1,721,046 research outputs found
Replication Data for: Finite state intensional semantics
No full textReplication material for a paper to be presented at the 12th International Conference on Computational Semantics, 19-22 September 2017 Montpellier (France). Includes code that interprets a logical language in finite state possible worlds semantics, and a categorial grammar that translates an English fragment into the logical language. Based on Xfst/Foma and Barker and Shan's reference parser for continuation categorial grammar
Replication Data for: Finite state intensional semantics
No full textReplication material for a paper to be presented at the 12th International Conference on Computational Semantics, 19-22 September 2017 Montpellier (France). Includes code that interprets a logical language in finite state possible worlds semantics, and a categorial grammar that translates an English fragment into the logical language. Based on Xfst/Foma and Barker and Shan's reference parser for continuation categorial grammar
Lexical Structure, Weightedness, And Information In Sentence Processing
Get PDFcorpus. We derive greater surprisals (Hale, 2001) for the unergative than the unaccusative case, which we interpret as supporting the account that reduced relative clauses with unergative verbs present the human comprehender with compounded surprise. Overall, the thesis argues that the proper explanatory role of lexical semantics in sentence processing is complexity-based, and not a special appeal to lexically-sensitive ameliorative effects (-Q verbs, as argued in Ross (1967), unaccusative verbs as argued by Stevenson and Merlo (1997)). The thesis argues that the human sentence processor is optimized for human language, and that these classical sentence processing cases are better considered as corner cases where the human sentence processor fails to leverage the lexicon. In such cases, lexical ambiguity compounds main structural ambiguity. This thesis argues for a resource-optimizing parallel parser where at each point in the sentence, processing resources are allocated proportionally to their probabilistic weight. Information-theoretic accounts of sentence processing (Hale, 2001, 2006; Levy, 2008) obtain on an account of the human sentence processor where resources are dynamically reallocated to more likely and less entropic hypotheses to insure that sentences are parsed as fast as possible with as few resources as possible. The thesis argues for a probabilistic lexical syntax, where the observations of professional lexical semanticists can be encoded in a grammar that can be empirically weighted by corpora and computed by a parser. The thesis strives to maximize parsimony and theory-independence in both the core and linking hypotheses: our theories are formulated as lexicalized mildly context-sensitive grammars, directly suggested by independently attested lexical semantics facts, and translatable into a variety of formalisms (Minimalist Grammars, Tree Adjoining Grammars, Generalized Phrase Structure Grammars); and our results are information-theoretic com- plexity metrics (entropy, surprisal), which are interpretable as theory-independent measurements, respectively, of ambiguity and surprise
Types Of Focus In Spanish: Exploring The Connection Between Function And Realization
Get PDFThis thesis revisits the divide between information focus and contrastive focus in Spanish. This divide is said to manifest itself not only in meaning differences, but in surface structure as well, resulting in different syntactic and phonological realizations depending on the intended pragmatic function of a focused sentence. According to previous accounts, an utterance in which focus is expressed by dislocating the relevant element to the end of the sentence is assumed to signify information focus, while a strategy utilizing a specific prosody change on the focused element is said to express contrast. This work argues, in opposition to previous assumptions from the literature, that there is not a strict divide between the realization of one kind of focus or another, and in fact, that these preconceived meaning divides are not themselves straightforward to characterize. Further, this work argues for the possibility that the choice of focus construction is highly influenced by speakers' communicative intentions and constraints. The conclusions reached in this thesis are the product of a combination of empirical and theoretical work. Empirical evidence is drawn from two sources: 1) data from an original elicitation experiment involving native speakers of Spanish producing focused constructions under different pragmatic situations; and 2) findings from the literature on on-line sentence processing studying focused constructions specifically. The first source of data points at the conclusion that the strict information-as-syntactic vs. contrast-as-phonological divide has no base in Spanish. The second source argues for the need for a more functionally-informed approach to focus constructions. A formal analysis of the data using the QUD framework also demonstrates that different kinds of focus can be represented under a single unified semantic approach
The Relevance Of Gradability In Natural Language: Chinese And English
