Galit Weidman Sassoon

​ Questions and their answers have been discussed in length over the past few years. In this paper we present and analyze a Hebrew hedger, ​ be-gadol ​ , roughly translated as ​ basically. ​ We use the literature on questions, answers and the relation between them to suggest that ​ be-gadol ​ is an item which conveys a restriction on the context of utterance. This restriction concerns the relation between answers to the QUD on an answerhood scale, which is characterized as involving two notions, informativity (Roberts 1996) and resolution, defined using tools from decision theories (Ginzburg 1995; van Rooij 2003). This significantly supports the linguistic reality of these notions.

There are two fundamentally different kinds of comparison: difference comparisons and contrast comparisons. Unlike adjective phrases, noun phrases can occur in contrast comparisons (such as This bird is more a duck than a goose), but not in difference comparisons (#This bird is more a duck than that one is), where the mediation of a partitive particle is necessary (as in more of a duck). The problem is that postulating either semantic gradability or even just ad-hoc, metalinguistic, gradable interpretations for nouns in order to capture the meaning of contrast comparisons results in wrong predictions for difference comparisons and for most other gradable constructions (#very duck, #too duck, #duck enough, #the most duck). This article presents an account that exploits the psychological notion of a contrast set to explain these data and to correctly predict the truth conditions and characteristic inference patterns of contrast comparisons. Two main conclusions are, first, that if adjectives are degree expressions, so are nouns, and second, that nouns form a different type of degree expression.

The degree approach is one of the leading frameworks for the analysis of gradability in language, providing fully compositional accounts for constructions with gradable predicates of different categories. However, in this framework, the interpretation of a gradable predicate is almost always modeled in terms of a single scalar dimension (degree relation). This paper presents a compositional account which models multidimensional adjectives and nouns in terms of sets of degree relations, while keeping the standard assumptions and analyses of degree morphemes intact. Degree constructions are proposed to involve counting of or quantification on dimensions.
Based on cognitive psychological research, nouns are assumed to be associated with dimension sets, like adjectives do. The approach correlates the acceptability of a given noun in degree constructions and the default way in which its dimensions combine into a single interpretation. Nouns at which the dimensions can combine via counting operations or quantifiers are predicted to be more acceptable in many degree constructions than nouns at which the dimensions combine via other operations. The proposal is motivated by introspective data and experimental evidence, which is briefly reviewed. Its broad consequences are discussed and directions for future research are proposed.

(From The proceedings of IATL 18)

In model theoretic semantics, we represent the core of predicate-sense by intension. Another notion, clusters of characteristic properties, serving as conceptual guidelines that help us identify the denotation in each context of use, has intrigued scholars from a variety of disciplines for years. Model theoretic theories often appeal to clusters of properties / features /criteria / propositions etc. However, there is no systematic account of the role of clusters. Stipulations are made in each case separately regarding the presence of clusters and/or their various effects. I propose a detailed formal model, which incorporates two kinds of clusters into our denotational representation of Predicate meaning. For example, the interpretation of the predicate chair includes in each context a set of individuals (chairs), necessary properties (like piece of furniture or solid) and stereotypical characteristics of chairs (like: has a back, four legs, is used to sit on it etc.) In this paper, I illustrate the use of this model in semantic analysis with one case study: contextual restrictions in universal generalizations with every, any and generic a.

(From proceedings of Israel Association of Theoretical Linguistics 21)

Parts 1-3 present and criticize Partee and Kamp’s 1995 well known analysis of the typicality effects. The main virtue of this analysis is in the use of supermodels, rather than fuzzy models, in order to represent vagueness in predicate meaning. The main problem is that typicality of an item in a predicate is represented by a value assigned by a measure function, indicating the proportion of supervaluations in which the item falls under the predicate. A number of issues cannot be correctly represented by the measure function, including the typicality effects in sharp predicates; the conjunction fallacy; the context dependency of the typicality effects etc. In Parts 4-5, it is argued that these classical problems are solved if the typicality ordering is taken to be the order in which entities are learnt to be denotation members (or non-members) through contexts and their extensions. A modified formal model is presented, which clarifies the connections between the typicality effects, predicate meaning, and its acquisition.

(From the proceeding of Sinn und Bedeutung 13)

This paper presents a novel semantic analysis of unit names and gradable adjectives, inspired by measurement theory (Krantz et al 1971). Based on measurement theory's typology of measures, I claim that different predicates are associated with different types of measures whose special characteristics, together with features of the relations denoted by unit names, explain the puzzling limited distribution of measure phrases.

