a), “Some Problems for Case Grammar”, in Report of the 22nd Annual Roundtable Meeting on Linguistics and Language Studies, R. J. O’Brien (ed.), Georgetown University Press, Washington, DC, pp. 35-­‐56. Fillmore, C. J. (1971b), “Types of Lexical Information”, in Semantics, D. Steinberg and L. Jakobovits (eds.), Cambridge University Press, Cambridge, pp. 370-­‐392. Fillmore, C. J. (1977), “The Case for Case Reopened”, in Grammatical Relations, P. Cole & M. Sadock (eds.), Academic, New York, San Francisco, London. Fillmore, C. J. (2008), “The Merging of ‘Frames’”, in Frames, Corpora, and Knowledge Representation, R. Rossini Favretti (ed.), Bononia University Press, Bologna, pp. 1-­‐
12. Fodor, J. A. (1983), Modularity of Mind: An Essay on Faculty Psychology, MIT Press, Cambridge Mass. Geeraerts, D. (2010), Theories of Lexical Semantics, Oxford University Press, Oxford. Giovanardi, C. (1986), “Subordinazione: di + infinito”, in Parallela 2. Aspetti della sintassi dell’italiano contemporaneo, K. Lichem et al. (eds.), Atti del III Convegno italo-­‐
austriaco SLI, Graz, 28-­‐31 Maggio 1984, Tubingen, Narr, (1986), pp. 105-­‐ 118. Goldberg, A. E. & Jackendoff, R. (2004), “The English Resultative as a Family of Constructions”, Language 90, pp. 532-­‐568. 109 Graffi, G. (1994), Sintassi, Il Mulino, Bologna. Green, G. (1973), “A Syntactic Sincretism in English and French”, in Issues in Linguistics, B. Kachru, R. B. Lees, Y. Malkiel, A. Pietrangeli and S. Saporta (eds.), University of Illinois Press, Urbana, IL, pp. 257-­‐278. Greimas, A. J. (1966), Sémantique structurale: recherche de méthode, Libraire Larousse, Paris. Grimshaw, J. (1979), “Complement Selection and the Lexicon”, in Linguistic Enquiry Volume, 10-­‐2, pp. 279-­‐326. Grimshaw, J. (1990), Argument Structure, The IMT Press, Cambridge Mass. Gropen, J., Pinker, S., Hollander, M. & Goldberg, R. (1989), “The Learnability and Acquisition of the Dative Alternation in English”, Language 65, pp. 203-­‐257. Gruber, J. (1965), Studies in Lexical Relations, Doctoral Dissertation, MIT, Cambridge MA. Gruber, J. (1976), Lexical Structures in Syntax and Semantics, North Holland, Amsterdam. Guerssel, M., Hale, K., Laughren, M., Levin, B. and White Eagle, J. (1985), “A Cross-­‐
linguistic study of Transitivity Alternations”, in Papers from the Parasession on Causatives and Agentivity, Chicago Linguistic Society, Chicago IL, pp. 48-­‐63. Hale, K. L. & Keyser, S. J. (1987), “A View from the Middle”, Lexicon Project Working Papers 10, Center for Cognitive Science, MIT, Cambridge MA. Hall, B. (1965), “Subject and Object in English”, Doctoral Dissertation, MIT, Cambridge MA. Haspelmath, M. (1993), "More on the typology of inchoative/causative verb alternations", in Causatives and transitivity. (Studies in Language Companion Series, 23.), B. Comrie & M. Polinsky (eds.), Benjamins, Amsterdam, pp. 87-­‐120. Hensman, S. & Dunnion, J. (2004), “Automatically Building Conceptual Graphs Using VerbNet and WordNet”, in Proceedings of the 3rd International Symposium on Information and Communication Technologies, Las Vegas, NV, pp. 115-­‐120. Hjelmslev, L. (1957), “Pour une sémantique structurale”, in L. Hjelmslev, Essais linguistiques (1959), Les Editions des Minuit, Paris. Hjelmslev, L. (1961), Prolegomena to a Theory of Language (1943), The University of Wisconsin Press, Madison. 110 Hovy, E., Marcus, M., Palmer, M., Ramshaw, L. & Weischedel, R. (2006), “OntoNotes: The 90% solution”, in Proceedings of HLT-­‐NAACL06, pp. 57–60. Huang, Y. (1995), “On Null Subjects and Null Objects in Generative Grammar”, Linguistics 33, pp. 1081–1123. Jackendoff, R. S. (1972), Semantic Interpretation in Generative Grammar, MIT Press, Cambridge Mass. Jackendoff, R. S. (1976), “Toward an Explanatory Semantic Representation”, Linguistic Inquiry 7, pp. 89-­‐150. Jackendoff, R. S. (1983), Semantics and Cognition, MIT Press, Cambridge MA. Ježek, E. (2003), Classi di Verbi tra Semantica e Sintassi, Edizioni ETS, Pisa. Ježek, E. (2005), Lessico, Il Mulino, Bologna. Keller, F. & Lapata, M. (1998), “Object Drop and Discourse Accessibility”, in Proceedings of the 17th West Coast Conference on Formal Linguistics, K. Shahin, S. Blake and E. Kim (eds.), Stanford: CSLI Publications, pp. 362-­‐374. Koening, J. & Davis, A. R. (2006), “The KEY to Lexical Semantic Representations”, in Journal of Linguistics 42, pp. 71-­‐108. Lakoff, G. (1977), “Linguistic Gestalts”, in Papers from the Thirteenth Regional Meeting of the Chicago Linguistic Society, Chicago Linguistic Society, Chicago, pp. 236-­‐287. Lenci, A. (2009), “Argument alternations in Italian verbs: A computational study”, in Linguaggio e cervello – Semantica / Language and the brain – Semantics, V. Bambini, I. Ricci, P. M. Bertinetto & Collaborators (eds.), Atti del XLII Convegno della Società di Linguistica Italiana (Pisa, Scuola Normale Superiore, 25-­‐27 settembre 2008), vol. 2, Bulzoni, Roma. Lenci, A., Lapesa, G. and Bonansiga, G. (2012), “LexIt: A Computational Resource on Italian Argument Alternations”, in Proceedings of the Eight International Conference on Language Resources and Evaluation (LREC’12), Istanbul, 23-­‐25 May 2012, pp. 3712-­‐3718. Lenci, A. (in press), “Carving Verb Classes from Corpora" in Word Classes, R. Simone and F. Masini (eds.), John Benjamins, Amsterdam-­‐Philadelphia. 111 Lepschy, A. L. & Lepschy, G. C. (1986), La Lingua Italiana. Storia, varieta’ dell’uso, grammatica, Bompiani, Milano. Levin, B. (1993), English Verb Classes and Alternations, The University of Chicago Press, Chicago. Levin, B. & Rappaport Hovav, M. (1998), “Building verb meaning”, in The Projection of Arguments: Lexical and Compositional factors, M. Butt & W. Geuder (eds.), CSLI Publications, Stanford, pp. 97-­‐134. Levin, B. & Rappaport Hovav, M. (2005), Argument Realization, Cambridge University Press, Cambridge. Malchukov, A. & Ogawa, A. (2011), “Towards a typology of impersonal constructions. A semantic map approach”, in Impersonal Constructions. A cross-­‐linguistic perspective, A. Malchekov & A. Siewierska (eds), John Benjamins, Amsterdam-­‐