Get PDFThis dissertation explores how the syntax and semantics of gradability contribute to the understanding of other linguistic phenomena. Within this research agenda, I examine three different topics instantiating the interaction between gradability and other linguistic notions. The exploration is important not only for studying the linguistic properties of these phenomena, but also for understanding theoretical issues behind them. Chapter 1 provides an overview of the basics of degree semantics and the interval-based ontological formalization of degrees. Chapter 2 presents a detailed description and syntactic-semantic analysis of the construction in which a possessive verb takes a surface degree expression. During the description and analysis, the construction is compared to other degree constructions. The interpretation of the construction can be derived from the function of the possessive verb interacting with the interval-based representation of degrees. My proposal provides additional support to the small clause-based analysis of possessive verbs. Chapter 3 deals with the interaction between nominal and adjectival gradability, through examining size adjectives used as degree modifiers for gradable nouns. In particular, I examine the status of the Bigness Generalization, which says that only positive size adjectives can degree-modify gradable nouns. I show that the generalization is not categorical, because negative size adjectives indeed can be degree modifiers for gradable nouns in certain contexts. In the chapter I propose an analysis of the phenomenon at the semantic-pragmatic interface that explains the violable nature of the Bigness Generalization. Chapter 4 explores the relevance of gradability to the interpretation of weak generic sentences best represented by Dutchmen are good sailors. Interpreting such a sentence requires restricting the domain of individuals that the sentence quantifies over and determining what the underlying predicate is. The first aspect makes reference to a contextual standard associated with the population denoted by the subject, and the second aspect makes reference to a different standard, which is associated with the population alternative to the denotation of the subject
Towards an articulatory model of tone: a cross-linguistic investigation
Get PDFThis thesis examines the role of timing information in phonological representation, focusing on how tone aligns with segmental material. On the basis of three acoustic studies, I present a novel gestural model of tone representation, where tone gestures are durationally underspecified and receive their timing information from the constellation of segmental gestures they are coordinated with. I also argue that the distributional and temporal characteristics of tone are the direct result of gestural coordination: phonological association can be analyzed as the existence of coordinative relationships between a tone gesture and a constellation of segmental gestures, and the precise nature of that coordinative relationship produces the cross-linguistically variable acoustics. Chapter 1 delineates two major approaches to tone representation: Autosegmental(-Metrical), which references point-like features and nominal time, and Articulatory Phonology, which uses gestures that unfold over time and space. I discuss the different ways in which each theory arrives at phonetic realization from underlying representation, as well as their differing notions of overlap. Chapter 2 presents the results of an acoustic study on contour tones in Thai (Tai-Kadai), a tone language where the mora serves as a licensing unit for tone. Counter previous hypotheses, tonal extrema do not map to moraic edges: both moraic edges and tone extrema instead independently refer to the syllable as a unit of timing. Based on these results, I argue that the apparent acoustic mismatches can be straightforwardly derived from the application of different coordinative modes between tonal and segmental gestures, while maintaining the phonological licensing relationship between moras and tones. The data also suggests that the F0 targets of tone gestures play a role in tone timing, indicating that there is an interaction between the tone gesture and segmental gestures when determining duration. Chapter 3 presents the results of an acoustic study focused on the realization of the falling accent in the Belgrade and Valjevo dialects of Serbian (Indo-European---Slavic), a tone language where one syllable per word is specified for tone. I compare the alignment and duration of pitch excursions across varying phonetic and phonological properties of the syllable onset of the tone-bearing syllable. I show that the duration of pitch excursions increases with phonetically longer syllable onsets, which indicates that tone gestures are durationally underspecified and receive their timing information from the constellation of segmental gestures they are coordinated with. The two dialects also exhibit distinct patterns of both the duration and the alignment of the pitch excursion, and I argue that this is due to differences in the type of coordination used. Chapter 4 focuses on the rising accents of the same dialects of Serbian, crucially examining the Valjevo dialect, which routinely retracts the pitch peak into the syllable preceding the H-bearing syllable. Despite this phonetic retraction, the patterns of alignment parallel those observed for the falling accent. This indicates that the tone gesture is still receiving timing information from the H-bearing syllable, and as such is still coordinated to it. Based on these results, I argue for the availability of gestural target coordination, in addition to gestural onset coordination. In Chapter 5 I synthesize the findings from the experimental chapters to present a gestural model of tone representation, and discuss its implications for Articulatory Phonology and avenues for future research