(From proceedings of Israel Association of Theoretical Linguistics (IATL) 24)
What should an adequate representation of individuals (elements of the domain of discourse) be like, within vagueness models with degrees? This paper explores the hypothesis that individuals are distinguished by their property values, i.e. the extents to which they satisfy gradable properties. First, as Lewis (1986) argues, cross-world identity is intuitively implausible between individuals differing in their property values, e.g., their height, weight, etc. ('intrinsic properties'). However, cross-world identity is intuitively plausible between individuals sharing the same property values (the same heights, weights, etc.) , even if they differ along extrinsic, relational properties, e.g., if they are considered 'tall' in one world, but not in another, due to variance in the cutoff point of tall across the two worlds. A representation of individuals by their property values captures the intuitively sharp distinction between these two cases. Second, this proposal captures the intuitive difference between cases we tend to call 'ignorance' and cases we tend to call 'vagueness' (for Williamson 1994, cases of 'accidental' versus cases of 'inherent' ignorance), which are usually modeled with the same formal means. While vagueness/inherent ignorance (about the truth value of statements like Dan is tall) arises due to partial information regarding cutoff points of vague predicates like tall, accidental ignorance (say, about the truth value of statements like Dan is two meters tall or Dan is taller than Sam) arises due to partial information regarding property values (e.g., the height) of referents of arguments like Dan ('discourse entities'). A representation of individuals by their property values, then, captures the intuitive distinctions between both phenomena.

(Synthese (2010) 174:151–180 DOI 10.1007/s11229-009-9687-5)
This paper presents a novel semantic analysis of unit names (like pound and meter) and gradable adjectives (like tall, short and happy), inspired by measurement theory (Krantz et al. In Foundations of measurement: Additive and Polynomial Representations, 1971). Based on measurement theory’s four-way typology of measures, I claim that different adjectives are associated with different types of measures whose special characteristics, together with features of the relations denoted by unit names, explain the puzzling limited distribution of measure phrases, as well as unit-based comparisons between predicates (as in the table is longer than it is wide). All considered, my analyses support the view that the grammar of natural languages is sensitive to features of measurement theory.

(Proceedings of the 17th Amsterdam colloquium conference on Logic, language and meaning)
This paper provides an analysis of statements with predicates of personal taste (tasty, fun, etc.) Rather than directly relativizing semantic interpretation to a judge (cf., Lasersohn, 2005), this paper aims to capture the phenomenon called ‘faultless disagreement’ (the fact that one can deny a speaker’s subjective utterance without challenging the speaker‘s opinion) by means of pragmatic restrictions on quantification domains. Using vagueness models, a statement like the cake is tasty is analyzed as true in a partial context c iff it is true in the set of completions t consistent with c (Kamp, 1975), wherein tasty denotes different, contextually possible, taste measures (Kennedy, 1999). Phrases like for me restrict the set of completions to those with taste measures consistent with the speaker’s taste. Faultless disagreement naturally follows assuming speakers accommodate or reject implicit restrictions of this sort (Lewis, 1979).

(Natural Language Semantics 06/2010; 18(2):141-181. DOI:10.1007/s11050-009-9052-8 · 0.92 Impact Factor)
This paper provides a new account of positive versus negative antonyms. The data includes well known linguistic generalizations regarding negative adjectives, such as their incompatibility with measure-phrases (cf., two meters tall/ *short) and ratio-phrases (twice as tall/ #short), as well as the impossibility of truly cross-polar comparisons (*Dan is taller than Sam is short). These generalizations admit a variety of exceptions, e.g., positive adjectives that do not license measure phrases (cf., #two degrees warm/ cold), and rarely also negative adjectives that do (cf., two hours late/ early). Furthermore, new corpus data is presented, regarding the use of twice with positive and negative adjectives. The analysis the paper presents supposes that grammar associates gradable adjectives with measure functions – mapping of entities to a set of degrees, isomorphic to the real numbers (Kennedy 1999). On this analysis, negative adjectives map entities to values that are linearly reversed and linearly transformed in comparison with their values in the positive antonyms. As shown, the generalizations, as well as their exceptions, directly follow. Negative polarity is explained in terms of function-reversal, and non-licensing of measure-phrases is explained in terms of transformation by an unspecified value.

