Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
The locative alternation: unaccusative constructions and subject position
(USC Thesis Other)
The locative alternation: unaccusative constructions and subject position
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
THE LOCATIVE ALTERNATION: UNACCUSATIVE CONSTRUCTIONS AND SUBJECT POSITION by Roberto Mayoral Hernández ______________________________________________________________________ A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (LINGUISTICS, HISPANIC LINGUISTICS) August 2008 Copyright 2008 Roberto Mayoral Hernández ii Acknowledgements I would like to thank all the people who have helped me to begin and finish this dissertation, sometimes by assisting me with their academic brilliance, and other times by providing me with emotional support. Let’s begin in a chronological fashion. First of all, I would like to thank my parents. They were patient with me when I battered them with questions that required their native intuitions. They also encouraged me to pursue a career so far away from home. I would also like to thank my brother and my sister for their patience and support. A special mention goes to my high school Latin teacher, María José Sarabia, because in her classes, I began to appreciate the beauty of grammar. I will be forever indebted to Violeta Demonte. I would not have reached this far without her support and encouragement. I began studying the locative alternation under her expert tutelage. She inspired me to pursue an academic career in linguistics and has always given me her opinion, based on her deep knowledge of grammar. Now in the US, the two co-chairs of my dissertation committee, María Luisa Zubizarreta and Carmen Silva-Corvalán, deserve a central position in this section. María Luisa has been my academic advisor for the entire duration of the PhD. She has always had the ability to shine the light of knowledge when I was in darkness. When things seemed to make little sense, she discovered the way and guided me along the path. Her close guidance gave form to this dissertation. Carmen was always there when the iii structures seemed to shake and loose stability, to show that variation is not completely random and can be quantified. She offered me her insightful comments that guided me through the apparently winding roads of variation. She awakened in me the interest for these twisting roads and the shady places in grammar. Mario Saltarelli has always given me his support, from the very beginning. He is an expert in every subject. I admire his vast knowledge of linguistics and his openness to explore grammar from any possible view, as well as his approachability and innate ability to provide balance to apparent extremes. I would like to express my sincere gratitude to Jean Roger Vergnaud. My conversations with him have always been deeply enlightening and have made me think deeper into the structure of language. His joviality illuminates the whole department. I would also like to thank all the professor at USC and UAM, who taught me different aspects and views within the field of linguistics. A special mention goes to Jack Hawkins, who directed some of the projects I was involved in and showed me the weight of typological and psycholinguistic research. Elsi Kaiser taught me the details of psycholinguistics and offered me her expert guidance and advice. Also, thanks to Roumyana Pancheva, Elliott Moreton, Utpal Lahiri, Audrey Li and Bonnie Glover Stalls for everything I learned with them. At the UAM, I am indebted to Olga Fernández Soriano, who guided my first steps into the GB framework, together with Violeta Demonte. iv Thanks to Elena Guerzoni, for her native intuitions on Italian and patience with my questions; Tomomasa Sasa, for Japanese; Asier Alcázar, for Basque; Alba Morera and Eva Pons, for Catalan; and Marianne Mans, for the Dutch data. I would like to thank all my fellow linguistics students in GSIL and HILSA, in particular: Asier Alcázar, Emily Hinch Nava, Mónica Cabrera, Lindsey Chen, Omar Beas, Rebeka Campos, Álvaro Cerrón-Palomino, Aslı Ci ğer, Jerry Liu, Michal Martínez, Isabelle Roy, Michael Rushforth, Ana Sánchez Muñoz and Tomomasa Sasa. Thanks for your friendship, patience and interesting conversations we had, and for being there. Your comments and opinions have enriched my research. My thanks also go to the staff in the Linguistics Department at USC: Joyce Perez, Karma Dolma and Frankie Hayduk. My special gratitude goes to Gayle Vierma, USC Spanish Basic Language Program Director, for all her support, advice and friendship. Thanks for caring about me. I would like to acknowledge the financial support of the La Caixa Fellowship, the Del Amo Foundation, the Linguistics Department, and the College of Letters, Arts and Sciences. Finally, I would like to thank all my friends, who were by my side here in Los Angeles and in Spain. v Table of Contents Acknowledgements List of tables List of figures Abstract Chapter 1: Preliminaries to the locative alternation and its main properties 1.1. The locative alternation 1.1.1. Spray/load alternation 1.1.2. Clear alternation (transitive) 1.1.3. Wipe alternation 1.1.4. Swarm alternation 1.1.5. Clear alternation (intransitive) 1.1.6. A lexical analysis of the locative alternation: Levin & Rappaport 1998 1.2. Syntactic asymmetries in the alternating constructions 1.2.1. Thetic and categorical variants 1.2.2. The case of Basque 1.2.3. Should the thetic / categorical distinction determine the syntactic structure? 1.3. The holistic effect 1.4. Two types of intransitive verbs: unaccusative and unergative 1.4.1. Unaccusativity in Spanish Chapter 2. The locative alternation in Spanish 2.1. Types of locative alternation in Spanish 2.1.1. Verbs of the type of cargar “load”, empacar “pack” and plantar “plant” 2.1.2. Verbs of the type of barrer “sweep”, disculpar “forgive” and recoger “put in order” 2.1.3. Verbs of the type of arder “burn”, resonar “resound” and pulular “swarm” 2.1.4. Verbs of the type of despejar “clear”/borrar “erase” ii viii x xi 1 3 5 6 6 7 7 8 10 10 12 21 31 35 40 46 46 47 49 51 53 vi 2.2. Syntactic analyses of the locative alternation in Spanish 2.2.1. Mateu 2002: Argument Structure: relational construal at the syntax-semantics interface 2.2.1.1. The system 2.2.1.2. The locative alternation 2.2.1.3. A different type of locative alternation: Verbs involving Manner 2.3. The semantic components of the locative alternation in Spanish Chapter 3. An unaccusative account of the locative alternation 3.1. The unaccusative hypothesis 3.2. The universal structure of alternating verbs 3.2.1. The swarm verbs as unaccusatives: Theoretical advantages 3.2.1.1. Unergativity, unaccusativity and the swarm type 3.2.2. The notion of internal and external causation: transitive and intransitive counterparts of the locative alternation 3.3. Locative alternation in Spanish: subtypes and special features 3.3.1. Verbs of the type of cargar “load”, empacar “pack” and plantar “plant” 3.3.2. Verbs of the type of barrer “sweep”, disculpar “forgive” and recoger “put in order” 3.3.3. Verbs of the type of despejar “clear” and borrar “erase” 3.3.4. Verbs of the type of arder “burn”, resonar “resound” and pulular “swarm” 3.4. On the status of alternating prepositions 3.4.1. Prepositional aspectology 3.4.2. The syntax of alternating prepositions 3.5. Why do verbs alternate? Chapter 4. On subject position, unaccusativity and the swarm type in Spanish 4.1. Introduction 4.2. On the relation between verb types and subject position in Spanish 4.2.1. The variationist approach 4.2.2. Subject position in Spanish 4.2.3. Experiment 1: Mayoral Hernández 2004c 4.2.4. Experiment 2: Mayoral Hernández 2006 4.2.4.1. Research hypothesis and factors 4.2.4.2. Methodology 4.2.4.3. Results 4.2.4.4. Verbs of Light and Sound Emission in Spanish: the test 4.2.4.5. Verbs of light and sound emission: Results 55 55 55 58 61 66 78 78 83 97 98 105 108 108 110 114 114 116 117 120 123 131 131 132 133 134 137 140 141 141 142 145 146 vii 4.3. A variationist analysis of subject position: swarm verbs 4.3.1. Methodology 4.3.2. Factors and research hypotheses 4.3.3. Results: On verb types 4.3.3.1. Brillar 4.3.3.2. Arder 4.3.3.3. Abundar 4.3.3.4. Hervir 4.3.3.5. Pulular 4.3.3.6. Summary 4.3.4. Results: Presence of preverbal XPs 4.3.5. Results: Presence of postverbal XPs 4.3.6. Results: Type of subject 4.3.7. Results: Type of clause 4.3.8. Results: Sentence polarity 4.3.9. Results: Type of alternant 4.4. Discussion and analysis of the results Chapter 5. Conclusions Appendix References 149 150 151 156 156 159 161 162 164 164 165 167 169 170 173 174 175 181 188 192 viii List of Tables Table 1: Influence of subjects on the position of adverbials, including wh- Subjects Table 2: Verb types and subject distribution Table 3: Subject distribution in unergative and transitive verbs Table 4: Subject distribution in unergative and unaccusative verbs Table 5: Subject position: VLSEs with and without locative alternation Table 6: Subject position by Verb type: VLSEs and unaccusative verbs Table 7: Subject position by Verb type: VLSEs and unergative verbs Table 8: Alternating vs. non-alternating constructions with brilla Table 9: Subject position by Verb type: alternating brilla and unergative verbs Table 10: Subject position by Verb type: alternating brilla and unaccusative verbs Table 11: Subject position by Verb type: non-alternating brilla and unergative verbs Table 12: Subject position by Verb type: non-alternating vs. alternating arde Table 13: Subject position by Verb type: arder vs. unaccusatives Table 14: Subject position by Verb type: non-alternating vs. alternating abundar Table 15: Subject position by Verb type: non-alternating vs. alternating hervir Table 16: Subject position by Verb type: hervir vs. unaccusatives 137 138 143 144 147 147 148 157 158 158 159 160 161 162 163 163 ix Table 17: Subject position by verb type: alternating vs. non-alternating pulular Table 18: Subject position by Presence of preverbal XP, all verbs Table 19: Subject distribution in unergative and transitive verbs Table 20: Subject position by Type of verb, with overt preverbal XP Table 21: Subject position by Presence of postverbal XP Table 22: Subject position by Type of Verb, with overt postverbal XP Table 23: Subject position by Type of Subject Table 24: Subject position by Type of clause; unaccusative and swarm type verbs Table 25: Subject position by Type of clause; unergative verbs Table 26: Subject position by Sentence polarity Table 27: Subject position by Type of alternant 164 166 166 167 168 168 170 171 172 173 174 x List of Figures Figure 1: Spray/load Figure 2: barrer “sweep”, disculpar “forgive” and recoger “put in order” Figure 3: arder “burn” resonar “resound” Figure 4: Distribution of PPs in swarm verbs Figure 5: List of variables and their values Figure 6: Types of locative alternation and features 74 75 76 129 151 184 xi Abstract The study of transitivity alternations plays a central role in linguistic research, because it can provide us with valuable information about the interface between the lexicon and the syntax. The main focus of this dissertation is to find the intrinsic features that characterize the locative alternation, i.e. to separate idiosyncratic and language specific properties from those that are inherent to this construction. Special attention is given to the locative alternation in Spanish. I first provide an introduction to the most prominent features analyzed in the literature, such as the holistic effect and other asymmetries. However, the holistic effect cannot be hold as a feature of alternating verbs, however, since it can only be applied to a small group of verbs. These asymmetries do not hold crosslinguistically either, so they must be discarded. The division of Spanish alternating verbs into subgroups serves as a means to identify what properties have are idiosyncratic. In particular, the prepositional object has a fundamental role in this construction; it is conceptually necessary and provides the argument structure to the sentence. Based on the centrality of the preposition, I develop a system the lower PP generates the argument structure of an unaccusative construction. A fundamental assumption is that all alternating constructions are build on an unaccusative construction with the structure [ VP D [V [PP]]], which may either imply a notion of change of state or xii location when the PP is complex, or a pure state or location when the PP is simple. Here, I postulate that the controversial swarm type verbs, a subtype within the locative alternation, which has been described as both unergative and unaccusative in the literature, have unaccusative properties when they enter the locative alternation. In order to support such claim, I develop a test that can differentiate between unaccusative and unergative verbs in Spanish by measuring subject placement. The results indicate that swarm verbs behave as unaccusative when they participate in the locative alternation, because they show a higher percentage of postverbal subjects. These results support my classification and the description of alternating verbs as underlying unaccusatives. 1 Chapter 1 Preliminaries to the locative alternation and its main properties The analysis of alternations has been a common trend in the linguistic field because they seem to indicate that there is an overt relationship between the syntactic structure and some meaning components. Some radically different approaches have tried to account for the behavior of alternating verbs. On one extreme, a purely lexicalist approach will predict that the syntactic structure is semantically determined (Pinker 1989, Jackendoff 1990, Levin & Rappaport 1995). On the other extreme, one could assume that the semantic interpretation is derived from the underlying syntactic representation without the intervention of the lexicon (Borer 2005). Somewhere in the middle, in this paper I will defend the existence of syntactic constructions 1 (Goldberg 1995) that are intimately related to specific semantic interpretations (Mateu 2002, Zubizarreta & Oh 2004, 2007). The central theme of this dissertation is the locative alternation. All the verbs that display this alternation select two internal arguments that can be realized either as a direct object or as a prepositional phrase. These two arguments always bear the same thematic roles, which I refer to as locatum and location following Levin 1993. The locatum is the 1 The definition of construction has taken different forms in the linguistic literature after Goldberg’s 1995 influential work. Here I will not address the issue of how many constructions exist or what their specific features are. Instead, I assume Zubizarreta & Oh’s 2007 framework as it relates to unaccusative constructions. Grosso modo, they are meaningful primitive structures that pertain in the interface between lexicon and syntax, which corresponds to the l-syntax in Hale & Keyser’s framework (H&K 1993 and 2001 inter alia). 2 constituent that experiences a change of location, so it is a theme. The location is the place where the locatum will end up, so it is the goal. The syntactic flexibility displayed by alternating verbs has raised curiosity among linguists, and it has lead to abundant linguistic research, especially of English. As in a parable, this dissertation tells the story of the locative alternation in Spanish, in search of its own identity. This search will lead us to important discoveries about its deepest origins as an unaccusative construction and about how the subject position test becomes an important tool that will help the protagonist to reassure its newly found identity. Thus, a part of this research will aim at understanding some of the idiosyncratic properties of unaccusative constructions and their relation with subject position in Spanish. Additionally, we will discover that there exist similar structures in other languages, and that all of them share the same intrinsic properties. In fact, I will postulate that many of the features that characterize this alternation in Spanish may be universal, specifically its unaccusative properties. This dissertation is organized into five chapters. Chapter 1 provides an introduction to the locative alternation and its main properties, while pointing out several issues underlying current analyses of this construction. I also include a brief introduction to the unaccusative hypothesis, which will become a central topic in this dissertation. In chapter 2, I first present some of the most relevant studies of the locative alternation in Spanish and some issues. Then I provide a classification of verbs that participate in the locative alternation in Spanish, based on corpus data, and describe their main properties. In Chapter 3, I present a syntactic analysis of this alternation based on the idea that it is 3 built on an unaccusative construction, based on Zubizarreta & Oh 2007. Chapter 4 provides corpus evidence of the classification of the swarm type verbs as unaccusative. In order to do so, I carry out a variationist analysis of subject position in Spanish that will be used as a test for unaccusativity. The last chapter, chapter 5, presents a recapitulation of the main issues examined in the dissertation. In this chapter, I also address some issues posed by such languages as German and Dutch. In the present chapter, section 1.1 provides a general description of the locative alternation and of Levin’s 1993 classification of its subtypes. Section 1.2 focuses on some syntactic asymmetries that differentiate between the two variants of the alternation, as presented in Basilico 1998, while posing some questions regarding the crosslinguistic validity of Basilico’s observations. Section 1.3 discusses the holistic effect. Finally, section 1.4 provides an introduction to unaccusativity. 1.1. The locative alternation Numerous current studies tend to accept that the meaning of a verb determines, to a great extent, the number of arguments that it takes, as well as the structure that it generates. Several linguistic analyses suggest that subcategorization frames can be derived from the semantic requirements of certain meaning components (Levin & Rappaport 1998, Jackendoff 1990,…). This statement is based on proposals stating that the ideal lexical entry of a word should just include its idiosyncratic characteristics. The challenge for scholars working in this area is to determine those meaning components, or semantic templates, that are associated with a certain kind of syntactic structure. 4 Levin & Rappaport 1998, Jackendoff 1990 and Levin 1993, among others, have studied different alternations in order to determine the common meaning components that enable verbs to display a specific type of alternation. These authors adopt different lexical-semantic approaches, although, there is no agreement about the basic LCS (Lexical Conceptual Structure) associated with specific types of alternation, such as for example the spray/load type. This lack of agreement results in a multiplicity of different proposals, which proves that there is still much research needed in this field before we are able to separate the idiosyncratic aspects of meaning from those that are structural. The particular lexicalist views adopted by these authors (section 1.1.6 summarizes Levin and Rappaport’s 1998 lexicalist analysis of the locative alternation) face the problem of polysemy. Grosso modo, they propose the existence of different lexical conceptual structures (LCSs) for every possible syntactic configuration alternating verbs may appear in. Therefore, if a given verb can occur in different constructions, does it imply that there is a different lexical entry for each of them? And, if that is the case, is there a limit to the number of lexical entries? How can the system be constrained? But what is the locative alternation? Levin 1993 presents a comprehensive list of verbs that participate in this alternation and divides them into five subgroups, depending on their argument structure. The first three involve the syntactic alternation between the two internal arguments of transitive verbs. In particular, the direct object can also appear as a prepositional object, maintaining its theta role, while the prepositional object becomes the direct object, as in (1a) and (1b). The last two classes are formed by 5 intransitive verbs which show a similar alternation, but this time between the subject and the prepositional object, as in (1c) and (1d). (1) a. John loaded the potatoes into the truck. b. John loaded the truck with the potatoes. c. The bees are swarming in the garden. d. The garden is swarming with bees. Due to the interchangeability of the direct object, or the subject of intransitive types, and the prepositional object, the latter is considered as an argument of the verb. The following five sections present Levin’s classification of verbs that display the locative alternation. 1.1.1. Spray/load alternation The verbs included in this type are transitive verbs which select a DO and a PP. In the locative variant 2 , the DO is the locatum argument (or theme) and the PP is the location argument (or goal). In the with variant, the DO is the location argument and the PP (always introduced by with) is the locatum argument 3 . 2 In the locative variant, the direct object, or the subject in the intransitive classes, bears the locatum or theme theta-role, while the PP represents the location or goal. 3 I am using the terms locatum and location following Levin (1993). These terms are only used as a descriptive tool. I am not adopting a theory of thematic roles to describe the properties of the lexicon (see Rappaport & Levin 1998 for further discussion about the adequacy of a lexical theory based on theta-roles). 6 (2) a. John loaded potatoes (locatum) into the truck (location). (locative variant) b. John loaded the truck (location) with potatoes (locatum). (with variant) c. John sprayed the paint (locatum) onto the wall (location). d. John sprayed the wall (location) with the paint (locatum). 1.1.2. Clear alternation (transitive) This group, which also has an intransitive version, included in section 1.1.5, is only formed by four verbs: clear, drain, empty and clean. In the locative variant the direct object is the locatum while the PP functions as the location. In the of variant the order is the opposite. (3) a. John cleared dishes (locatum) from the table (location). (locative variant) b. John cleared the table of dishes. (of variant) 1.1.3. Wipe alternation The structure of the two variants of the verbs included in this group is not parallel. In the locative variant (4a), there is a DO whose theta-role is the locatum, and a PP that bears the location θ-role. However, the other variant (4b) has only a DO with location role, but no overt locatum PP. 7 (4) a. John wiped the fingerprints (locatum) off the table (location). (locative variant) b. John wiped the table (location) (*of fingerprints (locatum)). (locative object variant) 1.1.4. Swarm alternation This type of locative alternation is displayed by intransitive verbs. In the locative variant the subject is locatum and the PP is location. The order in the with variant is the opposite, but the preposition must always be with, as in (5). (5) a. Mosquitoes (locatum) are swarming in the garden (location). (locative variant) b. The garden (location) is swarming with mosquitoes (locatum). (with variant) 1.1.5. Clear alternation (intransitive). As in the case of the transitive clear alternation, this group is composed by a very small number of verbs, only clear, drain and empty. In the locative variant, the subject is the locatum argument, while the location role is displayed by the PP. The of variant has a subject which bears the location role and an object introduced by the preposition of which bears the locatum theta-role. However, there is not agreement between native speakers about the possibility of including the PP in the of variant. (6) a. The clouds (locatum) cleared from the sky (location). (locative variant) b. The sky (location) cleared (?of clouds (locatum)). (of variant) 8 1.1.6. A lexical analysis of the locative alternation: Levin & Rappaport 1998 Levin and Rappaport 1998 also provide a lexical description of the verbs that participate in the locative alternation of the spray/load class. They use lexical conceptual structures (LCSs) similar to those proposed by Hale & Laughren 1983 and Hale & Keyser 1986 and 1987. They adopt the predicate decomposition approach, where arguments are not identified by the use of theta-roles, but by using variables included in lexical conceptual substructures. Thus theta-roles are not considered as primitives. Predicate decomposition approaches assume the existence of primitive predicates, such as ACT or BECOME, a universal set of lexical semantic templates, such as (6a), and an open- ended set of constants. In (7a), the lexical conceptual template provides the argument structure for the sentence and part of its meaning, by being associated with a specific combination of predicates. In particular, (7a) represents a change of state where the actor x causes y to become something, i.e. a final state. The constant adds a lexical item to the template, and either it fills an argument position or functions as a modifier of a predicate. These authors suggest that each variant of the alternation has a slightly different LCS, related by the presence of the same constant (LOAD). The purpose of this dichotomy is to account for the difference in meaning that both variants of the same alternation show. The LCS in (7b) represents the locative variant, while (7c) shows the LCS of the with variant. 9 (7) a. [[x ACT] CAUSE [y BECOME [ ] STATE ]] b. [[x ACT] CAUSE [y BECOME P Loc z] [LOAD] MANNER ] c. [[x ACT] CAUSE [z BECOME [ ] STATE WITH-RESPECT-TO y] [LOAD] MANNER ] The constant [LOAD] indicates the manner in which the action came about and it represents the component of meaning that makes load different from any other verb. This analysis implies that the reason for the alternation is the existence of two different LCSs associated with the same constant (LOAD). The template that represents the with variant instantiates a change of state, so it follows that the z variable in (7c), which denotes the entity that experiences a change of state, has to be associated with the direct object. (7b) represents a change of location, where the direct object is realized by the argument that experiences the change of location, i.e. the locatum or goal. Furthermore, unlike (7b), the LCS in (7c) implies that the location argument is completely affected, so that it becomes completely loaded. The phenomenon through which the change of state variant implies complete affectedness of the location argument is called the holistic effect, which is explained in section 1.3. In this section, we have seen that Levin 1993 proposes five classes of locative alternating verbs. Every class is formed by verbs which display exactly the same kind of alternation. What allows the alternation is the existence of two different LCSs. In the next chapter I will present the different types of locative alternating verbs in Spanish and show that it is necessary to provide a different classification of verbs with locative alternation. 10 1.2. Syntactic asymmetries in the locative alternation In this section, I discuss Basilico’s 1999 description of the locative alternation, which is based on the distinction between thetic and categorical judgments or statements. This author postulates that this distinction applies to the locative alternation, and that it implies the existence of syntactic differences between the two variants of this alternation. Before going into more detail, I provide a brief description of the thetic/categorical distinction based on research developed in Kuroda 1972. 1.2.1. Thetic and categorical variants The distinction between thetic and categorical judgments was brought into linguistics by Kuroda 1972: -A thetic judgment entails the presentation of an object, which can be an entity or an event -A categorical judgment involves the presentation of an object and also a property to be affirmed or denied of the object. Kuroda observed that the distinction between thetic and categorical judgments is realized overtly in Japanese by the focus marker –ga and the topic marker –wa respectively, as in the following sentences: 11 (8) a. John-wa imo-o truck-ni tsunda John-Top potato-acc. truck-loc/dat load-past Speaking of John, he loaded potatoes into the truck. b. John-ga imo-o truck-ni tsunda John-Foc. potato-acc. truck-loc/dat load-past It’s John who loaded potatoes into the truck. Basilico 1998 applied this differentiation to several diathesis alternations, including the locative alternation. He claimed that the alternants with the locatum argument functioning as DO, i.e. the locative variant, represented a thetic judgment, while the other alternant represented a categorical judgment: (9) Thetic alternants: a. The farmer loaded a bale of hay into the truck. El granjero cargó un fardo de heno en el camión. (10) Categorical alternants: a. The farmer loaded the truck with a bale of hay. El granjero cargo el camión con un fardo de heno. Let’s summarize in a chart what the features associated with thetic and categorical statements are: 12 (11) Thetic statements Categorical statements Entails the presentation of an object, which can be an entity or an event Involves the presentation of an object and also a property to be affirmed or denied of the object. Focus marker Topic marker Japanese -ga Japanese -wa Locatum as DO variant (locative variant) Location as DO variant (with-variant) In section 1.2.2, I present the system proposed in Basilico 1998 to describe the locative alternation in order to introduce certain syntactic asymmetries purportedly associated to the difference between thetic and categorical judgments. 1.2.2. Basilico 1999 Basilico 1998 describes three diathesis alternations (creation/transformation, locative and dative) based on their interpretation as thetic or categorical statements (distinction brought into linguistics by Kuroda 1972 and Sasse 1987). A thetic judgment entails the presentation of an object, which can be an entity or an event, while a categorical judgment involves the presentation of an object and also a property to be affirmed or denied of that object. In (12) I include the trees that this author proposes for the two alternants. I will attempt to determine to what extent the differentiation between thetic and categorical statements can provide us with a description of the locative alternation. I will also point out some flaws in Basilico’s argumentation. Applying the VP External Object Hypothesis (Chomsky 1991, Borer 1993 and others), Basilico (1998) proposes two different structural positions for objects, depending on the interpretation of the predication as thetic or categorical. Basically, this author assumes that thetic alternants have an internal object VP originated as [Spec, VP], which moves to [Spec, TransP] (Transitivity Phrase, following Jelinek 1995) to receive Case, since it is Trans the category responsible for accusative case assignment. The categorical alternants have an external object, originated in Spec, TransP, where it is Case-marked. It is also proposed that the internal V is able to assign an inherent case, following Belletti 1988, and that movement to Spec, Trans “is motivated to satisfy a strong [nominal] feature (Chomsky 1995) associated with the head of Trans” (op. cit.: p.545). Finally, TransP is a complement of a higher VP (or vP) with agentive content and able to assign an agent theta-role. (12) a. Thetic Alternant VP 3 Subj V’ 3 V TransP 3 Obj i Trans’ 3 Trans VP 3 t i V’ 3 V Complement 13 14 b. Categorical Alternant VP 3 Subj V’ 3 V TransP 3 Obj Trans’ 3 Trans VP ! V’ 3 V Complement The use of two different structures explains the asymmetries between the two alternants of the above mentioned alternations. In particular, Basilico mentions scope asymmetries with respect to (i) quantified NPs and (ii) frequency adverbs −frequently, occasionally −; (iii) the ability to appear as the postcopular constituent in there existential sentences; (iv) locative inversion, and (v) specificity constraints. Here, I include some examples involving locative alternation (from Basilico 1998) and its translation into Spanish. (13) Thetic alternants a. The farmer loaded a bale of hay into the truck. El granjero cargó un fardo de heno en el camión. 15 b. There was hay loaded onto the truck. Había heno cargado en el camión. c. How quickly i can there be [paint sprayed onto that wall t i ]? ¿Con qué rapidez i puede haber [pintura rociada en la pared i ]? d. A chorus of voices echoed in each hall (scope ambiguity). Un coro de voces resonaba en cada pasillo. Categorical alternants e. The farmer loaded the truck with a bale of hay. El granjero cargó el camión con un fardo de heno. f. There was a truck loaded with hay. Había un camión cargado con heno. g. *How quickly i can there be [a wall sprayed with paint t i ]? ¿Con qué rapidez i puede haber [un muro rociado con pintura i ]? h. A hall echoed with every voice in the choir (no scope ambiguity). Un pasillo resonaba con cada voz del coro. Basilico (op. cit.) explains that only thetic alternants (14a) allow for a scope ambiguity with quantified noun phrases within the VP. However, categorical alternants (14b) allow only the interpretation where the direct object has wide scope with respect to the prepositional object. Thus, sentence (14b) would only allow an interpretation in which there is only one truck and all the hay is loaded into it. 16 (14) a. The farmer loaded a bale of hay into every truck (ambiguous). b. The farmer loaded a truck with every bale of hay (not ambiguous). (15) a. El granjero cargó un fardo de heno en cada camión. b. El granjero cargó un camión con cada fardo de heno. The sentences in (15a) and (15b) are the Spanish translations of the sentences in (14a) and (14b) respectively. However, unlike Basilico’s claim for English, both variants of the alternation in Spanish may be ambiguous, and both quantifiers can take scope over each other. Sentence (15a) favors, however, an interpretation in which there are several bales of hay, since it is difficult to conceive the same bale of hay filling different trucks. In the same way, (15b) favors an interpretation in which there is only a truck and several bales of hay, because the bales of hay tend to be much smaller than a truck. This shows that pragmatics is the only factor operating here, and not the syntactic structure. Basilico observes a similar phenomenon in sentences with passive voice and PP preposing. The author noted that when thetic alternants are passivized and the PP is left- dislocated to preverbal position, as in (16a), the NP that remains within the VP must take narrow scope with respect to the prepositional complement. Therefore, it is only possible a reading where there is one single truck, and everything will be loaded into it. Again, this is not true for Spanish, since the thetic variant in (16b), which is the Spanish translation of (16a), allows for both narrow and wide scope of the direct object NP, where 17 there might be one (narrow scope) or several (wide scope) trucks. Nevertheless, as in the case of sentence (15b), the favored reading is that where there is only one truck. (16) a. On a truck was loaded every bale of hay. b. En un camión fue cargado cada fardo de heno. Another syntactic property observed in Basilico (op. cit.) that seems to affect the English variants of the locative alternation is that when frequency adverbs are inserted in the structure, the object can take either narrow or wide scope with respect to preverbal adverbials, both in thetic and categorical statements ((17a) and (17b) respectively). However, when the adverb appears after the direct object but before the prepositional phrase, only the thetic alternants (18a) allow an ambiguity, while categorical alternants (18b) are forced to have a reading in which the direct object takes scope over the frequency adverbial. (17) a. While we watched, the cook frequently sprinkled a bit of cheese on a pizza. b. While we watched, the cook frequently sprinkled a pizza with a bit of cheese. (18) a. While we watched, the cook sprinkled a bit of cheese frequently on a pizza. b. While we watched, the cook sprinkled a pizza frequently with a bit of cheese. 