Equational Reasoning for Verified Cryptographic Security
No full text150 pagesModern software systems today have increasingly complex security requirements – such as supporting privacy-preserving computations, or resistance against quantum attackers – that are fulfilled by advanced forms of cryptography. At the same time, these advanced forms of cryptography often have subtle security proofs that require careful auditing. To ensure security, it is thus crucial to formally verify the security of the underlying cryptography, and to do so in a manner that is approachable to cryptographers. This thesis explores the use of equational reasoning to conduct machine-checked security proofs. Equational reasoning is pervasive in cryptography, as it underlies the concepts of game-hopping hybrids and the simulation paradigm; thus, optimizing formal tools for equational reasoning delivers machine-checked proofs closer to their on-paper counterparts.We first present AutoLWE, a prover for cryptographic primitives that sup- ports reasoning about lattices. AutoLWE is built around deducibility, which (semi-) automatically applies hardness assumptions by partitioning the security game into an application of the hardness assumption with a context. Using AutoLWE, we deliver very short proofs of several representative constructions, including Public-Key Encryption, Identity-Based Encryption, and Inner Product Encryption. We then present IPDL, a simple calculus and equational logic for distributed, interactive cryptographic protocols in the computational model. The purpose of IPDL is to prove simulation results between real and idealized protocols in the style of Universal Composability (UC) [Can01]. IPDL does so by restricting its attention to straight-line protocols, a particularly simple but expressive subset of protocols. Using IPDL, we deliver short proofs of multiple case studies, including a semi-honest multiparty computation protocol over general circuits [GMW87], and an n-party coin toss protocol [Blu83]
Alternatives, Exclusivity and Underspecification
No full text418 pagesThis dissertation serves as both a broad analysis of the structure and generation of alternative sets and a detailed proposal for the lexical semantics of exclusive operators across the world's languages. Much of the dissertation revolves around English exclusive just. Just is particularly elucidative of the analysis I propose due to extreme variability; however, there are exclusive operators with similarly interesting patterns across many languages, and the analysis applies equally to them. At the broad level, I propose that the generally accepted view that exclusive operators like only are (merely) focus-sensitive is an incomplete account of the distribution of interpretations available for exclusivity. I propose that non-canonical exclusive operators like just demonstrate that focus is only one of several mechanisms by which alternative sets can be generated, and furthermore that focus follows the same basic template as generalized alternative set generation. Lastly, I take seriously the concerns raised in Zimmermann 2017 about the failure to generate the correct truth conditions when operators like only are taken to be intensional operators. I argue that a structured proposition approach allows us to retain some of the key insights of Rooth 1992 while avoiding the pitfalls of quantification over propositions. In the process, I provide a single semantics which allows any element to be the locus of the alternatives. The semantics of exclusives that I propose is of a polymorphic type with a restriction that its first argument be a potential argument of its second. These correspond to the focus and background components of a structured proposition. It also must compose with a contextually determined scale, which will vary depending on the lexical requirements of the particular exclusive. This semantic entry allows exclusive operators like just to share a core semantics with only, capturing their similarities while also staying broad enough to account for all of the variability in uses of non-canonical exclusives. In particular, the uses of exclusives I am concerned with are grouped loosely into several categories: canonical exclusives (quantification over overt elements), quantification over causation, quantification below the event, quantification over degrees and granularity, and metalinguistic quantification over discourse and potential questions. I define several parameters by which exclusive operators may vary. These parameters include scale type (e.g., entailment vs. evaluative), adjunction site, whether there is an additional lexical type restriction, whether its first argument must be overtly expressed, and whether the quantification is truth-conditional. This allows us to define the category of exclusives that pattern like merely as those that are required to adjoin above vP, be ordered by entailment, and yield a truth-conditional quantification. Only, on the other hand, is like merely except it is not restricted by scale