Philadelphia. Miller, G. A. (1990), “WordNet: An On-­‐line Lexical Database”, International Journal of Lexicographic 3(4), pp. 235-­‐312. Matthews, P. H. (1997(2007)), The Concise Oxford Dictionary of Linguistics, Oxford University Press, Oxford. Moretti, G. B. & Orvieto, G. R. (1983), Grammatica Italiana, vol. III, Il Verbo: morfologia e note generali di sintassi, Benucci, Perugia. Moro, A. (2006), I confini di Babele. Il cervello e il mistero delle lingue impossibili, Longanesi, Milano. Moro, A. (2012), Parlo dunque sono, Adelphi Edizioni SPA, Milano. Olbertz, H., Hengeveld, K., Garcia, J. S. (eds.) (1998), The Structure of the Lexicon in Functional Grammar, John Benjamins, Amsterdam-­‐Philadelphia. Palmer, M., Gildea D. & Kingsbury, P. (2005), “The Proposition Bank: A Corpus Annotated with Semantic Roles”, Computational Linguistics 31(1), pp. 71-­‐106. Piñón, C. (2001), “A finer look at the causative-­‐inchoative alternation”, in Proceedings of Semantics and Linguistic Theory 11, R. Hastings, B. Jackson and Z. Zvolensky (eds.), CLC Publications, Cornell University, Ithaca, New York. 112 Porzig, W. (1934), “Wesenhafte Bedeutungsbeziehungen”, in Beitrage zur Geschichte der deutschen Sprache und Literatur, 58, pp. 70-­‐97. Plato (1921), Plato in Twelve Volumes, Vol. 12, translated by Harold N. Fowler, Harvard University Press, Cambridge Mass.; William Heinemann Ltd, London. Pustejovsky, J. (1995), The Generative Lexicon, The MIT Press, Cambridge Mass. Rappaport Hovav, M. & Levin, B. (2002), “Change of State Verbs: Implications for Theories of Argument Projection”, BLS 28, Berkeley Linguistics Society, Berkeley, pp. 269-­‐280. Ravin, Y. (1990), Lexical Semantics without Thematic Roles, Clarendon Press, Oxford. Ramchand, G. & Reiss, C. (eds.) (2007), The Oxford Handbook of Linguistic Interfaces, Oxford University Press, Oxford. Robins, R. H. (1967), A Short History of Linguistics, Longman, London. Roland, D. & Jurafsky, D. (2002), “Verb sense and verb subcategorization probabilities”, in The Lexical Basis of Sentence Processing: Formal, Computational, and Experimental Issues, S. Stevenson and P. Merlo (eds.), John Benjamins, Amsterdam, pp. 325-­‐346. Salvi, G., Renzi, L. & Cardinaletti, A. (1988) (eds.), Grande Grammatica Italiana di consultazione, Vol. I: La Frase. I Sintagmi Nominale e Preposizionale, Il Mulino, Bologna. Salvi, G., Renzi, L. & Cardinaletti, A. (1991) (eds.), Grande Grammatica Italiana di consultazione, Vol. II: I Sintagmi Verbale, Aggettivale, Avverbiale. La Subordinazione, Il Mulino, Bologna. Salkoff, M. (1983), “Bees are swarming in the garden”, Language 59.2, pp. 288-­‐346. Schäfer, F. (2009), “The Causative Alternation”, Language and Linguistics Compass 3/2, pp. 641–681. Schulte im Walde, S. (2009), “The Induction of Verb Frames and Verb Classes from Corpora”, in Corpus Linguistics. An International Handbook, A. Lüdeling & M. Kytö (eds.), Mouton de Gruyter, Berlin, pp. 952-­‐972. 113 Serianni, L. (1989 (2006), Grammatica italiana. Italiano commune e lingua letteraria, UTET, Torino. Steiner, G. (1975), After Babel: Aspects of Language and Translation, Oxford University Press, Oxford. Stowell, T. (1981), Origins of Phrase Structures, Doctoral Disseratation, MIT, Cambridge MA. Shi, L. & Mihalcea, R. (2005), “Putting Pieces Together: Combining FrameNet, VerbNet and WordNet for Robust Semantic Parsing”, in Proceedings of the Sixth International Conference on Intelligent Text Processing and Computational Linguistics, Mexico City. Swier, R. & Stevenson, S. (2004), “Unsupervised Semantic Role Labelling”, in Proceedings of the 2004 Conference on Empirical Methods in Natural Language Processing, Barcelona, Spain, pp. 95-­‐102. Swift, M. (2005), “Toward Automatic Acquisition from VerbNet for Spoken Dialog Processing”, in Proceedings of Interdisciplinary Workshop on the Identification and Representation of Verb Features and Verb Classes, Saarbrücken, Germany. Talmy, L. (1985), “Lexicalization Patterns: Semantic Structure in Lexical Forms”, in Language Typology and Syntactic Description 3: Grammatical Categories and the Lexicon, T. Shopen (ed.), Cambridge University Press, Cambridge, pp. 57-­‐149. Talmy, L. (1991), “Path to Realization – Via Aspect and Result”, BLS 17, Berkeley Linguistics Society, Berkeley, pp. 480-­‐519. Tesnière, L. (1959), Eléments de Syntaxe Structurale, Libraire C. Klincksieck, Paris. van Valin, R. & Foley, W. (1984), “Role and Reference Grammar” in Syntax and Semantics, vol. XIII: Current Approaches to Syntax, E. Moravcsik & J. Wirth (eds.), Academic Press, New York. Van Valin, R. & La Polla, R. J. (1997), Syntax: Structure, Meaning and Function, Cambridge University Press, Cambridge. Vendler, Z. (1957), “Verbs and Time”, in Linguistics in Philosophy, pp. 97-­‐121, Cornell University Press, Ithaca. 