18 Once more, the effect of the frequency adverbial on scope is not clear in Spanish. The sentences in (19), which are the translations of those in (18), favor the interpretation in which there is only one pizza, while the other reading is hard to get in both of them. (19) a. Mientras mirábamos el cocinero espolvoreaba un poco de queso frecuentemente sobre una pizza. b. Mientras mirábamos, el cocinero espolvoreaba una pizza frecuentemente con un poco de queso. Finally, there is another asymmetry that has been suggested in the literature to affect the two variants of the alternation. Basilico mentions that in English there is a difference in specificity between both variants. This, Basilico claims, can be seen overtly in the intransitive types of locative alternation, as in the swarm type. In (20), bare plural subjects of the locative variant allow an existential interpretation, while the with variant (21) can only have subjects with a generic interpretation. (20) a. Bees swarmed in the garden b. Voices echoed in the hall. (21) a. Gardens swarmed with bees. b. Halls echoed with voices. 19 The above generalization is not observed in Spanish either. Spanish does not allow bare plural subjects in preverbal position, but the sentences in (22) contain a relative clause whose meaning is similar to those in (21) and whose relative pronoun, which functions as subject, has a bare noun as antecedent. These Spanish sentences have an existential antecedent with a non-generic interpretation, which shows that the generalization established for English does not hold for Spanish either. The crucial difference between Spanish and English lies on the fact that in the former language it is always possible to get the same meanings in both variants of the alternation, which might support the idea that there are no structural differences between them. (22) En este monasterio hay pasillos que resuenan con las voces de los muertos. In this monastery there are halls that resound with the voices of the dead. The syntactic phenomenon analyzed in this section might suggest that the two variants of the alternation have the same syntactic structure, in opposition to the view adopted by other authors about the locative alternation (Mateu 2002, Basilico 1998,…). However, consider the sentences in (23), which seem to show an apparent distinction between the two variants of the alternation 4 . 4 This fact was pointed out to me by Jean Roger Vergnaud. 20 (23) a. Juan cargó su libro con fórmulas innecesarias. John loaded his book with unnecessary formulae. b. *Juan cargó formulas innecesarias en su libro 5 . John loaded unnecessary formulae in his book. These Spanish sentences seem to indicate that only the with-variant can allow a metaphorical use of the verb cargar “load”. There are two possible ways to account for this distinction. On the one hand, one could hypothesize that this difference is universal, and that it reflects some primitive difference between the two variants of the alternation that should therefore hold crosslinguistically. On the other hand, it is possible to assume that Spanish, and perhaps other genetically related languages like French, have randomly endowed one of the two variants of the alternation with metaphoric meaning. It is obvious that the meaning of load in (23a) and (23b) differ from the standard one, and would be somewhat equivalent to verbs such as endow or provide or fill. If we assume this approach, the expectation would be that different languages could also have a metaphoric meaning associated with the locative variant of the alternation, which seems to be impossible in Spanish. And indeed this seems to be the case. Native speakers of English allow for the English counterpart of Spanish (23b), as shown in (24b): (24) a. John loaded his book with unnecessary formulae. b. John loaded unnecessary formulae in his book. 5 There is disagreement between native speakers regarding the acceptability of these two sentences or the existence of a difference between them. 21 The English sentences show that this apparent difference in Spanish cannot be universal, but language specific, and, therefore, idiosyncratic. In this section, I have argued that it might not be appropriate to consider some syntactic differences discussed in the literature as key features of the locative alternation, since they do not hold crosslinguistically. An analysis that intends to discover the intrinsic universal properties of the locative alternation should only include crosslinguistic commonalities as a defining component. In the following section, I investigate if there is any language that displays the thetic/categorical distinction in the locative alternation constructions through the use of overt morphological or syntactic markers. 1.2.3. Should the thetic / categorical distinction determine the syntactic structure? As I mentioned above, Basilico applied this dichotomy to the locative alternation, but he never explained why or what the consequences derived from such interpretation are. In this section, I commit myself to the search of an overt marking for the two types of judgments in the locative alternation, in order to determine if this differentiation deserves further attention. Considering the characteristics represented in the chart in (11), repeated below in (25), one might predict that if there were a language with overt focus and topic marking in the direct object, it would be possible to verify whether there are restrictions to focalization or topicalization associated with the thetic and categorical differentiation. 22 (25) Thetic statements Categorical statements Entails the presentation of an object, which can be an entity or an event Involves the presentation of an object and also a property to be affirmed or denied of the object. Focus marker Topic marker Japanese -ga Japanese -wa Locatum as DO variant (locative variant) Location as DO variant (with-variant) The Basque language, spoken in the North of the Iberian Peninsula, happens to have morphologically overt focus marking for objects and an additional syntactic process of topicalization, which involves movement of the topic element to the right of a focalized element (Ortiz de Urbina 1989, Alcázar 2003). However, only deictic pronouns show overt focus marking in Basque, so I will look for sentences which show locative alternation in Basque and I will substitute the direct object NP by a deictic pronoun, which will be focalized and topicalized. The Basque verbs load and spray show locative alternation, so I have selected sentences containing these two verbs and applied the aforementioned tests. The relevant examples are included in (26), for the verb load, and in (27) for the verb spray. Consider the following: -X sentences, only marked with a letter without any apostrophe, include two NPs, where the first one functions as a direct object (DO) and the second one as a prepositional object (PO). 23 -X’ sentences have a deictic pronoun as DO, without any focus marker or topicalization process, plus an NP that functions as a PO. -X’’ show sentences with overt focus morphology in the DO argument, which is a deictic pronoun in absolutive case, plus an NP that functions as a PO. -X’’’ sentences include a deictic pronoun which has been topicalized to the beginning of the sentence, plus an NP that functions as a PO. - a sentences, including those marked with one or more apostrophes, take a direct object which bears the locatum θ-role, and have been claimed by Basilico (1998) to express a thetic judgment (expressed with –ga in Japanese). They represent the locative variant. - b sentences, including those marked with one or more apostrophes, take a DO with a location role, and are supposed to express a categorical judgment. They represent the with-variant. (26) a. Jon-ek kamioi-an patata-k karga-tu zituen. Jon-Erg.Sg truck-Loc.Sg patata-Abs.Pl load-Per have.3Sg_3Pl “John loaded potatoes into the truck” a’. Jon-ek kamioi-an haiek karga-tu zituen. Jon-Erg.Sg truck-Loc.Sg those.Abs.Pl load-Per have.3Sg_3Pl “John loaded those (ones) into the truck” 24 a’’. Jon-ek kamioi-an haie-xe-k karga-tu zituen. Jon-Erg.Sg truck-Loc.Sg those.Abs.Pl.Foc load-Per have.3Sg_3Pl “John loaded those (ones) into the truck” a’’’. ??Haiek Jon-ek kamioi-an karga-tu zituen. them.Abs.Pl(TOP) Jon-Erg.Sg truck-Loc.Sg load-Per have.3Sg_3Pl “John loaded those (ones) into the truck” b. Jon-ek kamioi-a patata-z karga-tu zuen. Jon-Erg.Sg truck-Abs.Sg patata-Inst.Pl load-Per have.3Sg_3Sg “John loaded the truck with potatoes” b’. Jon-ek hura patata-z karga-tu zuen. Jon-Erg.Sg that.Abs.Sg patata-Inst.Pl load-Per have.3Sg_3Sg “John loaded that (one) with potatoes” b’’. Jon-ek (patata-z) hura-xe karga-tu zuen (patata-z). Jon-Erg.Sg potato-Inst.Pl that.Abs.Sg.Foc load-Per have.3Sg_3Sg potato-Inst.Pl “John loaded that (one) with potatoes” b’’’. Hura Jon-ek patata-z karga-tu zuen. that.Abs.Sg (TOP) Jon-Erg.Sg potato-Inst.Pl load-Per have.3Sg_3Sg “John loaded that (one) with potatoes” 25 (27) a. Jon-ek pintura horma-n zipriztin-du zuen. Jon-Erg paint wall-Loc.Sg spray-Per have.3Sg_3Sg “John sprayed the paint on the wall” a’ Jon-ek hura horma-n zipriztin-du zuen. Jon-Erg that.Abs.Sg wall-Loc.Sg spray-Per have.3Sg_3Sg “John sprayed that (one) on the wall” a’’ Jon-ek (horman) hura-xe zipriztin-du zuen (horman). Jon-Erg wall-Loc.Sg that.Abs.Sg.Foc spray-Per have.3Sg_3Sg wall-Loc.Sg “John sprayed that (one) on the wall” a’’’ ??hura Jon-ek (horman) zipriztin-du zuen (horman). that.Abs.Sg (TOP) Jon-Erg wall-Loc.Sg spray-Per have.3Sg_3Sg wall-Loc.Sg “John sprayed that (one) on the wall” b. Jon-ek horma pintura-z zipriztin-du zuen. Jon-Erg wall-Abs.Sg paint.Instr.Mug 6 spray-Per have.3Sg_3Sg “John sprayed the wall with paint” 6 “Mug” stands for mugagabe, which is an unspecified number. 26 b’. Jon-ek hura pintura-z zipriztin-du zuen. Jon-Erg that.Abs.Sg paint.Instr.Mug spray-Per have.3Sg_3Sg “John sprayed that (one) with paint” b’’. Jon-ek (pinturaz) hura-xe zipriztin-du zuen (pinturaz). Jon-Erg paint.Instr.Mug that.Abs.Sg spray-Per have.3Sg_3Sg paint.Instr.Mug “John sprayed that (one) with paint” b’’’. hura Jon-ek pintura-z zipriztin-du zuen. that.Abs.Sg Jon-Erg paint.Instr.Mug spray-Per have.3Sg_3Sg “John sprayed that (one) with paint” The previous examples show that all the variants allow focusing of the DO. However, it is noticeable that only categorical alternants, the ones with the location realized as DO, accept topicalization. For some reason, the thetic alternants cannot be associated with topic marking. This could indicate that the thetic/categorical distinction can have influence in the structure of the locative alternating verbs. The a’’’ sentences are grammatically correct, which is shown through comparison with the b’’’ sentences, although it is harder for the native speakers of Basque to interpret a thetic alternant in a context where the object has been topicalized (recall that thetic statements are associated with focus, while categorical judgments are associated with topic). 27 This discussion might lead us to believe that the thetic / categorical distinction might have some influence on the verbs that enter the locative alternation, by restricting the existence of certain constructions. But to which extent can we postulate an underlying universal structure based on this distinction? As I mentioned before, it is possible to find languages with overt topic and focus markers, such as Basque, that show some difference between thetic and categorical structures. However, the influence of this phenomenon is so limited that it does not apply homogenously to all the verbs in a given language. In Basque, both categorical and thetic statements can be associated with a focused DO, which, in principle, sounds counterintuitive, since only the variants that imply a thetic judgment should allow for this possibility, if one accepts Basilico’s observations. Only topic marking, associated with categorical judgments by Kuroda 1972, seem to be degraded when associated with thetic statements. However, even in Basque, this only applies to the spray / load alternation. When we consider the intransitive types of locative alternation, like those displayed by arder “burn” or hervir “boil”, there is no difference between the thetic and categorical variants with respect to topic or focus. In fact, both topic and focus can be expressed by either thetic or categorical judgments (see examples below). (28) Topicalized Subject NP Argi-ak bere begi-etan distira-tzen zuen luz-Erg.sg su ojo-Loc.Pl brillar-Imp aux.pasado The light beams in his eyes. 28 (29) Structure with focused subject NP Argi-ak distira-tzen zuen bere begi-etan luz-Erg.sg brillar-Imp aux.pasado su ojo-Loc.Pl “The light beams in his eyes”. (30) Topicalized Subject NP Bere begi-ek argi-az distira-tzen zuten su ojo-Erg.pl luz-Instr brillar-Imp aux.pasado “His eyes beam with light” (31) Structure with focused subject NP Bere begi-ek distira-tzen zuten argi-az su ojo-Erg.pl brillar-Imp aux.pasado luz-Instr “His eyes beam with light” The same is true for Japanese, as shown in the examples in (32) and (33). As we saw before, Japanese has overt topic and focus markers, -wa and -ga respectively, for the subject argument. Following Basilico 1998, one might expect that the intransitive variants of the alternation, in which the subject alternates with a complement PP, will neither show a location subject (associated with categorical statements) with the focus marker –ga, which indicates a thetic statements, nor a locatum subject (associated with 29 thetic judgments) with the topic marker –wa, which indicates a thetic judgment. However this is not the case, since all the possibilities are available in Japanese. (32) a. hikari-wa kare-no me-no naka-de kagayai-ta. (talking about light) Light-subj he-gen eye-gen inside-in shine-pst A light beams in his eyes. b. hikari-ga kare-no me-no naka-de kagayai-ta. (emphasis on light) Light-subj he-gen eye-gen inside-in shine-pst A light beams in his eyes. (33) a. kare-no me-wa hikari-de kagayai-ta. (talking about the eye) He-gen eye-top light-inst shine-pst His eyes beam with a light. b. kare-no me-ga hikari-de kagayai-ta. (emphasis on the eye) He-gen eye-top light-inst shine-pst His eyes beam with a light. At this point, one might wonder if the contrast seen in the Basque examples in (26) and (27) is really due to the thetic / categorical distinction or to something else; for example, the idiosyncratic properties of the different case markers and prepositions. 30 More evidence against considering this difference as relevant for the grammar can be drawn from a detailed observation of the different types of locative alternation in Spanish, which I will introduce in detail in chapter 2. Following Basilico 1998, one might surmise that the fact that thetic judgments involve two XPs that are specifier and complement of the same projection would imply that, whenever one of them appears, the other one should be able to appear too, since they hang from the same XP. But this is not the case in Spanish, where, unlike in English, the appearance of a location PP is ungrammatical, or degraded, in certain subtypes of locative alternation. (34) a. El cielo se ha despejado de nubes The sky itself has cleared of clouds. The clouds cleared from the sky. b. Las nubes se han despejado (??del cielo). The clouds themselves have cleared from the sky. “The clouds cleared from the sky”. (35) a. Elena se alivió de la ansiedad que causa el despegue. Elena herself relieved from the anxiety that causes the take-off. “Elena relieved from the anxiety caused by the take off”. b. El ahogo que causa el despegue se alivió *de Elena. The anxiety that causes the take-off itself relieved from Elena. "The anxiety caused by the take-off was relieved”. 31 In both (34b) and (35b), which represent a thetic judgment, the overt expression of the location argument gives rise to ungrammaticality. This fact cannot be accounted for in terms of the distinction between thetic and categorical judgments as proposed by Basilico (1998). In section 1.2, I have described the analysis of the locative alternation proposed by Basilico 1998. After studying the data provided, I have concluded that the apparent syntactic asymmetries do not seem to be pervasive, and, therefore, should not be included as a universal property of the locative alternation. 1.3. The holistic effect The scholars that have studied the locative alternation have paid a great deal of attention to the properties of the holistic effect. Rappaport & Levin 1988 consider that the location argument receives an interpretation as fully affected by the action of the verb when it appears as a direct object. However, they mention that the locatum argument is never fully affected by the meaning of the verb. Thus, sentence (36a) should imply that the direct object is completely affected and the truck ends up full. By contrast, the direct object in (36b) is not necessarily affected by the verb and there might be remaining hay that was not loaded. (36) a. John loaded the truck with hay. b. John loaded hay into the truck. 32 This phenomenon is what has traditionally been called the holistic effect. However, not all the linguists that have studied this phenomenon agree with Rappaport & Levin’s 1988 intuition. Dowty 1991 and Moreno Cabrera 1998 maintain that the holistic effect affects the direct objects of both variants in the same way, since both are incremental themes in Dowty’s terms, and that the complete affectedness suggested by Rappaport & Levin is only derived from the use of different quantifiers and our knowledge of the world. Furthermore, Jackendoff 1990 and Di Tullio 2001 point out that not all alternating verbs display this effect, as in the case of verbs such as spray in English or rociar “spray” and salpicar “splash” in Spanish. In the same line as Dowty and Moreno Cabrera, Beavers 2006 systematically describes for the first time the affectedness properties of the two arguments involved in the locative alternation. He explains that the Direct Objects are always completely affected, while that is not the case for the oblique complements. However Beavers notices that several factors must be controlled to avoid interferences with other linguistic phenomena, specifically (a) the effects of the overall predicate and result state predicate modifiers, (b) the effects of context and conventionalization and (c) the effects of what Kiparsky 1998 refers to as quantitative indeterminacy (e.g. bare plural and mass noun objects). The following examples are extracted from Beavers 2006 (p. 48): 33 (37) a. John loaded the hay onto the wagon, but left some space for the grain. b. John loaded the hay onto the wagon, filling the wagon all up. c. #John loaded the wagon with the hay, but left some space for the grain. (38) a. John loaded the wagon with the hay, but left some hay to fill the truck. b. John loaded the wagon with the hay, moving every last straw. c. #John loaded the hay onto the wagon, but left some hay to fill the truck. Note that both internal arguments appear preceded by the, to avoid quantifier ambiguities. In all the examples provided the Direct Object needs to be fully affected, i.e., the wagon must be completely filled up, but not the oblique, since not all the hay needs to be loaded. Beavers also notices another property of the locative alternation: in the two variants, it is not possible “for the participant realized as an oblique to have been unchanged or unmoved in the event” (op. cit.: p. 48). However, I will redefine this last statement to fit also intransitive verbs such as beam or transitives such as exculpar “exculpate”, which alternate in Spanish (see chapters 2 and 3). The relevant generalization has to do with the internal aspect of the prepositions involved in the locative alternation: these prepositions require the theme to be in contact with the goal at some point, even if it is in an abstract way –as in the case of exculpar. This is shown in the following examples (op. cit.: p. 48): 34 (39) a. #John loaded the hay onto the wagon, but the wagon was empty afterwards. b. #John loaded the wagon with the hay, but none of the hay moved. Now, the total affectedness of the DO is not a specific property of the locative alternation, and non-alternating verbs such as put, fill, remove or cure also show this phenomenon, where the DO is completely affected, while the oblique need not be. Beavers provides the following examples (op. cit.: p.57-8): (40) a. The sheriff put the guns on the rack. b. *The sheriff put the rack with the gun. (41) a. The gardener filled the bucket with the water. b. *The gardener filled the water into the bucket. (42) a. The waiter removed the dishes from the table. b. *The waiter removed the table of the dishes. (43) a. The doctor cured John of the disease. b. *The doctor cured the disease from John. Beaver’s contribution to the knowledge about alternations lies on his description of the relationship between the DO and the oblique as far as their affectedness is 35 concerned. He proposes a scalar approach to measuring out where dynamic predicates are defined as a relation between an event, a theme, and a scale (ignoring other participants such as agents). On this approach, there is a homomorphism between the event and the scale such that a predicate that imposes a boundary on the scale indirectly imposes a boundary on the event and is therefore telic 7 . Thus the author assumes the presence of a scale argument for every dynamic predicate. Beaver’s analysis aims at explaining object oblique affectedness relationships, but it does not address the variation found within the different types of alternations or between the different subtypes of the locative alternation itself. He does not deal with the intransitive classes of the locative alternation either, such as clear intransitive or the swarm type. There is no explanation as to why some verbs alternate and others don’t. The evidence presented in this section shows that the holistic effect is not a property of the locative alternation, since it is present in other types of alternation and in non-alternating verbs (Beavers 2006). In this research, I am mainly concerned with the intrinsic properties of the locative alternation, and therefore there will be no further discussion about the holistic effect. 1.4. Two types of intransitive verbs: unaccusative and unergative This section serves as an introduction to one of the most central topics in my dissertation. Here I will briefly describe some crosslinguistic properties of these 7 This notion is similar to what Dowty 1991 calls Incremental Theme (IT). In the sentences John drunk a beer (IT) and John drank beer, only the direct object of the first one, a beer, is an IT, because there is homomorphism between the event and the direct object, so that as the event of drinking progresses, the amount of beer decreases accordingly. This is not true for the second sentence. 36 intransitive verbs to be able to differentiate them. I will also provide an introduction to unaccusativity in Spanish and the problems that arise when trying to distinguish between unaccusative and unergative verbs in this language. Perlmutter 1978 divided intransitive verbs into two subgroups: unaccusative and unergative. Both classes have distinct semantic and syntactic features which make them different from one another. Without going into much detail, Perlmutter associated unergative verbs with a notion of volition or causation, and unaccusative verbs with affected arguments. The differences existing between these two classes of verbs have explicit syntactic consequences. Specifically, this author provides examples to show that while unergative verbs can appear in the impersonal passive construction in Dutch (44), unaccusatives are not allowed (45). (44) Impersonal passive with unergative verb in Dutch. Er wordt hier veel geskied. “It is skied here a lot” (45) Active sentence (a) and impersonal passive (b) with unaccusative verb in Dutch a. Dat blok hout heeft goed gebrand. “That block of wood burned well” b. *Er werdt door dat blok hout goed gebrand. “By that block of wood it is burned well” 37 Later, Burzio 1981, 1986 described the properties of unaccusative verbs in Italian, and came to the realization that unaccusative and unergative verbs can be differentiated by using different syntactic tests: (46) 1. Unaccusative verbs select the auxiliary essere “be” for periphrastic forms. 2. Unergative verbs select the auxiliary avere “have” for periphrastic forms. 3. Only unaccusative verbs accept ne-cliticization. 4. The participle appearing in periphrastic forms of unaccusative verbs agrees with the subject, while there is no overt agreement with the subject of unergative verbs. Therefore, in languages that have two auxiliaries used to form periphrastic tenses, unaccusative verbs will select the auxiliary be, while unergative verbs will select have. The following sentences show examples from Dutch and Italian. Both languages use the auxiliary be, zijn and essere respectively, in unaccusative constructions, while unergative verbs select the auxiliary have, hebben and avere respectively. (47) a. Ik heb in totaal twee dagen gewerkt. (have) (Dutch) “I have worked for a total of 2 days” b. Ik ben gekomen. (be) “I have arrived” 38 (48) a. Giovanni è arrivato. (be) (Italian) “Giovanni has arrived” b. Giovanni ha telefonato. (have) “Giovanni has telephoned” The Italian partitive clitic ne can be used to substitute for the direct object of a transitive verb, as in (49), or the subject of an unaccusative verb, such as in (50). However, it cannot substitute for the subject of an unergative verb (51). This test seems to suggest that the subject of an unaccusative verb behaves in the same way as the object of a transitive verb, since the quantified object of a transitive verb can also be replaced by the pronominal form ne, but not its subject. (49) a. Giovanni inviterà molti esperti. “Giovanni will invite many experts” b. Giovanni ne inviterà molti. “Giovani will invite many (of-them)” (50) a. Arriverano molti esperti. “Many experts will arrive” b. Ne arriverano molti. “Many of them will arrive” 39 (51) a. Telefoneranno molti esperti. “Many experts will telephone” b. *Ne telefoneranno molti. “Many of them will telephone” In Italian, the past participle of periphrastic forms agrees with the subject in gender and number when the auxiliary be is used, i.e. with unaccusative verbs (52a). Unergative verbs, however, disallow this possibility (52b). (52) a. Maria è arrivata (Fem, sg) “Maria has arrived” b. Maria ha telefonato (Mas, sg) “Maria has telephoned” The literature dealing with unaccusativity is very extensive and it is beyond the scope of this paper to carry out a complete review of it. Nevertheless, it is crucial to provide a description of Spanish intransitive verbs in order to determine the right classification of the swarm type verbs, which becomes a crucial part of this dissertation in section 3.2 and chapter 4. As in Italian, there have been several tests aimed at differentiating unergatives from unaccusative verbs in Spanish. However, some of the most uncontroversial tests, 40 such as auxiliary selection, cannot be applied to Spanish, because only haber “have” can be used as an auxiliary in periphrastic forms. Mendikoetxea 1999 and the references therein are a good source for a detailed description of Spanish intransitive verbs. In section 1.4.1, I briefly present some of these tests and the problems that they present. 1.4.1. Unaccusativity in Spanish As discussed before, many languages make use of several morphological markings that have been associated with the unaccusative/unergative distinction. For example, unaccusative verbs select the auxiliary be for periphrastic forms, while unergatives select have in Romance languages such as Italian or French. Furthermore, Italian partitive clitic ne can only be used to substitute for the subjects of unaccusative verbs or the objects of transitives, but never for the subject of unergatives. Unlike Old Spanish, Spanish lacks overt morphological markings that could help us identify the different types of verbs, but Mendikoetxea 1999 provides a thorough description of Spanish unaccusative verbs. She divides them into two different groups: (i) verbs indicating a change of state or location, and (ii) verbs of existence and appearance. Change of state verbs in the first group are further divided into verbs of internal and external causation. She provides the lists included below. 41 (53) Change of state verbs with external causation a. Deadjectival verbs, without prefixes: secar “dry”, vaciar “empty”… b. Parasynthetic deadjectival verbs, with prefix a-: ablandar “soften”, alargar “lengthen”… c. Parasynthetic deadjectival verbs, with prefix en-: ensuciar “soil”, enfriar “cool” d. Denominal verbs, with prefixes a- / en-: arrodillar “kneel”, enroscar “curl”… e. Verbs ending in –ificar: purificar “purify”, solidificar “solidify”… f. Verbs ending in –izar: cristalizar “crystallize”, fosilizar “fossilize”… (54) Change of state verbs with internal causation a. Deadjectival verbs without prefix: mejorar “improve”, oscurecer “darken”… b. Parasynthetic deadjectival verbs, with prefix a-: adelgazar “grow thin”, aclarar “clear”… c. Parasynthetic deadjectival verbs, with prefix en-: ensordecer “go deaf”, envejecer “get old”… d. Other: encoger “shrink”, hervir “boil”, arder “burn”… (55) Verbs of existence and appearance a. Verbs of existence: existir “exist”, vivir “live”, permanecer “stay”… b. Verbs of absence or lack: faltar “lack”, escasear “be scarce”… c. Non-pronominal verbs of appearance: aparecer “appear”, resultar “result”… 42 d. Pronominal verbs of appearance: presentarse “present oneself”, manifestarse “manifest”, originarse “originate”… e. Verbs of disappearance: desaparecer “disappear”, morir “die”… f. Verbs of happening: ocurrir “occur”, pasar “happen”, suceder “happen”… Mendikoetxea also presents a list of tests that have been proposed to differentiate between unergative and unaccusative verbs. From a semantic point of view, she mentions that unaccusative verbs have affected subjects, while the subjects of unergative verbs are agentive. This is the traditional view adopted in the literature (Dowty 1991, Levin & Rappaport 1995). From a morphological point of view, it is necessary to note that Spanish only uses the auxiliary haber “have” to form periphrastic forms, which makes the auxiliary selection test impossible to apply (Sorace 2000). Spanish also lacks clitics such as the Italian partitive ne, which have been associated with direct objects and unaccusative subjects (Burzio 1981). Because Spanish lacks any overt morphological marking, there have been several syntactic tests used to distinguish between the two types of intransitives. 1. Unaccusative verbs, but not unergative, can head absolute participial clauses: (56) a. Muerto el perro, se acabó la rabia “Once the dog was dead, rabies was over” b. *Trabajado el profesor, se fue a dormir. “Once the teacher was worked, he went to sleep” 43 The previous examples show that while the unaccusative verb morir “die” can head an absolute participial clause, a typically unergative verb like trabajar “work” cannot. 2. The clitic se appears in the pronominal (unaccusative) variant of the causative alternation, so its presence can be associated with an unaccusative verb. (57) a. Causative variant: Pedro rompió la ventana. “Peter broke the window” b. Unaccusative variant: La ventana se rompió. “the window broke” 3. Unaccusative verbs allow for postverbal subjects with no determiner, unlike unergatives. (58) a. Existen problemas. “there exist problems” b. ?Duermen mujeres. “there sleep women” 44 The previous sentences show that only an unaccusative verb like existir “exist”, can appear with bare postverbal subjects, unlike unergative verbs, such as dormir “sleep”. Unfortunately, due to the lack of homogeneity within the unaccusative class, different verbs subgroups may react differently to the tests provided above. In fact, these tests can sometimes make the wrong predictions, as the following sentences show. 1. Some unaccusative verbs cannot appear in an absolute participial construction: (59) ?Existido el problema,… “once the problem was existed,…” 2. Not all unaccusative verbs participate in the causative alternation: (60) a. Los árboles florecen en primavera. “trees bloom in spring” b. *Juan florece los árboles. “John blooms the trees” 3. Some verbs that participate in the causative alternation do not select the clitic se: 45 (61) a. Juan hierve el agua. “John boils the water” b. El agua hierve. “The water is boiling” 4. Under certain conditions, as in the case of locative inversion 8 , unergative verbs can appear with bare postverbal subjects: (62) a. Aquí trabajan mujeres. “Here work women” (locative inversion) b. Me gusta el programa Menudas Estrellas porque cantan niños. “I like the program “Menudas Estrellas” because children sing” These data might apparently suggest that there is no reliable test for unaccusativity in modern Spanish. This is especially problematic for those groups of verbs whose behavior varies crosslinguistically. In this group, we should include the swarm subtype, which I will describe as unaccusative in this dissertation, as well as those placed by Sorace (2000) in the middle of her auxiliary selection hierarchy. In particular, in this dissertation I contend that all the verbs that enter the locative alternation share an underlying unaccusative construction, as I explain in chapter 3. 8 Note that Torrego 1989 proposes that when a typically unergative verb appears in a sentence with locative inversion, such as en la torre anidan cigüeñas “storks make their nests on the tower”, the construction will have unaccusative properties. 