HOW TO BEGIN? TOWARD A GENERAL THEORY OF ASPECTUAL VERBS By Shannon Cousins A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Linguistics—Doctor of Philosophy 2024 ABSTRACT Aspectual verbs, like begin, start, continue, finish, and end, have been shown in recent literature to participate in several linguistic phenomena, resulting in a complex array of theoretical puzzles. The majority of semantic inquiry into these verbs revolves around their transitive constructions, focusing on the compositional relationship between the aspectual verb and its semantic complement. Many previous accounts, which rely on processes like complement coercion, type-shifting, and lexical metonymy to explain specific verbal behaviors, assume that aspectual verbs are type-specific in their semantic selection of complements, only ever taking an event argument. However, Piñango and Deo (2016) introduce the first of many behavioral complexities exhibited by these verbs in their novel observations of their varied stative (non-eventive) interpretations, reflecting their underlying mereological nature. Based on this, their account, the Structured Individual Hypothesis (SIH), in essence, dissolves the event/entity type distinction held by previous researchers and assumes instead a type-flexible account for aspectual verb semantic composition. However, while the mereological framework assumed in the SIH can be considered fundamental to aspectual verb semantics across complement types, the set of novel empirical evidence put forth in this dissertation shows that neither the SIH nor any current theory is sufficient for a comprehensive theory of all aspectual verb behaviors. As these theories limit their scope to primarily transitive aspectual verb structures, they miss generalizations evident when considering the varied argument structures these verbs may appear in (including z began x, x began, x began with y, and z began x with y). Specifically, the novel data of aspectual verb selective transitivity alternations presented here, which indicates their underlying (anti-)causative nature, crucially challenges the generality of the SIH, in particular, as it provides significant support for the event/entity type distinction in aspectual verb composition. The linking of prior (anti-)causation accounts to aspectual verb behavior aids in motivating a monadic approach to their semantic argument structure, which assumes they take in a single argument in their base form. Furthermore, previous accounts suffer from a lack of the fine-grained semantic formalism necessary to account for the varied telicity and meanings exhibited across these verbs. As such, this dissertation has five main goals: (1) to provide a comprehensive review of prior accounts, (2) to examine the theoretical challenges to each in light of novel empirical data, (3) to link aspectual verb behavior to a monadic non-derivational non-lexical (anti-)causation account, (4) to investigate possible expansions of each prior account, and (5) to propose a set of theoretically well-motivated formal components that are argued to be necessary to incorporate in a comprehensive aspectual verb theory. Copyright by SHANNON COUSINS 2024 ACKNOWLEDGEMENTS As I stand on the threshold of this significant milestone in my academic journey, it’s with immense gratitude that I acknowledge the countless individuals who have supported me throughout the challenging but rewarding process of completing this dissertation. I firstly want to thank my graduate advisor, Dr. Brian Buccola, for consistently lending his expertise, feedback, encouragement, and patience, which have played a pivotal role in shaping this work and my academic confidence more broadly. I hope to one day become a similarly great teacher and semanticist. I also extend my appreciation to my esteemed committee members, Dr. Suzanne Wagner, Dr. Karthik Durvasula, and Dr. Alan Beretta, for their invaluable contributions to this work and to my graduate career as a whole. Beyond being amazing sounding boards, Dr. Wagner and Dr. Durvasula have been pillars of encouragement in my professional development and have fueled my determination in future endeavors. In his time as my primary advisor, Dr. Beretta has shown unwavering enthusiasm in his advisee’s projects, continually expressing his pride in our skills and accomplishments, which I could not be more thankful for. I must, of course, also express my deepest gratitude to my PhD cohort, Jason Smith, Dr. Naiyan Du, and Komeil Ahari, as well as my dear colleague, Dr. Ye Ma, for their inspirational perseverance, collaboration, and companionship throughout my time here. To my dear friends, whom I will address directly, I doubt I will ever be able to express how important you have been to me on this journey as I hope I can be for you now and in the future. Victoria, your presence through every step of this process has truly been my anchor. Amidst the chaos, it has been your patience, devotion, and commitment to my well-being that has kept me grounded and I love you dearly for it. Darby, Sarah, and Mitch, your friendship has been invaluable to me over the years, and I will forever cherish the support and joy you have surrounded me with. Yongqing, your positivity and optimism have remained contagious throughout the years, and you have helped to lift my spirits during the toughest of times. And Ben, your gentle encouragement and quiet belief in me regardless of the challenges I’ve faced have helped me immensely in getting to this point in my academic career. You have become a source of strength and solace, and I am v profoundly grateful for our relationship. To my family, your unending belief in my abilities has made this journey possible. You have helped me solidify my confidence with your encouragement throughout my