114 Vossen, P. (1995), Grammatical and Conceptual Individuation in the Lexicon, IFOTT, Amsterdam. Yi, S-­‐t., Loper, E. & Palmer, M. (2007), “Can Semantic Roles Generalize Across Genres?”, in Proceedings of HLT/NAACL-­‐2007, Rochester, NY, USA. XTAG Research Group (2001), “A Lexicalized tree adjoining grammar for English”, Technical Report, University of Pennsylvania. 115 APPENDIXES APPENDIX A – List of Subcategorisation Frames in Il Sabatini & Coletti (2012) Subcategorisation Frames Explanation Example 1) Sogg-­‐V 1. Perché tu tremi? Intransitive monovalent “Why are you shaking?” verb 2. La situazione sta precipitando “The situation is degenerating” 1. Domani mi chiameresti? 2) Sogg-­‐V-­‐Arg Transitive bivalent verb “Would you call me tomorrow?” 2. Io ignoro tutto di lui “I ignore everything about him” 1. Ti giudicavo più saggio 3) Sogg-­‐V-­‐Arg+Comp.Pred Transitive bivalent verb “I judged you wiser” with a predicative complement 2. Vi credevo sinceri “I considered you honest” 1. Daro un libro al mio amico 4) Sogg-­‐V-­‐Arg-­‐Prep.Arg Transitive trivalent verb “I will give a book to my friend” with a prepositional phrase 2. Ho mostrato il biglietto al controllare “I have shown the ticket to the controller” 1. Hanno spostato la sede da 5) Sogg-­‐V-­‐Arg-­‐Prep.Arg-­‐Prep.Arg Transitive quadrivalent Milano a Torino verb with two “They have moved the prepositional phrases headquarter from Milan to Turin 2. Mia madre ha tradotto un romanzo dal turco all’inglese “My mum translated a novel from Turkish into English 1. Sono arrivato primo 6) Sogg-­‐V-­‐Comp.Pred Intransitive verb with a “I have ranked first” predicative complement 2. I nostri accordi rimangono chiari “Our agreements remain clear” Intransitive bivalent 1. Mario ci sembra stanco 7) Sogg-­‐V-­‐Comp.Pred-­‐Prep.Arg verb with a “Mario seems tired to us” prepositional phrase and a predicative complement 2. L’affare ci parve una frode “The affair appeared like a fraud to us” 116 8) Sogg-­‐V-­‐Prep.Arg 9) Sogg-­‐V-­‐Prep.Arg-­‐Prep.Arg 10) Non.Sogg-­‐V 11) Non.Sogg-­‐V-­‐Prep.Arg Intransitive bivalent verb with a prepositional phrase 1. Gianluca crede a Babbo Natale “Gianluca believes in Santa Claus” 2. Oggi resto in ufficio “Today I will stay in the office” 1. Questo treno va da Milano a Intransitive trivalent Roma verb with two “This train goes from Milan to prepositional phrases Rome” 2. Raccontaci del tuo ultimo viaggio “Tell us about your last trip” 1. Piove forte Impersonal avalent verb “It rains heavily” 2. Sta tuonando “It is thundering” 1. Ci sembrava di impazzire Impersonal monovalent “It seemed to us to go crazy” verb with a prepositional phrase 2. Pare che i lavori siano terminate “It appears that the works have finished” APPENDIX B – List of the 1000 Top Frequent Italian Verbs Analysed in our Sample (see CD-­‐ROM) 117 APPENDIX C – List of Subcategorisation Frames identified in the annotation procedure Frame NP_subj#V-­‐si#{affinché}fin Example Si è sacrificato affinché studiassimo “He sacrificed-­‐pron so that we could study” NP_subj#V-­‐ne#{a}PP Gliene voglio perché non mi ha detto la verità “I bear him a grudge because he didn’t tell me the truth” NP_subj{che}fin#V#{da}PP Da ciò consegue che non e possibile partire “From this it follows that it is not possible to leave” NP_subj#V-­‐sene#{...}PP Se ne restò a casa/sulla montagna/in classe “He stayed at home/on the mountain/in class” NP_subj#V-­‐si#{in}inf Devi limitarti nel mangiare “You have to limit-­‐pron in eating” NP_subj#V#{per}inf Ha insistito per avere un colloquio “He insisted to have an interview” NP_subj{0}inf#V#NP_obj Sapere che sei malato ci affligge “To know that you are ill afflicts us” NP_subj#V#{a}inf Mio nipote ha imparato a camminare “My niece learnt to walk” NP_subj#V#{di}PP#{a}PP NP_subj#V#NP_obj#{da}PP Hanno domandato di te alla mamma “They asked about you to mum” Hanno espulso alcuni studenti da scuola “They expelled some students from school” NP_subj#V#{con}PP#{...}PP Non consento con te riguardo la vendita della casa “I don’t agree with you in relation to the selling of the house” NP_subj#V-­‐si#{come}fin Non mi spiego come tu sia bocciato “I don’t explain-­‐pron how you failed the exam” NP_subj#V#NP_obj#{di}PP NP_subj#V-­‐ci#{per}inf Mia nonna ci fornisce di molte provviste “Our grandmother provides us of many supplies” Ci vuole pazienza con i bambini “There needs-­‐pron patience with children” NP_subj#V#NP_obj+Poss NP_subj#V#{fra}PP La situazione impedisce la realizzazione del progetto “The situation prevents the realisation of the project” Maria oscilla ancora fra le due possibilità “Maria still swings between the two possibilities” NP_subj#V#{di}PP#{per}inf Luca approfitta sempre di sua madre per rimediare un passaggio “Luca always takes advantage of his mother to get a lift” NP_subj#V#NP_obj#{per}PP Prepara lo zaino per la palestra, Daniele! Prepare the bag for the gym, Daniele! NP_subj#V#NP_obj#{in}PP Mi hanno introdotto subito nel nuovo ambiente di lavoro “They immediately introduced me into the new work environment” NP_subj{di}inf#V#{a}PP Ti importa di fare una brutta figura? “Of looking foolish matters to you?” NP_subj#V#NP_obj#{a}inf Ti hanno autorizzato a rispondere alle telefonate? “Did they authorise you to take the phone calls?” NP_subj#V#{per}PP Gianna è impazzita per la morte della madre “Gianna got mad for the death of her mother” 118 NP_subj{0}inf#V “Affligge sapere che l’azienda sta fallendo” “It upsets (to) know that the business is collapsing” NP_subj#V#Cpred Flora ha debuttato come attrice “Flora debuted as actress” NP_subj#V#NP_obj#{su}PP Basi sempre le tue opinioni sulle parole di altri? “Do you always base your opinions on somebody else’s words?” NP_subj#V#{su}PP#{per}PP I miei capi premono su di me per un veloce miglioramento “My bosses pressures on me for a quick improvement” NP_subj#V#{a}PP#{...