46 Chapter 2 The locative alternation in Spanish Chapter 2 is dedicated to the study of the locative alternation in Spanish. In section 2.1, I present my classification of the Spanish subtypes, based on data extracted from the online Corpus de Referencia del Español Actual (CREA). Section 2.2 introduces some syntactic accounts of the locative alternation, paying special attention to Mateu’s 2002 account. Section 2.3 describes some semantic properties of alternating verbs. 2.1. Types of locative alternation in Spanish Levin 1993 explains that the locative alternation “is found with verbs that relate to putting substances on surfaces or things in containers, or to removing substances from surfaces or things from containers”. The alternation affects the location and locatum (substance or entity whose location is changed) arguments, which alternate as in Juan cargó el coche con su equipaje (John loaded the car with his luggage) / Juan cargó su equipaje en el coche (John loaded his luggage in the car), where el coche “the car” is the location and su equipaje “his luggage” the locatum. There can also be alternation between the subject and a prepositional object in intransitive verbs, as we saw for English in sections 1.1.4 and 1.1.5. However, she proposes five simple (undivided) different types of locative alternating verbs for English, see previous chapter, but I posit the existence of four complex types for Spanish, where every class is divided in subgroups. 47 2.1.1. Verbs of the type of cargar “load”, empacar “pack” and plantar “plant” These verbs alternate in the following way Juan cargó el camión con/de naranjas versus Juan cargó naranjas en el camión (see translation below). In this type, the arguments (direct and prepositional object) change their syntactic configuration, but they do not change their θ-role. The object of the first variant and the PO (Prepositional Object) of the second one are location, while the other argument is locatum. I consider POs to be PPs with argument properties. This type is subdivided into three different groups, depending on the possibility for the locatum argument to be headed by the prepositions con “with” (ii), de “of” (iii) or both (i). (i) rociar “spray”/ cargar “load” verbs. Their locatum can be preceded by both con “with” and de “of”. (63) a. Juan cargó el camión con /de naranjas. John loaded the truck with /of oranges. b. Juan cargó naranjas en el camión. John loaded oranges in the truck. (ii) empacar “pack” verbs. These verbs only accept the preposition con when the locatum is a Prepositional Object (PO). 48 (64) a. Empacó sus pertenencias en cajas. Packed-3ps his/her belongings in boxes. b. empacó una maleta con la ropa de Gertrudis. packed-3ps a suitcase with the clothes of Gertrudis. (iii) plantar “plant” verbs. This is a very small group, formed by verbs whose locatum argument can only be headed by the preposition de 9 . (65) a. Plantaron pinos en el monte. planted-3pp pine trees in the mountain b. Plantaron el monte de pinos. Planted-3pp the mountain of pine trees It is obvious that this is not a homogeneous group. There also seems to be a difference in meaning between the preposition de and con, which will be addressed in chapter 3. 9 Some speakers accept also the locatum preposition con “with” in the plantar verbs, which might suggest the elimination of this subtype and its inclusion within the cargar/rociar group. Because I created my classification based on data extracted from CREA, and I did not find any examples containing the verb plantar with the locatum preposition “con” I will keep the current classification, although this issue will deserve further attention in future research. 49 2.1.2. Verbs of the type of barrer “sweep”, disculpar “forgive” and recoger “put in order” These verbs show an alternation similar to the one displayed by clear and wipe in English, except for the disculpar verbs. Their alternation shows the following features: (i) barrer “sweep” verbs(“location de locatum” / ”locatum de location”). This class is equivalent to the English clear type. (66) a. El viento barrió de cenizas nuestras amadas calles. the wind swept of ashes our beloved streets. “The wind swept the ashes from our beloved streets”. b. El viento barrió de nuestras amadas calles las cenizas. the wind swept of our beloved streets the ashes. “The wind swept the ashes from our beloved streets”. (ii) disculpar “forgive” verbs (“location de locatum” / “locatum (de location)”). When preceded by the preposition de “of”, the location argument can only be interpreted as a complement of the noun (CN), never as a PO. The sentence in (67c) is ungrammatical because when the PP headed by de “of” is not preceded by an NP, it needs to be interpreted as an independent PP, as in sentence (66b), which is impossible in this group. 50 (67) a. disculpó a Felipe de sus horrendos crímenes. he/she excused to Felipe from his horrendous crimes. b. disculpó los horrendos crímenes de Felipe. he/she excused the horrendous crimes of Felipe. c. *disculpó de Felipe los horrendos crímenes. he/she excused of Felipe the horrendous crimes. (iii) recoger “put in order” verbs (“location (*de locatum)” / “locatum de location”): This class is similar to the English class of wipe verbs. The locatum argument can never appear with the preposition de, not even as a CN. (68) a. recogieron el cuarto (*de libros). They put in order the room (of books). b. recogieron de su cuarto los libros. They put in order from his room the books. In this section, I have proved the existence of a new subclass: the disculpar “forgive” type. This group is formed by many members, and it has characteristics that are in between class (i) and (iii). 51 2.1.3 Verbs of the type of arder “burn” / resonar “resound”/ pulular “swarm” This class is formed by intransitive verbs that can alternate the position of their Subject and PO arguments. This type is equivalent to the swarm class in English, which for the most part, is formed by emission verbs, such as shine or resound. It is subdivided into three groups, depending on the number of prepositions that can head the locatum argument when it functions as a PO: only con (ii), only de (iii), or both prepositions (i). (i) Arder “burn” verbs. The verbs included here show the following alternation: locatum en location / location con/de locatum. These verbs are atelic and durative, not indicating a terminative state. This group is formed by verbs of light emission, except for hervir “boil”, which can display different types of alternation: locative and causative (Pedro hirvió el agua “Pedro boiled the water”. / El agua hierve “water boils”). (69) a. Su Habano es luz que arde en el mar profundo. His Habano is light that burns in the see deep. b. La zona portuaria parecía arder con la luz de la ciudad. The zone harbor seemed to burn with the light of the city. c. La zona portuaria parecía arder de luz. The zone harbor seemed to burn of light. (ii) Resonar “resound” verbs. This class shows a structure similar to the previous one, but its locatum argument does not accept the preposition de. 52 (70) a. Los ecos de Bahía de Cochinos resonaban en las paredes. The echoes from Bahía de Cochinos echoed in the walls. b. Todas las mezquitas resonaron con exhortaciones a la unión nacional. All the mosques echoed with exhortations to the union national. (iii) pulular “swarm” verbs. This group is reduced to a small number of verbs, as we saw with the plantar verbs. Their locatum argument can only be headed by the preposition de “of” when it functions as a PO 10 . (71) a. esas criaturas que pululaban en el ministerio. Those creatures swarmed in the ministry. b. Las playas pululan de gente. The beaches swarm of people. The intransitive versions of the cargar “load” / rociar “spray” group are also included within this group. Verbs such as cargar are highly agentive, so their intransitive version usually has metaphoric meaning, as in sentence (72), extracted from the corpus CREA (Corpus de Referencia del Español Actual). 10 As in the case of the plantar verbs, the existence of speakers who accept these verbs with the locatum preposition con “with” might point towards the elimination of this class. 53 (72) a. La administración pública mexicana no es sino el resultado de un traslape the administration public Mexican not is but the result of an overlap de culturas y gobiernos que parece que se carga en la piel of cultures and governments that seem that CLITIC loads in the skin “The Mexican public administration is only the result of an overlap of cultures and governments that seems to be loaded on the skin”. b. y el aire se carga de sospechas and the air CLITIC loads of suspicions “and the air is loaded with suspicions” The sentence in (72a) represents the locative variant, while (72b) is the with variant with the preposition de “of”. All the verbs in this section are associated with either one or two locatum prepositions (con “with”, de “of” or both). 2.1.4. Verbs of the type of despejar “clear”/borrar “erase” This type is related to the English clear (intransitive) verbs, and is formed by intransitive verbs. However, the Spanish verbs included in this group show a structure that is different from English. In English, these verbs show the structure location (of locatum) / locatum from location, while the Spanish despejar verbs whose meaning 54 corresponds to the English clear display the following alternation location de locatum / locatum (de location) 11 . a. Despejar “clear” verbs. The meaning of these verbs is similar to the one of the clear (intransitive) verbs in English, but their structure is different, since the Spanish ones admit a locatum argument preceded by de., while English cannot have an of variant. (73) a. El cielo se ha despejado de nubes. The sky itself has cleared of clouds. b. Las nubes se han despejado (?del cielo). The clouds themselves have cleared from the sky. b. Borrar “erase” verbs. These verbs display a structure that is completely equivalent to that of the English clear (intransitive) verbs, but their meaning does not seem to be related. (74) a. No soporta que su memoria se borre (?de información). not stands that his memory itself erases (of information). b. La información se borró de la memoria del ordenador. The information itself erased from the memory of-the computer. 11 The brackets indicate that the argument included in them cannot be syntactically realized, or, at least, its inclusion is not accepted by some speakers. 55 In this section, I have shown that, even though Spanish shows an alternation structurally similar to the English clear intransitive verbs, the borrar “erase” type, their meanings do not correlate at all. By contrast, the Spanish verbs of the despejar “clear” type, which have equivalent meanings, display a different structure. 2.2. Syntactic analyses of the locative alternation in Spanish In section 2.1, I have divided the verbs that display locative alternation in Spanish into four complex groups, based on Levin 1993. In this section, I present Mateu’s 2002 analysis of the locative alternation because it explains different properties of this construction. 2.2.1. Mateu 2002: Argument Structure: relational construal at the syntax-semantics interface In his dissertation, Mateu proposes a system hat relies on the properties of the construction to derive different meanings, but it also needs the appearance of certain lexical-semantic features and, occasionally, some information about θ-roles. 2.2.1.1. The system In his system, he proposes three basic argument structure types (75). The lexical head x in (75a) is associated with an eventive relation, while the lexical head in (75b) represents a non-eventive relation, and (75c) is interpreted as a non-relational element. 56 (75) a. x gi x y b. x gi z x gi x y c. x The eventive relation (75a) can express two different semantic relations, depending on the syntax: (i) Source relation, when it is associated with a Functional projection that contains an external argument. It appears in transitive (76) and unergative constructions (77). (ii) Transitional relation, when there is no external argument. It appears in unaccusative constructions (78). Given this, he provides the following representation for transitive (76), unergative (77) and unaccusative (78) sentences respectively: 57 (76) Transitive structure F gi z 1 F gi F x 1 gi x 1 x 2 gi z 2 x 2 gi x 2 y 2 (77) Unergative structure F gi z 1 F gi F x 1 gi x 1 y 1 (78) Unaccusative structure x 1 gi x 1 x 2 gi z 2 x 2 gi x 2 y 2 There are properties that cannot be derived from the configurational disposition of elements in the syntactic structure. These are semantic properties that are encoded as binary features, represented in (79). (79) a. [+R]: positive semantic value associated to the source relation b. [-R]: negative semantic value associated to the source relation c. [+T]: positive semantic value associated to the transitional relation d. [-T]: negative semantic value associated to the transitional relation 58 e. [+r]: positive semantic value associated to the non-eventive relation f. [-r]: negative semantic value associated to the non-eventive relation (Op.cit.: 33) “The [+R] feature subsumes both the CAUSE function and the agentive {ACT/DO} function, while the [-R] feature subsumes the HAVE function (cf. transitive stative verbs like fear) and whatever function is assigned to non-agentive unergative verbs (e.g., verbs of smell emission).”… “Finally, [+T] and [-T] subsume the {GO/BECOME/CHANGE} and {BE/STAY} functions, respectively.”(op. cit.: 33) The [+r] and [-r] features are correlated to Hale and Keyser’s (1993) “terminal coincidence relation” (exemplified by to, out of and off of) and “central coincidence relation”, (exemplified by at, in and with) respectively. For example, the sentence in (77a) would have the representation in (77b). (80) a. John sent Peter to prison. b. [ F John [x 1 [+R] [x 2 Peter [x 2 [+r] prison]]]] 2.2.1.2. The locative alternation Mateu departs from the assumption that the locative variant (the one with “in”) expresses a change of location, while the “with” variant expresses a change of state. Both are instances of [V + Small Clause]: 59 (81) a. …Verb [ SC NP material PP locative ] b. …Verb [ SC NP locative A] (PP material ) c. ….Verb [SC NP Pred] In this theory, the PP headed by with is an adjunct, and not an argument. The A in (81b) is a function containing the end point of a change of state. In order to provide a descriptive account of this structure, the author proposes the tree in (82), which is a pruned version of the one in (83). (82) x 1 ei x 1 x 2 ! eg [+R] z 2 x 2 ! gi FIGURE x 2 x 3 ! gi [+r] x 3 y 3 ! ! [-r] GROUND (83) x 1 ei x 1 x 2 ! eg [+R] z 2 x 2 ! gi FIGURE x 2 x 3 ! gi [+r] z 3 x 3 ! gi FIGURE x 3 y 3 ! ! [-r] GROUND In (83), a causative verb subcategorizes for a birelational Path element headed by a terminal coincidence relation (recall that Terminal Coincidence Relation is exemplified by prepositions like to). The lower x 3 expresses a central coincidence relation, exemplified by prepositions like in or with. Only those verbs whose GROUND can be interpreted as the end point of a change of location or state will show locative alternation. In (84), I have included an example of verb whose GROUND expresses the physical end point of a change of location, while (85) shows a verb whose GROUND can only be interpreted as the abstract end point of a change of state, which would be equivalent to [LLENO/FULL] in the case of llenar/fill. (84) a. Juan echó las colillas en el suelo. Juan threw-out the stubs on the floor. x 1 ei x 1 x 2 ! ei [+R] z 2 x 2 ! ! gi echar las colillas x 2 x 3 ! gi [+r] x 3 y 3 ! ! [-r] ! en el suelo 60 (85) a. Juan llenó el depósito {con/de} agua. Juan filled the tank {with/of} water. x 1 ei x 1 x 2 ! ei [+R] z 2 x 2 ! ! gi llenar el depósito x 2 x 3 ! gi [+r] x 3 y 3 ! [-r] 2.2.1.3. A different type of locative alternation. Verbs involving Manner Mateu noticed that locative alternation is much more productive in Germanic than in Romance. See the examples in (86) and (87) for English and Spanish respectively. (86) a. Joe rubbed the fingerprints off the crystal ball. b. Joe rubbed the crystal ball. (87) a. *Joe frotó las huellas fuera-de la bola de cristal. Joe rubbed the fingerprints off the ball of crystal. a’. Joe quitó las huellas de la bola i frotándola i . Joe got+out the fingerprints from the ball rubbing-it b. Joe frotó la bola de cristal. 61 Talmy 1985, 1991, 2000 classified languages depending on how different semantic components are conflated into the verb. This author explained that Romance languages incorporate Motion with Path, while Germanic languages incorporate Motion with Manner. Adopting Talmy’s generalization, Mateu explains the difference between Spanish and English observed in (86) and (87) by assuming that in English, but not in Spanish, the PP headed by off does not saturate the phonologically null matrix of the main eventive head (see (88b)). So, in order to avoid problems at PF it is necessary to conflate an object indicating manner (see (88a)) into the null main eventive head. This subordinate object will be adjoined to the null eventive head, as in (89). (88) a. x4 eo x4 y4 ! ! [+R] ! [ ∅] RUB- 62 b. x 1 ei x 1 x 2 ! ei [+R] z 2 x 2 ! ! gi [ ∅] the fingerprints x 2 x 3 ! gi [+r] x 3 y 3 off ! ! [-r] the ball (89) x 1 ei x 1 x 2 qy ei x4 x1 z 2 x 2 gi ! ! gi x4 y4 [+R] the fingerprints x 2 x 3 ! ! gi [+R] [+r] x 3 y 3 rub ! ! [-r] the ball However, in verb-framed languages 12 like Spanish the directional element is lexically conflated into the main eventive head, saturating it, and this is the only possibility. The structure that we have in Spanish is given in (90), which corresponds to the Spanish sentence in (87b) Joe frotó la bola de cristal, and to the English one in (86b) 63 12 The division between verb-framed and satellite-framed languages is due to Talmy 1985, 1991, 2000. Joe rubbed the ball. Both sentences could be paraphrased by Joe provided the ball with a rub. (90) x 1 ei x 1 x 2 [+R] gi rub/frotar z 2 x 2 ! gi ball/bola x 2 y 2 [-r] He also mentions in a footnote that, even though some people have observed that verbs like limpiar, barrer or fregar in Spanish admit the alternation, he claims that there is a very fundamental difference between English and Spanish: in English the sentence *John swept the crumbs is ungrammatical (he claims), but in Spanish it is acceptable. As far as I know, this comment is wrong, after checking with native speakers of English and Spanish who disagree. Furthermore, consider the following sentences where the Spanish verb frotar shows locative alternation: (91) frotar: Variant with en: a. “…les frotó en una mano esas hojas y en la otra algunas de mi tabaco…” [1997. Granma Internacional, 12/1997, núm. 49 : “Vegas Robaina País: ...”. Cuba]. …he rubbed them on a hand those leaves and on the other some of my tobacco… 64 65 Variant with con: b. Les frotó una mano con esas hojas. He rubbed them one hand with those leaves. It is obvious that the verb frotar “rub” in Spanish participates in the locative alternation, together with many other verbs of its class. A theory that excludes them fails to explain the facts. Another problem that Mateu’s analysis faces is that he considers the PP in the locative variant to be argumental, while he argues that the one in the with-variant behaves as an adjunct. Citing Mulder (1992), he mentions that the adjunct nature of the with complement can be overtly seen in Dutch, since it can be extraposed and omitted (92a), and can be clefted (92b). (92) a. dat hij de tuin beplant (met tulpen). that he the garden BE-plants (with tulips). b. hij beplant de tuin en doet dat met tulpen. he BE-plants the garden and does that with tulips. However, Mateu seems to ignore that Spanish verbs like plantar (plant) seem to behave differently from their Dutch counterpart, since the appearance of a with-phrase is obligatory in Spanish. These Spanish verbs need to express the with-phrase overtly, as seen in (93b) and (93c), although it is not obligatory to include the locative PP in the 66 locative variant (93a). Because verbs like plantar (plant) need the presence of the with- phrase, sentence (93d), which is the Spanish translation of (92b), is ungrammatical too. (93) a. Juan plantó tres pinos (en el monte). John planted three pines in the mountain. b. *Juan plantó el monte. John planted the garden. c. Juan plantó el monte de pinos. Juan planted the mountain of pines. d. *él planta el jardín y lo hace de tulipanes. he plants the garden and itACC. does of tulips. An analysis that classifies the PPs in the locative and with-variants as argumental and adjunct respectively would need to elaborate on the reasons why Spanish verbs such as plantar (plant) require the presence of the locatum PP. In the following section, I show that these PPs are always conceptually obligatory and, therefore, should be analyzed as arguments of the construction. 2.3. The semantic components of the locative alternation in Spanish This section deals with idiosyncratic semantic features that characterize this alternation. The first thing that needs to be pointed out is that the two variants of the locative alternation have a slightly different meaning, as we have seen earlier (Mateu 67 2002, Levin 1993). The locative variant indicates mainly a change of location (94a), while the with variant indicates a change of state (94b). For example, (94a) focuses on the movement of hay into the truck, while (94b) focuses on the loading of the truck. Beavers 2006 relates this fact to the special relation existing between the verb and the direct object. This difference is the reason why some languages use two different verbs or structures to express what English or Spanish could convey using the same verb. We might think that one of the reasons that could enable the existence of alternations is that the speaker, or speech community, considers that two different events can be associated somehow by means of implications. For example, one might think that whenever something is displaced and taken into a certain container there are two different events going on: a change of state, where a container is getting full, and a change of location, where something is being moved. Some languages might overtly show this perceptual relationship by using the same lexical item for both events, while others might prefer to just center their attention on either the change of state or change of location as separate events. For example, while the verb load alternates in English and Spanish, this verb does not alternate in Japanese. (94) a. John loaded hay into the truck. CHANGE OF LOCATION b. John loaded the truck with hay. CHANGE OF STATE But what are the basic or primitive concepts that a speaker or community will consider as relevant when creating words? The literature on language acquisition can help 68 us in this respect. Smiley and Huttenlocher (1995) show that conceptual development can operate independently from input and that there are certain meaning primitives that seem to be relevant for children from very early ages, such as the concept of possession or the concept of change, which are considered central by children. More research on acquisition should shed some light on which are these relevant meanings. At this point, and based on the existing literature on the lexicon and language acquisition, I will assume that the notions of change of state and change of location (or the distinction between locations, states and their ability to be interpreted as scales) are linguistic primitives and relevant for our conceptualization of the world. As commented before, even if (94a) represents basically a change of location, in which the hay ends up in the truck, it is also possible to imply that the truck became loaded, even if not completely, in the process of putting things in it. The same happens with the sentence in (94b), which expresses basically a change of state, but implies that something, the load, had to move from one place to another in order to end up in the truck. So (94b) implies also a change of state. In languages like Spanish, which allows for the same kind of alternation shown in (94), normally known as the spray/load alternation, the basic change of state or change of location interpretation is neither indicated by the verb, which by itself could be associated to both concepts, as (95) shows, nor by constituent ordering, since it is very free in Spanish, as (96) demonstrates. 69 (95) a. Juan cargó en el camión el heno sobrante. John loaded in the truck the hay remaining. John loaded the remaining hay in the truck. b. Juan cargó el camión con el heno sobrante. John loaded the truck with the hay remaining. John loaded the truck with the remaining hay. (96) a. Juan cargó en el camión el heno sobrante. John loaded the remaining hay in the truck. b. Juan cargó el heno sobrante en el camión. John loaded the remaining hay in the truck. In fact, the only overt difference between (95a) and (95b) is the use of a different preposition 13 and a different argument structure. In this sense, the preposition provides the construction with its meaning and argument structure. The locative variant will focus on the locative meaning because of the presence of the locative preposition en “in”, while the with variant is rather associated with a notion of change of state or abstract transfer or possession, as I will discuss in chapter 3. 13 Of course, from a syntactic point of view we could consider that both (92a) and (92a) are derived from a process of topicalization or scrambling of the PP. This might imply slight meaning differences related to the organization of information in the sentence. However, note that those two sentences could express the unmarked position if the PP was lighter than the NP, due to the effect of weight (see Mayoral Hernández 2004b for further discussion on weight effects). 70 If we take into consideration that the verb cargar “load” can take as its direct object either a location or a locatum argument, we will have to assume one of the following possibilities: (97) Alternating verbs are able to assign different θ-roles. (98) The theta role is assigned by the construction. Considering that the only surface difference between the two variants of the alternation is the preposition, it would be logical to derive that the presence of one preposition or another will determine the interpretation of the sentence, and therefore, its argument structure. Another interesting fact that leads us to believe that the preposition plays a central part in the locative alternation is the fact that even when it is omitted, it is implicitly present. (99) a. Juan cargó el camión. John loaded the truck. b. Juan cargó el heno. John loaded the hay. (100) #John loaded the truck without adding any load. 71 The example in (99a) implies that something was loaded in the truck, since otherwise the notion of loading does not make sense, which is why (100) becomes deviant. The same happens in (99b), where it is implied that the hay was loaded in some container. This suggests again that the prepositional phrase enjoys an argumental status, since it is conceptually necessary (see also Beavers 2006). Nevertheless, it is also possible to find constructions in which the direct object is missing, like in (101), where the PP becomes the only overt argument. (101) Juan cargó con las cajas. John loaded with the boxes. “John carried the boxes” The sentence in (101) implies that John took the boxes and carried them around, but there is not a location argument implied necessarily. Constructions with verbs that take only a PP as a complement are not common at all in Spanish, compared to the abundance of phrasal and prepositional verbs in English. The central role that PPs play in the locative alternation, as seen above, and the facts described in (101) can reasonably lead us to the belief that the PPs in both variants are the ones that assign the θ-role, –and hence the argument structure– to the constructions with locative alternation, and that they have argumental status. We have established that both change of state and change of location are central and necessary concepts in the locative alternation, and that they are inalienably associated 72 to the presence of a preposition. But there are other meaningful features associated with this alternation in Spanish. The literature on this subject mentions that there are verbs that only have one of the two variants of the alternation, but it is not obvious why. These verbs might typically participate in the locative alternation in some languages but not in others: (102) a. John filled the bathtub with water. b. Juan llenó la bañera con agua. c. *John filled water in the bathtub. d. *Juan llenó agua en la bañera. (103) a. John removed the stains from the fabric. b. Juan quitó las manchas de la tela. c. *John removed the fabric of stains. d. *Juan quitó la tela de manchas. (104) a. John removed the cup from the table. b. Juan quitó el vaso de la mesa. c. *John removed the table of the cup. d. *Juan quitó la mesa del vaso. 73 In (102) the verbs fill and llenar focus mainly on a change of state, while the verbs quitar and remove focus on a change of location. However, it would be possible to imply that there is also a change of location in (102) and a change of state in (103), since the water will have to be moved into the bathtub and the fabric will look clean after removing the stains. Nevertheless, the change of state is not so clear in (104), since there might still be other objects on the table. It might be the case that these non-alternating verbs can only express either change of state (fill) or change of location (remove) by themselves, while the other meaning has to be derived from the context and our knowledge of the world. At this point, we might conclude that the grammar of English and Spanish only gives these verbs either a change of state or a change of location meaning. Still we can go deeper in our intent to determine if there are some features that can predict if a particular verb will alternate or not in Spanish. Let’s observe figure 1, which represents a set of verbs that participate in the locative alternation. In this figure, it is possible to observe that there are verbs with a very broad meaning, like llenar “fill” or poner “put”, which do not alternate. There are also verbs that form a subset with respect to the former ones and that show locative alternation. For example, cargar “load” means “to fill something through a process of loading”, while rociar “spray” means “to put something somewhere through a process of spraying”. It would seem that only those verbs with a richer meaning, i.e. the verbs that not only express a pure change of state or location, but a specific way in which that process takes place, can participate in the locative alternation in Spanish. In this sense, it seems natural to assume that the extra meaning that alternating verbs exhibit allows them to modify the pure change of state or location process, and in doing so, the extra meaning component endows those lexically rich verbs with the ability to display locative alternation (figure 1). Figure 1: Spray/load PONER “PUT” asperjar “spray” LLENAR “FILL” cargar “load” empacar “pack” imbuir “imbue” plantar “plant” pintar “paint” rociar “spray” espolvorear “dust” This same generalization can be extended to all the subtypes of locative alternation described in section 2. Figure 2 shows that the superordinate verb quitar “remove” does not alternate, but the lexically richer verbs included in its set do alternate. Something similar can be observed in the verbs that express light and sound emission, represented in figure 3. The verb lucir “light” expresses pure light emission, but the verbs included in its set add some extra meaning component. For example, the verb brillar “shine” implies a higher intensity of light or a process of twinkling, while arder “burn” implies that the light emission resembles the one in a burning event, with violent changes in intensity and direction of light. The verbs in the sonar “sound” set differ from their superordinate non alternating verb in that they do not indicate pure sound emission, but they add a specific way in which that sound emission takes place. For example, the verb 74 resonar “resound” entails that the sound resounds on a surface repeatedly, as in a tunnel. In all the figures represented here, the verbs in capitals do not display locative alternation, since their meaning is very weak, almost like a light verb, and therefore they are the superordinate verbs. Figure 2: barrer “sweep”, disculpar “forgive” and recoger “put in order” QUITAR “REMOVE” limpiar “clean” borrar “erase” barrer” sweep” podar “prune” disculpar “forgive” recoger “put in order” Nevertheless, it is also necessary to give an account of verbs of the kind of adornar “adorn”, which do not alternate in Spanish or English, although their meaning is richer than that of the superordinate verb llenar “fill”. In fact, this might help us to determine the nature of that richer meaning that is characteristic to the alternating verbs. Demonte 1991 observed that what enables locative alternation is not only the conceptual composition of verbs, but also their aktionsart properties. Demonte notes that verbs focusing on the process, like cargar “load”, enter into the locative alternation, while verbs focusing on the beginning of the process, like echar “ ≈ throw, drop”, or verbs that express a pure effect, like adornar “adorn”, cannot enter into the locative alternation. 75 Figure 3: arder “burn” resonar “resound” 76 LUCIR “LIGHT” SONAR “SOUND” arder “burn” resonar “resound” brillar “beam” centellear “sparkle” retumbar “reverberate” chispear “sparkle” tronar “thunder” A combination of all the features observed in this section, i.e. change of state plus change of location plus rich meaning (or manner) plus aktionsart (process), is able to constrain the system enough as to include only those verbs that indeed show alternation. However, its lack of specificity with respect to the definition of rich meaning leaves the door open for variation, which is a desired result. Basically, the only way to exclude verbs like llenar “fill” is to think, as we mentioned before, that they express a mere change of state. However, if one would endow this verb with richer lexical properties and imply that there is a process of filling something using specific procedures, we should expect such a verb to alternate. 77 (105) a. “...rápidamente llenó en la computadora los datos requeridos”. quickly filled(3.p.s.) in the computer the data required. He/she quickly entered the data in the computer. [CREA: 1997 Tomás A. Cabal. El Siglo, 16/06/1997 : “Verdades Ocultas”. Panamá]. In (105), the process of filling is extremely specific, since it involves a process (aktionsart) of typing (manner). This sentence would not be acceptable for many speakers, but others consider it grammatical. Those speakers that do not consider the process of typing as being derived from the verb meaning cannot accept this construction. However, the speakers that consider the process of typing as part of the process of filling accept the alternation. Considering the wide crosslinguistic variation regarding which verbs can alternate or not, it is necessary to imply that languages may have slightly different constraints that will either increase or decrease the number of alternating verbs. 