(Belgian Journal of Linguistics 12/2010; DOI:10.1075/bjl.25.06sas)
Classification of entities into categories can be determined based on a rule – a single criterion or relatively few criteria combined with logical operations like ‘and’ or ‘or’. Alternatively, classification can be based on similarity to prototypi- cal examples, i.e. an overall degree of match to prototypical values on multiple dimensions. Two cognitive systems are reported in the literature to underlie processing by rules vs. similarity. This paper presents a novel thesis according to which adjectives and nouns trigger processing by the rule vs. similarity systems, respectively. The paper defends the thesis that nouns are conceptually gradable and mul- tidimensional, but, unlike adjectives, their dimensions are integrated through similarity operations, like weighted sums, to yield an overall degree of match to ideal values on multiple dimensions. By contrast, adjectives are associated with single dimensions, or several dimensions bound by logical operations, such as ‘and’ and ‘or’. In accordance, nouns are predicted to differ from adjectives semantically, developmentally, and processing-wise. Similarity-based dimen- sion integration is implicit – processing is automatic, fast, and beyond speaker awareness – whereas logical, rule-based dimension integration is explicit, and is acquired late. The paper highlights a number of links between findings reported in the literature about rule- vs. similarity-based categorization and corresponding structural, distributional, neural and developmental findings about adjectives and nouns. These links suggest that the rule vs. similarity (RS) hypothesis for the adjective-noun distinction should be studied more directly in the future.

(From proccedings of Israel Association of Theoretical Linguistics (IATL) 29)

Dutch-speaking adults and children were tested to find out about their criteria to apply the relative adjective groot (En.: big, large). The hypothesis that leaps in the distribution trigger cutoff points between the items that will and will not be called groot was confirmed for the adults, but not for the children

(Utterance Interpretation and Cognitive models, Edited by Kissine Mikhail; Philippe De Brabanter, 01/2009: pages 127-178; Emerald publishing)

In model-theoretic semantics, we represent the core of predicate meaning by intension. Another notion, clusters of characteristic properties ('dimensions'), serving as conceptual guidelines that help us identify the denotation in each context of use, has intrigued scholars from a variety of disciplines for years. Typical examples are the prototype and exemplar theories, which are prevalent in the study of concepts in cognitive psychology. Model-theoretic theories often appeal to clusters of properties / features / criteria / propositions etc. However, there is no systematic account of the role of clusters. Stipulations are made in each case separately regarding the presence of clusters and/or their various effects. I propose a detailed formal model, which incorporates two kinds of clusters into our denotational representation of predicate meaning. For example, the interpretation of the predicate chair includes in each context in this model a set of individuals (chairs), necessary properties (like piece of furniture) and typical characteristics of chairs (like: has a back, four legs, etc.) I illustrate the use of this model in semantics with one case study: contextual restrictions in universal generalizations with every, any and generic a. I propose that these quantifying expressions access the cluster of their first argument and use it in different systematic ways (determined by their semantics) to construct their domain. In virtue of this feature, the present proposal captures the similarities, as well as the precise differences between the interpretations of statement with every, any and generic a (per a given predicate as their first argument).

This paper presents corpus-based evidence for a typology of multidimensional adjectives, like for example, healthy and sick. The interpretation of the latter is sensitive to multiple dimensions, such as blood pressure, pulse, sugar, cancer, etc. The study investigated the frequency of exception phrases, which operate on an implicit universal quantifier over adjectival dimensions, as in healthy, except for a slight cold, and not sick, except for high cholesterol. On the emerging typology, adjectives classify by the way their dimensions are glued together to create a single, uniform interpretation. The default interpretation of adjectives such as healthy involves implicit universal quantification over dimensions (dimension conjunction), while that of adjectives such as sick involves existential quantification (dimension disjunction). In adjectives like intelligent, the force of quantification over dimensions is context relative. Moreover, the paper presents support to the hypotheses that antonym polarity and modifier distribution guide our choice of quantifiers over dimensions in different adjectives. Thus, this research sheds new light on the nature of negative antonymy in multidimensional adjectives, and on the distribution of degree modifiers and exception phrases among multidimensional antonyms. Finally, it raises new questions pertaining to multidimensional comparisons.