life, which I have heavily relied on in the daunting hours of my time away from you. Lastly, I cannot forget my three feline companions, Lemon, Pigeon, and Poppy, who have been constant sources of comfort. Their unconditional love and often not-so-quiet presence have been a needed grounding on even the most challenging days. As I trek toward the conclusion of my PhD, I look fondly at the memories I’ve made along the way and consider them to be fundamental to the person I am today. Spanning back all the way to the beginning of this journey, I must finally thank the person who was the first to share with me his passion for linguistics and set me on this path, Dr. Greg Johnson, without whom I would never have come this far. Thank you all for the immeasurable roles you have played in helping me reach this goal. vi TABLE OF CONTENTS CHAPTER 1 OVERVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CHAPTER 2 EMPIRICAL AND THEORETICAL LANDSCAPE . . . . . . . . . . . 1 8 CHAPTER 3 NOVEL COUNTER-EVIDENCE: VARIED ARGUMENT STRUCTURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 CHAPTER 4 EXPLORING THEORETICAL POSSIBILITIES: SUPPORTING A MONADIC ACCOUNT . . . . . . . . . . . . . . . . . . . . . . . . . . 51 CHAPTER 5 INVESTIGATING AND REFUTING THEORETICAL ADAPTATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 CHAPTER 6 PROPOSAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 CHAPTER 7 DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 vii CHAPTER 1 OVERVIEW 1.1 Introduction Aspectual verbs, like begin, start, continue, finish, and end, have been shown in recent literature to participate in several linguistic phenomena, resulting in a complex array of theoretical puzzles. The majority of semantic inquiry into these verbs revolves around their transitive constructions (e.g. x began y). Certain constructions like Lia began the book (to mean Lia began reading the book), where no activity verb is given, yet one is still implicitly interpreted, have been of particular initial interest. This data prompted a Type-Shifting Hypothesis which assumes that aspectual verbs necessarily select for an event denoting complement argument and that any complement of a different type (e.g. entity denoting) must be coerced to conform to the verbs selectional restrictions (see McElree et al. 2001). This type-specific account is contrasted in Piñango and Deo (2016), who provide empirical evidence that aspectual verbs in these constructions may also denote non- dynamic, stative readings, as in This sign begins the running trail. They take this to motivate an account that highlights the underlying mereological meanings within aspectual verb semantics. Their Structured Individual Hypothesis (SIH) supports a type-flexible approach to aspectual verb predicate selection, dissolving the event/entity distinction between complement types. Mourounas and Williamson (2019) later attempt to adapt this same account to deal with variable predicate forms, namely when combining with definite and non-definite clauses. Though the SIH generalizes to the multiple meanings exhibited by aspectual verbs, Piñango and Deo (2016), and other recent accounts, including Mourounas and Williamson (2019), do not consider their different argument structures and syntactic realizations, resulting in missed behavior generalizations. For instance, neither this account nor any previous account discusses the selective transitivity alternations that aspectual verbs may undergo (resulting in x began), shown here to be due to their underlying (anti-)causation. Further, no account addresses the possibility of a three- argument aspectual verb predicate structure (z began x with y), which introduces an additional meaning sense for this verb class. This novel data is shown throughout this dissertation to pose 1 significant theoretical challenges to existing accounts. As such, this dissertation gives an extended argument against prior accounts for aspectual verbs, especially regarding the SIH, focusing primarily on three arguments concerning aspectual verb distributions and meanings: (1) transitive structures may denote causative meanings, (2) intransitive/anticausative uses are possible, and (3) the complements that show causative readings in the transitive are the ones which may undergo (anti-)causative alternations to appear in the intransitive, restricted to only event denoting arguments. This overall empirical paradigm for aspectual verb behaviors is exemplified in the following examples, which will be referred to and reiterated throughout this discussion. (1) Lia began the fight ⇒ Lia caused the fight to begin a. The fight began b. The fight begins with Lia (2) Lia began the book ⇏ Lia caused the book to begin a. *The book began b. The book begins with Lia Investigating the complex semantic selection exhibited in the bare intransitive (1a, 2a), along with the theoretical critiques put forth here, ultimately points to a proposal that reconsiders the adoption of a type-specific account, rooted in the event/entity complement type distinction. Further- more, the linking of prior (anti-)causation accounts to aspectual verb behavior aids in motivating a monadic approach to aspectual verb semantic argument structure, which assumes they take in a single argument in their base form. As such, prior accounts by Hovav and Levin (2012) and Hovav (2014) are taken as joint motivation for a non-lexical derivation between such (anti-)causative transitivity alternations.1 This provides additional aspects to an underlying theoretical framework for future accounts. 1Further clarifications and motivations for this non-lexical account are provided in Chapter 3. 