78 Chapter 3 An unaccusative account of the locative alternation In previous chapters, I have presented several accounts that try to explain the locative alternation construction. While doing so, I have separated the features that seem to be universal from those that are only language specific. In this chapter, I develop a syntactic account of this alternation that is based on Zubizarreta & Oh’s 2007 description of unaccusative directed motion constructions. 3.1. The unaccusative hypothesis This section provides an introduction to Zubizarreta & Oh’s 2007 analysis of unaccusative constructions, based on the system developed by Hale & keyser 2002. In the model developed by Hale & Keyser 2002, eventive structures are encoded in the syntax, by means of specifier-head and head-complement relations. The distinction between unaccusative and unergative verbs occupies a central place in their model. They propose that unaccusatives are associated with an l-structure that contains a specifier position. (106) [ VP D [ VP V XP] 79 Furthermore, the authors consider that the subject of unergative verbs is not an argument of the verb. In fact, it does not originate in the l-syntax (which represents the interface between syntax and the lexicon), but in sentential syntax (s-syntax). Unergatives are supposed to be denominal verbs, and they have the following structure. (107) [ V V N ] In their model, unergative verbs appear as the result of a conflation process where the noun complement provides the phonological content and the verb the categorial information. The following sentence represents the unergative verb sleep. (108) [ V V [sleep]] Unergatives are in many cases denominal verbs, as illustrated by he examples in (107) and (108). However, Zubizarreta & Oh contend that not all unergative verbs have to be denominal. For example, the authors postulate that manner of motion verbs, such as run or crawl, are not denominal. There is clear evidence from Spanish, where the noun corrida “run” is clearly deverbal instead, since it is formed out of the past participle of the verb correr “to run”. Zubizarreta & Oh endorse the view that the presence of an external argument is not regulated by the verb per se. However, in contrast to Hale & Keyser’s approach, Zubizarreta & Oh assume the existence of a vP, headed by a functional head v and 80 generated immediately above the l-structure of the verb, following Marantz 1984, Chomsky 1993 and Hale & Keyser 1993. The category v is equivalent to that referred to as “Voice” by Kratzer 1996 and as “Tr(ansitive)” node by Collins 1997. The function of the v projection is to license an external argument, while allowing us to structurally define the notion of transitivity, which is so intimately related to Case assignment, in terms of the sequence [v [V]]. In Zubizarreta & Oh’s system, the node v is introduced by a general rule of syntactic composition, included below: (109) a. v may be freely merged with a VP b. If VP lacks a Specifier, v must be merged with VP Given (109b), the unergative structures in (107) and (108) become (110a) and (110b) respectively, and are interpreted as activity-denoting events. (110) a. [ vP D [ v [ VP V D]]] b. [ vP John [ v [ VP V [laugh]]]] Building on Hale & Keyser, Zubizarreta & Oh also assume the existence of two distinct verb types: unergatives, already introduced above, and unaccusatives. The main difference between these two types is that unergatives (110) lack a specifier position in the l-syntax, unlike unaccusative verbs (111-113). 81 (111) [ VP D [V [XP]]] (112) a. [ D [V [ P [ P [ D ]]]]] b. [ VP John [ V went [ PP to [ PP P [the park]]]]]] (113) a. [ D [V [P [ D]]]] b. [ VP John [ V is [ P in [the park]]]] The structure in (111) represents the prototypical unaccusative construction. The sentences in (112) and (113) illustrate two subtypes of unaccusative verbs, where the nature of the XP complement differentiates between them. When the XP is formed by a complex preposition (109), the complement denotes a path with an endpoint, and the structure encodes movement towards a telos (i.e. directed motion). A change of state can also be conceptualized as an abstract path with an end point or final state, and be characterized as a case of directed motion (see Goldberg 1995; Jackendoff 1990, 1996; Krifka 1989; Tenny 1987, 1994, 1995). The construction in (112) represents an accomplishment in Vendler’s 1957 terms. Furthermore, the compositional nature of PPs has been studied for Dutch (Koopman 1997, den Dikken 2003), Italian (Folli 2001, Tortora 2005, 2006) and Spanish (Bosque 1993), and in all these languages directional PPs can take a locative PP as a complement, and not vice versa. On the other hand, when the XP complement denotes a location or a state the construction is interpreted as a state (113). Thus, the lower XP will be formed by a simple 82 PP. By observing the structures in (102) and (113) it is possible to identify both the semantics (accomplishment vs. state) and the syntax (unaccusative verbs) of the sentence. Regarding transitive constructions, it is possible to create transitive verbs by adding a vP shell to an unaccusative construction. In the following examples, a transitive construction is formed out of a change of location (114) and a change of state (115) predicate. (114) a. [ vP DP v [ VP DP [ V [ P [ P [ DP ]]]]] b. John sent the package to Paris. (115) a. [VP D [V A]] (change of state) b. The milk went sour. In (114) the lower PP is realized by a directional PP, i.e. a PP that takes another PP (location) as its complement. Therefore, this construction encodes directed motion, or a notion of transfer. I claim this is also the case with the locative alternation verbs, which behave as accomplishments. This would explain why some of them can also have an intransitive counterpart (the despejar/borrar type), as long as they can be associated with a notion of internal causation. I will not describe here the unaccusative sentences where the endpoint is realized as an adjectival phrase, such as in (115) and (116) for intransitive and transitive constructions respectively, since they are beyond the scope of this dissertation. 83 (116) a. [vP DP v [VP DP [ V [ AP]]] b. The cold froze the water. Once I have provided an introduction to the system that I have adopted, I will proceed with the description of the specific syntactic structures I propose for the locative alternation construction. 3.2. The universal structure of alternating verbs In this section, I develop an analysis of alternating verbs based on the assumption that they are all built on an unaccusative construction. As seen before, the literature dealing with locative alternation has frequently used the terms change of state and change of location to refer to the two variables of the alternation (Mateu 2002, Levin 1993, Levin & Rappaport 1998). In particular, the with variant is considered to imply a change of state, while the locative variant implies a change of location. (117) a. John loaded the truck with hay (CHANGE OF STATE). b. John loaded hay into the truck (CHANGE OF LOCATION). The reason for this dichotomy is that in sentences like (117a), where the direct object is the goal argument, there is an emphasis in loading the truck, i.e., in making the truck become loaded. A sentence like (117b), however, focuses on the process of loading 84 hay, i.e., in a change of location of the hay. This difference can be immediately related to the properties of unaccusative verbs described in section 3.1, where unaccusative constructions imply a change of state or location when the complement of V is realized by a complex preposition. In chapter 2, I have also described the central role of the preposition in both locative alternation variants and how it determines the argument structure of the sentence. I claimed that it is the preposition that projects its argument structure to the construction, and not the verb. Recall that this claim was derived from the fact that alternating verbs can take direct objects with different θ-roles, and that the preposition seems to be the one that will disambiguate the meaning, specifying the argument structure. In fact, the DO of the sentence in (118a) could receive two different interpretations, being either locatum or location. The presence of the preposition in (118b) and (118c) disambiguate the structure. (118) a. John loaded the car (location or locatum?). b. John loaded the car (locatum) in the wagon. c. John loaded the car (location) with all his stuff. All the previous considerations lead us to postulate that it is the preposition the one responsible for assigning the θ-roles to the construction and projecting its argument structure, as suggested in chapter 2. The verbs that display locative alternation seem to be unable to assign a specific θ-role by themselves. In the structures where no overt PP is expressed, we will assume that there is an implicit PP that assigns the thematic structure 85 to the sentence (recall the discussion in section 2.3, where it was noted that the meaning of the PP is always recoverable and is conceptually necessary). Another feature that is characteristic of this alternation is that the locatum argument needs to be in direct contact with the location argument, at least at a certain point. Therefore, the sentence in (119) is semantically deviant. This fact will be further explained in section 3.4. (119) #John loaded the hay onto the wagon, but the wagon was empty afterwards. Thus, let us postulate that in the locative alternation construction there is a PP projection low in the structure of the VP, in which the two internal arguments of the alternation originate and where the argument structure of the construction is determined. The suggested structure is represented in (120). (120) [ PP D [P XP]] For now, let’s call this projection a PP. It is important to bear in mind that it has predicate properties, since it behaves a s a small clause. Also consider that (116) can be a pruned version of a more complex PP structure (see Koopman 1997, Hale and Keyser 2002, den Dikken 2003 or Zubizarreta & Oh 2007 for a further discussion on this issue). If a state preposition is inserted, such as con “with”, the direct object NP will express a location, while the PP will represent the locatum or theme. When a locative preposition is 86 inserted, the direct object will bear a locatum θ-role and the NP in the prepositional phrase will indicate a location. Thus, the choice of preposition determines the theta- relations. Over the PP projection, there is a VP with a specifier position, as shown in (121). (121) [ VP D [V [ PP D [P XP]]]] It is now obvious that (121) is an unaccusative structure as described by Zubizarreta & Oh. At this point it is necessary to recall that, following Zubizarreta & Oh, when the XP in a structure such as (121) is realized by another PP (a location) or an AP (a state), the structure will exemplify a case of directed motion. This meaning is derived from the construction itself (122), and not from any property of the lexical items inserted in it: (122) Complex XP = directed motion Simple XP= state But what is the nature of the XP in the locative alternation? My claim is that it is a locative PP that functions as a complement of a higher directional PP, and therefore both variants of the locative alternation are instances of directed motion constructions as described in Zubizarreta & Oh. Evidence for this can be found in English, where complex prepositions such as into or onto can appear overtly (123). 87 (123) John loaded hay into the truck. Finally, it is necessary to insert a vP projection to introduce the external argument in the transitive variants of the alternation. As pointed out in section 3.1., this is a common derivation widely available in the language. Thus, the structure that I postulate has a lower PP with two layers, where the higher PP has a directional meaning, and the lower one has a locative or state meaning. According to Tortora 2005, Spanish would not allow two grammatical prepositions, such as “in” and “to”, to occur as a complex PP overtly. However, Bosque 1993 showed that in Spanish it is also possible to find complex prepositions formed by a locative and a directional P, such as a-trás “behind”, when they result form the compounding of a grammatical preposition, such as a “to”, and an adverbial preposition, such as tras “behind”. Whether the lower PP will be interpreted as a state or a location will depend on the preposition inserted in the construction: a locative preposition such as in will provide a locative meaning to its projection, while the insertion of a preposition such as with will be associated with a stative meaning (a notion of possession ≈ have, cf. section 3.4.). Therefore, whether the sentence will be interpreted as either a change of location or a change of state will depend on the preposition placed in the lower PP. Summarizing, the directed motion or change meaning characteristic of the locative alternation is derived from the construction itself, following Zubizarreta & Oh 2007. The complex PP of the transitive types of the alternation provides the argument structure to the sentence. The direct object originates in the spec position of the lower PP and will rise to [spec [VP]] to become the direct object. Thus we have structures (124) and (125). (124) Juan cargó el heno en el camión “John loaded the hay into the truck” v 3 D v Juan 3 v V 3 D V 5 3 el heno i V P dir ! 3 cargó D P dir ! 3 t i P P loc ! 3 ∅ D P loc ! 5 t i en el camión The so called change of state variant, or with variant, has the same structure. The only difference will be the preposition inserted, which will project its argument structure to the construction and the locative or stative (possessive) meaning. The labels P dir , P loc and P st represent Directional PP (such as to), Locative PP (such as in) and Stative PP (such as with) respectively, and have only been added to the tree as a label. However they 88 are not necessary, since the properties of the construction are derived from the syntactic relations existing between the different constituents, while meaning is derived from the argument structure of the lower PP. (125) Juan cargó el camión con el heno (John loaded the truck with the hay). v 3 D v Juan 3 v V 3 D V 5 3 el camión i V P dir ! 3 cargó D P dir ! 3 t i P P st ! 3 ∅ D P st ! 5 t i con el heno The only difference between English and Spanish in the previous structures is that in English the complex preposition can spell out. The intransitive types, such as intransitive clear and despejar, have exactly the same structure and properties, but they lack the higher vP. Since these verbs are unaccusative (or because they are inserted in an unaccusative construction) the 89 90 underlying object will rise to subject position to fulfill the EPP. This is also the case of the intransitive versions of the cargar “load” type presented in (72) and repeated in (126). (126) a. La administración pública mexicana no es sino el resultado de un traslape the administration public Mexican not is but the result of an overlap de culturas y gobiernos que parece que se carga en la piel. of cultures and governments that seem that CLITIC loads in the skin “The Mexican public administration is only the result of an overlap of cultures and governments that seems to be loaded on the skin”. b. y el aire se carga de sospechas and the air CLITIC loads of suspicions “and the air is loaded with suspicions” However, there is an exception. The verbs in the swarm type, excluding those that have a transitive counterpart (intransitive load), are not instances of directed motion or transfer, since they are states. Therefore they lack the higher directional PP. The relevant structures are provided in (127) and (128). (127) Los turistas pululan en las playas “the tourists swarm in the beaches” V 3 D V 5 3 los turistas i V P loc ! 3 pululan D P loc ! 5 t i en las playas (128) Las playas pululan de gente “The beaches swarm with people” V 3 D V 5 3 las playas i V P st ! 3 pululan D P st ! 5 t i de gente In this paper, I am going to follow Perlmutter and consider the swarm type verbs as unaccusative, at least when they enter in the locative alternation. Perlmutter does not mention the verb swarm, but includes emission verbs in the unaccusative class. For the most part, the swarm type is formed by emission verbs both in English and Spanish. The intransitive uses of the load class are clearly unaccusative, since they participate in the 91 92 causative alternation and have the clitic se. In the next section I provide further evidence supporting their unaccusative classification. Although the swarm type verbs are the intransitive counterpart of the spray/load type, as far as structure is concerned, there are some significant differences that must be addressed. Putting aside the intransitive variants of the load type, such as (126), the locatum of these verbs does not end up in the location as a result of a movement, and the relationship between the two internal arguments varies. For example, in emission verbs the location argument represents the point of departure of the light or sound emitted. This source relation is not derived from the meaning of the preposition however, but from the meaning of the verb. In a verb such as pulular “swarm”, the location argument is where the swarming event takes place, but there is no necessary movement towards the location. These meaning differences are idiosyncratic and do not describe any universal feature inherent to the locative alternation. In fact, some of these verbs (emission verbs) have been frequently described as activities that lack an end point or telos. However, I believe that when these verbs participate in the locative alternation they behave as states. This is also Sorace’s 2000 point of view. She states that “verbs in this class denote “stative”, non-dynamic activities, which require a low degree of volitional energy to be performed… and are characterized by a high degree of subject affectedness (for example, an emitter is totally defined by the emission process)” (op. cit.: p. 877). For example, if we apply the tests presented in Dowty 1979 to differentiate between activities and states, we find contradictory facts that 93 lead us to their classification as states, or perhaps a hybrid class 14 . I include Dowty’s tests below, where the symbol # indicates a deviant meaning. Only the first test seems controversial. I. Only non-statives occur in the progressive: Una extraña luz está brillando en sus ojos. A strange light is shining in his eyes. Las playas están pululando de gente. The beaches are swarming with people. Los turistas están pululando en las playas de Benidorm. The tourists are swarming in the beaches of Benidorm. II. Only non-statives occur as complements of force and persuade: #Juan forzó a sus ojos a brillar con una extraña luz. #Juan forced his eyes to shine with a strange light. # Juan forzó a las playas a pulular de gente. #Juan forced the beaches to swarm with people. # Juan forzó a los turistas a pulular en las playas de Benidorm. #Juan forced the tourists to swarm in the beaches of Benidorm. 14 Recall that my claim is that these verbs behave as unaccusatives when they enter the locative alternation, but not necessarily when they enter other constructions. 94 III. Only non-statives can occur as imperatives: #¡Brilla con una extraña luz! #Shine with a strange light. #¡Pulula de gente! #Swarm with people. #¡Pululad en las playas de Benidorm! #Swarm in the beaches of Benidorm IV. Only non-statives occur with the adverbs deliberately, carefully: #Una extraña luz brillaba en sus ojos cuidadosamente. #A strange light shines in his eyes carefully. #Las playas pululan de gente cuidadosamente. #The beaches swarm with people carefully. #Los turistas pululan cuidadosamente en las playas de Benidorm. #The tourists carefully swarm in the beaches of Benidorm. V. Only non-statives appear in Pseudo-cleft constructions: #Lo que la luz hizo fue brillar en sus ojos. #What the light did was to shine in his eyes #Lo que las playas hicieron es pulular de gente. #What the beaches did was to swarm with people 95 #Lo que los turistas hicieron es pulular en las playas de Benidorm. #What the tourists did was to swarm in the beaches of Benidorm. VI. Activities that occur in present tense have a habitual or frequentative meaning: #Una extraña luz brilla en sus ojos 15 . #A strange light shines in his eyes #Las playas pululan de gente. #The beaches swarm with people #Los turistas pululan en las playas de Benidorm. #The tourists swarm in the beaches of Benidorm. Although all these tests probably have exceptions, the first test is perhaps the most problematic. Verbs like love, which is typically a state, can easily appear in the progressive both in English and Spanish (129). (129) a. I’m loving it. (McDonald’s slogan) b. Te estoy amando locamente, pero no sé cómo te lo voy a decir. (lyrics from song by Las Grecas) “I am loving you crazily, but I don’t know how I’m going to tell it to you”. 15 All these sentences are grammatically correct and possible, but are normally interpreted as a particular event, and not a habitual action. 96 Still the sentences included sound unnatural in Spanish, at least to my ears. My intuitions are supported by data from the corpus CREA and the Davis Corpus (corpusdelespanol.org). I have not found any single example of brillar “shine” or resonar “resound” displaying locative alternation in a progressive tense. I have found however some instances of these two verbs in gerund constructions, but they have different properties from progressive tenses, and have rather been associated with absolute participials. The rest of the tests are very straightforward, and, in the best case scenario, they sound even funny, which support my classification of these verbs as stative predicates when they enter in the locative alternation construction. Now, my analysis might imply that the swarm type verbs could have a transitive counterpart, by adding a causative v. In principle, nothing in my analysis would avoid this, and in fact there might be cases of emission verbs used as transitives, like the sentece in (130). (130) Everything about you resonates happiness. (Lyrics of the song “Bliss”, by Muse) Even so, the transitive uses of these stative verbs are uncommon both in English and Spanish. In section 3.2.2, I relate this fact to the notion of internal and external causation. 97 In this section, I have shown that all the verbs that display the locative alternation are built on an unaccusative construction, including stative verbs of the swarm type, such as emission verbs. The only difference between these stative verbs and the rest of alternating verbs is that the PP complement of stative verbs is only formed by a locative or stative PP, while the rest of alternating verbs have a combination of a directional and a locative PP. This structural difference characterizes the swarm type as stative verbs, while the rest of alternating verbs, including the intransitive versions of cargar “load”, behave as accomplishments. 3.2.1. The swarm verbs as unaccusatives: Theoretical advantages In section 3.2, I have suggested that the swarm type verbs, including emission verbs, behave clearly as unaccusative in Spanish, while in languages like Italian they can also present unaccusative behavior (Sorace 2000). The inclusion of the swarm type in the unaccusative class also presents several theoretical advantages. As I suggested already, if the swarm type verbs are analyzed as unaccusatives, the intransitive verbs that enter the locative alternation would all be unaccusative. All intransitive verbs that enter in the locative alternation are clearly unaccusative (presence of se, participate in causative alternation, etc), with the exception of the stative swarm type verbs, which include emission verbs (EV). In my analysis, all verbs entering the locative alternation would either be transitive or unaccusative, and the alternation would happen between an (underlying) object and a PP. From a theoretical point of view, one might conclude that the only difference between transitive and intransitive alternating 98 verbs would be the presence of a causative projection v in the transitive variants. I have presented a more detailed analysis of this theory in section 3.2. However let me recall that I will consider the stative swarm type verbs as unaccusative when they participate in the locative alternation, but not necessarily in other constructions, following the line of research in Sorace 2000, where instead of a clear-cut classification, there is a hierarchy of verbs ranging from purely unaccusative to purely unergative. Of course, this does not apply to the intransitive uses of the cargar “load” verbs, which have overt unaccusative morphology (presence of the clitic se, participate in the causative alternation, etc.). The following section deals with the analyses that have considered the swarm type verbs as unergative (Levin and Rappaport 1995) or have claimed that the swarm alternation, in which emission verbs are included, should not even be considered as cases of locative alternation (Dowty 2000). 3.2.1.1. Unergativity, unaccusativity and the swarm type Some researchers have opposed Perlmutter’s 1978 original classification of emission verbs, a subgroup within the swarm type, as unaccusative. In particular, Levin and Rappaport 1995 (L&R) propose that they behave as unergative rather than unaccusative. Furthermore Dowty 2000 proposes that the swarm alternation class, which is the one where EVs are included, should not be considered as a case of locative alternation. Let’s review first Dowty’s 2000 proposal. In Dowty 2000, he carries out a detailed description of the semantic properties of the swarm type verbs in English. This author suggests that the properties of this class 99 differ from those of the transitive alternating verbs, like the load/spray class. For example, he points out that all the verbs in the transitive spray-load alternation are telic. This telicity is intimately related to the aforementioned holistic effect. However, as Jackendoff 1990 points out, the group of transitive alternating verbs is not homogeneous, because verbs like spray do not imply a completely affected object, or incremental theme. Dowty also explains that the transitive variants of the alternation imply a change of location into or out of a space or enclosure, but this would not be true for the intransitive swarm type. Nevertheless, this change of location into or out of a space cannot be central or necessary for the alternation to happen. In fact, verbs like send or throw imply a change of location of the type described by Dowty, but they never show locative alternation. Dowty’s research shows that the class of alternating verbs is not semantically homogeneous, but this is true for both the transitive and the intransitive alternating verbs. Therefore, one should not forget that these verbs, while not others, show an explicit syntactic alternation which needs to be accounted for. It is clear that there are semantic differences between the verbs, and that the swarm type differs from other verbs included in the alternation. It is also true that the transitive variants also differ from each other in different ways. What they obviously share is their ability to alternate. Iwata 2005 provides a very detailed account of the different semantic properties of alternating verbs, but what I am trying to describe in this dissertation is what all these verbs have in common, not what differentiates them from each other. 100 Going back to the subject of unaccusative versus unergative verbs, Perlmutter 1978 classified EVs as unaccusatives, because they cannot appear in the impersonal passive construction in many languages, such as Dutch. By contrast, unergative verbs can appear in this construction. L&R op. cit. accept the validity of this diagnostic, but they warn us that emission verbs can also appear in the impersonal passive construction if one finds an example in which “the emission of the stimulus is understood to be intentional”, as in (131). Such sentence becomes grammatical when one translates the Dutch noun krengen as “nasty woman” instead of “carcasses”. This fact was pointed out by Zaenen 1993. (131) Er werd door krengen gestoken. There is by the nasty women/carcasses stunk “There is stunk by the nasty women/*carcasses.” (Zaenen 1993: 139, (37)) From the previous discussion we can derive that emission verbs in Dutch may have both unaccusative and unergative properties, because, under certain circumstances, they can pattern together with both unaccusative and unergative verbs. But L&R also note that, in Dutch, emission verbs appear with the auxiliary hebben “have”, which characterizes transitive and unergative verbs. Because of this, they conclude that they should be described as unergative. However, this is not completely true. The sentence in (132) shows how a typically alternating emission verb such as 101 weerklinken “resound” can also appear with the auxiliary zijn “be” in periphrastic forms in Dutch. (132) Wanneer de laatste tonen van ‘Octopussy’ zijn weerklonken, stroomt de zaal dan ook snel leeg “When the last notes of ‘Octopussy’ have resounded, the room rapidly empties.” (Peeters, E. (2006): The Wedding Present, http://www.kindamuzik.net/live/the- wedding-present/the-wedding-present) Thus, emission verbs behave as unaccusative because they normally cannot enter in the impersonal passive construction and their meaning does not imply volition or causation (Perlmutter 1978). They can also pattern together with unergative and unaccusative verbs as far as auxiliary selection is concerned. Another language examined by L&R is Italian. When talking about emission verbs, they mention that “the Italian counterparts of these verbs invariably select the auxiliary avere ‘have’, rather than the unaccusative auxiliary essere ‘be’” (L&R 1995). However, this statement is again incorrect. The sentences in (133) show that the auxiliary essere “be” can also be used with emission verbs in Italian. In fact, Sorace 2000 explicitly mentions that the behavior of emission verbs “is highly variable in Italian” (Sorace 2000: p. 877). (133) a. Il canto della fanciulla è risuonato per tutta la casa. the chanting of-the girl is resounded by all the house. “The girl’s chanting has resounded around the whole house.” 102 b. Il canto della fanciulla ha risuonato per tutta la casa. the chanting of-the girl has resounded by all the house. “The girl’s chanting has resounded around the whole house”. Again, what we can derive from the previous data is that emission verbs can behave both as unaccusative or unergative. The swarm type verbs do not seem to show a clear meaning difference related to the presence of a different auxiliary. There is no aspectual difference between the two options because these verbs are always atelic. Furthermore, the data presented by Sorace 2000 seems to suggest that stative unaccusatives do not respond very well to the auxiliary selection test. This author shows that while stative verbs such as exist, which is normally considered unaccusative, typically select the auxiliary essere “be” in Italian, they consistently select avoir “have” in French. German and Dutch also seem to have a clear preference for the auxiliary have with stative verbs, except for some specific verbs. This data indicates that the auxiliary selection test may not be appropriate for stative verbs. Finally, let me examine the case of Basque. Basque is a non-Indoeuropean language spoken in the north of Spain and the south of France. It has a very rich morphology and two verbal auxiliaries which behave in a similar manner as be and have in languages like Dutch or Italian. The auxiliary izan “be” can only be used with unaccusative verbs, while ukan “have” can be used with both transitive and unergative verbs. L&R mention that many verbs that are traditionally classified as unergative in languages like Dutch, English or Italian are expressed periphrastically in Basque, by a 103 construction headed by a light verb egin “do/make” and preceded by a noun. So, for example, English laugh would correspond to Basque barre egin “laugh do”. Interestingly, these authors also explain that this is the case of emission verbs in Basque, indicating that they behave as unergative verbs. Of course, another possibility would be to assume that those verbs are not unergative, but transitive (see examples (134) and (135)). This seems to be the most accepted option nowadays, not only because the morphology supports this conclusion (a transitive verb is that which takes a complement NP) but also because these verbs pattern together with transitive verbs in different syntactic diagnostics. For example, unlike Romance, Basque unergatives may participate in the absolute participial (brillados los ojos) and adnominal constructions (los ojos brillados). These two constructions are only possible with transitive or unaccusative verbs in languages like Spanish (see Alcázar 2005 and 2006, and the references therein, for a description of unergative verbs in Basque) (134) a. Argi-ak bere begi-etan diztira egi-ten du light-erg.sg her eye-in.pl brightness do-imp have.3sg.3sg “The light shines in her eyes” b. Bere begi-ek argi-az diztira egi-ten dute her eye-in.pl light-instr.sg brightness do-imp have.3pl.3sg “Her eyes shine with the light” (135) a. Argi-ak bere begi-etan diztiratzen du light-erg.sg her eye-in.pl shine-imp have.3sg.3sg “The light shines in her eyes” 104 b. Bere begi-ek argi-az diztira-tzen dute her eye-in.pl light-instr.sg shine-imp have.3pl.3sg “Her eyes shine with the light” The previous examples show two different ways of expressing the emission verb shine in Basque in a construction with locative alternation. In the sentences in (134) the verb is expressed through the use of a light verb plus a noun. It is also possible to express a similar meaning without using a periphrastic form, using instead a simple verbal form, as in (135). Alcázar 2006 suggests that both forms, the periphrastic and the non- periphrastic, can be better described as transitive verbs. In this way, it would be possible to explain the existence of adnominal and absolute participial constructions that in principle should only be possible with transitive or unaccusative verbs. Recall that in this dissertation I defend the idea that the transitive variants of the locative alternation, such as load or clear, are also built upon an underlying unaccusative construction. 16 The previous data shows that there is enough crosslinguistic evidence to support an unaccusative classification of emission verbs. Relying on a single unaccusativity test, such as auxiliary selection, without taking into consideration other constructions might lead us to contradictory conclusions. Only when different constructions are analyzed it becomes evident that emission verbs show unaccusative, or transitive, properties in all languages described, Romance, Germanic or Basque. At this point, it becomes obvious that there is a close relationship between unaccusativity and alternating verbs. In the next section, I address the issue of why some verbs can occur in transitive and intransitive variants of the locative alternation, such as 16 Here I will not provide an account of the locative alternation in Basque, and how its particular morphology represents this alternation, since that is beyond the scope of this dissertation. 