}PP Questi soldi bastano a tutti voi fino a fine mese “This money is enough for all of you until the end of the month” Non-­‐NP_subj#Cpred Sembra estate “It seems summer” Non-­‐NP_subj#V#{...}PP Piove in casa/dentro la macchina “It rains into the house/inside the car” NP_subj#V-­‐ci#NP_obj Ci ho messo due ore a finire i compiti “I put-­‐pron to hours to finish my homework” NP_subj{0}inf#V#{...}PP Imparare l’inglese ti servira’ “(To) learn English will be-­‐pron useful to you” NP_subj#V#NP_obj Il generale ha condotto una notevole spedizione “The general conducted a remarkable campaign” NP_subj#V-­‐ci#{a}inf Ci vuole poco ingegno a capire come stanno le cose “Little intelligence is needed-­‐pron to understand how things are” NP_subj{che}fin#V Succede che qualcosa vada male nella vita “It happened that somebody goes wrong in life” NP_subj#V-­‐si#{davanti a}PP Non mi umilio davanti a nessuno “I don’t humiliate-­‐pron before anybody” NP_subj#V-­‐ci#{...}PP#{...}PP Ci corre una bella differenza tra il promettere e l’ottenere/dal promettere all’ottenere “A great difference stands-­‐pron between promising and obtaining/from promising to obtaining” NP_subj#V#{in}PP La macchina è affondata nella neve “The car sunk in the snow” NP_subj#V#{di}inf Finalmente ho capito di avere bisogno di aiuto “Finally I’ve understood to need help” NP_subj#V-­‐si#{...}PP I manifestanti si stanno raccogliendo in piazza/davanti al comune “The protesters are gathering-­‐pron in the square/before the town hall” Non-­‐NP_subj#V Sta piovendo fortissimo “It is raining heavily” NP_subj#V#{su}PP#{per}inf Il mio professore preme su di noi per fare bella figura “The professor pushes on us to make a good impression” NP_subj#V-­‐si#{di}inf Vi siete proposti di pulire il giardino “You put forward-­‐pron to clean the garden” NP_subj#V#{a}inf#{a}PP La maestra ci ha insegnato a scrivere “The teacher taught us to write” NP_subj#V#{0}inf Io odio stirare! “I hate (to) iron!” NP_subj#V#NP_obj#{che}fin Scommetterei i miei risparmi che non si laureerà mai “I would bet my savings that he will never graduate” 119 NP_subj#V#{da}PP#{...}PP La rondine è volata dal tetto all’/sull’albero “The swallow has flown from the roof to/onto the tree” NP_subj#V#NP_obj Mio nonno raccontava storie bellissime “My grandfather told beautiful stories” NP_subj#V-­‐si#{su}PP La mia idea si fonda sulla mia personale esperienza “My idea based-­‐pron on my personal experience” NP_subj#V-­‐si Ti preoccupi troppo “You worry-­‐pron too much” NP_subj#V#{da}PP Il sole sorge da Ovest “The sun rises from the West” NP_subj#V-­‐si#NP_obj#{da}PP Mi sono cavato i soldi dalla tasca! “I pulled-­‐pron the money from the pocket!” NP_subj#V#NP_obj#{...}PP NP_subj#V#NP_obj#{fra}PP Hai lanciato la palla sul tetto/contro la macchina “You threw the ball on the roof/against the car” Lo calcolo fra i miei migliori amici “I count him among my best friends” NP_subj#V-­‐si#{che}fin Mi attendevo che tu intervenissi “I expected-­‐pron that you intervened” NP_subj#V-­‐sela Te la stai proprio godendo! “You are really enjoying-­‐pron!” NP_subj#V#{con}PP Hai litigato di nuovo con tuo fratello? “Have you fought again with your brother?” NP_subj#V-­‐si#Cpred_Adv Comportatevi bene bambini! Behave-­‐pron properly kids! NP_subj#V-­‐si#{a}PP Si richiamarono al simile caso dell’anno scorso “They referred-­‐pron to the similar case of last year” NP_subj#V#{...}inf Il filo serve a/per cucire “The thread is necessary-­‐pron to/for knitting” NP_subj#V-­‐ci#{in}PP Ci vanno due uova nella ricetta “There go-­‐pron two eggs in the recepy” NP_subj#V#{con}PP#{che}fin Hanno convenuto con noi che la votazione non è valida “They agreed with us that the election is not valid” NP_subj#V#{di}PP#{per}PP Il tuo amico approfitta della tua disponibilità per i suoi interessi “You friend takes advantage of your availability for his own interests” Non-­‐NP_subj#V-­‐cene#{di}PP Ce ne vuole di pazienza! “There is-­‐pron need of patience!” NP_subj#V#Cpred#{a}PP Questa notizia mi suona strana “This piece of news sounds strange to me” NP_subj#V#NP_obj+Poss I miei genitori hanno stabilito le regole della casa “My parents established the rules of the house” NP_subj#V#NP_obj#{da}PP#{...}PP Gli operai hanno scaricato la merce dal camion in cantina/al primo piano “The workers have unloaded the goods from the lorry into the cellar/to the first floor” NP_subj#V-­‐si#{fra}PP I borsaioli si confusero tra i passeggeri del tram “The pickpockets mixed up-­‐pron among the passengers of the tram” NP_subj#V#{in}inf Ho esitato nello scegliere la professione “I hesitated in choosing my job” NP_subj#V#NP_obj#{con}PP Abbiamo discusso il progetto della casa con l’ingegnere “We have discussed the project of the house with the engineer” 120 NP_subj#V#Cpred_Adv#{...}PP L’errore è valso da monito per/a tutti “The mistake functioned as warning for/to all of us” NP_subj#V-­‐si#{contro}PP La moto si è schiacciata contro il muretto “The motorbike crashed-­‐pron against a wall” NP_subj#V#NP_obj#{...