2 As no current semantic account for aspectual verbs can (at this point) predict the behaviors discussed, a set of possible theoretical expansions of these previous approaches are investigated here and ultimately refuted. This set of novel empirical evidence coupled with the theoretical analyses considered here ultimately contributes to my argument that no current aspectual verb theory, nor intuitive adaptations of them, is sufficient to account for the complex compositional behavior these verbs show. As such, this dissertation has five main goals: (1) to provide a comprehensive review of prior accounts, (2) to examine the theoretical challenges to each in light of novel empirical data, (3) to link aspectual verb behavior to a monadic non-derivational non-lexical (anti-)causation account, (4) to investigate possible expansions of each prior account, and (5) to outline a set of theoretically well-motivated formal components that are argued to be necessary to incorporate in a comprehensive aspectual verb theory. The structure of this dissertation is as follows: The remainder of this chapter discusses relevant concepts integral to the analyses reviewed in later chapters; Chapter 2 reviews the theoretical and empirical landscape provided by primary aspectual verb theories, focusing on Type-Shifting accounts and the SIH along with their respective theoretical critiques; Chapter 3 introduces the novel evidence of aspectual verb (anti-)causative alternations as well as other complex constructions and discusses the empirical challenges posed by these varieties to prior accounts; Chapter 4 investigates and debunks a possible theory-preserving avenue for the SIH, further motivating a monadic approach to aspectual verb semantics; Chapter 5 tests other intuitive adaptations of previous theories in order to account for the reviewed variable behaviors, which are shown to likewise be insufficient; Chapter 6 provides a synthesized proposal of foundational theoretical claims based on the evidence and critiques reviewed throughout; and lastly, Chapter 7 includes a final discussion and concludes the dissertation. 1.2 Relevant Concepts Before delving into the complex linguistic nature of aspectual verbs, it is first useful to review some crucial theoretical concepts and terms that will be used throughout the following discussion. Presented here is such a review, first considering the nature of the eventualities these verbs describe, 3 followed by a review of mereological (part-whole) relations, which they have been shown to express. 1.2.1 Eventualities and Lexical Aspect A crucial part of studying aspectual verb behavior is understanding the nature of the events they describe. Within semantic theories, verbs are thought to internally make reference to some kind of eventuality (using Bach’s (1981) term), often represented within the verbs’ denotation, so that the meaning of the verb is directly tied to the shape of the eventuality it expresses (Davidson 1967). Different verbs are considered to make reference to different kinds of eventualities, tradi- tionally categorized by Vendler (1967)2 across four classes: states, activities, achievements, and accomplishments, defined and exemplified below. • State: an event which contains no internal structure or change, without an implicit endpoint e.g. Lia loves Mary, The book is pretty good, Mary thinks shes nice • Activity: an ongoing event with internal change, but no implicit temporal endpoint e.g Lia walked along the path, Mary painted happily, She spoke to her mom • Accomplishment: an ongoing event with internal change and an obligatory implicit temporal endpoint e.g. Mary ate the apple, Lia sang the song, They told her a story • Achievement: an event with no internal structure or duration, but may denote change, with an implicit endpoint e.g. Lia arrived home, Mary reached the summit, I recognized her These definitions revolve around the verbs’ lexical aspect (or ‘aktionsart’), which deals with the internal temporal structure of the verbal eventuality, i.e. how that event extends over time. Importantly, this is distinct from grammatical aspect, which relates an event or subevent to some reference time in the grammar via tense morphemes. The lexical aspectual properties of verbal 2Though many varying classification systems like Vendler’s have been proposed, for the sake of this review, his system is used as it is the most heavily adopted within the field. However, an adapted account given by Piñón (1997) will be reviewed and considered in Chapter 2 4 events are characterized and distinguished in terms of dynamicity - whether an event denotes some internal change of state - and telicity - whether the event has an implicit endpoint (Jackendoff 1991). A reliable diagnostic for a predicates internal telicity is whether it may combine with an adverbial phrase headed by either the prepositions for or in. For example, the atelic phrase walked in (3) is compatible with a for-phrase, expressing duration of an event, but is not compatible with an in-phrase, as it does not express any end-point of the event. Conversely, the telic predicate ate an apple in (4) is compatible with an in-phrase but not with a for-phrase. (3) Lia walked (*in ten minutes / for ten minutes) (4) Mary ate an apple (in ten minutes / *for ten minutes) This telicity test further indicates that the meaning of a verbal eventuality is dependent not only on the verb itself but on the predicate clause as a whole; factors like adverbial modification and the nature of the thematic participants in the composition interact with the interpreted aspectual properties of the verb. Similar to other semantic arguments in our ontology, events are considered to be ‘objects’ in the same sense as truth values, ordinary entities, possible worlds, degrees, etc., and thus may be input into linguistic compositions in a similar functional way. These verbal eventualities encode