105 clear, while others can only appear as either transitive or intransitive, relating this phenomenon to the notion of internal and external causation. 3.2.2. The notion of internal and external causation: transitive and intransitive counterparts of the locative alternation. In the previous discussion, I have postulated that all alternating verbs behave as unaccusative directed motion constructions, except for the swarm type which are states. Thus the intransitive clear subtype and the intransitive versions of the load type are also cases of directed motion, which lack an external argument, and, therefore, a vP shell. Here I claim that what makes verbs alternate between the transitive and intransitive versions of the locative alternation is their ability to be associated with internal and/or external causation 17 (see Levin & Rappaport 1995 for further details on the notion of causation). In particular, I claim that those verbs that are difficult to associate with a notion of internal causation can only display the transitive variants of the alternation, while those verbs that can also be interpreted as internally caused may also occur in intransitive constructions. For instance, a verb such as load is normally associated with external causation, and therefore with a transitive construction, because things do not typically load into containers by themselves. However, a verb such as clear can easily have an external cause, as in “someone cleared the dishes from the table”, as well as an internal cause, as in “the clouds cleared from the sky”. 17 However, not all verbs that are externally caused are derived from an unaccusative base, for example eat, smoke, etc. 106 Further evidence for the notion of internal causation comes from the fact that verbs that can have an intransitive counterpart (despejar “clear”) only allow this intransitive form when they can be associated with internal causation. Compare (136a) and (136b). (136) a. John cleared the table of dishes vs. #the table cleared of dishes b. The wind cleared the clouds from the sky vs. the clouds cleared from the sky The sentences in (136a) show that even a verb that allows for an intransitive counterpart, such as clear, cannot take the intransitive form in contexts where a volitional agent is conceptually necessary. However, the intransitive sentence in (136b) is grammatical because it does not require an external agent. The swarm type class deserves a special mention because of its peculiar properties. All these verbs have an internal cause that makes them able to shine, resound or swarm, and no external agent can imbue such property in an object that does not have it initially. Furthermore, emission verbs and all those included in the swarm type are normally states, as discussed in section 3.2. They do not denote a change of state, but a pure state or location. Even if we added an external argument still we would not have a change of state/location predicate, since the lower XP is formed by a simple PP. I explained in section 3.2 that the notion of directed motion or change of state, i.e. the notion of change, is derived from the presence of a complex PP in the structure, which is not present in stative predicates, which only have the lower stative/locative PP, since the 107 directional P is not present. Recall that the swarm type verbs are states, which therefore lack a directional PP. Further evidence in favor of the grammatical distinction between internal and external causation comes from the psycholinguistics literature. McKoon & McFarland 2000 carried out two experiments using corpora and acceptability judgments and determined that sentences with externally caused verbs take more time to process (cause + change of state). Sentences with internally caused verbs are processed faster. This experiment suggests that there is a generalized mechanism to add subjects to verbs, and that when the subject involves external causation mental processing requires extra time, which might indicate that the structure is more complex. Internally caused subjects are also semantically restricted by the verb. 18 The structure I propose might imply that it is structurally possible to add an external argument to stative verbs, such as emission verbs. The only thing that impedes this is the fact that the swarm type verbs can hardly be associated with an external argument, due to the meaning of these verbs. However, it is possible to find, under certain conditions, emission verbs used in transitive forms. For example, the Merriam- Webster’s online dictionary includes an entry for a transitive resonate, defined as “to subject to resonating”. The sentences in (137) were extracted from internet using the search engine Google. 18 Demonte 2002 avoids using the traditional verb classifications in favor of a basic distinction between internal and externally caused verbs. 108 (137) a. the clock resonates a friendly bell alarm b. Nakatani strikes his cymbal with wire brush and repeatedly resonates a large gong. Verbs such as swarm or shine also have transitive counterparts in the Merriam- Webster’s online dictionary. However, the transitive uses of these verbs are not frequent, and sometimes they have a rather metaphoric meaning. 3.3. Locative alternation in Spanish: subtypes and special features In this section I address some peculiarities that characterize the different subtypes of the locative alternation in Spanish and propose some possible explanations. Because the particular features of these subtypes seem to be purely specific to Spanish, I am not going to devote a great deal of detail to this section. 3.3.1. Verbs of the type of cargar “load”, empacar “pack” and plantar “plant” These verbs have an agentive external argument and two internal arguments, as we saw in section 2.1. Their location argument indicates the destination point of the change of location, and therefore the telos, which must be reached necessarily, as seen in section 6 (Beavers 2006). The group is divided in three subclasses, depending on their ability to appear with specific STATE prepositions: de “of”, as in plantar “plant”; con “with”, as in empacar “pack”; or with both, as in cargar “load”. The preposition con “with” can indicate either a type or set, or a subtype or subset, i.e. certain elements of a set. In this way, con can have both a generic (set) and a specific (subset) meaning (138). 109 (138) a. Juan cargó el camión con heno (set, type). Juan loaded the truck with hay. b. Juan cargó el camión con 20 kilos de heno (subset, specific). Juan loaded the truck with 20 kilos of hay. By contrast, the preposition de “of” can only have a generic meaning (type, set), it can only refer to the type of set, not to a subset (139). (139) a. Juan cargó el camión de heno (type). Juan loaded the truck of hay. John loaded the truck with hay. b. *Juan cargó el camión de 20 kilos de heno (specific). Juan loaded the truck of 20 kilos of hay “John loaded the truck with 20 kilos of hay”. Hence, the preposition de can only be followed by collective or mass nouns, never by a count quantified noun. The reason why some verbs can appear with both prepositions, the subclass of cargar “load”, is that these verbs can have a locatum PP with both a generic and a specific reading. However, the verbs of the group of empacar “pack” can only have a locatum PP with a specific meaning, which excludes the appearance of the preposition de “of”. Finally, verbs like plantar require the appearance of a generic locatum PP, which enables the appearance of de “of”. The reason why the 110 preposition con “with” cannot appear with the plantar “plant” type of verbs must be due to the fact that its basic unalienable meaning is that of specificity, while the generic meaning must be derived. This could be accounted for from a syntactic point of view stipulating the existence of intermediate projections within the PP, something similar to aspectual projections, but at the PP level (see the references mentioned above dealing with PP structure). However the fact that the plantar subclass is only formed by two verbs might suggest two possible explanations: (i) it might be possible to find sentences where these verbs appear with the locatum preposition con “with”, even if I did not find any, or (ii) the selection of the preposition de “of” might be a lexical idiosyncrasy of these verbs. Should we modify a syntactic structure to accommodate facts that can only describe two verbs in the system, or should we consider that the difference is due to the lexical properties of certain verbs or prepositions? At this point, without excluding the validity of any of these possibilities, I am going to keep the simple structure suggested for PPs in this chapter, since it is more minimal. Furthermore, the different use of this preposition cannot be generalized to languages like English, that only allow the locatum preposition with, or French, where only de “of” is possible. 3.3.2. Verbs of the type of barrer “sweep”, disculpar “forgive” and recoger “put in order” The verbs included in this class are transitive verbs that take two internal arguments: a direct object and a PP. They also have an agentive external argument. What makes this group different from the one in 6.1 is the fact that the location argument 111 indicates the point of departure of the change of location, and more abstractly a negation or removal of a property. There is only a preposition associated with this alternation in Spanish, de “of, from, off”, although its meaning corresponds to different prepositions in English. In Spanish there is a preposition de that can be identified with the English of and that takes a locatum complement. There is another de whose meaning is equivalent to the English prepositions from or off and that takes a location complement. Even if their phonological form is exactly the same, their meaning and thematic structures are different. Because of this, I will still assume the trees in (125) and (124) for the change of location and change of state variants respectively. Let’s call the STATE preposition de st and the LOCATION one de loc . There is another peculiarity regarding the argument structure of the verbs included in this subtype of locative alternation. The verbs of the type of barrer “sweep” can overtly express both the location and locatum arguments by means of an argumental PP, exactly in the same way as the English verb clear. However, the verbs of the type of disculpar “forgive” cannot express the location argument as a PP. This structure does not appear in English at all. Plus there is a third subclass in this alternation, formed by the verbs of the type of recoger “put in order”, characterized by its inability to express the locatum argument by means of a PP. The thematic structure of the recoger class corresponds to the English wipe class. When observing the form of the Spanish verbs that enter into this type of alternation, it is worth noting that many of them are formed by the overt addition of a preposition, attached to the verb by prefixation. In fact, one might be tempted to postulate 112 that what makes impossible the addition of a PP with locative meaning in the structure of the disculpar verbs is that the locative preposition de LOC has been overtly incorporated to the verb, which is why, being absent from the structure it cannot assign Case to the location argument in the change of location variant. Some examples of overt incorporation of the preposition into the verbs of the disculpar type are shown in (140). This incorporation happened somewhere in the history of Spanish or was even present in Latin, and it is still possible to see the effects of this syntactic incorporation. (140) desvalijar “plunder”, disculpar “forgive”, dispensar “exempt”, exculpar “exculpate”, exonerar “exonerate”. The structure of the recoger verbs is similar to the one shown for the disculpar class, but the recoger class must have incorporated the STATE preposition de ST instead, which is the reason why they cannot express a locatum PP, as it is the case in English. However, this would not explain why the verbs in this group that are not formed by the addition of a preposition, such as curar “cure” in the disculpar “forgive” verbs or filtrar “filter” in the recoger “put in order” class, still do not allow the overt expression of one of their arguments 19 . The incorporation theory cannot explain either why some of the verbs of the barrer “sweep” type, which allow the expression of both the locatum and location arguments as a prepositional object, include the prefix de- too. Some examples are the verbs desalojar “vacate” and descargar “unload”. 19 Recall that the disculpar verbs do not allow the overt expression of their location argument as a prepositional object (PO), while the recoger type does not allow a PP that functions as a locatum. See chapter 2 for details. 113 Regarding de-prefixation, Zubizarreta & Oh 2007 mention that this morphological process is lexically restricted. The prefix de- can be added to certain nouns, such as hueso “bone” or agua “water”, to form the verbs deshuesar “debone” and desaguar “drain”, but it cannot combine with nouns such as gas “gas” or zapato “shoe”. However, this prefix can be productively combined with verbs formed via prefixation of the locative preposition en- to a noun, as in (141). (141) a. de(s) +en + sillar el caballo (remove the saddle from the horse) b. de(s) +em + botellar el vino (remove the wine from the bottle) c. de(s) +em +paquetar el regalo (remove the gift from the package) d. de(s) +en +vainar el cuchillo (remove the knife from its sheath) e. de(s) + en +capuchar al prisionero (remove the hood from the prisoner) f. de(s) +en +mascarar al culpable (remove the mask from the culprit) Moreover, verbs like embotellar (loc. prefix+bottle) and des-em-botellar (neg. pref. + loc. pref. +bottle) seem to have different syntactic properties. (142) a. Embotellamos el vino en una botella de vidrio. (We) loc. pref + bottled the wine in a glass bottle b. Desembotellamos el vino. (*de una botella de vidrio). (We) neg. prefix+loc. pref.+bottle the wine (*out of the glass bottle) 114 The previous examples show the variable behavior of these prefixes, which seem to require different analyses. Here I will leave this issue for future research due to our lack of understanding of prepositions in general and prefixes in particular. 3.3.3. Verbs of the type of despejar “clear” and borrar “erase” This group is formed by the unaccusative variants of the disculpar/recoger class. They are intransitive with overt unaccusative morphology in Spanish, since they are pronominal with no agent entailed and require the presence of the clitic se. The verbs of the despejar “clear” type cannot express location through a PP, while the recoger class cannot express the PP locatum. See the discussion on this issue in section 3.3.2, where I introduce some of the problems related with preposition incorporation. The syntactic structure of these verbs will be identical to their transitive counterparts. However, because these verbs are unaccusative, the vP projection should not be present and their inability to assign case will cause the movement of the DO to subject position. 3.3.4. Verbs of the type of arder “burn”, resonar “resound” and pulular “swarm” This group is formed by verbs that show the intransitive counterpart of the cargar “load”, empacar “pack” and plantar “plant” class, as far as the use of prepositions is concerned. While the intransitive uses of the cargar type indicate accomplishments (143) like the transitive counterparts, most of the verbs in this group, the swarm type, express only a pure state/location (143b), as seen in section 3.2. 115 (143) a. La imagen que se adora, que se interioriza, se impregna en lo más cercano e íntimo de uno mismo. “The image that you worship, that you interiorize, becomes impregnated in the deepest and most intimate part of yourself”. b. El aire que nos separa se impregna con aquel olor. “The air that separates us becomes impregnated with that smell”. (144) a. Su Habano es luz que arde en el mar profundo. His Habano is light that burns in the see deep. b. La zona portuaria parecía arder con la luz de la ciudad. The zone harbor seemed to burn with the light of the city. While some of these stative verbs are clearly unaccusative in Spanish (following Mendikoetxea 1999), such as hervir “boil” and arder “burn”, which enter the causative alternation, the status of other verbs in this group is not clear. Emission verbs were classified by Perlmutter 1978 as unaccusative. However, there is not agreement about this and Mendikoetxea considers them as unergative in Spanish, lacking any morphological clue. With respect to the meaning of pulular, Mendikoetxea comments that the non- agentive verbs that express manner of motion seem to behave as unaccusative in Spanish. The verb pulular “swarm” focuses on the manner, not on agentivity. In fact, it can be 116 used to describe the running of liquids. The online Diccionario de La Real Academia Española relates pulular to bullir, and this one to hervir “boil” which is overtly unaccusative. The subclass of verbs analyzed here shows the same distribution of the prepositions con “with” and de “of” already observed in the spray/load type. It is interesting to note that the distribution of these verb types in subtypes (see appendix), considering the prepositions they select, can be semantically driven. The verbs that accept the two state prepositions (de and con) are light emission verbs, except for hervir “boil”. However, the verbs that can only take con are all sound emission verbs, with the exception of abundar “abound”. Finally, the verbs that take only the preposition de all represent instances of some undetermined type of movement, with no direction. 3.4. On the status of alternating prepositions Another phenomenon that deserves our attention is the fact that only a specific set of prepositions can appear in this alternation. In particular, only locative prepositions and the stative prepositions with and of. This contrasts with the prepositions that can appear in other types of unaccusative constructions, such as the dative alternation, which are mainly directional. Section 3.4 provides a description of some semantic properties of these prepositions, and a possible syntactic analysis that reflects these semantic differences. A complete explanation of the grammatical status of prepositions is beyond the scope of this research. 117 3.4.1. Prepositional aspectology There seems to be a fundamental similarity that joins together all the prepositions that appear in the locative alternation, while differentiating them from others. Beavers 2006 explains how the two arguments that occur in the locative alternation need to be in contact for those sentences to be true. It is not possible to have a loading event where the locatum argument is never in contact with the location, for example, where the hay never reaches the truck. This is not the case when the preposition to appears. In a sentence such as John sent a letter to Mary, although Mary is the intended recipient it is still possible to conceive that the letter may not reach its goal. Therefore, it is possible to say that John sent a letter to Mary, but she never received it, and the sentence will still be true. I have already discussed this issue in chapter 2. Zubizarreta & Oh’s description of unaccusativity predicts that ditransitive verbs, such as those that participate in the dative alternation, where there is again an implied notion of transfer or change of state/location, should be described as underlying unaccusatives. Both the locative and the dative alternation should then share a basic unaccusative structure such as the one proposed in this chapter. In the system I develop here, where the lower PP determines the argument structure and some semantic properties of the construction, the difference between the locative alternation construction and the dative construction will be derived from the meaning of the prepositions inserted. I believe that what makes the dative and the locative alternation different is the aspect of the prepositions that appear in these constructions. Described in terms of 118 boundedness, I postulate that the prepositions in, with and of are bounded, because the two arguments involved need to be in direct contact with each other at a certain time. See the following examples: (145) a. Peter is with John b. Peter is in France. c. El coche es de Peter. The car is of Peter “The car is Peter’s.” d. Peter went to France. The sentences in (145) indicate that the prepositions “with”, “in” and de “of” require an overt relation between the subject and the predicate. Sentence (145a) indicates that Peter and John are together, not that they might have the potential of being together; similarly, (145b) implies that Peter is in France, not that he might be there; (145c) also requires the possession of the car by Peter. In (145d) though, France is an intended goal, but it might never be reached. Sentences (145a), (145b) and (145c) would no be true if Peter was not together with John, Peter was not in French territory and Peter did not possess a car respectively. By contrast, a sentence such as Peter went to France but he never got there because his car crashed is perfectly possible. Now consider which preposition characterizes the locative and the dative constructions. In this chapter, I have postulated that the only preposition that is necessary 119 in the locative alternation is the lower stative/locative PP. The higher directional PP will only appear in those sentences that imply a change. I believe that it is not a coincidence that languages like Spanish only use these stative prepositions overtly in the locative alternation, unlike in English, where the complex prepositions into and onto may appear. Conversely, in the dative alternation the only preposition possible in Spanish is the directional a “to”. Something similar happens with the preposition from, which is both directional and locative. While in and with specify the goal where something ends up or is located, from specifies the point of departure instead, plus a direction away from it. I will consider that the prepositions that are intrinsically bounded (in, of, with, from) can appear in the locative alternation. They can be marked with a feature +bounded [+ β] (Chen 2008). The preposition to appears in constructions that indicate directed motion, and it does not require a necessary contact between the theme and the goal, so it can be described as – bounded [- β]. This distinction might imply that the only syntactic difference between certain alternations, such as the transitive locative alternation versus the dative alternation or the directed manner of motion construction (Zubizarreta & Oh 2007) can only be derived from the presence of by the feature [± β] of the obligatory PP that characterized each construction. Specifically, the locative alternation always requires the overt presence of the lower locative/stative P, which has the feature [+ β]. Conversely, both the dative and the directed motion constructions only require the overt appearance of the preposition to, which is [- β]. These obligatory prepositions are the ones that introduce the argument 120 structure. When the directed (manner of) motion construction includes a locative preposition, which, again, is not obligatory for this construction to exist, the lower P, which is the complement of the directional PP and is typically empty, will be filled with the optional locative preposition to imply the arrival to the goal. Unlike the locative alternation, the argument structure of the directed (manner of) motion constructions cannot be determined by the lower stative P. Here I would like to suggest that the theme will initially appear as the complement of P dir instead of as the specifier of the lower P loc . Thus while the relevant structure for the locative alternation is [ PPloc DP [P [DP]]]] the manner of motion constructions are [ PP P dir [ PPdir DP [P [ PP DP [P loc [DP]]]]. 3.4.2. The syntax of alternating prepositions In sections 3.1 and 3.2, I have already discussed the basic structure of the complex and simple prepositions that appear in the locative alternation, although I paid special attention to the locative prepositions (in, on) and their compounds with the directional P to, which forms into and onto in the directed motion or transfer constructions (transitive alternating verbs plus the clear and load intransitive subtypes). As far as the source prepositions from and off are concerned, I claim that they are also formed compositionally from a locative preposition that incorporates into a directional preposition, exactly in the same way as into/onto. As in the case of into and onto, they indicate a bounded locative relation between the locatum and location arguments with obligatory contact, in this case at the point of departure. The source prepositions indicate a movement away from the location argument, which clearly reveal 121 their complex nature, both locative and directional. Unlike into and onto, the source prepositions from and off are not morphologically divisible in two basic prepositions. Recall that these complex prepositions can only appear in those constructions that are accomplishments, i.e. in the transitive versions of the locative alternation and in the intransitive version of the clear and load types. Because these prepositions require inalienable direction, they can never appear in stative constructions, where there does not exist a notion of change. Regarding the difference between the stative prepositions with and in/on, I believe that they might be semantically and syntactically related. Intuitively, I think that while the locative prepositions indicate a locative state, the preposition with indicates the possession of some property. Taking Freeze 1992 as a point of departure, it might be possible to find parallels between the predicates be vs. have, and the prepositions in/on vs. with. Freeze indicates that there is a crosslinguistic relation between the predicate be and the possessive have. This author observes that these two notions are expressed by the same verb in many languages, where possession is also indicated by the use of the copular verb be. He explains these similarities by syntactically deriving the verb have from the predicate be. The relation between be and have can be extended to the prepositions in/on and with, and their corresponding Spanish versions en and con/de respectively. The sentences in (146) show clear similarities. Note however that Spanish has two copular verbs be: ser and estar. 122 (146) a. Juan está en Francia. “John is in France” b. Juan está con la gripe. Juan is with the flu “Juan has the flu”. In sentence (146a) the verb estar “be” followed by the locative preposition en “in/on” is used to indicate a location. In sentence (146b) the same verb followed by the preposition con “with” indicates some type of property that the subject has. Observe that the English verb “have” can be used to translate (146b). Furthermore, there are cases where the prepositions en and con seem to alternate freely or without a necessary meaning difference, as in the sentences in (147), extracted form the corpus CREA. (147) a. Hierve en deseos y aspiraciones. He burns with desires and aspirations. b. abunda en amores It abounds with love. c. Su pregunta resuena en múltiples ecos, como si hubiera valles y montañas. His question resounds with multiple echoes, as if there were valleys and mountains. 123 In these sentences the locative preposition en “in”, instead of a location, takes a locatum argument as its complement, instead of a location. This is possible because they are related at an abstract level. In this dissertation, I have claimed that both of them represent some type of state. I believe that it would be compatible with my present account an expansion of the prepositional domain to derive the stative preposition with, with a meaning similar to have, from the locative preposition in, following parallelisms with the verbal domain (Freeze 1992), as long as the expansion takes place below the P dir level. See Koopman 1997, den Dikken 2003, Folli 2001, Tortora 2005, 2006 and Bosque 1993 for other compositional accounts of PP structure. Although this idea deserves to be explored further, I leave it for future research. 3.5. Why do verbs alternate? Up to this point, I have provided a syntactic description of the verbs that display the locative alternation. A complete list of all the subtypes is included in the appendix. Now some relevant questions arise: how can we constrain the system to include only alternating verbs? And what are the syntactic differences between alternating and non- alternating verbs? Some of the most famous cases of non-alternating verbs are put and fill and their Spanish counterparts poner and llenar. The verb put can only appear in the locative variant, while fill appears in the change of state variant. I include a list of some of the verbs that do not display locative alternation in (148) and (149) 124 (148) Change of location verbs (put): poner “put”, colocar “place”, instalar “install”, alojar “lodge”, montar “mount”, posicionar “position”, establecer “set”, dejar “rest”, suspender “suspend”, colgar “hang”, meter “put in”, verter “pour”, derramar “pour”, enrollar “roll”, incrustar “inlay”. (149) Change of state verbs (fill): llenar “fill”, adornar “adorn”, vendar “bandage”, bañar “bathe”, atar “bind”, bloquear “block”, bombardear “bombard”, enmoquetar “carpet”, ocultar “cloak”, taponar “clog”, taponar “clog”, atascar “clog”, forrar “coat”, contaminar “contaminate”, cubrir “cover”, decorar “decorate”, ensuciar “dirty”, embellecer “embellish”, dotar “endow”, enmarcar “frame”, inundar “flood”, infectar “infect”, infestar “infect”, impregnar “impregnate”, enmascarar “mask”, contaminar “pollute”, saturar “saturate”, sazonar “season”, rodear “surround”, empapar “soak”. As far as their syntactic structure is concerned, it is clear that these verbs do not present an ambiguity between two possible interpretations when the PP is not present. For example, in the American army bombarded the city, the direct object the city can only be interpreted as location, and never as locatum, even when the lower PP (such as with nuclear weapons) is not present. This indicates that it is the verb the one responsible for theta-role assignment. This fact can be observed in the syntactic structure by generating 125 the direct object directly in the [spec [V]] position. The rest of the structure remains the same, as in (150). (150) Juan puso el heno en el granero “Juan put the hay in the barn” v 3 D v Juan 3 v V 3 D V 5 3 el heno i V P dir ! 3 puso P dir 3 P P loc 5 en el granero Exactly the same structure will be available for the verbs of the fill type. But how can we determine which verbs alternate? This is a difficult question and I think that learners use several clues to determine this, and not necessarily the same ones. An experiment conducted by Braine & Brooks 1995 with adults and preschool children show interesting results. When presented with novel verbs that could or could not display the causative alternation, around 50% of adults preferred always the variant given in the input. The other half either always preferred one of the variants of the alternation or used both variants with all verbs, irrespective of the context in which the verb was introduced. 126 Children always used the intransitive version when asked about themes, and the transitive version when asked about the agent, so they overgeneralized. This study clearly shows the extreme variation affecting the use of alternating verbs. One possibility to account for this might be that speakers only produce the structures that they have previously heard. Alternating verbs are not very common in Spanish because of their meaning (see the list of verbs in appendix 1, which includes verbs such as taracear “inlay” and rielar “glimmer”, which are not used in everyday speech) and many speakers do not have intuitions about their usage. However, I think that there are some clues that may help speakers decide if verbs can or not alternate. The first one is the meaning of the verb. It seems that light verbs are only associated with a specific structure. That is the case of put, sound, shine and remove, which cannot alternate. However, there is a high percentage of denominal verbs that show alternation, such as those in (151). (151) asperjar (spray) tr Denominal in Latin <aspergo –are < aspergo -inis f. [sprinkling , spray] cargar (load) tr Denominal in Spanish < n. carga espolvorear (dust) tr Denominal in Spanish < n. polvo pulverizar (pulverize) tr Denominal in Spanish < n. polvo rociar (spray) tr Denominal in Spanish < n. rocío salpicar (splash) tr/int Denominal in Spanish < n. sal & v. picar < n. pico 127 sembrar (sow) tr Denominal in Latin >lat. semin āre < semen –inis (seed) tiznar (spot, stain) tr Denominal in Spanish < v. tizonear < n. tizón (partially burned stick). It is also interesting that both English and Spanish show only three cases of alternating deadjectival verbs: aliviar –from lat. levis -e “light”–; vaciar “empty”; limpiar “clean/clear”. However, there are many deadjectival verbs that cannot alternate (inter alia aclarar “rinse”, llenar “fill”, depurar “depurate”, embellecer “embellish”, enriquecer “enrich”, establecer “establish”, liberar “free”, purificar “purify”, desvanecer (from lat. v. evanesco < adj. vanus = empty)). Many alternating verbs also tend to be characterized by repetitive events or a distributive meaning that was already observed by Jackendoff 1990 in verbs such as spray or sprinkle. These are verbs that can establish a “measurable scale” relation between the location (truck, plants…) and the locatum, namely an “amount” relation; e.g. “amount of sprinkled water on the plant”, “amount of loaded hay on the truck”. It is precisely for this reason that the Location can function as a scale that “measures out the event” (see Beavers 2006 for details on the notion of measuring out). Verbs such as put lack this property. This is also related to the fact that many non-alternating denominal verbs are formed from terms that refer to specific objects, such as bomb, bandage, block, carpet, cloak, coat, cover, frame, and mask. Alternating denominal verbs, on the other 128 hand, tend to be derived from nouns such as load, spray, sprinkle, pile, jam, dust,… which refer to quantity nouns or to liquids. In section 2.3, I mentioned that alternating verbs seem to have a rich meaning that enables them to alternate. Rodríguez Ramalle 2006, when describing emission verbs, postulates that they have a manner component that makes their alternation possible. L&R 1998 also included a manner component in the lexical conceptual structure of the verbs that enter the locative alternation. However, Talmy 1985, 1991 and 2000 classified Romance languages as conflating path, and not manner. Zubizarreta & Oh 2007 also postulate the presence of a manner component in the directed motion construction in Germanic languages which makes possible the productive creation of manner of motion constructions, through a productive compounding process, using typically unergative verbs such as dance in John danced to the kitchen. Grosso modo, the absence of this manner component and the lack of compounding in Romance makes this type of constructions impossible in Spanish. However, if we understand Talmy’s generalization not as an absolute rule, but as a relative one, we might be able to reconcile Rodríguez Ramalle and L&R’s observation, with Talmy’s generalization and its application in Zubizarreta & Oh. In particular, I would like to suggest that in the same way as English has verbs with the path incorporated, such as enter, Spanish may have some verbs that express manner. However, this does not mean that the process of manner incorporation is productive in Spanish, or that paths can freely conflate with verbs in English. Instead, I suggest that because the incorporation of manner is not productive in Spanish, because Romance lacks the compositional algorithm that gives rise to productive compounding, the number of alternating verbs should be much smaller than in English. And indeed this seems to be the case. For example, there are approximately 178 verbs of the swarm type in English, following Levin’s 1993, while only about 28 in Spanish. Interestingly, those swarm type verbs that enter the locative alternation in Spanish seem to have a clear preference for the locative variants, with the preposition en “in”, rather than the with variant. Figure 4 includes the most frequently occurring swarm type verbs in Spanish and it is part of the variationist experiment developed in the in chapter 4 (see chapter 4 for a detailed description). Figure 4: Distribution of PPs in swarm verbs 7.1% 83.6% 9.3% Other locatives en con/de Figure 4 explicitly shows how the with variant only appears a 9.3% of the cases, which indicates that the distribution of the locative and with variants is not completely 129 130 homogeneous. This easily follows from the non-productivity of manner incorporation in Spanish. All these observations leave room for language variation, due to the lack of a strict system to recognize alternating verbs. I think this result is desirable because of the great variation observed between native speakers of the same language and crosslinguistically. 