}inf Mi hanno accusato di/per aver favorito il candidate “They accused me of/for favouring the candidate” NP_subj#V#{come}fin#{a}PP Mi domandaste come facessi ad essere sempre felice “You asked me how I could be always happy” NP_subj#V-­‐sela#{con}PP Il governo dovrà vedersela con il deficit “The government will have to deal-­‐pron with the deficit” NP_subj#V#{con}PP#{su}PP Convengo con te sull’urgenza dei provvedimenti “I agree with you on the urgency of the measurements” NP_subj#V-­‐si#{di}PP Non mi sorprendo più di niente “I am not surprised-­‐pron of anything anymore” NP_subj#V-­‐la#{di}inf Piantiamola di prenderci in giro “Let’s stop of fooling ourselves!” NP_subj#V#Cpred_Adv Questo telefono funziona male “This phone works badly” NP_subj#V#{su}PP La cupola pesa su possenti colonne “The dome weighs on the mighty columns” NP_subj#V#{che}fin#{a}PP Mi ha raccontato che la serata non e’ proseguita bene “She told (to) me that the night didn’t go on well” NP_subj#V-­‐sene#{di}PP Se ne sbatte di tutti “He doesn’t care of anyone” NP_subj#V-­‐si#{in}PP Il provvedimento si inquadra in un vasto piano di interventi “The measurement is placed-­‐pron within a broad plan of intervention” NP_subj#V#NP_obj#{a}PP Ho scritto una lettere al direttore “I wrote a letter to the chief” NP_subj#V#NP_obj#{di}inf Ho avvisato mio padre di essere in ritardo “I have told my father to be late” NP_subj#V-­‐ci#{...}PP Mi ci vorrebbe un po’ riposo “A little rest would be-­‐pron necessary for me” NP_subj#V#NP_obj#{per}inf Ha speso tutti i suoi risparmi per comprare la macchina “He spent all his savings to buy a car” NP_subj#V-­‐si#{con}PP#{per}PP Mi compiaccio con te per l’esito dell’esame “I congratulate-­‐pron with you for the outcome of the exam” NP_subj#V-­‐si#{per}inf Mi impegnerò per migliorare “I will commit-­‐pron to improve” NP_subj#V-­‐si#Cpred La lana si è mantenuta morbida “The wool maintained-­‐pron soft” NP_subj#V#Cpred_Adv#{a}PP La ragazza ci servirà come segreteria “The girl will be necessary to us as secretary” NP_subj#V-­‐si#{da}inf Mi guardo sempre dal giudicare gli altri “I always refrain-­‐pron from judging other people” NP_subj#V#{di}inf#Cpred Lucia ha creduto opportuno agire in questo modo “Lucia believed opportune (to) act in this manner” NP_subj#V#ger Il relatore ha esordito cantando “The speaker started off singing” NP_subj#V#{che}fin Questo dimostra che avevo ragione io “This proves that I was right” NP_subj#V#NP_obj#ger Avete occupato spazio inutilmente sistemando i mobili così “You took up space in vain arranging the furniture like this” 121 NP_subj#V-­‐ci Lo zio non ci sente “My uncle doesn’t hear-­‐pron” NP_subj#V#{con}PP#{di}inf Convennero con me di agire segretamente “They agreed with me to act secretly” NP_subj#V#{...}PP La serata è finita a casa di Marco/con l’arrivo dei miei genitori “The evening ended at Marco’s place/with my parents’ arrival” NP_subj#V#{a}PP#{di}PP Narraci del tuo ultimo viaggio “Tell (to) us of your last trip” NP_subj#V-­‐si#NP_obj#{con}PP I figli si contendono l’azienda con il padre “The sons contend-­‐pron the business with the father” NP_subj#V-­‐si#NP_obj#{...}PP Non coprirti il viso con le mani! “Don’t cover-­‐pron your face with your hands!” NP_subj#V#NP_obj#{contro}PP Gli indigeni scagliavano frecce contro i bianchi “Native people threw arrows against white people” NP_subj#V#NP_obj#Cpred Ti hanno eletto presidente “They have elected you president” NP_subj{che}fin#V#{a}PP Mi risulta che tu mi abbia cercato “It appears to me that you have looked for me” NP_subj#V#{contro}PP La barca ha sbattuto contro gli scogli “The boat slammed against the cliffs” NP_subj#V#{da}PP#{a}PP Il mio gatto è salito dal terrazzo al tetto “My cat climbed from the balcony to the roof” NP_subj#V-­‐si#NP_obj#{a}PP Si puntò la rivoltella alla tempia “He pointed-­‐pron the gun to his head” NP_subj{che}fin#V#NP_obj Mi stupisce che tu non ci abbia pensato prima “That you have not thought about it before surprises (to) me” NP_subj{di}inf#V NP_subj#V Accade di fare tardi la sera “It happens to be late at night” È morto mio nonno “My grandfather died” NP_subj#V-­‐si#NP_obj#Cpred Ti sei dimenticato di nuovo la luce accesa “You have forgotten the light on once again” NP_subj#V#{di}inf#{a}PP Ti suggerisco di andare e parlare con il preside “I suggest to you to go and talk to the principal” NP_subj#V-­‐sene#Cpred Me ne sono rimasto tranquillo “I stayed-­‐pron calm” NP_subj#V-­‐si#{da}PP#{...}PP Ci siamo spostati dalla cucina alla camera “We moved-­‐pron from the kitchen to the bedroom” NP_subj#V#{come}fin Ho notato come tu ti sia arrabbiato “I noticed how you got angry” NP_subj#V#{che}fin#Cpred Credo appropriato che tu gli dica la verità NP_subj#V#NP_obj#Cpred_Adv Mia sorella ha un dottore per marito “I believe appropriate that you tell him the truth” “My sister has a doctor as husband” NP_subj#V#{0}inf#Cpred NP_subj#V#{che}fin#{da}PP Giudico corretto avvisare le persone per tempo “I judge honest to inform people in advance” Da quello che ha detto deduco che non accetterà la proposta “From what you said I deduce that he won’t accept the offer” NP_subj#V-­‐si#{per}PP Questo romanzo si caratterizza per lo stile inconfondibile “This novel characterises-­‐pron for the unmistakable style” 122 NP_subj#V-­‐si#{a}inf NP_subj#V-­‐si#{con}PP NP_subj#V-­‐la NP_subj{che}fin#V#Cpred#{a}PP NP_subj{0}inf#V#{a}PP NP_subj#V#NP_obj#{addosso a}PP NP_subj{0}inf#V#Cpred#{a}PP NP_subj#V-­‐si#{da}PP NP_subj#V#{di}PP NP_subj#V-­‐si#NP_obj NP_subj#V#{a}PP NP_subj#V-­‐ne#{di}PP NP_subj#V-­‐sene NP_subj{di}PP#V NP_subj#V-­‐si#{0}inf NP_subj{come}fin#V#{a}PP Ci siamo fermati a fare benzina “We stopped-­‐pron to fill the tank” L’automobile si è