specified relations between sentential participants, assigning them thematic roles, e.g. an agent (sentient actor of the event), theme (argument undergoing/affected by the event), cause (the causer of the event), goal (the target of the event), etc. Within Neo-Davidsonian semantics, these thematic roles are part of the event structure and serve to further define the predicate meaning (Parsons 1990). For the most part, while verbs may often shift in grammatical aspect (tense shifts), it is thought that only in certain structural cases may a verb undergo a change in its lexical aspect (Thomp- son 2006). Though, as will be shown, in both transitive and complex intransitive constructions, aspectual verbs may appear with multiple aspectual meanings, denoting either a bounded telic achievement, an unbounded atelic state, or an ambiguity between them, depending on the interpre- 5 tation of the eventuality participants. This phenomenon will be examined further in section 2.3 and throughout. 1.2.2 Mereological Semantics In mereological semantics, certain objects can be conceptualized as being made up of parts; pages are parts of a book, individuals are part of a group, etc. and language is sensitive to these different kinds of part structures. Moltmann (1997) understands such entities as integrated wholes - distinct spatiotemporal entities that are made up of related parts, which may themselves reflect integrated wholes. The parts of an object form a specified part structure, which can be interpreted across different conceptual dimensions, including time, space, individuals, etc. For example, a concert event can be understood as reflecting an interval of time along a temporal dimension, made up of sub-interval performances. Similarly, a book can be understood as reflecting an object in space along a spatial dimension, made up of smaller objects (chapters) in space. However, the part structure of an integrated whole can vary in two ways; (i) a singular entity/event may have different sets of parts in different pragmatic/situational contexts, and (ii) a singular entity/event can have different sets of parts along different dimensions. For example, a book may be segmented by pages or by chapters depending on context, and may similarly have a differently structured order depending on the point of reference. Further, a concert may be interpretable as an extended eventuality segmented based on either songs or performers. Krifka (1998) assumes that all ontological dimensions – time, space, eventualities, and entities – are characterizable by the same underlying mereological algebraic structure and thus can be considered within the same mereological framework. This perspective is followed by Piñango and Deo (2016), who motivate an account which interprets aspectual verbs as denoting mereological part relations between their thematic participants, and further instantiates a presupposition on the verb which invokes selectional restrictions involving the part structure of the complement argument. This will be discussed further in the following chapters. With this review of fundamental concepts, we may now turn to the current theoretical and 6 empirical landscape regarding the complexities associated with aspectual verb behaviors. The fundamental accounts given by type-shifting literature and Piñango and Deo (2016) are individually investigated and assessed in terms of theoretical strength and generality. 7 CHAPTER 2 EMPIRICAL AND THEORETICAL LANDSCAPE 2.1 Introduction This chapter focuses on two linguistic behaviors aspectual verbs take part in that have been the primary areas of focus in recent literature; namely, their ‘complement coercion’ constructions and their multiple mereological meanings. Following a description of each behavioral phenomenon is an outline of the prominent theories applied to account for them; the Type-Shifting Hypothesis (see Pustejovsky and Bouillon 1995) and the Structured Individual Hypothesis (Piñango and Deo 2016), respectively. As will be shown, while the former theory may effectively capture the complement coercion (henceforth, CC) construction data, it alone cannot capture the multiple mereological senses of the verb outlined in Piñango and Deo (2016). While the presented data empirically challenges the generality of the Type-Shifting Hypothesis, the Structured Individual Hypothesis is similarly critiqued on several theoretical bases, as discussed in 2.3.3. In what follows, each phenomenon and their respective theories are discussed, along with their competing theoretical implications for aspectual verbs broadly. 2.2 Complement Coercion Constructions 2.2.1 The Phenomenon One of the possible meanings that aspectual verbs like begin, end, continue, etc. may have is one that is eventive, specifically denoting the temporal ‘edges’ of some durative event it combines with, expressing the aspectual point at which it begins, continues, or concludes. This meaning sense has been extensively researched, prompting the theoretical perspective that aspectual verbs have a singular lexical entry which takes in some durative event argument to modify (Freed 2012; Pustejovsky 1991). In transitive aspectual verb constructions, this event argument may be provided explicitly by an embedded activity verb, e.g. packing in I began packing my suitcase or running in She finished running the marathon. Consider now a similarly well-researched1 semantic behavior 1See Baggio et al. 2010; Brennan and Pylkkänen 2010; Frisson and McElree 2008; Frisson and Pickering 2001; Jackendoff 1997; Katsika et al. 2012; Kuperberg et al. 2010; McElree et al. 2001; Pickering and Frisson 2001; Pickering et al. 2006; Piñango et al. 2006; Pustejovsky 1991; Pustejovsky and Bouillon 1995; Traxler et al. 2005 8 of aspectual verbs exemplified below in (5), where