131 Chapter 4 On subject position, unaccusativity and the swarm type in Spanish 4.1. Introduction In previous chapters, I have already indicated how all alternating verbs seem to have unaccusative properties. I have explained how a particular subgroup, the swarm type, has been classified both as unaccusative (Perlmutter 1978) and unergative (Levin & Rappaport 1995) in the specialized literature. This section is specifically designed to provide further evidence that supports a constructional view where all alternating constructions are always unaccusative. This does not require that these verbs have always unaccusative properties in every context. On the contrary, I will show how some of these verbs, such as brillar “shine”, can behave as either an unergative or an unaccusative verb, depending on the construction where they are inserted. In this chapter, I develop a corpus study to show that the swarm verbs have unaccusative properties whenever they appear in the locative alternation construction. First I will discuss the properties of Spanish unaccusative verbs and show that they lack an overt morphological marking that could differentiate them from unergative verbs. Then, I present a variationist analysis of subject position in Spanish to demonstrate that the position of overt subjects can be used as a reliable test to distinguish between the three main types of verbs: transitive, unergative and unaccusative. Finally, I apply the subject position test to show that all the swarm type verbs studied here behave as 132 unaccusative when they appear in the locative alternation construction, supporting the claims made in the previous chapter. 4.2. On the relation between verb types and subject position in Spanish The main purpose of this chapter is to provide further evidence that support an unaccusative classification of the swarm type verbs. In chapter 3, I provided crosslinguistic evidence that supports this claim. This section provides experimental evidence of the unaccusative classification of the swarm verbs in Spanish, by using the subject position test developed in Mayoral Hernández 2004c, which is presented in detail in section 4.2.3. This test is based on the following observation: Spanish unaccusative verbs tend to have a higher percentage of postverbal subjects when compared to unergative verbs, which, by contrast, show a clear preference for preverbal subjects. In Mayoral Hernández 2004c, I carried out an experiment, using the variationist approach, whose goal was to determine if there is a relation between verb types and subject position. In this section I will first provide an introduction to the variationist method. Then, I will briefly describe the properties of Spanish subjects. In the rest of this section, I will present several experiments that I have conducted to determine if there exists a reliable test that can be used to differentiate between different verb types, or constructions in Spanish. 133 4.2.1. The variationist approach The variationist approach should be understood as a methodology used for linguistic research. It is not framework specific, although it was probably first employed for sociolinguistic studies (Labov 1969, 1994). This method utilizes quantitative analyses of data and probability theory to extract regularities within a system. It is not designed to study categorical rules, but variable rules. For instance, it cannot be used to study the distribution of nasal + <b> within a word in the Spanish spelling system, because, by convention, the grapheme <b> can only be preceded by the nasal <m> (combinar ‘combine’ vs. *conbinar), unless of course one wants to study misspellings. For this reason, it is difficult to find a probability of 1.0, i.e. 100%, when studying linguistic variation, since there should always be room for arbitrary speaker preferences. Studies of variation make use of two types of statistics: descriptive and inferential. Descriptive statistics, such as the chi-square test ( χ 2 ), are used to present and describe data. Inferential statistics, such as a regression analysis, is used to draw conclusions derived from the observation of the data and make predictions, so it goes beyond descriptive statistics. The factor that is being analyzed in any given study, such as subject position, is called the dependent variable. This factor is then cros-tabbed with other factors to examine if they are related. Those other factors are called independent variables. In a linguistic study of subject position, dependent variable, in Spanish, one might want to consider several other factors, independent variables, that might influence the position of the subject in the sentence, such as information structure and focus (Zubizarreta 1998), sociolinguistic variables (Silva-Corvalán 1982, Silva Corvalán & 134 Sánchez Walker 2007), presence of co-occurring adverbials (Mayoral Hernández 2004a and 2004b),… Variation studies take advantage of new advances in mathematics and statistics applied to social sciences. There are two main software tools used in this field. One of them is SPSS (Statistical Package for the Social Sciences), the other is GoldVarb. While the first one includes many different statistical tests, such as t-tests, ANOVA and different types of regression analyses, the latter is specifically designed for sociolinguistic research to calculate probabilities using binomial and multinomial logistic regressions. In my research, I use the program SPSS to display crosstabulations, which is a contingency table that shows the joint distribution of two categorical variables, and then I apply the Pearson’s chi-square test to find out if the null hypothesis the null hypothesis is rejected. A probability value lower than 0.05 (p < 0.05) has been adopted as the threshold for statistical significance. 4.2.2. Subject position in Spanish Spanish is a pro-drop subject language, which implies that subjects can be omitted. When they appear overtly, they must agree in number and person with the verb. This agreement relationship is marked by overt verbal and nominal morphology. The following examples (152) show how the verb amar “love” has a different ending depending on the person and number of the subject. 135 (152) a. Yo amo “I love” b. Tú amas “you love” c. Él ama “he loves” d. Nosotros amamos “we love” e. Vosotros amáis “you love” f. Ellos aman “they love” As far as subject position is concerned, agreeing subjects can appear in preverbal or postverbal position, irrespective of the type of verb. The following examples, which have been extracted from the online Corpus de Referencia del Español Actual (CREA), show sentences that contain transitive (155), unergative (154) and unaccusative (153) verbs with both preverbal and postverbal subjects. The verb appears marked as V, while the subject is marked with the letter S. (153) a. Postverbal subject with the unaccusative verb llegar “to arrive”: En febrero llega (V) el grupo Queen a la Argentina (S). “In February the band Queen arrives in Argentina” b. Preverbal subject with unaccusative verb: En 1896, Picasso llega a Barcelona. “In 1896, Picasso arrives in Barcelona.” 136 (154) a. Postverbal subject with the unergative verb trabajar “to work”: En el coro trabaja (V) un maestro de tracerías y taraceas (S). “In the choir works a master of tracery and marquetry” b. Preverbal position with unergative verb: El minero trabaja durante toda la noche … “The miner works during the whole night …” (155) a. Postverbal subject with the transitive verb empujar “to push”: Sólo empuja (V) el libro desde el borde de la mesa, aquel que tensiona la espalda (S) “Only the one that bends his back pushes the book from the table end” b. Preverbal position with transitive verb: Un niño chino empuja su carrito en la ciudad de Kunming. ‘A Chinese child pushes his cart in the city of Kunming.’ Although overt subjects can appear both in preverbal or postverbal position, their ordering is not completely random, and ongoing research shows that it depends on various grammatical and sociolinguistic factors (Silva-Corvalán 1992, Zubizarreta 1998, Mayoral Hernández 2004c, Hinch 2007). 137 4.2.3. Experiment 1: Mayoral Hernández 2004c This experiment was designed as a complement to a larger one, where I studied the position of Spanish frequency adverbials. In Mayoral Hernández 2004b I pointed out that frequency adverbials tend to appear in complementary distribution with overt subjects, as seen in the following table. Table 1: Influence of subjects on the position of adverbials, including wh- Subjects 20 Subject position Position of adverbials Omitted Postverbal Preverbal Wh- Subject Total Preverbal Adverbial 291 50.4% 80 62.5% 131 30.6% 100 50.5% 602 45.2% Postverbal Adverbial 286 49.6% 48 37.5% 297 69.4% 98 49.5% 729 54.8% Total 577 100.0% 128 100.0% 428 100.0% 198 100.0% 1331 100.0% Chi-Square test Value df Asymp. Sig. (2-sided) Pearson Chi-Square 60.883 a 3 .000 Likelihood Ratio 62.116 3 .000 This table shows that adverbs and subjects tend to appear in complementary distribution. Whenever there is a preverbal subject, the adverb will tend to appear in postverbal position, and vice versa. However, the literature dealing with subject position and verb types has observed that postverbal subjects seem to be associated with 20 A wh-Subject is a wh- word (or qu- word in Spanish) that functions as a subject in the sentence, such as the relative pronouns who or que in the sentences I saw the person who stole my purse and vi a la persona que me robó el bolso. They were assigned a code that was different from other subjects because their position is obligatorily clause initial. 138 unaccusative verbs (Torrego 1989). If this is true, it might be possible that the apparent effect that subject position has on adverbials is only due to the type of verb used in the sentence. In order to clarify this possible issue, I recoded my corpus to add information about the type of verb. The results of this last analysis are presented in Mayoral Hernández 2004c. The general idea is that unaccusative verbs should have a higher percentage of postverbal subjects when compared to unergative, transitive and copulative verbs. The results indicated that while copulative, transitive and unergative verbs showed a similar subject distribution, unaccusative verbs showed a significantly higher percentage of postverbal subjects (p< 0.05), as seen in table 2. This result can be expected from the fact that in a pro-drop language like Spanish, where subject position is variable, subjects with object properties, i.e. the subjects of unaccusative verbs, will tend to appear in postverbal position more frequently. This prediction was supported by the data, shown on the following table. Table 2: Verb types and subject distribution Type of verb Position of subject Intransitive Copulative Transitive Unaccusative Total Postverbal 6 12.2% 12 15.4% 23 11.5% 25 33.8% 66 16.5% Preverbal 43 87.8% 66 84.6% 177 88.5% 49 66.2% 335 83.5% Total 49 100.0% 78 100.0% 200 100.0% 74 100.0% 401 100.0% 139 Table 2 that intransitive (unergative), transitive and copulative verbs tend to appear in preverbal position in about 85 percent of the cases, and the difference between them is not statistically significant (p = 0.678). Unaccusative verbs showed a lower percentage of preverbal subjects (62.2%), which caused a statically significant difference with respect to the other three types of verbs (p < .0001). The fact that unaccusative verbs have a higher percentage of postverbal subjects follows from the assumption that they are underlying objects (see discussion in section 1.4), and the unmarked position of objects in Spanish is postverbal. Therefore, they seem to be hybrid linguistic constructs, with subject and object properties. However, the analysis in Mayoral Hernández 2004c raises some questions, due to the way in which the experiment was designed. First of all, the experiment was originally intended to study the position of frequency adverbials in Spanish, and subject position was only one of the several independent variables included. Because of this, the rich Spanish verbal morphology was not controlled for, and there is no clear understanding of the consequences this might have had on the results. As an example, one might expect that subject position could vary greatly depending on grammatical person, in particular if we compare first person singular with third person. In fact, the only first person subject possible is the personal pronoun yo “I”, which always represents given information (Prince 1981). Considering that Spanish has a preference for placing old information in preverbal position (Silva Corvalán 2003), and personal pronouns, by necessity, tend to be old information, it becomes obvious that the lack of control of this grammatical feature might influence the results. 140 Another possible problem that might derive from the methodology adopted by Mayoral Hernández op. cit. is that the classification of verbs as unaccusative or unergative, necessary for the coding of data, was based on Mendikoetxea’s 1999 non- exhaustive list of unaccusative verbs. Therefore, decisions had to be made regarding the classification of some verbs as either unaccusative or unergative when they did not appear in Mendikoetxea’s list. The classification adopted might have been perfectly accurate, but again, one might question this procedure after taking into account that the right classification of unaccusative and unergative verbs is still controversial, due to the lack of reliable tests (see section 1.4 for further discussion). Due to the shortcomings found in the methodology adopted in my previous work, I decided to design a new study where all the aforementioned factors could be controlled for. This new study is the subject of the following section, which develops an experiment designed to take more independent factors into account. 4.2.4. Experiment 2: Mayoral Hernández 2006 As noted before, subject position might be a good indicator of verb class in Spanish, since it might be able to differentiate between unergative and unaccusative verbs. The experiment that will be described in this section is a multifactorial variationist analysis that was specifically designed to test if transitive, unergative and unaccusative verbs have a unique subject distribution. 141 4.2.4.1. Research hypothesis and factors This research focuses on Subject Position in Spanish, the dependent variable. Mayoral Hernández 2006 is mainly concerned with the relationship between verb types and subject position. Therefore, the only independent variable analyzed is Verb Type, i.e. whether verbs are transitive, unergative or unaccusative. The research hypothesis states that unaccusative verbs will show a significantly higher percentage of postverbal subjects when compared to unergative and transitive verbs. 4.2.4.2. Methodology In this study, I extracted a total of 450 sentences from the online corpus CREA (Corpus de Referencia del Español Actual), which contains written texts in Spanish. Only those verbs whose classification is uncontroversial were selected. In particular, the verb empuja ‘he/she pushes’ represented transitive verbs, while trabaja ‘he/she works’ and llega ‘he/she arrives’ represented unergative and unaccusative verbs respectively. These verbs show a very stable classification crosslinguistically. In those languages that have overt unaccusative morphology, such as Italian, Dutch or French, the verb arrive selects the auxiliary be in periphrastic forms, its subject can be substituted by the clitic ne in Italian, etc. Similarly, the verb work shows very clear unergative properties across languages. Verbal morphology was strictly controlled too, since only the third person singular of the present simple tense, indicative mood, active voice was included. This corpus was selected because it contains mostly written data extracted from different 142 sources (see Newmeyer 2003 for discussion of the features of spoken language). The 450 sentences comprise the data that was finally used for the analysis, although a much bigger corpus was originally extracted. Additionally, the three verbs analyzed here were not considered when they appeared in constructions that do not represent the target verb type they were selected for. Therefore, the transitive uses of trabajar, the periphrastic construction llegar a + infinitive “get to”, the intransitive uses of empujar, etc. were not included in the final corpus. Many sentences were excluded for different reasons. Subject is defined as the argument that agrees with the verb. Sentences containing no overt subject were obviously omitted as well as subjects realized by a relative pronoun, since their position must always be clause initial, irrespective of the verb. Only affirmative and negative sentences have been included in the analysis, because interrogative and exclamative sentences are associated with different processes of topicalization and subject inversion which would bias the results of this research. The program SPSS was used as a tool to carry out the statistical analysis. The results are presented in crosstabulation tables, with the Pearson’s chi-square value and the probability estimate. 4.2.4.3. Results In table 3, I compare transitive and unergative verbs in order to find out if their subject distribution is similar. Indeed, it shows that there is a statistically significant 143 difference (p < .0001) between transitive and unergative verbs as far as subject position is concerned. Table 3: Subject distribution in unergative and transitive verbs (p < .0001) Type of Verb Transitive Unergative Total Count 5 32 37 Postverbal Column % 3.0% 21.5% 11.8% Count 160 117 277 Subject Position Preverbal Column % 97.0% 78.5% 88.2% Count 165 149 314 Total Column % 100.0% 100.0% 100.0% While both types of verbs clearly display a higher percentage of preverbal subjects, this preference is much more accentuated in the case of transitive verbs. It is interesting to note that Mayoral Hernández 2004c did not find a significant difference between these two types of verbs. The difference in the results might be linked to the way in which I separated unergative from unaccusative verbs in the former experiment. Sorace 2000 noticed that while some verbs show a fairly stable crosslinguistic behavior, as far as auxiliary selection is concerned 21 , most of them experience different degrees of variation. Thus, she proposed an auxiliary selection scale instead of a clear-cut division. Similarly, Torrego 1989 showed that, under certain conditions, typically unergative verbs behaved as unaccusatives. Zubizarreta & Oh 2007 also explained that manner of motion verbs, which are typically unergative, can show unaccusative properties when they appear in a sentence with a PP in Germanic languages such as English, because they imply a notion of change. Mayoral Hernandez 2004c included a list of randomly selected 21 Recall that auxiliary selection is closely linked to the unaccusative / unergative distinction. 144 sentences, with no control of the type of verb included. However, this analysis only includes verbs with an uncontroversial crosslinguistic status. This is possibly the reason why the verbs included in the former analysis showed less difference with respect to subject position, while Mayoral Hernández 2006, which only included prototypical verbs, displayed a more marked distinction. As expected, the comparison between unergative and unaccusative verbs also indicated a statistically significant difference between them (p < .0001), since unaccusatives show no clear preference for either preverbal or postverbal subjects. Table 4: Subject distribution in unergative and unaccusative verbs (p = .0001) Type of Verb Unergative Unaccusative Total Count 32 62 94 Postverbal Column % 21.5% 45.6% 33.0% Count 117 74 191 Subject Position Preverbal Column % 78.5% 54.4% 67.0% Count 149 136 285 Total Column % 100.0% 100.0% 100.0% A detailed analysis of these results seems to indicate that when only prototypes are selected and other factors are controlled there is a clear difference between the three types of verbs, as far as their subject position is concerned (p < 0.05). However, in a corpus containing many different verbs, like the one used in my previous research, the boundaries between verb types will be partially blurred. This result should be related to the fact that the boundaries between transitive and intransitive verbs and between unergative and unaccusative verbs are not so clear and more research is needed to create reliable classifications (see Campos 1999 for transitivity versus intransitivity). 145 Nevertheless, the current analysis coincides with Mayoral Hernández 2004c when observing the clear distinction between unaccusative and unergative verbs. Therefore, there seems to be enough evidence to conclude that a quantitative analysis of subject position in Spanish can successfully differentiate between unergative and unaccusative verbs. In the next section I will use this test to shed some light on the correct classification of verbs of sound and light emission (VLSEs), which are included within the swarm type. 4.2.4.4. Verbs of Light and Sound Emission in Spanish: the test Most of the verbs included in the swarm class are emission verbs. I have already pointed out the difficulties that one might find when trying to classify swarm verbs as unaccusative or unergative. This difficulty has been highlighted by the lack of agreement between researchers dealing with this issue. The verbs of light and sound emission (VLSEs) have been classified both as unaccusative (Perlmutter 1978) and unergative (Mendikoetxea 1999). Instead of arguing for a clear-cut distinction between unaccusative and unergative verbs, Sorace 2000 proposed an “auxiliary selection hierarchy”. She includes emission verbs somewhere near the middle of the scale, in between pure unaccusative and unergative verbs. In chapter 3, I defended their classification as unaccusative and provided evidence from crosslinguistic comparison and their semantic properties. 146 In this section, based on Mayoral Hernández 2004c, I apply the subject position test to VLSEs, in order to provide further evidence defending their classification as unaccusative verbs. I added 274 sentences more to the corpus used in the previous section for a total of 724 tokens. These sentences were also extracted from CREA and contain the verbal forms resuena ‘it resounds’ and resuenan ‘they resound’. Recall that, in experiment 2, I have only included the 3 rd person singular form of the selected verbs (llega, resuena and empuja). The third person plural form resuenan has been included here in order to increase the number of tokens of the verb resonar “resound”, because CREA did not contain enough sentences with the singular form, resuena, to create reliable statistics. In the coding, I specified when the verb appears in the locative alternation construction. The purpose of this annotation is to observe whether subject distribution is the same in alternating and non-alternating constructions. The research hypothesis states that these verbs should have a subject distribution that is similar to unaccusative verbs, so they should show no clear preference for either preverbal or postverbal subjects, as seen in the previous section. 4.2.4.5. Verbs of light and sound emission: Results Table 5 compares the behavior of VLSEs that appear in the locative alternation construction, and that therefore contain an argument PP, with other uses in which there is only one argument, the subject. The very high probability value (p = 0.384) indicates that there is no statistically significant difference between these two groups of verbs. 147 Therefore, the 274 tokens containing this verb will be recoded together for the rest of the analysis. Table 5 also indicates that the verb resonar “resound” shows no clear preference for either preverbal or postverbal subjects, irrespective of the construction where it is inserted. Table 5: Subject position: VLSEs with and without locative alternation (p = .384) Type of Verb VLSE with loc. altern. VLSE without loc. altern. Total Count 81 56 137 Postverbal Column % 47.9% 53.3% 50.0% Count 88 49 137 Subject Position Preverbal Column % 52.1% 46.7% 50.0% Count 169 105 274 Total Column % 100.0% 100.0% 100.0% Table 6: Subject position by verb type: VLSEs and unaccusative verbs (p = .400) Type of Verb VLSE. Unaccusative Total Count 137 62 199 Postverbal Column % 50.0% 45.6% 48.5% Count 137 74 211 Subject Position Preverbal Column % 50.0% 54.4% 51.5% Count 274 136 410 Total Column % 100.0% 100.0% 100.0% When the verb resonar “resound”, now recoded into one variable, is compared with the verb llegar “arrive”, a typical unaccusative, the high probability value (p = .400) indicates that they have a similar subject distribution, since neither of them shows a clear 148 preference for either preverbal or postverbal subjects, as seen in table 6. Therefore, VLSEs and unaccusative verbs pattern together, as far as subject position is concerned. However, when comparing unergative verbs with VLSEs (table 7), the results yield a very low p value (p = .0001), which shows that VLSEs and unergative verbs have a different subject distribution. In fact, while unergative verbs clearly prefer preverbal subjects, this preference cannot be observed in VLSEs, whose syntactic subjects show no clear preference for either preverbal or postverbal position. Therefore, emission verbs cannot be described as unergative in Spanish; they pattern together with unaccusative verbs with respect to subject position. Table 7: Subject position by verb type: VLSEs and unergative verbs (p = .0001) Type of Verb VLSE Unergative Total Count 137 32 169 Postverbal Column % 50.0% 21.5% 40.0% Count 137 117 254 Subject Position Preverbal Column % 50.0% 78.5% 60.0% Count 274 136 423 Total Column % 100.0% 100.0% 100.0% These results seem to suggest that the emission verb resonar “resound” can be better described as an unaccusative verb but, can we generalize this observation to the rest of alternating verbs? The following experiment includes all the alternating verbs that have a significant amount of tokens in the corpus CREA. It has been designed to support the claims made in this section, by providing evidence from other alternating emission verbs and adding a tighter control of other syntactic factors. 149 4.3. A variationist analysis of subject position: swarm verbs The data observed in the previous experiments reveal two important things. On the one hand, there seems to be a relation between verb types and subject position in Spanish. In particular, while transitive and unergative verbs have a higher percentage of preverbal subjects, unaccusative verbs show no clear preference. Thus, one might derive that the subject position test can be used to identify verb types. On the other hand, the verb of sound emission resonar “resound” patterns together with unaccusative verbs, as far as subject position is concerned, which might indicate that this verb is indeed unaccusative. The statistically significant difference between unergative and unaccusative verbs is supported by several experiments. While Mayoral Hernández 2006 selected only one verb to represent each verb type, previous research (Mayoral Hernández 2004c) included a wider range of verbs for every verb type. The results from both analyses coincided in the fact that there is a significant difference between unergative and unaccusative verbs as far as subject position is concerned. This evidence demonstrates that the subject position test is a reliable means of identifying verb classes. In section 4.3, I present a more complex experiment that includes five more verbs of the swarm type and a more detailed coding, specified in figure 5. The only criterion used to select the newly included verbs form the list of all swarm type verbs (see appendix 1), which includes VLSEs verbs, is that they should have more than 200 tokens in the corpus CREA, to ensure the validity of the statistics used. This means that only six verbs fulfilled this requirement, one of them being resonar “resound”. 150 In the following sections I will present a new experiment whose purpose is to add further evidence supporting the classification of swarm verbs as unaccusative. 4.3.1. Methodology This experiment takes as a point of departure the corpus used in Mayoral Hernández 2006. It includes a total of 2141 sentences that contain the following verbs: empujar “push”, representing transitive verbs; trabajar “work”, prototypical unergative; llegar “arrive”, typical unaccusative; and the swarm type verbs resonar “resound”, abundar “abound”, arder “burn”, brillar “shine”, hervir “boil” and pulular “swarm”. Following the procedure adopted in Mayoral Hernández 2006, the only tense used is the third person singular of the present tense, indicative mood in the active voice was selected. As in the case of resonar “resound”, I have included the third person plural forms of verbs pulular “swarm” and abundar “abound” to increase the number of tokens of these two verbs and thus enable reliable statistics. All the tokens that appeared in the corpus CREA that contained these verbs in the forms specified are included in this research. The original number of sentences extracted from CREA was far more extensive, but I excluded all those that were ambiguous or did not have overt subjects 22 . Interrogative and exclamative sentences are not included in the final corpus, because, as mentioned before, they involve different processes of subject inversion. In section 4.2.4.2 there is a more detailed description of other methodological aspects. 22 Most of the sentences excluded from the analysis lack a subject. Additionally, ambiguous sentences are also eliminated. A sentence may be ambiguous for many reasons; for example, in “Actor Dos se para, resuena el silbato, se sienta”, from CREA, the NP el silbato “the whistle” may be analyzed as either the direct object or the subject of the verb resonar “resound”, which is normally intransitive but can also be used as a transitive verb in formal language. 151 4.3.2. Factors and research hypotheses As in the previous experiments, the independent variable is subject position. The complete list of codes and variables used for this research appears in figure 5. Figure 5: List of variables and their values 1. Dependent variable: Position of Subjects 23 • Before XP in Preverbal position b • Adjacent to the left of the verb l • Adjacent to the right of the verb d • After XP in Postverbal position a 2. Type of verb • Unaccusative U • Intransitive (unergative) I • Transitive with overt DO T • Transitive with pronominal DO t • Transitive with DO duplicated (clitic & DO NP) R • Verbs of Light/sound emission (Loc. Alternation) E • Verbs of Light/sound emission (no Loc. Alternation) e • Verbs of Light/sound emission (others) v 3. Number • Singular u • Plural m 4. Presence of Preverbal XP • Argumental (DO or IO or alternating PP) O • Other XP (Adjunct) X • CP C • No XP N 23 Clitics do not count as XPs and have been excluded because their position is not variable: they always appear in preverbal position, adjacent to finite verbs; they appear as an enclitic particle at the end of imperatives and non-finite forms. 