scontrata con un camion “The car crashed-­‐pron with a lorry” Alla fine l’ho spuntata! “I finally made-­‐pron it” Spetta a noi fare la spesa “It is up to us to do the food shopping” Spetta a noi fare la spesa “It is up to us to do the food shopping” Gli è piombata addosso una disgrazia “A disgrace fell onto him” Ci appare scontato rimandare la seduta “Postponing the meeting seems granted to us” Ti sei allontanato dall’argomento “You moved-­‐pron from the topic” Dubita delle tue parole “She doubts (of) your words” Mi sono messo il cappotto “I put-­‐pron the coat on” A Mario la tua offesa brucia ancora “Your insult still burns to Mario” Ne va della nostra immagine “It is about our imagine” La mia memoria se ne sta andando “My memory is failing-­‐pron” Si tratta del nostro futuro “It is a matter of our future” Mi sono lasciato ingannare “I left be-­‐pron fooled” Mi dispiace se credi che io abbia approfittato di tuo fratello “If you believe that I have taken advantage of your brother it upsets (to) me” 123 APPENDIX D – Argumental Sentence Alternations a. Alternations involving argumental sentences in a complement position Alternation Example Verbs Augurare, garantire, NP_subj#V#{che}fin#{a}PP 1. Ti auguro che tutto vada bene ricordare, assicurare, / “I wish you that everything go well” promettere, gridare, NP_subj#V#{di}inf#{a}PP confidare, annunciare, 2. Ti auguro di fare un buon viaggio raccontare, chiedere, “I wish you to have a pleasant journey” comandare, raccomandare, dire, dichiarare, permettere, scrivere, giurare, confessare, comunicare, riferire Ignorare, disporre, NP_subj#V#{che}fin 1. La situazione impone che tutti scoprire, tacere, / partecipino imporre, badare, NP_subj#V#{di}inf “The situation requires that everyone escludere, vedere, get involved” riscoprire, protestare, stabilire, intuire, 2. La situazione impone di essere uniti gridare, temere, “The situation requires to stick confidare, sopportare, together” constatare, supporre, tollerare, annunciare, pensare, ammettere, deliberare, immaginare, dubitare, fingere, sapere, convenire, ipotizzare, nascondere, smentire, esigere, ritenere, sperare, pretendere, dire, aspettare, riconoscere, dichiarare, credere, dimenticare, sostenere, aggiungere, prescrivere, ottenere, decidere, negare, accettare, dimostrare, sognare, testimoniare, rivelare 1) La maestra ha notato che gli studenti erano stanchi Provare, sottolineare, NP_subj#V#{che}fin “The teacher noticed that the students notare, concepire, / were-­‐indic tired” ricordare, vedere, NP_subj#V#{come}fin giudicare, imparare, 2) La maestra ha notato come gli sapere, intendere, studenti fossero stanchi stabilire, prevedere “The teacher noticed how the students were-­‐subj tired” 124 NP_subj#V#{che}fin#{a}PP / NP_subj#V#{come}fin#{a}PP
}PP 1) Il poliziotto ha spiegato agli studenti che l’autocontrollo è vitale “The policeman has explained that self-­‐control is vital to the students” 2) Il poliziotto ha spiegato agli studenti come l’autocontrollo sia vitale agli “The policeman has explained how self-­‐control be-­‐subj vital to the students” NP_subj#V-­‐si#{che}fin 1) Il calciatore si dispiace di aver perso / “The footballer is sorry to have lost” NP_subj#V-­‐si#{di}inf 2) Il calciatore si dispiace che la squadra abbia perso “The footballer is sorry that the team lost-­‐subj” NP_subj#V#{come}fin#{a}PP 1) Ti spiego come si fa / “I will explain to you how it is done” NP_subj#V-­‐si#{come}finPP 2)Ora mi spiego come mai non mi saluta “Now I explain myself how he doesn’t say hi to me” NP_subj#V#{di}inf#{a}PP 1) Permettere al figlio di uscire / “To let one’s son to go out” NP_subj#V-­‐si#{di}infPP 2) Si può permettere di andare in vacanza dove vuole “She can afford to go on holiday wherever she wants” Spiegare, mostrare Convincersi, augurarsi, assicurarsi, illudersi, ricordarsi, sorprendersi, immaginarsi, dispiacersi, attendersi, accorgersi, sognarsi Spiegare, domandare Proporre, permettere, imporre, augurare, assicurare, offrire, ricordare, impedire, rimproverare, risparmiare 2. Alternations involving argumental sentences in the subject position NP_subj{che}fin#V 1) Pare che nessun risultato sia Sembrare, / ottenibile accadere NP_subj{di}inf#V “Seems-­‐impers that no result is achievable” 1) Pare di non ottenere nessun risultato “Seems-­‐impers to not achieve any results” parere, 125 NP_subj{0}inf#V / NP_subj{che}fin#V NP_subj{0}inf#V#{a}PP / NP_subj{che}fin#V#{a}PP 1) Bisogna raccogliere i soldi “Needs-­‐imper to collect the money” 2) Bisogna che i soldi siano raccolti “Needs-­‐imper that the money be collected” 1) A noi conviene che tu parta subito “To us mattersimper that you leavesubj immediately” 2) A noi conviene partire subito “To us mattersimper to leave immediately” Occorrere, bisognare Dispiacere, convenire, risultare 126 APPENDIX E – Alternations involving Noun or Prepositional Phrases (NP or PP) a. Alternations causing a change in the transitivity of the verbs Alternation Example 1)Hai buttato i sassi dalla finestra NP_subj#V#NP_obj#{da}PP “You threw stones from the / window” NP_subj#V-­‐si#{da}PP 2) Lui si è buttato dal trampolino “He jumped-­‐pron from the trampoline” NP_subj#V#NP_obj#{di}PP / NP_subj#V-­‐si#{di}PP NP_subj#V#NP_obj#{a}PP / NP_subj#V-­‐si#{a}PP 1) Convincere i colleghi della validità della teoria “To convince the colleagues of the validity of the theory” 2) Mi sono convinto del mio errore “I convinced-­‐pron of my mistake” 1) Ha abituato i figli allo studio “She accustomed the