no event or activity is explicitly provided in the semantic structure, yet one is still implicitly (understood, but unexpressed) interpreted. This is compared to the example in (6) where the event/activity is denoted explicitly in the composition from the event nominal complement. (5) The author began/finished/continued the book. reading/writing (6) The boxer began/finished/continued the fight. The resulting interpretation in (5) has an implicit predicate phrase of reading/writing, that incorporates the complement, the book, as its direct object. However, the default nominal inter- pretation of the book is not of an activity or an event, but of an entity. As such, under a theory which assumes aspectual verbs take in only an event denoting complement, it is unclear what processes underlie this apparent event extrapolation from a default entity denoting complement. Compare this sentence to the one in (6), where the lexical meaning of the complement, the fight, as an event nominal, naturally denotes an event argument (Borer 2003; Grimshaw 1990). In this sentence, the resulting compositional eventive meaning comes about straightforwardly given the default event interpretation of the complement. The phenomenon demonstrated in sentences like (5) initially prompted researchers (Jackendoff 1997; Pustejovsky 1991; Pustejovsky and Bouillon 1995) to motivate an argument for a necessary enriched form of semantic composition in these specific CC constructions. This process involves the default interpretations of semantic elements in the sentence being modified in order to adhere to the aspectual verb’s hypothesized semantic restrictions, as examined below. 2.2.2 Type-Shifting Hypothesis Pustejovsky and Bouillon (1995) first applied the idea of enriched semantic composition to account for the process of apparent event extrapolation in these aspectual verb constructions. To review, a process of semantic enrichment introduces semantic structure that is not explicitly represented in the surface form (Jackendoff 1997; Pustejovsky 1991; Pustejovsky and Bouillon 1995). In other words, additional meaning is extrapolated implicitly from the conceptual structure 9 that is not present in the lexical surface structure (e.g. the implicit event in (5)). In the case of these aspectual verb constructions, this enrichment has been implemented in terms of a coercion process (Partee and Rooth 1983; Partee 1990), wherein there exists a coercion trigger and some object that is coerced via a coercion operator in the composition. The fundamental trigger of coercion is a mismatch between the selectional properties of a ‘selector’ in the semantic composition and the semantic properties of the ‘selected’ element, with the presupposition that any element that has selectional restrictions, like aspectual verbs are assumed to have, has the potential to impose a coercion trigger (Lauwers and Willems 2011).2 In the case of sentences like (5), the aspectual verb is considered to be the selector while the complement is the selected element, following typical semantic function application. As stated, on this view of aspectual verbs, their default semantic selection is restricted to an event denoting argument, as in (6). This assumption is rooted firstly in the assumption that they denote a singular lexical entry (as opposed to distinct entries for distinct compositional verbal restrictions), and secondly on the fact that in transitive structures, these verbs often take event denoting verbal arguments, as in Lia began snoring, where the verbal expression is eventive. Indeed, a preference towards the eventive reading as the default interpretation of these verbs is further supported in the Acceptability Judgment task reviewed in Chapter 3. Thus, when aspectual verbs combine with an entity denoting complement, there is a mismatch between the verb’s hypothesized selectional restrictions and the entity denotation of the complement, acting as a trigger for a coercion process. Specifically, Pustejovsky and Bouillon (1995) argue that the complement undergoes an operation in which its semantic type is coerced from an entity to an event, hence the nomenclature of complement coercion (CC) to describe such phenomena. This coercion is thought to result in the repair of the mismatch between the verb’s selectional restrictions and the complement’s lexical properties, resulting in the implementation of additional semantic structure to restore meaning. 2As recognized in Lauwers and Willems (2011), these processes fall along the same conceptual lines as other theorized notions of recovery such as pragmatic accommodation (Goldberg 1995), which seem to require similarly theoretically slippery operations. 10 To explain the process by which an event is extrapolated in these CC constructions, Pustejovsky and Bouillon (1995) rely on the lexical representation of the entity denoting complement, specifi- cally claiming that the qualia structure it expresses includes telic and agentive properties that are accessible in the composition. The qualia structure of a lexical item accounts for a wide variety of linguistic features intrinsic to the lexical item, specifying its acceptable forms of composition and interpretation including any argument selection, adjectival modification, and type coercion (Pustejovsky (1991) and Pustejovsky and Bouillon (1995)). In the case of these complement types, the telic properties extrapolated from their qualia structures describe the function/purpose of the complement while the agentive properties describe how it comes to be (Pustejovsky and Bouillon 1995). Consider the book; the qualia function of it is to be read, so, on this account, the implicit activity verb in the telic structure is extrapolated as read. It also comes to be by the activity of