152 5. Presence of Postverbal XP • Argumental (DO or IO or alternating PP) o • Other XP (Adjunct) x • CP c • No XP n 6. Type of subject • Demonstrative/Possessive pronoun or adjective D • Personal pronoun P • Proper noun G • Clausal S • Indefinite article i • Definite article K • Other Y 7. Type of clause • Main clause p • wh- (Qu-) clause s • Subordinate clauses with conjunction k 8. Sentence polarity • Affirmative A • Negative - 9. Type of alternant • con c • en e • de d • other locatives L • not applicable h As in most variationist studies, I have included variable values that may not be used in the present analysis. As far as the dependent variable is concerned, the only two values that I analyze in this dissertation are preverbal and postverbal, which require the recoding of the four values that appear in figure 5 into only these two. On the one hand, one of the reasons for this recoding is purely methodological, and it is due to the lack of enough tokens to make 153 reliable statistics for certain positions, such as postverbal subjects with transitive verbs. On the other hand, from a theoretical point of view and for the purpose of this dissertation, I am only interested in finding out if subject position is predominantly preverbal (and therefore associated with transitives and unergatives) or postverbal (connected to unaccusative properties) in order to differentiate unergatives from unaccusative verbs. I have also recoded the first independent variable, type of verb. All the transitive constructions have been recoded into one single code, because there was no statistically significant difference between them. The category Verbs of Light/sound emission (others) includes those sentences that can alternate but that appear with none of the standard prepositions en “in”, con “with” or de “of/from”. It also includes sentences that apparently have the appropriate locative alternation structure (a subject, a verb, and a locative XP, for example) but that for some reason cannot alternate. For instance, in example (156) the locative phrase is an adverb, and, therefore, even if it has a possible alternating structure [Verb + Subject + Locative], adverbs cannot function as a subject, which makes the alternation impossible. Because of this, I annotated the sentence (156) with the code “v”. (156) porque arde la rebelión abordo (CREA) “because the rebellion bursts onboard” 154 I added the variable number because, as in the case of resonar “resound”, I had to include the plural form of three of the swarm type verbs to compile a significant amount of tokens. Variables 4 and 5 are especially important because they can be used to control the presence of other XPs in the sentence. In Mayoral Hernández 2004a and 2004b, I show how the human processor normally prefers a balanced distribution of weight between the preverbal and postverbal domains, at least in Spanish, where order is more variable than in English. The type of subject variable includes information about the morphological realization of the subject phrase. For example, the code definite article represents a subject phrase headed by a definite article, such as el coche “the car”, while I use the code Demonstrative/Possessive pronoun or adjective for phrases headed by possessive or demonstrative adjectives 24 . The type of clause variable identifies the clause where the alternating verb appears. The code “s” is used to identify subordinate clauses headed by a complementizer with a syntactic function. So this code is only used for relative clauses, where the relative pronoun may function as subject, adjunct, etc, such as the pronoun that in the sentence I ate the ice-cream that i you prepared t i , where it functions as a direct object of prepare. The code “k” identifies all subordinate clauses headed by a complementizer without a 24 I use these terms in the traditional way. (1) I like this car, vs. I like this. (2) I like my car, vs. this car is mine. In the sentences in (1) this is a demonstrative adjective in the first one and a demonstrative pronoun in the second one. In the sentences in (2), my is a possessive adjective, while mine is a possessive pronoun. 155 syntactic function, such as the complementizer that in the sentence I said that I want an ice-cream. The variable sentence polarity is designed to determine if this feature can influence subject position. The last variable, type of alternant, only applies to alternating verbs, and it signals the type of preposition that heads the argument PP. The code “locatives” identifies argument PPs headed by locative prepositions that are not en “in”, such as por “by”, sobre “on, over” and dentro de “inside of”. The code “h”, meaning not applicable, comprises those sentences with no alternation, such as those containing the verbs llegar, trabajar and empujar, as well as non-alternating cases of the other verbs. Every independent variable is associated with at least one hypothesis. Of the multiplicity of possible tests that could be done with all this annotation, I will focus only on the ones listed in (157). (157) 1. Type of verb: all the verbs that appear in a construction that displays the locative alternation will have unaccusative properties, i.e. the subjects will have a distribution similar to that of llegar “arrive”. 2. Number: this factor should not have any influence on the position of overt subjects. 3. Presence of Preverbal XP: subjects will tend to be postverbal. 4. Presence of Postverbal XP: subjects will tend to be preverbal. 156 5. Type of subject: those subjects that can be associated with new information will tend to appear postverbally. Given information will tend to appear in preverbal position. 6. Type of clause: there will be a higher percentage of postverbal subjects when they appear in a sentence headed by a relative pronoun (Gutiérrez-Bravo 2006). 7. Sentence polarity: this factor should be associated with a higher number of preverbal subjects (see section 4.4). 8. Type of alternant: argument PPs that are headed by a locative preposition should have a higher percentage of preverbal occurrences when compared to the other PPs, since they may experience locative inversion, which implies that the locative PP will appear in preverbal position while the subject will become postverbal (Torrego 1989, Fernández Soriano 1999). 4.3.3. Results: On verb types 4.3.3.1. Brillar The analysis of the data containing the verb brillar “shine” indicates that there is no statistically significant difference between sentences where this verb enters the locative alternation and those classified as “Verbs of light/sound emission (others)”, where brillar occurs in non-canonical types of alternation 25 . so I have recoded the codes E and v together before proceeding any further. 25 Recall that “Verbs of light/sound emission (others)” include verbs that could show locative alternation but which either do not occur with the standard prepositions or cannot alternate for other reasons (see example 156 in section 4.3.2. 157 However, the comparison between the alternating and non-alternating uses of brillar shows a statistically significant difference between them, as table 8 illustrates. Subjects tend to appear in preverbal position (70.3%) when this verb occurs in a non- alternating construction. The opposite happens when it appears in the locative alternation construction, where subjects tend to be postverbal (56.7%). The difference between the two constructions is statistically significant (p < .0001). Table 8: Alternating vs. non-alternating constructions with brilla Type of Verb Non-alternating brilla Alternating brilla Total Count 99 72 171 Postverbal Column% 29.7% 56.7% 37.2% Count 234 55 289 Subject position Preverbal Column% 70.3% 43.3% 62.8% Count 333 127 460 Total Column% 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square 28.619 1 .000 Likelihood Ratio 28.009 1 .000 These observations anticipate the results of tables 9, 10 and 11. When comparing the alternating cases of brillar with a typical unergative verb, in table 9, the difference is statistically significant (p < .0001), indicating that they do not pattern together. However, when alternating brillar is compared with an unaccusative verb (table 10), it is possible to observe that both of them have a much higher percentage of postverbal subjects than unergative verbs. However, because alternating brillar has a more accentuated percentage of postverbal subjects than the unaccusative verb llega “arrives”, the p value becomes 158 non significant. In any case the difference, if from a theoretical point of view unaccusative verbs should have a higher percentage of postverbal subjects than unergatives, the verb brilla should clearly be considered unaccusative when it appears in alternating constructions. However, when it appears in any other construction, it seems to behave as an unergative verb, as table 11 shows. Table 9: Subject position by verb type: alternating brilla and unergatie verbs Type of Verb Alternating brilla Unergative Total Count 72 32 104 Postverbal Column% 56.7% 21.3% 37.5% Count 55 118 173 Subject Position Preverbal Column% 43.3% 78.7% 62.5% Count 127 150 277 Total Column% 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-square 36.670 1 .000 Likelihood Ratio 37.355 1 .000 Table 10: Subject position by verb type: alternating brilla and unaccusative verbs Type of Verb Alternating brilla Unaccusative Total Count 72 61 133 Postverbal Column% 56.7% 45.2% 50.8% Count 55 74 129 Subject Position Preverbal Column% 43.3% 54.8% 49.2% Count 127 135 262 Total Column% 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-square 3.467 1 .063 Likelihood Ratio 3.475 1 .062 159 Both the unergative verb trabajar and the non-alternating brillar have a clear preference for preverbal subjects (table 11), although it is interesting to note that the p value is slightly above .05, which is the significance threshold. Table 11: Subject position by verb type: non-alternating brilla and unergative verbs Type of Verb Non-alternating brilla Unergative Total Count 99 32 131 Postverbal Column% 29.7% 21.3% 27.1% Count 234 118 352 Subject position Preverbal Column% 70.3% 78.7% 72.9% Count 333 150 483 Total Column% 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square 3.689 1 .055 Likelihood Ratio 3.797 1 .051 This discussion leads to the conclusion that brillar displays unergative properties when it does not alternate, since it prefers preverbal subjects (70.3%), but when it appears in the locative alternation it behaves as an unaccusative, with a high percentage of postverbal subjects (56.7%). 4.3.3.2. Arder For this verb, I have also recoded “v” (other alternating construction) with those coded as “E”, since the number of cases where arder appears in non-canonical alternating constructions is only 15. 160 In the case of this verb, it does not matter whether it appears in an alternating or non-alternating construction. Subject distribution is balanced between preverbal and postverbal occurrences. Table 12: Subject position by verb type: non-alternating vs. alternating arde Type of Verb Non-alternating arder Alternating arder Total Count 66 37 103 Postverbal Column % 49.3% 50.7% 49.8% Count 68 36 104 Subject position Preverbal Column % 50.7% 49.3% 50.2% Count 134 73 207 Total Column % 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square .039 1 .844 Likelihood Ratio .039 1 .844 From table 12, it can be assumed that arder “burn” behaves as an unaccusative verb as far as subject position is concerned, since neither of them shows a clear preference for either preverbal or postverbal subjects. Table 13 clearly indicates that the verb arder always behaves as an unaccusative verb, because there is no statistically significant difference between these two verb types (p = .408). As expected, when the verb arder is compared with an unergative verb the difference turns out statistically significant (p < .0001). As a way of conclusion, the subject position test classifies the verb arder together with unaccusative verbs, irrespective of the construction where it may appear. 161 Table 13: Subject position by verb type: arder vs. unaccusatives Type of Verb Arder Unaccustive Total Count 103 61 164 Postverbal Column % 49.8% 45.2% 48.0% Count 104 74 178 Subject position Preverbal Column % 50.2% 54.8% 52.0% Count 207 135 342 Total Column % 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square .685 1 .408 Likelihood Ratio .685 1 .408 4.3.3.3. Abundar Due again to the low number of ambiguous alternating constructions, I have coded together v and E. The result of the comparison between alternating and non- alternating occurrences of abundar “abound” yields similar percentages for both groups. Both of them seem to favor postverbal subjects, as seen in table 14. If we assume that unaccusative verbs will either have a similar distribution of subjects across the preverbal and postverbal domains or have a higher percentage of postverbal subjects, we can conclude that this verb seems to behave as an unaccusative verb regardless of the construction where it occurs. 162 Table 14: Subject position by verb type: non-alternating vs. alternating abundar Type of Verb Non-alternating abundar Alternating abundar Total Count 131 152 283 Postverbal Column % 53.9% 60.6% 57.3% Count 112 99 211 Subject position Preverbal Column % 46.1% 39.4% 42.7% Count 243 251 494 Total Column % 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square 2.230 1 .135 Likelihood Ratio 2.232 1 .135 4.3.3.4. Hervir The verb hervir “boil” participates in the causative alternation, so it can occur in both transitive and intransitive (unaccusative) constructions. Apart from these constructions, it can also appear in locative alternation constructions of the swarm type. Because of its participation in the causative alternation it has traditionally been classified as an unaccusative verb (Mendikoetxea 1999). The transitive uses of this verb have not been included in this analysis. Table 15, where “v” (1 token) and “E” were recoded together, shows no statistically significant difference between constructions with and without locative alternation (p = .282), since there is no clear preference for either preverbal or postverbal subjects. 163 Table 15: Subject position by verb type: non-alternating vs. alternating hervir Type of Verb Non-alternating hervir Alternating hervir Total Count 56 24 80 Postverbal Column % 54.4% 45.3% 51.3% Count 47 29 76 Subject position Preverbal Column % 45.6% 54.7% 48.7% Count 103 53 156 Total Column % 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square 1.156 1 .282 Likelihood Ratio 1.157 1 .282 Table 16: Subject position by verb type: hervir vs. unaccusatives Type of Verb Hervir Unaccusative Total Count 80 61 141 Postverbal Column % 51.3% 45.2% 48.5% Count 76 74 150 Subject position Preverbal Column % 48.7% 54.8% 51.5% Count 156 135 291 Total Column % 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square 1.077(b) 1 .299 Likelihood Ratio 1.078 1 .299 The comparison with the unaccusative verb llegar “arrive”, in table 16, indicates no difference between hervir “boil” and unaccusative verbs with respect to subject placement whenever hervir is used in an intransitive construction, as indicated by the high probability value (p = 299). However, the difference between unergative verbs and hervir is statistically significant, as expected (p < .0001). Thus, the verb hervir seems to pattern together with unaccusative verbs. 164 4.3.3.5. Pulular Table 17 presents the subject distribution for the verb pulular “swarm”. The first column (“non-altern. pulular”) contains those sentences where this verb does not display the locative alternation. The second column (“Altern. Pulular”) displays the sentences where the verb appears in the canonical (“E”) and non-canonical (“v”) locative alternation construction. The chi-square test indicates that there is no statistically significant difference between the factors studied (p = .887), since all of them have a similar subject distribution. The preference for postverbal subjects (63.3%) clearly suggests that this verb has unaccusative properties. Table 17: Subject position by verb type: alternating vs. non-alternating pulular Type of Verb Non-altern. pulular Altern. pulular Total Count 13 49 62 Postverbal Column % 61.9% 63.6% 63.3% Count 8 28 36 Subject Position Preverbal Column % 38.1% 36.4% 36.7% Count 21 77 98 Total Column % 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square .021 1 .884 Likelihood Ratio .021 1 .884 4.3.3.6. Summary The analysis of subject position indicates that all the swarm type verbs analyzed here pattern together with unaccusative verbs as far as subject position is concerned. This 165 is always true when these verbs appear in the locative alternation construction. The verb brillar “shine” shows this same tendency, but it behaves as an unergative verb when it does not alternate. The analyzed verbs include light emission verbs (brillar “shine”, arder “burn”), sound emission verbs (resonar “resound”) and other representative verbs (pulular “swarm”, hervir “boil”, abundar “abound”). All of them show an unaccusative behavior when they occur in the locative alternation construction. In section 4.3.3 to 4.3.9, I discuss other factors that may influence subject position, and that therefore, need to be controlled. 4.3.4. Results: Presence of preverbal XPs The function of this independent factor is to control if there is any phrase in preverbal position that is not a subject (subject information is already coded as the variable type of subject. Mayoral Hernández 2004c showed that the presence of preverbal and postverbal constituents have an effect on subject position, which tend to appear in complementary distribution with any co-occurring XP. And as expected, table 18 shows that subject position is affected by the presence of a preverbal XP, depending on its syntactic nature. In particular, subjects will tend to be postverbal when there is a preverbal XP functioning as an argument (DO, IO and argumental PP in verbs with locative alternation). By contrast, subjects will tend to be preverbal when there is no co- occurring preverbal XP or there is a preverbal subordinate clause (CP). When there is an adjunct in preverbal position, subjects show a slight preference to be postverbal. 166 Table 18: Subject position by Presence of preverbal XP, all verbs Presence of preverbal XP CP No prev. XP Argument Adjunct Total Count 13 529 145 246 933 Postv Column % 28.3% 36.4% 88.4% 51.7% 43.6% Count 33 926 19 230 1208 Subject Position Prev Column % 71.7% 63.6% 11.6% 48.3% 56.4% Count 46 1455 164 476 2141 Total Column % 100.0% 100.0% 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square 182.042 3 .000 Likelihood Ratio 193.556 3 .000 Most importantly, table 20 shows that when there is an overt preverbal XP, the subjects of transitive and unergative verbs tend to be preverbal, while unaccusatives and the swarm type verbs tend to be clearly postverbal. Nevertheless, even unergative and transitive verbs show a noticeable increase of postverbal subjects when a preverbal XP is present, when one compares to table 3, repeated here as table 19, which contains the uncontrolled results. Table 19: Subject distribution in unergative and transitive verbs (p < .0001) Type of Verb Transitive Unergative Total Count 5 32 37 Postverbal Column % 3.0% 21.5% 11.8% Count 160 117 277 Subject Position Preverbal Column % 97.0% 78.5% 88.2% Count 165 149 314 Total Column % 100.0% 100.0% 100.0% 167 Table 20: Subject position by Type of verb, with overt preverbal XP 26 Type of Verb Swarm type Unerg Trans Unacc Total Count 362 13 2 27 404 Postv. Column % 64.3% 32.5% 5.7% 56.3% 58.9% Count 201 27 33 21 282 Subject position Prev. Column % 35.7% 67.5% 94.3% 43.8% 41.1% Count 563 40 35 48 686 Total Column % 100.0% 100.0% 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square 59.328 3 .000 Likelihood Ratio 63.822 3 .000 Thus, these results do not change the difference in behavior between verb types. Transitive verbs and unergatives prefer preverbal subjects. Unergative verbs and the swarm type have a lower percentage of preverbal subjects. 4.3.5. Results: Presence of postverbal XPs This independent variable is intended to control for the presence of any co- occurring postverbal phrases (XPs). Table 21 shows that the presence of an overt XP in postverbal position is automatically associated with preverbal subjects. When there is no postverbal XP, however, subjects tend to be postverbal. 26 Note that there are less than 5 tokens of transitive verbs with postverbal subjects, which may influence the chi-square test results. Irrespective of the χ 2 value, subject placement clearly separates unaccusatives and the swarm type verbs, which prefer postverbal subjects, from unergatives and transitives, with a preference for preverbal subjects. 168 Table 21: Subject position by Presence of postverbal XP Presence of postverbal XP CP No postv. XP Argument Adjunct Total Count 22 628 77 206 933 Postv Column % 45.8% 77.1% 21.0% 22.6% 43.6% Count 26 186 290 706 1208 Subject Position Prev Column % 54.2% 22.9% 79.0% 77.4% 56.4% Count 48 814 367 912 2141 Total Column % 100.0% 100.0% 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square 612.871 3 .000 Likelihood Ratio 639.920 3 .000 Table 22: Subject position by Type of Verb, with overt postverbal XP Type of Verb Swarm type Unerg Trans Unacc Total Count 281 7 2 15 305 Postv. Column % 28.6% 6.4% 1.3% 18.3% 23.0% Count 702 103 150 67 1022 Subject position Prev. Column % 71.4% 93.6% 98.7% 81.7% 77.0% Count 983 110 152 82 1327 Total Column % 100.0% 100.0% 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square 75.929 3 .000 Likelihood Ratio 102.835 3 .000 Table 22 shows that even unaccusatives and the swarm type verbs, normally associated with a higher percentage of postverbal subjects, tend to have preverbal subjects when there is an overt postverbal XP, although the difference between the groups is still significant, and we are still able to differentiate between verb types, irrespective of the occurrence of postverbal XPs (p = .0001) 169 4.3.6. Results: Type of subject This variable only resulted significant with unaccusatives and the swarm type verbs. Both transitive and unergatives still show a marked preference for preverbal subjects in all the conditions, especially transitives. Due to this manifest preference for preverbal subjects, several cells in the postverbal domain had less than 5 tokens, an undesirable amount for statistical purposes. It would be necessary to compile a larger corpus that includes more sentences with unergative and transitive verbs, to ensure that there are 5 or more tokens in every cell. Therefore, the tables included in this section only contain unaccusative and swarm type verbs, since the previous data have shown that these are the ones that allow a higher syntactic flexibility regarding subject placement. The following table does not include subjects realized by a clause ([ Subject el que me dijo la información] se fue pronto “the one who told me the information left soon”) or a personal pronoun, because of the low number of tokens in the corpus (14 and 12 respectively). Table 23 does not include subjects coded with the value “other” either, because this code comprises a heterogeneous group of nouns, such as bare NPs, subjects headed by numerals and other quantifiers, which did not fit within any of the other values within the independent variable type of subject. 170 Table 23: Subject position by Type of Subject Type of Subject Demons- trative Proper noun Inde- finite Definite Total Count 48 32 79 611 770 Postv. Column % 28.7% 25.2% 64.8% 51.2% 47.9% Count 119 95 43 582 839 Subject position Prev. Column % 71.3% 74.8% 35.2% 48.8% 52.1% Count 167 127 122 1193 1609 Total Column % 100.0% 100.0% 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square 69.935 3 .000 Likelihood Ratio 72.375 3 .000 The results displayed in table 23 indicate that the type of subject influences subject position (p < .0001). If the subject is realized as a proper noun or headed by a demonstrative or possessive determiner, it will tend to be preverbal. When subjects are headed by an indefinite article, they will tend to appear in postverbal position. Finally, if they are headed by a definite article they will tend to show a uniform distribution across the preverbal and postverbal domains. The effects observed in this section cannot be used to refute the validity of the subject position test, because it did not have a significant influence in unergative and transitive verbs, which always prefer preverbal subjects, irrespective of subject type. 4.3.7. Results: Type of clause This variable is intended to classify the nature of the clause or sentence where the verbs analyzed here are included. For example, Gutiérrez-Bravo 2003, inter alia, observes that relative clauses are associated with a higher percentage of postverbal 171 subjects. Transitive verbs have been excluded again, since they still have a marked preference for preverbal subjects, always higher than 90% of the sentences have preverbal subjects, irrespective of the type of clause where the verb occurs. The results presented in table 24 only include unaccusatives and the swarm type verbs, because they have a homogeneous subject distribution across preverbal and postverbal domains (49.1% postverbal subjects vs. 50.9% preverbal). Table 24: Subject position by Type of clause; unaccusative and swarm type verbs Type of clause Completive Main Relative Clause Total Count 74 549 273 896 Postv. Column % 41.8% 42.6% 76.3% 49.1% Count 103 740 85 928 Subject position Prev. Column % 58.2% 57.4% 23.7% 50.9% Count 177 1289 358 1824 Total Column % 100.0% 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square 131.258 2 .000 Likelihood Ratio 136.472 2 .000 When only these verbs are selected, the type of clause where they appear seems to influence subject position (p < .0001). In particular, relative clauses greatly favor postverbal subjects (76.3%), while in the other two types of clauses there is a slight tendency for subjects to precede the verb. As far as unergative verbs are concerned, table 25 illustrates that they show an obvious preference for preverbal subjects, except when they appear in a relative clause, where the trend reverts. While there is an obvious and statistically significant difference 172 between unaccusative verbs (including the swarm type) and unergatives in completive sentences and main clauses (p = .001 and p = .0001 respectively), there is no significant difference between them when they occur in a relative clause, since all of them appear with postverbal subjects (p = .115). This might suggest that the influence of this factor should be controlled for when comparing unaccusative and unergative verbs. However, note that, by random chance, the total number of unaccusative verbs that occur in relative clauses is 19.6% in my entire corpus, compared to a very similar 20% of unergatives. This fact probably means that the exclusion of these tokens from this experiment would not alter the results already observed or the claims made in this dissertation regarding the validity of the subject position test as a means to identify verb classes. Table 25: Subject position by Type of clause; unergative verbs Type of clause Completive Main Relative Clause Total Count 2 11 19 32 Postv. Column % 7.7% 11.7% 63.3% 21.3% Count 24 83 11 118 Subject position Prev. Column % 92.3% 88.3% 36.7% 78.7% Count 26 94 30 150 Total Column % 100.0% 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square 39.612 2 .000 Likelihood Ratio 34.112 2 .000 173 4.3.8. Results: Sentence polarity The research dealing with information structure has analyzed negation as a focal element in the sentence (Lambrecht 1994), which might have influence on word order in Spanish (Silva Corvalán 1994, Zubizarreta 1998). Indeed, sentence polarity has an effect on subject position. Table (26) only includes unaccusative and swarm type verbs, due again to their syntactic flexibility, regarding subject position. My corpus does not have enough negative sentences with unergative (7 negative sentences) or transitive (1 negative sentence) verbs to make any claims about the influence of negation on these two types of verbs. This necessarily means that this factor cannot have interfered with my results. As far as unaccusatives and the swarm type verbs are concerned, table 26 illustrates that while affirmative sentences seem to have a neutral subject distribution, negative sentences favor preverbal subjects. This result is statistically significant (p = .019). Table 26: Subject position by Sentence polarity Sentence polarity Negative Affirmative Total Count 29 867 896 Postverbal column % 36.3% 49.7% 49.1% Count 51 877 928 Subject position Preverbal column % 63.8% 50.3% 50.9% Count 80 1744 1824 Total column % 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square 5.548 1 .019 Likelihood Ratio 5.625 1 .018 174 4.3.9. Results: Type of alternant This section is only concerned with the swarm type verbs that display the locative alternation, so it cannot modify in any way my claims about the validity of the subject position test. The variable “type of alternant” represents the possible prepositions that can head the argument PP that characterizes the locative alternation constructions. The results displayed in table 27 show two important results. On the one hand, it is obvious that argument PPs headed by the preposition en “on, in, at” constitute the majority of cases (83.6%). On the other hand, the sentences that contain an argument PP headed by en greatly favor the presence of postverbal subjects. Table 27: Subject position by Type of alternant Type of Alternant con de en Other locatives Total Count 8 2 352 23 385 Postv. Column % 20.0% 10.0% 65.2% 50.0% 59.6% Count 32 18 188 23 261 Subject position Prev. Column % 80.0% 90.0% 34.8% 50.0% 40.4% Count 40 20 540 46 646 Total Column % 100.0% 100.0% 100.0% 100.0% 100.0% Value df Asymp. Sig. (2-sided) Pearson Chi-Square 55.241 3 .000 Likelihood Ratio 56.795 3 .000 If the stative prepositions con “with” and de “of” are recoded into one single variable, the results are still statistically significant (p < .0001). The recode of the two 175 stative prepositions into one is statistically plausible because there is no significant difference between them, as far as subject position is concerned (p = .327). 4.4. Discussion and analysis of the results In previous chapters, I proposed that the locative alternation is indeed an unaccusative construction. Based on Hale and Keyser 2002, I defended that the transitive sentences that display this alternation are also built on an unaccusative construction that indicates a change of state, as put forward by Zubizarreta & Oh 2007. The intransitive verbs that display this alternation exhibit unaccusative properties, because they either participate in the causative alternation or in other unaccusative constructions (Mendikoetxea 1999). However, there is a group of alternating intransitive verbs whose classification as either unaccusative or unergative is controversial. In order to provide a classification for these controversial verbs, which display the swarm type alternation, it was necessary to develop a reliable test that could differentiate between the two types of intransitives: unergatives and unaccusatives. In this chapter, I have presented several studies where I propose a test to differentiate between verb types by looking at subject position in Spanish. My investigation shows that the subjects of unergative verbs tend to be clearly preverbal, while unaccusatives show no clear preference for either preverbal or postverbal subjects. At this point the study of subject position in Spanish becomes central for this dissertation, because it may provide a strong argument in favor of my analysis of the locative alternation as an unaccusative construction. In Mayoral Hernández 2006 I 176 already pointed out that the verb resonar “resound”, one of the verbs in the swarm group, seemed to behave as an unaccusative verb as far as subject position is concerned. In this dissertation, I have expanded Mayoral Hernández 2006 to control for other intervening factors that may have an influence on subject position. I have also extended the original corpus to include other representative swarm type verbs. Here, I discuss the results of this latter analysis. For the purpose of this dissertation, the most important factor is the type of verb variable. The results support Mayoral Hernández 2006 and the unaccusative hypothesis defended in this dissertation, because all the verbs analyzed here show unaccusative properties when they enter the locative alternation. This does not mean that these verbs can only appear in unaccusative constructions. For example, the verb brillar “shine” has unergative properties when it does not alternate, but the rest of the verbs studied always show an unaccusative behavior. These results imply that what gives unaccusative properties to the locative alternation is the construction itself, not necessarily the verbs inserted in it. This is also in agreement with constructional views of grammar, where argument structure is not necessarily determined by the lexical conceptual structure of the verb (Zubizarreta & Oh 2007, Goldberg 1995). It is interesting to note that this is not a new claim. There exists crosslinguistic evidence that supports this statement. For example, it has been noted in the literature that manner of motion verbs, which normally have unergative properties, can appear in unaccusative constructions when there is a goal PP in the sentence. 177 (158) a. Wij zijn naar de stad gelopen. “We have walked to the city” b. Wij hebben de hele dag gelopen. We have walked all day. The Dutch sentences in (158) (from Shetter et al. 2002) show that the verb lopen “walk”, which normally selects the auxiliary have (associated with unergative verbs) in periphrastic forms, selects the auxiliary be (associated with unaccusative constructions) when the end point is specified. Although the verb lopen “walk” typically denotes an activity, it becomes an accomplishment when it appears in a construction that specifies a goal. Likewise, the presence of other phrases in the sentence has a significant effect on subject position. Subjects tend to appear in complementary distribution with respect to other co-occurring XPs. Unaccusative verbs, however, always display a significantly higher percentage of postverbal subjects when compared to unergative and transitive verbs. To my knowledge, there is no literature that deals with this phenomenon. A possible way of interpreting the fact that subjects tend to appear in complementary distribution with other XPs in the sentence is the separation of weight across the verbal domains 27 . In the same way as heavy elements tend to appear at the end of the sentence in VO languages in order to facilitate processing (Hawkins 1994, 1999, 2000, 2001), it might be the case that having a balanced distribution of constituents across the preverbal 27 This idea was suggested by John Hawkins in personal communication. 178 and postverbal domains facilitates processing in some way. However, more research is needed to understand this phenomenon. The type of subject variable also proved statistically significant. Abundant research shows that there is a relation between information structure and word order, especially in languages with ample syntactic flexibility like Spanish (Prince 1981, Silva- Corvalán 1994, Zubizarreta 1998). Other factors that may influence subject position are definiteness and specificity (Prince 1992, Leonetti 2004). The data included here shows how proper nouns and subjects headed by a demonstrative have the higher percentage of preverbal subjects, which might be linked to specificity and referenciality. Subject types that appear more frequently in postverbal position are indefinite nouns, which tend to be new information. A detailed discussion of this issue is beyond the scope of this dissertation. The interesting point to note is that even in the cases where swarm type verbs reach their highest percentage of preverbal subjects, i.e. when the subject is realized by a proper noun, the percentage of preverbal subjects (74.8%) does not reach the 78.5% mean that characterizes unergative verbs. It would be necessary to extract more tokens containing unergative and transitive verbs to be able to do reliable statistics about the interaction of subject position and subject type with these two types of verbs. However, it is possible to hypothesize that proper nouns and phrases headed by a demonstrative or possessive would also be associated with a higher percentage of preverbal subjects. In the case of unergatives it should be around 90%. At this point, this will have to be left for a future project. 