children to studying” 2) Si sono abituati a una nuova vita “They have accustomed-­‐pron to a new life” 1) Il provvedimento ha integrato i NP_subj#V#NP_obj#{in}PP portatori di handicap nella società “The measurement has integrated / disabled people within society” NP_subj#V-­‐si#{in}PP 2) Mi sono integrato nel nuovo ambiente di lavoro “I have integrated-­‐pron in the new workplace” Verbs Rilanciare, buttare, escludere, difendere, gettare, dividere, staccare, spostare, lanciare, sciogliere, allontanare, sollevare, distrarre, trarre, ritirare, levare, sfilare, separare, liberare, ritrarre, riparare, salvare Convincere, privare, fornire, ricoprire, riempire, caricare, svuotare, dotare, investire, colmare, incaricare, circondare, coprire Disporre, affidare, donare, mostrare, alternare, opporre, adattare, abituare, mescolare, affiancare, costringere, agganciare, esporre, consacrare, adeguare, accordare, unire, iscrivere, preparare, sottrarre, paragonare, attaccare, raccomandare, presentare, votare, appassionare, associare, allineare, vendere, avvicinare, indirizzare, dichiarare, dare, predisporre, sommare, sottoporre, accostare, concedere, rivolgere, consegnare, interessare, dedicare Proiettare, integrare, immergere, situare, rinchiudere, specializzare, inserire, inquadrare, trasformare 127 NP_subj#V#NP_obj#{su}PP / NP_subj#V-­‐si#{su}PP 1) Hanno basato i calcoli sui dati ufficiali “They based the evaluations on official data” 2) L’accusa si basa sulle testimonianze “The prosecution bases itself on the depositions” 1) I contadini scambiano grano NP_subj#V#NP_obj#{con}PP con petrolio / “Farmers swap wheat with petrol” NP_subj#V-­‐si#{con}PP 2)Mi sono scambiato con il compagno di banco “I swapped-­‐pron with my deskmate” I. DIRECT REFLEXIVES: 1) Spesso i miei genitori giustificano mia sorella “Often my parents justify my sister 2) Spesso si giustifica “Often he justifies-­‐pron II. RECIPROCAL REFLEXIVES: NP_subj#V#NP_obj / 1) Ho incontrato Luca tanto NP_subj#V-­‐si tempo fa “I met Luca a long time ago” 2) Ci siamo incontrati tanto tempo fa “We met-­‐pron a long time ago” Fondare, proiettare, basare, concentrare Conciliare, combinare, alternare, scambiare, confrontare, mescolare Isolare, lavare, rinnovare, liberare, negare, allontanare, scoprire, schierare, esprimere, giustificare, assicurare, uccidere, ammazzare, allenare, licenziare, accettare, consolare, contraddire, tormentare, interrogare, valorizzare, ferire, escludere, spiegare, umiliare, nascondere Conoscere, sospettare, rispettare, combattere, fronteggiare, controllare, attirare, ritrovare, rivedere, stimare, sfidare, abbracciare, rincorrere, inseguire, sfiorare, scegliere, dividere, stringere, picchiare, disturbare, odiare, respingere, frequentare, temere, incontrare, sposare, vedere, trovare, baciare, soccorrere Chiudere, spaventare, III. CAUSATIVE-­‐INCOHATIVE confondere, emozionare, ALTERNATION: abbassare, restringere, staccare, ridurre, intrecciare, 1) Lucia ha spezzato un biscotto sbloccare, scatenare, “Lucia broke a biscuit” rovinare, turbare, piegare, conservare, spezzare, 2) La corda si è spezzata spaccare, rompere “The rope has broken-­‐pron” 128 NP_subj#V#NP_obj#{...}PP / NP_subj#V-­‐si#{...}PP I. DIRECT REFLEXIVES: 1) I fondatori hanno stabilito la sede dell’azienda a Roma / nella capitale “The founders have established the headquarters of the business in Rome / in the capital” 2) Ho deciso di stabilirmi a Milano “I have decided to establish-­‐pron in Milan” II. CAUSATIVE-­‐INCOHATIVE ALTERNATION: 1) Ho rovesciato la minestra sul tappeto “I spilt the soup on the carpet” 2) Il latte si è rovesciato per terra “The milk spilt-­‐pron onto the floor” III. INTENSIVE REFLEXIVES: (1) Fissa sempre lo sguardo sulla mia ragazza “He always fixes his gaze on my girlfriend” (2) Si è fissato con quella ragazza / in un progetto “He fixed-­‐pron with that girl / in a project” b. Alternations involving two intransitive variants NP_subj#V 1) Nel dopoguerra, molti / furono costretti ad emigrare NP_subj#V#{...}PP “During the post-­‐war period many people were forced to emigrate” 2) I miei nonni emigrarono in Germania “My grandparents emigrated to Germany” Girare, disporre, calare, orientare, buttare, muovere, trascinare, gettare, voltare, rilanciare, avviare, spingere, stabilire, spostare, lanciare, dirigere, piazzare, sistemare, ambientare, stendere, mettere, rivolgere, abbandonare, porre, informare, accordare Infilare, collocare, versare, aggiungere, aprire, insinuare, avvolgere, rovesciare, spargere, imprimere, stampare Fissare Tornare, girare, correre, giocare, finire, accorrere, uscire, battere, emigrare, terminare, perdere, bastare, combattere, risalire, arrivare, durare, oscillare, riuscire, volare, nascere, giungere, ricadere, cadere, scivolare, avanzare, votare, venire, crescere, piombare, sorgere, comparire, reagire, saltare, slittare, lavorare, picchiare, salire, vagare, precipitare, rientrare 129 APPENDIX F – Alternations involving an argumental sentence and a phrase complement (NP or NP) a. Alternations affecting a complement position Alternation NP_subj#V#NP_obj#{a}PP / NP_subj#V#{di}inf#{a}PP Verbs 1) Ordinò una marcia forzata ai soldati “He ordered a forced march to the soldiers” 2) Ordinò agli studenti di preparare il testo entro una settimana “She ordered to prepare the text within a week to the