writing, hence the implicit verb in the agentive structure is extrapolated as write. In such cases, the aspectual verb may access either metonymic meaning of the complement. In this sense, the coercive type-shifting process in CC constructions is dependent on the qualia structure of the entity complement of the verb, namely whether it may achieve a telic eventive interpretation. Another verb class that has been shown to exhibit similar behavior patterns and as such has been considered alongside aspectual verbs are psychological (henceforth, psych) verbs like enjoy, prefer, tolerate, resist, endure, savor, etc. The verbs in this class also seem to participate in transitive CC constructions, where no embedded activity verb is expressed, but one is implicitly interpreted, as exemplified in (7) and (8) (Katsika et al. 2012). (7) Mary tried the ice cream. (8) Lia enjoyed the book. tasting/eating reading/writing Though the class of CC verbs had initially been considered to include both aspectual and psych verbs, Katsika et al. (2012) distinguish the typical CC operation and the one apparent in the psych verb examples above by appealing to the verbs’ differing argument structures. They argue that, in contrast to aspectual verbs, the external argument of psych verbs is that of a subject experiencer, 11 with their complement “entailed to be either a target of emotion or subject matter of emotion,” adopted from Pesetsky (1996). Given that the subject of emotion may manifest as either an entity (as above), event (e.g. tolerate the performance), or state-of-affairs (e.g. situations, processes), they claim that while psych verbs may take an event denoting complement, this is not necessarily encoded as a semantic selectional restriction as is thought to be the case for aspectual verbs. Rather, they attribute the operation of apparent event extrapolation in psych verb constructions not to a coercion of the complement from one type to another, but to a pragmatic inference process necessary to infer a relevant activity of targeting the subject-experiencer’s emotion to the complement.3 It is these behavioral facts that have prompted much of the early experimental research on aspectual verbs. Indeed, this data has been argued to persist cross-linguistically, analyzed in (American) English, Mandarin (Hsu and Hsieh 2013; Ma et al. 2022; Song 2014; Xue and Liu 2021; Zuo-ya 2011), German (Zarcone et al. 2017, 2015), and Icelandic (Jalbert et al. 2020). Given that this coercive Type-Shifting Hypothesis necessarily assumes aspectual verbs to be selective in the semantic type they take in, specifically restricted to events, this framework considers them to be type-specific. The competing framework promoted in Piñango and Deo (2016), to be discussed in 2.3.2, though also assuming a singular lexical entry for aspectual verbs, argues against a type-specific perspective on two counts: Firstly, they demonstrate stative (non-eventive) data showing that aspectual verbs can in fact take in entity complements that retain their entity readings. Secondly, and more relevant to the discussion presented here, they assume a type-flexible account for aspectual verb complement selection, treating both entity and event complement types the same compositionally and conflating their distinction in terms of aspectual verb selectional restrictions. This theoretical difference is further highlighted and challenged by the novel empirical data of intransitive aspectual verb constructions to be analyzed in Chapter 3. Barring the complexities and issues with over-generalization in the Type-Shifting Hypothesis, it provides a well-adopted theoretical perspective in accounting for transitive eventive aspectual verb 3Importantly, psych verbs have been shown to have differences in online processing costs when compared to aspectual verbs (Katsika et al. (2012)), further suggesting that a distinct and potentially more complex theoretical approach is necessary for either class. 12 sentences like (5). However, as will hopefully be evident from the work presented here, this account must be contextualized and assessed within the entire breadth of aspectual verb behaviors. Type- shifting Hypotheses hold an underlying claim that aspectual verbs are restricted only to eventive interpretations; an account which, given the observations made in Piñango and Deo (2016) of stative interpretations, will be shown to be relatively too narrow. We turn now to a review of this mereological account. 2.3 Structured Individual Arguments 2.3.1 The Phenomenon Another semantic behavior of aspectual verbs to address stems from the relatively recent observation in Piñango and Deo (2016) that they may not only be interpreted as expressing a telic achievement eventuality, as in CC constructions, but may also have generic stative readings where the verb instead denotes an unbounded atelic state eventuality. Thus, prior theoretical accounts for these verbs which rely only on their eventive/agentive readings, as in Type-Shifting Hypotheses, unintentionally miss the generalizations necessary to capture this data. Shown below are possible stative sentences including aspectual verbs begin, continue, and end.4 (9) The porcelain doll begins/continues/ends the row of toys. (10) This little sign begins/continues/ends the hiking trail. (11) The chapter on poverty begins/continues/ends the book. (12) A piece by Vivaldi begins/continues/ends the concert. In each of these examples, the eventuality being expressed is stative, containing no internal change or specified duration. These stative constructions highlight a multitude of mereological part-whole relations that aspectual verbs may denote. In eventive constructions (e.g. Mary began climbing the mountain), the aspectual verb is thought to denote an instantaneous sub-event (a beginning event) that is located adjacent to the boundaries of a whole durative event (a climbing event), where Mary is the agent (to be discussed further in 2.3.3.1). In contrast, the part-whole 4For ease, the aspectual verbs in these sentences are given in present tense, which has been shown to pattern more with stative readings in English. 