Are complex predicates – in particular, negated (e.g., not expensive), conjunctive (e.g., expensive and time consuming) and disjunctive predicates (e.g., tall or bald) – associated with a graded structure, namely a mapping of entities to degrees? On the one hand, most up to date semantic theories of gradability and comparison in natural language disregard this question. On the other hand, contemporary fuzzy logical theories provide compositional rules to construct a degree function for a complex expression based on the degree functions of its constituents. These composition rules have been found useful for a variety of practical applications. The question is then whether these rules can correctly represent the interpretation of complex natural language expressions and its relation to the interpretation of their constituents. The relevance of this question is enhanced by recent findings from a variety of studies (Ripley, 2010; Serchuk et al., 2010; Alxatib and Pelletier 2010), according to which high percentages of subjects count contradictory predicates such as tall and not tall as true of borderline cases (neither short nor tall entities). While these findings stand in sharp contrast to predictions of vagueness-based theories of adjectives, they are in accord with the predictions of a fuzzy analysis, as extensively argued by Kamp and Partee (1995). Given these new findings, then, the fact that fuzzy analyses allow for non-zero truth values to contradictions can no longer count against them (for a more detailed discussion see Sauerland, this volume). It is therefore increasingly important to test other predictions of applications of fuzzy analyses to natural language conjunctions and disjunctions. To this end, this paper discusses preliminary results based on a questionnaire eliciting judgments from 35 Hebrew speakers. The results suggest that, counter the predictions of fuzzy analyses, comparative and equative morphemes cannot apply to conjunctions and disjunctions of gradable adjectives.

(From proceedings of Amsterdam Colloquium 19)

A compositional analysis of a range of readings of comparison constructions, as well as the positive form is proposed, which, unlike previous accounts, is compatible with multidimensional adjectives and has the power to explain differences between them and nouns. To this end, adjectives are represented as properties of dimensional quantifiers, namely of sets of gradable properties; e.g., healthy  λGQ. n-many(λF.F is a health dimension, λF.GQ(F)), where many denotes a cardinality function and n sets up a standard. Comparison morphemes either set the standard of many or of the dimensions. Consequences are discussed for our understanding of the adjective-noun distinction and for the analysis of gradable morphology.

(From Proceedings of SALT 21)
This paper investigates core semantic properties that distinguish between different types of gradable adjectives and the effect of context on their interpretation. We contend that all gradable adjectives are interpreted relative to a comparison class (van Rooij 2011), and that it is the nature of the comparison class that constitutes the main semantic difference between their subclasses: some adjectives select a class comprised of counterparts of the individual of which the adjective is predi- cated, while others select an extensional category of this individual. We propose, following Kennedy (2007), that the standard of membership is selected according to a principle of economy whereby an interpretation relative to a maximum or a minimum degree within a comparison class takes precedence over one relative to an arbitrary point. This proposal captures so-called “standard shift” effects, that is, the influence of context on the interpretation of gradable adjectives from all subclasses, whether in their positive form or when modified by degree adverbials. Additionally, this proposal captures cases of apparent lack of context sensitivity (e.g. intuitive inference patterns, unacceptability of for-phrases, etc.). Finally, we hypothesize that the type of comparison class is aligned with the well known distinction between stage-level and individual-level predicates.

(From the proceedings of Sinn und Bedeutung 15)
Negative degree questions such as How short is John?, and negative equatives such as John is as short as Mary imply that John is short. Existing theories explain this ‘norm-related’ implication by means of (i) a competition between unmarked and marked antonyms, and (ii) by introducing a standard- variable in parallel with the standard analysis of the positive form John is tall. This paper argues against these principles and in favor of an analysis, whereby the zero on the measurement scales of norm-related adjectives is relative, rather than absolute. This principle captures the fact that norm-related implications arise with many positive adjectives (alongside with their negative antonyms) and are cross-linguistically tied with non-licensing of measure phrases.

This paper investigates core semantic properties that distinguish between different types of gradable adjectives and the effect of context on their interpretation. We contend that all gradable adjectives are interpreted relative to a comparison class (van Rooij to appear), and that it is the nature of the comparison class that constitutes the main semantic difference between the different subclasses of such adjectives: some select a class comprised of counterparts of the individual of which the adjective is predicated, while others - an extensional-category of this individual. The role of the context is to determine the elements that comprise the comparison class. It is proposed, following Kennedy (2007), that the standard of membership is selected according to a principle of economy whereby an interpretation relative to a maximum or a minimum endpoint within a comparison class takes precedence over one relative to an arbitrary point. This proposal captures so-called “standard shift” effects, that is, the influence of context on the interpretation of gradable adjectives from all subclasses, in their positive form and when modified by degree adverbials. Additionally, this proposal captures cases of apparent lack of context sensitivity (e.g. intuitive inference patterns, unacceptability of for-phrases, etc.) Finally, we show that the type of comparison class is aligned with the well known distinction between stage-level and individual-level predicates.