type. Alone, however, is type-restricted to quantify over individuals in a way that neither only nor merely is. And what makes just so interesting is that it is highly underspecified and therefore not restricted along any of the parameters that I have identified. It has polymorphic type, can occur in a variety of syntactic environments, be ordered by any scale type, 'associate' with covert elements, and yield non-truth-conditional quantifications. I derive the requirement of exclusive operators like only to operate over focus alternatives from the adjunction site parameter. I argue that focus constructions are only available at the matrix level; thus, exclusives that adjoin within the vP layer do not associate with focus. This is made very clear by the distribution of meanings of Serbian exclusives. The ordinary exclusive, samo `only', associates with focus and does not agree, while the morphologically related agreeing form, sam-o/acannot associate with focus. Rather, its alternatives are derived via enrichment of the logical form by existential entailments of the prejacent, quantifying over elements of the same type as the subject of its agreement. I argue that the construction by itself in English is similar: its alternatives are generated by existentially entailed eventualities, and quantification is restricted to the coreferent of its pronominal. Overall, this dissertation examines the connections between exclusive operators and a variety of linguistic phenomena, including scalarity, tense, causation, emphasis, reflexivity, coreference, syntactic structure, and morphological composition. Ultimately, I hope to present a unified account of alternatives in a way that both captures the similarity between the quantification by various particles and explains their variation
A Type-Based Approach To Adjectival Distribution
Get PDFMost English adjectives can appear either prenominally or as predicates. a. the red rose b. (1) Roses are red. There are also adjectives that can appear only in one position or the other, but not in both: (2) a. * the awake child b. (3) The child is awake. a. the former president b. * The president is former. The primary goal of this dissertation will be to address two questions that arise from this data. First, what is the relationship between prenominal red and predicative red? Second, what causes some adjectives, including awake and former, to be ungrammatical in prenominal and predicate positions, respectively? To answer these questions, I will appeal to the theory of semantic types and posit a type-shifting operator that is always present when an adjective appears in predicate position. The vast majority of English adjectives, including red, can either occur in their basic form prenominally or combine with the operator to produce viable predicates. Adjectives like awake cannot occur without the operator and as such are restricted to predicate positions. Adjectives like former can combine with the operator but the resulting predicates are ruled out by a constraint on trivial predication
How reasonable is "reasonable"? A study of U.S. jury instructions
No full textIn the U.S., juries composed of ordinary citizens make decisions that have profound implications for defendants’ assets, liberty, and even their right to live. To aid the decision process, the judge typically reads oral instructions to the jury. Unfortunately, these instructions abound with linguistic complexities that impede jurors’ understanding and can lead to inconsistent decisions (Charrow & Charrow 1979; Tiersma 1999, i.a.). This dissertation specifically addresses problems with the word “reasonable”. Numerous studies have grappled with “reasonable doubt” (Solan 1999; Dhami et al. 2015; i.a.). This study is unique because it analyzes “reasonable” in a broad range of contexts beyond “reasonable doubt”. To understand the importance of “reasonable" and how juries interpret the word, I utilized empirical, experimental, and theoretical approaches. I used Java to calculate the frequency of “reasonable” in eight codes of jury instructions. I searched through 62,000 lines of instructions and found 2,816 instances of “reasonable”. There are 82 different “reasonable” phrases in the civil instructions, and 42 in the criminal instructions -- suggesting that the interpretational problems span well beyond “reasonable doubt”. The high frequency of “reasonable” underscores the need for jurors to be able to evaluate the word more consistently. I also designed a survey using hypothetical cases and excerpts from actual jury instructions. I surveyed 65 Cornell undergraduates and 70 Cornell Law students. The findings indicated that there were demographic, contextual, and educational effects on participants’ interpretations -- evidence that “reasonable” does not lend itself to consistent interpretation. Finally, I offer a theoretical linguistics account, based on Kratzer 1981/2008. Using insights from this theoretical account and from a pilot study I ran prior to the experiment, I argue for the addition of scales for “reasonable” in jury instructions to help jurors evaluate reasonableness in a comparable way. My research recommends that jurors evaluate reasonableness along particular scales depending on the context: for example, “reasonable cost” might be assessed according to how typical the cost is, whereas “reasonable fear” might be assessed according to the plaintiff’s risk of injury
- …