179 Regarding the type of clause variable, researchers have noticed that relative clauses tend to have postverbal subjects (Gutiérrez-Bravo 2003), although, to my knowledge, it had never been quantified before. This research provides evidence supporting such statement. Even when this factor is controlled for, the results indicate that unaccusative and swarm type verbs always have a higher percentage of postverbal subjects than unergative verbs. Regarding sentence polarity, the results associate the presence of negation with a higher percentage of preverbal subjects. Lambrecht 1994 associated negation with focus structures. In particular, negation divides the sentence between new information, which is anything within its scope, and old information, which is whatever is placed out of the scope of negation. This implies that the only position for old information in the sentence becomes the preverbal position, since negation in Spanish always appears as a clitic attached to the left of the verb, and can only be separated from it by clitic pronouns. In this situation, any subject expressing old information must precede the verb. This phenomenon should increase the percentage of preverbal subjects, even with unaccusative verbs, which have an intrinsic tendency to appear postverbally 28 . As for the last factor, the type of alternating PP was used in chapter 3 to suggest that the locative alternation is not a productive construction in Spanish, since even those verbs that can alternate, frequently appear in only one of the variants (normally the locative variant). However, it would be necessary to include more data to make reliable 28 The relation between focus and negation was pointed out to me by Carmen Silva-Corvalán in personal communication. 180 predictions, because I have not studied the distribution of variants with the transitive types of the locative alternation, such as clear or load. Summarizing, I have shown that the swarm type verbs pattern together with unaccusatives as far as subject position is concerned. My analysis controls for the presence of other variables to be able to avoid the influence of other intervening factors. The data supports the classification of the swarm type verbs as unaccusative when they participate in the locative alternation, even if some of them may also appear in unergative constructions, like brillar “shine”. At this point I can safely put forth that what links together all alternating verbs is primarily their unaccusative properties. 181 Chapter 5 Conclusions In this chapter, I briefly summarize the content of this dissertation and provide an overview of the major contributions to the study of the locative alternation. The study of argument alternations is relevant for the linguistic research because it implies interface effects between the lexicon and the syntax. In particular, it is believed that the meaning of some verbs, or the constructions where they appear, is what enables the syntactic alternation of their arguments. The typical example of locative alternation is given below in (159). (159) a. Juan cargó el camión con el heno. (CHANGE OF STATE variant) LOCATION LOCATUM 29 John loaded the truck with the hay. b. Juan cargó heno en el camión. (CHANGE OF LOCATION variant) LOCATUM LOCATION John loaded hay into the truck. 29 The term Locatum refers to the argument that is displaced, and it can be equivalent to Theme. Location refers to the argument that represents the place where the Locatum ends up, and it can be equivalent to Goal. 182 The locative alternation is identified because it involves a locative argument (location) and a them argument (locatum) that need to be in touch at any point for the sentence to be true (Beavers 2006). In the examples in (159), the truck has to contain hay at some point for the sentence to be true. Levin 1993 postulated the existence of five subtypes within this alternation. Three of them involve transitive verbs, such as the load type in (159), where the alternation occurs between the direct object and the prepositional object (PO). There exist two more subtypes, which are intransitive and involve the alternation between the subject and the PO, such as in (70), repeated here as (160), which shows the verb arder “burn” included within the swarm subtype. (160) a. Los ecos de Bahía de Cochinos resonaban en las paredes. The echoes from Bahía de Cochinos echoed in the walls. b. Todas las mezquitas resonaron con exhortaciones a la unión nacional. All the mosques echoed with exhortations to the union national. These types of alternations have received several analyzes, from both a syntactic and a lexical point of view, but they all have failed in finding the common denominator that links alternating verbs together. In chapter 1, I also addressed several purported asymmetries that seem to differentiate between the two variants of the alternation. The so-called holistic effect has 183 enjoyed special attention in the field; it claims that only in the change of state variant the direct object receives an interpretation where it is fully affected. The direct object the truck is supposed to be completely full in (159a), while the direct object of (159b) may not be completely affected. However, I presented Beaver’s 2006 analysis where he explains that this effect is only illusory, because if one adds a definite article to hay in (159b), as in John loaded the hay into the truck, the direct object the hay becomes completely affected. This means that when the presence of determines is controlled for, there seems to be no difference regarding affectedness. Basilico 1998 also pointed out that there seem to be syntactic asymmetries between the two variants of the alternation, the change of state and the change of location, related to their interpretation as categorical and thetic statements respectively. I contended that the author’s observation does not hold for Spanish. Furthermore, I showed that in languages that have overt focus and topic markers, associated with thetic and categorical judgments respectively, such as Japanese and Basque, they can normally be used with both variants of the alternation. In chapter 2, I presented a classification of verbs that display the locative alternation in Spanish, based on data extracted from the online corpus CREA. This corpus contains written language mostly, but this is justified by the meaning of the verbs that form part of the locative alternation in Spanish, which are highly infrequent in spoken language, such as pulular “swarm”, taracear “inlay”, entreverar “intermingle” or exonerar “exonerate”. I proposed the existence of four classes, instead of the five classes in which Levin 1993 divides the verbs that enter the locative alternation in English. My 184 division is based on the use of the prepositions that can be used in every subtype. In figure 6, I include the four types and their main features. Figure 6: Types of locative alternation and features Transitivity Semantics Loc. Prep. State Prep. 1. rociar ‘spray’ cargar ‘load’ Transitive Goal en ‘in, on’ con, de ‘with’ 2. despejar ‘clear’ Transitive Point of departure de 'from, off’ de ‘of’ 3. arder ‘burn’ resonar ‘resound’ Intransitive Goal/state en ‘in, on’ con, de ‘with’ 4. despejar ‘clear’ borrar ‘erase’ Intransitive Point of departure de 'from, off’ de ‘of’ In section 2.2, I presented Mateu’s 2002 analysis of the locative alternation in Spanish, where I noted that the existence of the frotar “rub” verbs, included in the group I call despejar in figure 6, cannot be accounted for. In section 2.3, I described several semantic components of the locative alternation. On the one hand, I highlighted the centrality of the preposition, whose presence is always conceptually necessary. On the other hand, I introduced two more intrinsic features of this alternation: the verbs that enter the locative alternation need to have a rich manner component (Rodríguez Ramalle 2006) and focus on the process (Demonte 1991). In chapter 3, I contend that what links alternating verbs together is that they are inserted in an unaccusative construction, following the system developed by Hale & Keyser 2002 and Zubizarreta &Oh 2007. In particular, I propose the structure in (111), repeated below in (161): 185 (161) [ VP D [V [XP]]] Whenever the XP is realized as a complex PP, the structure will indicate directed motion with end point, and therefore an accomplishment, while if the XP is simple, it will represent a state. Thus, all the transitive types of the alternation and their intransitive counterparts are accomplishments that include a complex PP. There is a subgroup of alternating verbs, the swarm type, that are simple states, since they do not imply a notion of change. The swarm verbs have simple stative PPs as their complements. I claim that the centrality of the preposition I observed in chapter 2 arises from the structure. I have already noted that sentences such as John loaded the truck are ambiguous, since the truck can be interpreted as either the location (John loaded the truck with hay) or the locatum (John loaded the truck in the ferry).Here, I claim that the lower PP is the one responsible for providing the argument structure to the construction, and not the verb. Therefore both internal arguments, the locatum and the location, originate as specifier and complement of the lower PP, in a predication relation. The lower PP has always stative properties, and it can denote either a location or a state. The addition of a higher PP adds the directed motion meaning to the unaccusative construction (Zubizarreta & Oh 2007). Because the description of the swarm verbs as unaccusative is controversial (Perlmutter 1978, Mendikoetxea 1999, Levin & Rappaport 1995), I provided several tests that indicate that they behave as unaccusative when they appear in the locative alternation 186 construction. This is not the same as claiming that they always behave as unaccusatives. Instead I postulate that it is the locative alternation construction, which is intrinsically unaccusative, is the one responsible for the unaccusative properties observed in these verbs. I also claim that the lack of a productive way of incorporating manner that is typical of Romance languages (because of lack of VP compounding in Zubizarreta & Oh’s system) causes the lower number of alternating verbs in Spanish, compared to English. Because the need to identify the swarm type verbs as unaccusatives is fundamental for my analysis, due to their controversial classification, I carried out a variationist analysis of these verbs in chapter 4. In this thesis I defend that it is possible to differentiate between unergative and unaccusative verbs in Spanish by analyzing subject position. In particular, I show that while unergative verbs such as trabajar “work” are associated with a high percentage of preverbal subjects (around 78%), unaccusatives have a significantly higher percentage of postverbal subjects (50%), due to the theme properties of unaccusative subjects (Dowty 1991). In my experiment, I selected the most frequent swarm type verbs and coded different factors that might influence subject position in Spanish. The results indicate that all these verbs have a significantly higher percentage of postverbal subjects when compared to a typically unergative verb. This is always true when they enter the locative alternation. However, the verb brillar “shine” shows higher percentages of preverbal subjects when it occurs in other non-alternating constructions, such as la luna brilla mucho “the moon shines a lot”, where there is no locative argument. This variable 187 behavior supports the idea or unaccusative constructions, rather than a categorical classification of these verbs as unaccusative or unergative. Finally, my research has highlighted the importance of prepositions in alternating constructions. It is still necessary to learn about prepositional aspect and its influence in grammar. This issue remains for future research. 188 Appendix: list of alternating verbs and summary of their properties 1. Verbs of the type of rociar/cargar. Semantic Features 1. Their location argument indicates the destination point of the change of location. Valency Ditransitive verbs θ-structure CHANGE OF STATE: Location con/de (with/of) locatum CHANGER OF LOCATION: Locatum en (in) location a. rociar (spray)/ cargar (load) verbs: (1) abarrotar (crowd) (2) asperjar (spray) (3) atiborrar (cram) (4) cargar (load) (5) embeber (soak up) (6) entreverar (intermingle) (7) espolvorear (dust) (8) imbuir (imbue) (9) impregnar (impregnate) (10) imprimir (imprint) (11) llenar (“fill”, in the sense of llenar datos “filling data”) (12) marcar (mark) (13) pintar (paint) (14) pringar (spread) (15) pulverizar (pulverize) (16) rociar (spray) (17) salpicar (splash) (18) sembrar (sow) (19) tiznar (spot, stain) (20) untar (smear) Group 1 Special features: PP locatum can be introduced by both con (with) and de (of). b. empacar (pack) verbs: (1) bordar (embroider) (2) cultivar (cultivate) (3) empacar (pack) (4) empapar (soak) (5) ensartar (string) (6) frotar (rub) (7) grabar (engrave) (8) incrustar (inlay) (9) pinchar (pierce) (10) restregar (scrub) Group 2 Special features: PP locatum can only be introduced by con (with). c. plantar (plant) verbs: (1) plantar (plant) (2) taracear (inlay) Group 3 Special features: PP locatum can only be introduced by de (of). 189 2. Verbs of the type of barrer (sweep), disculpar (forgive) and recoger (put in order) Semantic Features 1.The location argument indicates the point of departure of the change of location, and more abstractly a negation or removal of a property. The ones highlighted have an intransitive variant that also have locative alternation, type 4. Valency Ditransitive verbs θ-structure CHANGE OF STATE: Location de (of) locatum CHANGER OF LOCATION: Locatum de (of) location a. barrer (sweep) verbs (1) barrer (sweep) (2) borrar (erase) (3) desalojar (oust) (4) descargar (unload) (5) despejar (clear) (6) drenar (drain) (7) enjugar (wipe) (8) evacuar (evacuate) (9) exorcizar (exorcize) (10) exprimir (squeeze) (11) limpiar (clean) (12) purgar (purge) (13) saquear (loot) (14) separar (separate) (15) vaciar (empty) Group 1 Special features: Both location and locatum arguments can appear as a PP. b. disculpar (forgive) verbs (1) absolver (absolve) (2) aliviar (relieve) (3) curar (cure) (4) despojar (strip of) (5) desvalijar (plunder, rob) (6) disculpar (excuse) (7) dispensar (exempt) (8) exculpar (exculpate) (9) exonerar (exonerate) (10) podar (prune) Group 2 Special features: The location argument cannot appear as a PP. c. recoger (put in order) verbs (1) destilar (distil) (2) recoger (pick up, put in order) (3) escardar (weed out) (4) aventar (winnow) (5) entresacar (pick out) (6) filtrar (filter) Group 3 Special features: The locatum argument cannot appear as a PP. 190 3 Verbs of the type of arder (burn)/resonar (resound) Semantic Features 1. Their location argument indicates the place where the action described by the verb occurs. Valency Unaccusative verbs θ-structure CHANGE OF STATE: Location (Subject) V con/de (with/of) locatum CHANGER OF LOCATION: Locatum (Subject) V en (in) location a. Arder (burn) verbs: (1) abundar (abound in) (2) arder (burn) (3) bullir (boil) (4) brillar (shine) (5) centellear (sparkle) (6) chispear (spark) (7) chisporrotear (crackle) (8) destellar (flash) (9) fulgurar (gleam) (10) hervir (boil) (11) parpadear (blink) (12) refulgir (gleam) (13) relampaguear (flash, lighten) (14) resplandecer (blaze) (15) rielar (shimmer) (16) rutilar (glitter) Group 1 Special features: PP locatum can be introduced by both con (with) and de (of). b. Resonar (resound) verbs (1) chasquear (crack) (2) crepitar (crackle) (3) fluir (flow) (4) palpitar (throb) (5) resonar (resound) (6) retumbar (boom) (7) rezumar (exude) (8) tronar (thunder) (9) zumbar (buzz) Group 2 Special features: PP locatum can only be introduced by con (with). c. pulular (swarm) verbs (1) erizarse (bristle) (2) hormiguear (swarm) (3) pulular (swarm, abound) Group 3 Special features: PP locatum can only be introduced by de (of). 191 4. Verbs of the type of despejar (clear)/borrar (erase). Semantic Features 1.The location argument indicates the point of departure of the change of location, and more abstractly a negation or removal of a property. They all have a transitive, agentive counterpart that also alternates. Valency Unaccusative verbs θ-structure CHANGE OF STATE: Location (Subject) de (of) locatum CHANGER OF LOCATION: Locatum (Subject) de (of) location a. Despejar (clear) verbs. (16) aliviar (relieve) (17) curar (cure) (18) despejar (clear) (19) limpiar (clean) (20) vaciar (empty) Group 1 Special features: Both location and locatum arguments can appear as a PP. b. Borrar (erase) verbs (17) borrar (erase) (18) destilar (distil) Group 2 Special features: The location argument cannot appear as a PP. 192 References ALCÁZAR, A. (2003). Split-Intransitivity in Basque and the Unaccusative Hypothesis. 19 th NWLC (forthcoming). ALCÁZAR, A. (2005): Against an ontological commitment to unergative verbs, Proceedings of CLS40 (in press). ALCÁZAR, A. (2006) "Transitive intransitives: Basque unergatives revisited," in Proceedings of the Fourth Cambridge Postgraduate Conference in Language Research. Cambridge: Cambridge University Press. (to appear) AOUN, J and LI, A. (1989): Constituency and scope. Linguistic Inquiry 20, 141-172. BAKER, M. (1997): Thematic roles and syntactic structure. In Haegeman, L. (ed.). Elements of grammar. Dordrecht: Kluwer. BAKER, M; JOHNSON, K; and ROBERTS, I (1989): Passive arguments raised. Linguistic Inquiry 20, 219-252. BASILICO, D. (1998): Object position and predication forms. Natural Language and Linguistic Theory, 16, 3 BEAVERS, J. T. (2006): Argument/oblique alternations and the structure of lexical meaning. Unpublished PhD Dissertation, Stanford University. BELLETTI, A. (1988): The case of unaccusatives, Linguistic Inquiry 19, 1-34 BORER, H. (1993): The projection of arguments. University of Massachusetts Occasional Papers in Linguistics 17, 19-47. BORER, H. (2005): Structuring Sense: An Exo-Skeletal Trilogy. Book II: The Normal Course of Events, Oxford University Press. BOSQUE, I. (1993): Preposición tras preposición, in Homenaje a Ramón Trujillo, Tenerife, Montesinos, 13-145 (published 1996). BRAINE, M.D.S., and BROOKS, P.J. (1995): Verb argument structures and the problem of avoiding an overgeneral grammar. In Beyond Names for Things: Young Children's acquisition of Verbs: M. Tomasello and W.E. Merriman, (eds.) . Hillsdale, N.J.: Erlbaum, 353-376. 193 BURZIO, L. (1981). Intransitive Verbs and Italian Auxiliaries. Ph. D. thesis, MIT, Cambridge, MA. BURZIO, L (1986): Italian syntax. Dordrecht, Reidel. CAMPOS, H (1999): Transitividad e intransitividad. In Bosque, Ignacio and Demonte, Violeta (eds.) (2000): Gramática Descriptiva de la Lengua Española, Madrid, Espasa. CINQUE, G. (1999): Adverbs and functional heads: A crosslinguistic perspective, New York and Oxford, Oxford University Press. CHEN, L. (2008): On Mass, Collectives and Pseudopartitives. Doctoral dissertation. CHOMSKY, N (1965): Aspects of the theory of syntax. MIT Press, Cambridge, MA. CHOMSKY, N (1991): Some notes on the economy of derivation and representation. In Robert Freidin (ed.), Principles and parameters in comparative grammar, MIT Press, Cambridge, Massachusetts, pp. 417-454. CHOMSKY, N. (1993): A Minimalist Program for Linguistic Theory. In K. Hale and S.J. Keyser, View from Building 20, Current Studies in Linguistics, MIT Press, 1-52. CHOMSKY, N (1995): The minimalist program, MIT Press, Cambridge, Massachusetts. COLLINS, C. (1997): Argument sharing in serial verb construction. Linguistic Inquiry 28: 461-497. DEMONTE, V. (1991): Detrás de la palabra. Estudios de gramática del español. Madrid: Alianza Universidad. DEMONTE, V. (2000): La interficie léxico-sintaxis. Un esquemático estado de la cuestión. I Jornadas de Lingüística Contemporánea. UNED. DEMONTE, V. (2002): Preliminares de una clasificación léxico-sintáctica de los predicados verbales del español, in S. Grosse and A. Schonberger (eds.), Ex oriente lux: Festchrift für Eberhard Gärtner zu seinem 60. Geburtstag, Frankfurt am Main: Valentia 12-144 DEN DIKKEN, M. (1995): Particles. On the Syntax of Verb-Particle, Triadic, and Causative Constructions. Oxford University Press. 194 DI TULLIO, Á. L. (2001): Los difusos límites de la “alternancia locativa” en español., in Veiga and Pérez (ed), Lengua española y estructuraas gramaticales, Verba 28:131- 140. Diccionario de la Real Academia Española, available online at www.rae.es. DOWTY, D. (1979): Word Meaning and Montague Grammar, Reidel, Dordrecht. DOWTY, D. (1991): Thematic Proto-roles and argument selection, Language, 67, 3. DOWTY, D. (2000): "The Garden Swarms With Bees" and the Fallacy of "Argument Alternation“, in Polysemy: Theoretical and Computational Approaches, ed. by Yael Ravin and Claudia Leacock, 111:28. Oxford: Oxford University Press. FERNÁNDEZ SORIANO, O. (1999): Two types of impersonal sentences in Spanish: Locative and dative subjects, Syntax, 2:2, p. 101-140. FREEZE, R. (1992): “Existential and other locatives”. Language 68.3: 553-595 GOLDBERG, A. (1995): Constructions: a construction grammar approach to argument structure. Cognitive Theory of Language and Culture, Chicago: University of Chicago Press. GUTIÉRREZ-BRAVO, R. (2003): Subject inversion in Spanish relative clauses: a case of prosody-induced word order variation without narrow focus. In Geerts, T. & H. Jacobs (eds.) Romance languages and linguistic theory 2003, Amsterdam: John Benjamins. HALE, K.L. and KEYSER, S.J. (1986): Some transitivity alternations in English. Lexicon Project Working Papers, 7. Center for Cognitive Science. Massachusetts Institute of Technology. Cambridge. HALE, K.L. and KEYSER, S.J. (1987): A view from the middle. Lexicon Project Working Papers, 10. Center for Cognitive Science. Massachusetts Institute of Technology. Cambridge. HALE, K.L., and KEYSER, S.J. (1993): On Argument Structure and the Lexical Expression of Syntactic Relations. In K. Hale and S.J. Keyser, View from Building 20, Current Studies in Linguistics, MIT Press, 53-110. HALE, K.L. and KEYSER, S.J. (1998): The basic elements of argument structure. In Harley, H (ed.) MIT Working papers in Linguistics 32, Papers from the UPenn/MIT roundtable on argument structure and aspect. MIT, Cambridge, MA. 195 HALE, K.L. and KEYSER, S.J. (1999): A response to Fodor and Lepore, ‘Impossible words’. Linguistic Inquiry 30.3: 453-466 HALE, K.L. and KEYSER, S.J. (2002): Prolegomenon to a theory of Argument Structure. Cambridge, Mass. MIT Press HALE, K.L. and LAUGHREN, M (1983): The structure of verbal entries: Preface to dictionary entries of verbs. Warlpiri Lexicon Project. Massachusetts Institute of Technology. Cambridge. HAWKINS, J. (1994): A performance theory of order and constituency, CUP, Cambridge. HAWKINS, J. (1999): Processing complexity and filler-gap depencencies across grammars. Language. 75.2. 224-285. HAWKINS, J. (2000): The Relative order of Prepositional Phrases in English: Going beyond Manner- Place- Time. Language Variation and Change, 11, 231-266. HAWKINS, J. (2001): Why are categories adjacent?, Journal of linguistics, 37.1, 1-34 IWATA, S. (2005): Locative alternation and two levels of verb meaning, Cognitive Linguistics 16.2, 355–407 JACKENDOFF, R. (1990): Semantic structures. Cambridge, Mass.: MIT Press. JACKENDOFF, R. (1996): The Proper Treatment of Measuring Out, Telicity, and Perhaps Even Quantification in English, Natural Language and Linguistic Theory 14, 305-354. JAKOBSON, R (1936): Beitrag Zur Allgemeinen Kasuslehre. In Jakobson, R. Selected writings, vol. 2, Word and Language, The Hague, Mouton (1971). JELINEK, E (1995): Argument structure in some native American languages. Unpublished manuscript, University of Arizona. KIPARSKY, P. (1998): Partitive case and aspect. In M. Butt andW. Geuder, eds., The Projection of Arguments: Lexical and Compositional Factors, pages 265–307. Stanford, CA: CSLI Publications. KRATZER, A. (1996): Severing the external argument from its verb. In J. Rooryck & L. Zaring (eds.) Phrase Structure and the Lexicon. Dordrecht: Kluwer, 109-137. 196 KRIFKA, M. (1989): Nominal Reference, Temporal Constitution and Quantification in Event Semantics, in R. Bartsch, J. van Benthem, and P. van Emde Boas, eds., Semantics and Contextual Expression, Foris, Dordrecht, 75-115. KOOPMAN, H. (1997): Prepositions, postpositions, circumpositions and particles: The structure of Dutch PPs. Ms., UCLA. KURODA, S (1972): The categorical judgement and the thetic judgement. Foundations of Language 9, 153-185. LABOV, W. (1969): Contraction, Deletion, and Inherent Variability of the English Copula. Language. 45: 715-762. LABOV, W. (1994): Principles of linguistic change. Internal factors. Blackwell. Cambridge. LARSON, R. (1988): On the double object construction. Linguistics Inquiry 19, 335-392. LARSON, R. (1990): Double objects revisited: a reply to Jackendoff. Linguistic Inquiry 21, 589-632 LEONETTI, M. (2004): Specificity and Differential Object Marking in Spanish. Catalan journal of linguistics, 3. LEVIN, B (1993): English verb classes and alternations. Chicago: The University of Chicago Press. LEVIN, B. & RAPPAPORT, M. (1995): Unaccusativity. At the Syntax-Lexical Semantics Interface. Cambridge, Mass: MIT Press. LEVIN, B and RAPPAPORT, M (1998): Morphology and lexical semantics. In: The handbook of Morphology (1998), Edited by Spencer& Zwicky. Blackwell handbooks in Lingusitics. MARANTZ, A. (1984): On the Nature of Grammatical Relations, MIT Press, Cambridge, Massachusetts. MATEU, J (2002): Argument Structure: relational construal at the syntax-semantics interface. Doctoral dissertation. Universidad Autónoma de Barcelona. MAYORAL HERNÁNDEZ, R. (2004a): On the position of frequency adverbials in Spanish, USC Working Papers in Linguistics. 197 MAYORAL HERNÁNDEZ, R. (2004b): Importance of weight and argumenthood on the ordering of adverbial expressions, WCCFL 23 Proceedings. Somerville, MA: Cascadilla Press. MAYORAL HERNÁNDEZ, R. (2004c): A typological approach to the ordering of adverbials: weight, argumenthood and EPP. BIDE 2004 Proceedings. In press. MAYORAL HERNÁNDEZ, R. (2006): A Variationist Approach to Verb Types and Subject Position in Spanish: Verbs of light and sound emission, presentation at LSRL 2006. MCKOON ,G. & MACFARLAND T. (2000): Externally and Internally Caused Change of State Verbs Language, Vol. 76, No. 4. (Dec., 2000), pp. 833-858. MENDIKOETXEA, A (1999): Construcciones inacusativas y pasivas. In Bosque, Ignacio and Demonte, Violeta (eds.) (2000): Gramática Descriptiva de la Lengua Española, Madrid, Espasa. MORENO CABRERA, J. C. (1998): Las alternancias locativas en el marco de la semántica de los esquemas eventivos. IV Jornadas de Lingüística. Universidad de Cádiz. MULDER, R. (1992): The aspectual nature of syntactic complementation. Doctoral dissertation. Leiden: HIL Dissertation Series. NEWMEYER, F. J. (2003): Grammar is grammar and usage is usage. Language, 79.4 ORTIZ DE URBINA, J. (1989). Parameters in the Grammar of Basque. Foris, Dordrecht. PINKER, S. (1989): Learnability and Cognition: The Acquisition of Argument Structure. Cambridge, MA: MIT Press. PESETSKY, D (1995): Zero Syntax: Experiencers and Cascades. Cambridge, Mass: MIT Press. PERLMUTTER, D (1978): Impersonal passives and the unaccusative hypothesis. Proceedings of the Fourth Annual Meeting of the Berkley Linguistics Society, 157- 189. PRINCE, E. (1981): Toward a taxonomy of given-new information. In P. Cole (Ed.), Radical pragmatics. New York: Academic Press. 223-256 198 PRINCE, E. (1992): The ZPG letter: Subjects, Definiteness, and Information Status. In Thompson, Sandra; Mann, William (eds.). Discourse Description: Diverse Linguistic Analyses of a Fund-raising Text. Philadelphia: John Benjamins, pp. 295-325. PUSTEJOVSKY, James (1995): The generative lexicon. Cambridge (Mass.). MIT Press. RAPPAPORT, M and LEVIN, B (1988): What to do with θ-roles. In Wilkins, W (ed.). Syntax and Semantics, volume 21: Thematic relations. San Diego, CA: Academic Press. REAL ACADEMIA ESPAÑOLA: (CREA) data base [online]. Corpus de referencia del español actual. <http://www.rae.es> RODRÍGUEZ RAMALLE, T. M. (2006): La posición de sujeto y otras propiedades sintácticas de los verbos de emisión, Actas del XXXV Simposio Internacional de la Sociedad Española de Lingüística, 1594-1611 ROBERTS, Ian (1985): The representation of implicit and dethematized subjects, Doctoral dissertation, University of Southern California, Los Angeles. SASSE, H (1987): The thetic/categorical distinction revisited. Linguistics 25, 511-580. SEILER, H (1983): Possession as an operational dimension of language. Tübingen, Gunter Narr Verlag Tübingen. SHETTER, W. Z and VAN DER CRUISSE-VAN ANTWERPEN, I. (2002): Dutch. An essential grammar. London, New York: Routledge SILVA-CORVALÁN, C. (1982): Subject expression and placement in Mexican- American Spanish, in Amastae, Jon, and L. Elias-Olivares (eds.). Spanish in the United States: Sociolinguistic Aspects, N. Y., Cambridge U. Press, 93-120. SILVA-CORVALÁN, C. (1994): Language contact and change: Spanish in Los Angeles. Oxford: Clarendon Press. SILVA-CORVALÁN, C. (2003): Otra mirada a la expresión del sujeto como variable sintáctica. Lengua, variación y contexto: Estudios dedicados a Humberto López Morales, ed. by Francisco Moreno Fernández et al., 849-60. Madrid: Arco/Libros. SILVA-CORVALÁN, C. & SÁNCHEZ-WALKER, N (2007): “Subjects in early dual language development: A case study of a Spanish-English bilingual child”. In Spanish in Contact, Potowski, Kim and Richard Cameron (eds.), 3–22. 199 SMILEY, P. and HUTTENLOCHER, J. (1995): Conceptual development and the child’s early words for events, objects, and persons. In Tomasello, and Merriman (Eds.) (1995): Beyond names for things: young children’s acquisition of verbs. Lawrence Erlbaum Associates, Publishers. New Jersey. SORACE, A (2000): Gradients in auxiliary selection with intransitive verbs. Language. 76. 859-890. TREMBLAY, M (1991): Alternances d’arguments internes en français et en anglais. Revue Québécoise de Linguistique, 20, 39-54. TALMY, L. (1985): Lexicalization patterns: Semantic structure in lexical forms. In Timothy Shopen (ed.), Language Typology and Semantic Description, Vol 3: Grammatical Categories and the Lexicon, 36-149. Cambridge: Cambridge University Press. TALMY, L. (1991): Path to realization: a typology of event conflation. Proceedings of the Berkeley Linguistics Society, 17, 480-519. TALMY, L. (2000): Toward a cognitive semantics. Cambridge, MA: MIT Press. TENNY, C.L. (1987): Grammaticalizing Aspect and Affectedness, Doctoral dissertation, MIT, Cambridge, Ma. TENNY, C.L. (1994): Aspectual Roles and the Syntax-Semantics Interface, Kluwer,Dordrecht. TENNY, C. L. (1995): How Motion Verbs are Special: The Interaction of Semantic and Pragmatic Information in Aspectual Verb Meanings, Pragmatics and Cognition 3, 31- 37. TORREGO, E. (1989): Unergative unaccusative alternations. MIT Working Papers in Linguistics 10: 253-272. TORTORA, C. (2005): The preposition's preposition in Italian: Evidence for boundedness of space, in R. Gess & E. Rubin (eds.) Current Issues in Linguistic Theory, pp. 307-327. Amsterdam: John Benjamins. TORTORA, C. (2006): On the aspect of space: The case of PLACE in Italian and Spanish, in N. Penello & D. Pescarini (eds.) Atti dell’undicesima giornata di dialettologia (Quaderni di lavoro ASIS, v. 5), pp. 50-69. VENDLER, Z. (1957): Verbs and Times, Philosophical Review 56, 143-160. 200 ZAENEN, A.. 1993. Unaccusativity in Dutch: Integrating Syntax and Lexical Semantics. In J. Pustejovsky (Ed.) Semantics and the Lexicon. Dordrecht: Kluwer Academic Publishers. ZUBIZARRETA M. L. (1998): Prosody, Focus, and Word Order. Linguistic Inquiry Monograph 33. Cambridge, MA: MIT Press ZUBIZARRETA, M. L. and Oh, E. (2004): The lexicon-syntax interface: the case of motion verbs. Manuscript. ZUBIZARRETA, M. L. and Oh, E. (2007): On the Syntactic Composition of Manner and Motion, Linguistic Inquiry Monograph, MIT Press.
Abstract (if available)
Abstract
The study of transitivity alternations plays a central role in linguistic research, because it can provide us with valuable information about the interface between the lexicon and the syntax. The main focus of this dissertation is to find the intrinsic features that characterize the locative alternation, i.e. to separate idiosyncratic and language specific properties from those that are inherent to this construction. Special attention is given to the locative alternation in Spanish.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Register and style variation in speakers of Spanish as a heritage and as a second language
PDF
Agreement on the left edge: the syntax of left dislocation in Spanish
PDF
Adding and subtracting alternation: resumption and prepositional phrase chopping in Spanish relative clauses
PDF
Prosody in contact: Spanish in Los Angeles
PDF
The Spanish feminine el at the syntax-phonology interface
PDF
Prosodic recursion and syntactic cyclicity inside the word
PDF
A reduplicative analysis of sentence modal adverbs in Spanish
PDF
The morphosyntax of states: deriving aspect and event roles from argument structure
PDF
Building adjectival meaning without adjectives
PDF
Minimal contrast and the phonology-phonetics interaction
PDF
Building phrase structure from items and contexts
PDF
Narrowing the focus: experimental studies on exhaustivity and contrast
PDF
A corpus-based discourse analysis of Korean grammatical constructions: Focus on the multifold functions and meanings of the pragmatic construction e kaciko
PDF
The poetics of reemergence: psetry, subjectivity, and political violence in the neoliberal age
Asset Metadata
Creator
Mayoral Hernandez, Roberto
(author)
Core Title
The locative alternation: unaccusative constructions and subject position
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Linguistics (Hispanic Linguistics)
Publication Date
06/20/2008
Defense Date
04/30/2008
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
language variation,locative alternation,OAI-PMH Harvest,Spanish,subject position,syntax
Language
English
Advisor
Silva-Corvalan, Carmen (
committee chair
), Zubizarreta, Maria Luisa (
committee chair
), Saltarelli, Mario (
committee member
)
Creator Email
mayoralh@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-m1281
Unique identifier
UC1219202
Identifier
etd-MayoralHernandez-20080621 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-76144 (legacy record id),usctheses-m1281 (legacy record id)
Legacy Identifier
etd-MayoralHernandez-20080621.pdf
Dmrecord
76144
Document Type
Dissertation
Rights
Mayoral Hernandez, Roberto
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Repository Name
Libraries, University of Southern California
Repository Location
Los Angeles, California
Repository Email
cisadmin@lib.usc.edu
Tags
language variation
locative alternation
subject position
syntax