students” Rimproverare, imporre, risparmiare, garantire, ricordare, giurare, promettere, gridare, proporre, ordinare, confidare, proibire, annunciare, raccontare, chiedere, augurare, comandare, offrire, raccomandare, sussurrare, dire, predicare, vietare, dichiarare, permettere, suggerire, denunciare, scrivere, assicurare, confessare, concedere, comunicare, consigliare, riferire, impedire Augurare, mostrare, garantire, ricordare, assicurare, promettere, gridare, confidare, ripetere, annunciare, raccontare, chiedere, comandare, raccomandare, dire, dichiarare, segnalare, spiegare, permettere, denunciare, proporre, scrivere, giurare, confessare, concedere, insegnare, comunicare, riferire Disporre, autorizzare, invitare, stimolare, abituare, indurre, motivare, incoraggiare, costringere, condannare, esercitare, educare, ammettere, trattenere, spedire, sollecitare, delegare, forzare, ridurre, obbligare, mettere, destinare NP_subj#V#NP_obj#{a}PP / NP_subj#V#{che}fin#{a}PP 1) Ho promesso aiuto a un compagno “I promised help to a friend” 2) Ti prometto che arriverò in orario “I promise you that I will arrive on time 1) La scuola educa i giovani al rispetto degli altri “School educates youngsters to the respect for others” 2)Educate vostro figlio a essere tollerante “Educate your son to be tolerant” 1) Hai avvisato i parenti della partenza? “Did you warn the parent about the departure?” 2) Ho avvisato tutti che la lezione è sospesa “I have warned everybody that the lesson is cancelled” NP_subj#V#NP_obj#{a}PP / NP_subj#V#NP_obj#{a}inf NP_subj#V#NP_obj#{che}fin / NP_subj#V#NP_obj#{di}PP Example Avvisare, convincere 130 NP_subj#V#{a}PP / NP_subj#V#{a}inf 1) Rinunciò a un progetto “I gave up on a project” 2) Rinunciò a partire “She gave up on going away” 1) Abbiamo obbligato il NP_subj#V#NP_obj#{a}PP nemico alla resa / “We forced the enemy to NP_subj#V-­‐si#{a}inf defeat” 2) Mi sono obbligato a finire il libro entro l’anno “I have forced-­‐pron to finish the book within the year” 1) Preparo i bambini per la scuola NP_subj#V#NP_obj#{per}PP “I prepare the kids for school” / 2)Preparati per andare a cena NP_subj#V-­‐si#{per}inf “Prepare-­‐pron to go for dinner” 1) Ci organizziamo per NP_subj#V-­‐si#{per}PP un’escursione / “We organise-­‐pron for a field NP_subj#V-­‐si#{per}inf trip” 2) Organizzatevi per uscire presto il mattino “Organise-­‐pron to leave early in the morning” 1) Mi sono dimenticato del tuo NP_subj#V-­‐si#{di}PP compleanno / “I have forgotten (of) your NP_subj#V-­‐si#{di}inf birthday” 2) Mi dimenticavo di farti gli auguri “I forgot-­‐pron to wish you happy birthday” 1) Ci assicureremo della verità NP_subj#V-­‐si#{che}fin delle affermazioni / “We will verify-­‐pron the truth of NP_subj#V-­‐si#{di}PP the statements” 2) Ti assicuri che tutto sia in ordine? “Will you verify-­‐pron that everything is in order?” Contribuire, arrivare, rinunciare, scappare, pervenire, provvedere, giocare, venire, badare, pensare, tenere, ritornare, mirare, aspirare, concorrere, tendere Disporre, convincere, costringere, obbligare, esercitare, indurre, abituare, ridurre, trattenere Preparare, sacrificare Organizzare, sacrificare preparare, Curare, convincere, accusare, assicurare, pentirsi, sorprendere, occupare, stupire, ricordare, accontentare, dimenticare, incaricare, accorgersi, vantare Convincere, stupire, assicurare, ricordare, sorprendere, accorgersi 131 NP_subj#V-­‐si#{a}PP / NP_subj#V-­‐si#{a}inf 1) Non mi presterei mai a queste cose “I would never lend-­‐pron to these things” 2) Ti presteresti a sostenere le spese” “Would you lend-­‐pron to support the expenses” b. Alternations taking place in the subject position 1) Dalle tue parole non NP_subj#V#{da}PP consegue niente di buono / “From your words nothing NP_subj{che}fin#V#{da}PP good follows” 2) Da ciò consegue che non è possibile partire “From this follows-­‐impers that leaving is not possible” 1) A nessuno importano queste sciocchezze! NP_subj#V#{a}PP “To nobody matter these silly / things!” NP_subj{che}fin#V#{a}PP 2) Ai ragazzi importa che tu sia qui con noi “To the guys matters-­‐impers that you are here with us” 1) Mi preme solo la tua felicità NP_subj#V#{a}PP “To me matters-­‐impers only your / happiness” NP_subj{0}inf#V#{a}PP 2)“A noi tutti preme essere presenti” “To all of us matters-­‐impers be-­‐inf present” c. Alternations requiring predicative complements 1) Vi credevo sinceri NP_subj#V#NP_obj#Cpred “I believed you sincere” / NP_subj#V#{0}inf#Cpred 2) Ho creduto opportuno agire in questo modo “I believed right act-­‐inf in this way” Preparare, rilanciare, costringere, determinare, obbligare, prestare, rassegnare, abbassare, rimettere, disporre, adattare, abituare, ridurre, indurre Conseguire, risultare Risultare, capitare, importare, dispiacere, sfuggire, convenire Spettare, risultare, capitare, dispiacere, premere, convenire Credere, considerare, giudicare 132 NP_subj#V#NP_obj#Cpred / NP_subj#V#{che}fin#Cpred NP_subj#V#{0}inf#Cpred / NP_subj#V-­‐si#Cpred NP_subj#V#{che}fin#Cpred / NP_subj#V-­‐si#Cpred 1) Lo considero il miglior giocatore del mondo “I consider him the best player in the world” 2) Considero sconveniente che tu rimanga qui “I consider inappropriate that you remain here” 1) Ho giudicato opportuno agire cosi “I judged adequate act-­‐inf like this” 2) Si giudica furbo “To judge oneself smart” 1) Credo giusto che tu vada in persona “I believe right that you go in person” 2) Mario si crede il più bravo di tutti “Mario believes himself the best” Credere, considerare, giudicare Credere, considerare, giudicare Credere, considerare, giudicare 133 
Scarica

a descriptive analysis of argument alternations