13 relations exhibited in (9)-(12) relate the structure of aspectual verbs’ thematic participants to one another, namely between the subject part and the complement whole. In the cases of the complements in (9)-(11), their entity meanings are preserved, where the verbal predicate similarly indicates some sub-part at the boundaries of the entity whole. Due to this, Piñango and Deo (2016) refer to such stative readings as constitutive, where a sub-part constitutes a part of the whole. For example, in (9), the subject the porcelain doll is understood as the initial/medial/final itemized part of the row of toys. Similarly, in (10), this little sign is a part of the hiking trail that serves to mark its external and internal spatial boundaries. In (11), the chapter on poverty can be interpreted as a specific portion of information represented as an initial/medial/final informational part of the book.5 Finally, in (12), a piece by Vivaldi denotes a temporal interval at the initial/medial/final boundary of the durative event denoted by the concert, similar to the meanings in CC constructions, though, crucially this predicate denotes a state. Moreover, examples (9), (10), and (11) contain entity denoting verbal complements, whereas (12) contains an event denoting complement. As such, the sentences in (9), (10), and (11) each construct the part structure of the complement along some non-temporal semantic dimension, whereas (12) represents it along a temporal dimension, reflecting that these aspectual verbs may describe part relations between participants across multiple mereological domains (i.e. space and time). This observation relates to the multiple meanings that Piñango and Deo (2016) argue aspectual verbs are able to access given more general selectional requirements. Furthermore, in line with the dependencies on the complement’s qualia structure in CC construc- tions, the specific interpreted part structure of the complement is dependent on the interpretations of both the complement and the subject, given that different parts may make up their whole enti- ties/events in variable ways (see Frazier and Rayner (1990)). Consider again the sentences in (5b), The author began the book and The boxer began the fight. In this case, the complement (book/fight), 5These constitutive transitive constructions may be considered to be marginally questionable to some readers due to the salience of the eventive verbal interpretation of aspectual verbs. However, experimental evidence presented in the latter half of Chapter 3 provides support for their availability among native American English speakers. 14 seemingly regardless of its default type, may be construed either with an eventive agentive reading, i.e. as the theme of a predicate of events, or with a stative mereological interpretation, i.e. as a whole object made up of parts denoted by the subject. The eventive reading is notably more salient; so salient, in fact, that some readers may find it difficult to access the constitutive reading. Conversely, with a sentence like (11), the availability is reversed, given that chapters aren’t the kinds of things to agentively read/write books. Consider (5), The author began/finished/continued the book; in the eventive agentive reading in the CC construction, the subject (the author) is interpreted as the agent of a sub-event part extrapolated from the complement theme (book), where the aspectual predicate denotes a telic achievement. However, in the stative reading of (5), the author is instead understood as a sub-part of the informational whole that is represented by the book, e.g. an author’s note to the reader. The same variability in meaning can be observed in (6), The boxer began/finished/continued the fight; consider another reading where the boxer is understood not as the agent of a fighting event, but e.g. as the first fighter in a lineup of fights for that night. On this meaning, the complement fight is interpreted as a temporal structure ordered by individualized fighters. Thus, it can be said that the fight began with the boxer as the first fighter in an event consisting of many other fights, resulting in a stative reading where the boxer is understood as representing the initial/medial/final part of the temporal whole that is represented by the fight.6 This data shows that both the availability of mereological stative meanings in these transitive constructions and the exact interpretation of the part-whole relation mapped between the subject and complement are contextually dependent on the semantic denotation of both thematic participants, which manifest either as a thematic agent of a subevent or as a subpart of a whole. These observations from Piñango and Deo (2016) show that in certain structures, aspectual verb predicates have an ambiguity in their event structure, stemming from differences in the interpretations of their thematic participants along with how the verb relates them to one another. This ambiguity is analyzed further by the account given in Piñango and Deo (2016), which contextualizes prior aspectual verb accounts 6This began with... structure, shown to paraphrase the stative transitive, will be discussed in more detail in section 2.4.3. 