Adjectives are typically felicitous in within-predicate comparisons—constructions of the form ‘X  is  more  A  than  y’, as in This is bigger than that, but are often infelicitous in between-predicate comparisons—constructions  of  the  form  ‘X  is  more  A  than  (y  is)  B’,  as  in  *Tweety is bigger than (it is) heavy. Nouns, by contrast, exhibit the inverse pattern. The challenge is to account for their felicity in between-predicate comparisons, as, for instance, in This bird is more a duck than a goose, while capturing their infelicity in within-predicate comparisons, as in #This bird is more (of) a duck than that one. Postulating either semantic gradability, or even only ad-hoc, meta- linguistic gradable interpretations for noun to capture the meaning of between-noun comparisons results in wrong predictions for within-noun comparisons and other gradable constructions (#very duck; too duck). The paper presents a solution to this problem, using the psychological notion of a contrast-set. The solution correctly predicts inference patterns and truth value judgments pertaining to between noun constructions.

(BRILL., ISBN: 9789004248588)
This book presents a study of the connections between vagueness and gradability, and their different manifestations in adjectives (morphological gradability effects) and nouns (typicality effects). It addresses two opposing theoretical approaches from within formal semantics and cognitive psychology. These approaches rest on different, apparently contradictory pieces of data. For example, for psychologists nouns are linked with vague and gradable concepts, while for linguists they rarely are. This difference in approach has created an unfortunate gap between the semantic and psychological studies of the concepts denoted by nouns, as well as adjectives. The volume describes a wide range of relevant facts and theories. Psychological notions such as prototypes and dimensions are addressed with formal rigor and explicitness. Existing formal semantic accounts are examined against empirically established cognitive data. The result is a comprehensive unified approach. The book will be of interest to students and researchers working on the semantics and pragmatics of natural languages and their cognitive basis, the psychology of concepts, and the philosophy of language.

(LOFT 6, 2004)

Typicality is extensively investigated in a variety of disciplines (Philosophy, Psychology, Linguistics, AI, etc.) However, despite the wide range of evidence for the existence of typicality effects, it is still not clear what the correct account for them is and how they relate to predicate meaning. I will present here Partee & Kamp's 1995 influential model theoretic analysis of typicality, which is (like most accounts of typicality) formulated in terms of predicate prototypes and degree functions. I will point out several problems with this kind of formulation (such as wrong predictions, compositionality problems, the complicated taxonomy of predicate types assumed by the theory etc.) Finally, I will attempt to show that they are all solved, if a typicality ordering of a predicate P in a context c (≤P) is taken to reflect gradual addition of entities into the set of contextually relevant Ps, [P]+c (and non-Ps, [P]-c), as information accumulates.

Review at the linguist list of the book:
EDITORS: Aarts, Bas; Denison, David; Keizer, Evelien; Popova, Gergana
TITLE: Fuzzy Grammar, a reader
PUBLISHER: Oxford University Press 
YEAR: 2004
Announced at http://linguistlist.org/issues/15/15-231.html

Multidimensionality in the grammar of gradability Galit Weidman Sassoon (20.7K words) In standard approaches to gradability, the interpretation of a gradable predicate is typically assumed to be generated by a single scalar dimension (Kennedy 1999). However, people seem to use multiple dimensions in a variety of cognitive tasks. This paper addresses the grammatical manifestations of this phenomenon. It presents data to the effect that dimension-sets and dimension-binding operations in the lexical representations of predicates affect the grammatical status, meaning composition, and truth conditions of utterances. The account is motivated by introspective data and a review of corpus patterns and experimental results. Multidimensional adjectives are quantificational; e,g,, to be healthy, one must be healthy in all respects (have no relevant disease). Given this meaning, questions arise as to what the interpretation of very healthy or healthier could be, and how it may be composed. The paper shows that degree-constructions with multidimensional adjectives involve counting of or quantification over dimensions. Data from the nominal domain further motivates this view. Based on psychological research, nouns are assumed to be associated with dimension-sets, like adjectives are. It is shown that when the semantic representations of nouns involve counting of or quantification over dimensions, their acceptability in degree-constructions improves. Moreover, the acceptability of nominal comparisons correlates robustly with that of constructions involving explicit quantification over dimensions. The paper presents a fully compositional account of multidimensional gradability using the standard semantic approach to gradability (Heim 2000; von Stechow 2009) plus a new respect-accessing operator.

This paper presents a new approach to morphological gradability, which extends existing ones to cover manifestations of gradability in nouns and multidimensional adjectives. The approach correlates between the type of categorization criterion characteristic of a given noun or adjective and its acceptability with gradable morphemes like more. A preliminary experimental study is reported that confirms the predicted correlation, and concrete directions for future research are proposed.
[To appear in: Yoad Winter and James Hampton (Eds.) Concept composition and experimental semantics & pragmatics. Springer]