15 within these stative mereological behaviors. Also gleaned from this data is how it challenges prior type-specific accounts on aspectual verb behavior like the Type-Shifting Hypothesis, which necessarily assumes aspectual verbs in such transitive constructions are restricted to an event denoting complement argument. The apparent mereological meanings present in (9)-(12), along with the variable meanings accessed in (5b), reflect part-whole relations across a range of dimensions, regardless of the default type of the complement; both entity and event denoting complements may be input into the predicate composition. While this poses a challenge for a type-specific stance (and further calls to question the generality of the Type-Shifting Hypothesis), Piñango and Deo (2016) explain such data by assuming a type-flexible account for aspectual verb composition, as discussed below. 2.3.2 Structured Individual Hypothesis Based on these phenomena, Piñango and Deo (2016) proposed the Structured Individual Hy- pothesis (SIH), which applies a multi-dimensional mereological semantics to aspectual verb de- notations, crucially incorporating their part-whole makeups within the mereological framework in Krifka (1998). This account approaches the underlying aspectual predicate structure more gen- erally, adopting Krifka’s idea of a generalized axis argument in defining the lexical semantics and selectional restrictions of verbal predicates. An axis is defined as a linearly ordered set of conceptual entities in any ontological domain (time, spatial paths, events, and objects) - in this way, the conceptual structure of the verbal predicate can generalize across multiple dimensions (Deo et al. 2013; Gawron 2009). These axes reflect a specified part structure of some entity/event along an ontological domain. Piñango and Deo (2016) adopt a subclass of Krifka’s axioms of such axes, a one-dimensional di- rected path structure (DPS), as most relevant to aspectual verb composition. Such one-dimensional DPSs are defined as follows, where for each two convex, non-overlapping parts x and y within a domain, it holds that either x precedes y or y precedes x. (13) ∀ x,y ∈ D[¬ x⊗y → x ≤ y∨y ≤ x] This results in a convex linear structure whose adjacent parts (x, y) are totally ordered by the 16 precedence relation (≤), which may be interpretable across multiple contextually derived domains. As Piñango and Deo (2016) put it, this structure reflects a conceptual ‘line’ which connects the edges of a one-dimensional DPS. For example, this may be conceptualized as a temporal precedence between subevents in a sum event, e.g. a concert event linearly structured by separate performances, which constructs a one-dimensional path along a temporal dimension. This can also be conceptualized as a spatial ‘line’, e.g. the connection between the U.S. cities of Lansing and New Orleans comprised of spatial subparts ordered by precedence, dependent on the point of reference (from Lansing vs. from New Orleans), constructing a one-dimensional path along a spatial dimension. Krifka (1998) provides the example of a book as a DPS, whose adjacent ordered subparts may be conceptualized as chapters or even pages. In this sense, the precedence relation orders such DPSs relative to their respective informational or spatial subparts. Similarly, a line of people constitutes a one-dimensional DPS, linearly ordered by individuals depending on the point of reference (whoever ‘begins’ the line to whoever ‘ends’ it). Piñango and Deo (2016) conceptualize the ontological domains along which axes may fall, namely the temporal, spatial, eventive, and individual domains, as one-dimensional DPSs in their own right and follow that connections exist between them via homomorphic partial functions that map objects from one domain to objects in another. A homomorphism is a structure-preserving functional mapping, i.e. a function from a structure A to a structure B that preserves the structures of both A and B. This set of functions include typical thematic role functions which map individuals to events along with temporal and spatial trace functions which map individuals and events to their respective temporal or spatial point. Most relevant to aspectual verb composition are those functions that map entities to one-dimensional DPSs. On this account, aspectual verbs serve to constrain the relations between these ontological domains. Such one-dimensional DPS axes (henceforth, axes, for simplicity) and the entities that relate to such axes via homomorphic functions are called structured individuals by Piñango and Deo (2016). The SIH claims that aspectual verbs impose selectional restrictions in that they select a structured individual (SI) as their complement, define in (14), explained in (15), and some entity that may be 17 mapped onto it as their subject. (14) ∀x𝜏[ struct-indf ⟨ 𝜏,𝜎⟩ (x) ↔[axis(f (x)) ∧ ∀y , z ≤ x [y ≤ z → f (y) ≤ f (z)]]] (15) An individual x of any type 𝜏 is taken to be a structured individual relative to a function f of any type (𝜏, 𝜎) iff f (x) is an axis and f is a homomorphism from the part structure of x to the axis f (x) adapted from Piñango and Deo (2016) The makeup of this axis is contextually dependent on the SI complement’s denotation and qualia structure. In order to achieve the multiple interpretations apparent in transitive aspectual verb sentences, Piñango and Deo (2016) assume that the axis of the aspectual verb complement falls along an appropriate dimension lexically encoded on the aspectual verb. The mechanisms of this are summarized below. Aspectual verbs are said to select for a general SI argument based on a lexically encoded presupposition. If satisfied, the SIH claims that a homomorphic function is applied to the SI complement in order to map it to possible dimensions for the axis to be interpreted along. Thus, the generalized lexical denotation for begin under this framework is given in (16) and explained in (16b). (16) a. [[𝑏𝑒𝑔𝑖𝑛]] = 𝜆𝑥𝜏 𝜆𝑦𝜎 : struct-indfc (x). ∃f ’[f ’(y)