In this paper I propose a general approach to the study of constraints on cooccurrence of grammatical categories and present one case study of a functionally infelicitous combination from the domain of TAM categories, the present perfective. It is argued that constraints on co-occurrence of particular categories can be accounted for in terms of local markedness and markedness hierarchies. This approach lends itself naturally for formalization in Optimality Theoretic terms. It was further shown that both production optimization (OT syntax) and comprehension optimization (OT semantics) is needed to model syntagmatic interaction of grammatical categories.
I defend an aspectual operator approach of the perfective/imperfective distinction against a coercion approach, as, for example, proposed for French by de Swart (1998). I propose an analysis that follows de Swart on many points, but keeps temporal and aspectual contributions separate. I argue that such an analysis has a larger cross-linguistic coverage than one that combines the two in a single operator. The argumentation is based on the aspectual system of Ancient Greek, but holds for any language in which temporal and aspectual information are encoded in separate morphemes, and in which the opposition perfective/ imperfective is not restricted to the past tense. In addition, I show that a coercion analysis is problematic for French as well.
This paper deals with different components of aspectual interpretation in Adyghe, a polysynthetic North-West Caucasian language, and hierarchical relations among them. Following Tatevosov (2002), I propose a classification of Adyghe predicates into actional classes, and then show how this classification can account for the distribution of temporal adverbials. I argue that temporal adverbials in Adyghe are able to shift the lexically specified actional characteristic of the predicate (coercion in the sense of de Swart 1998) and thus constitute a separate level of aspectually relevant operators intermediate between lexical and grammatical aspect.
In this paper, we examine failed attempt and partial success interpretations of accomplishment verbs cross-linguistically. We observe that accomplishments differ systematically as to which of these readings they can produce. Relying on Rothstein’s (2004) theory of accomplishments, we propose that this diversity can be accounted for through properties of the relation between subevents in the accomplishment event structure.
This paper describes the grammatical uses of the Burmese motion verbs la ‘come’ and thwà ‘go’. The verbal markers -la ‘come’ and thwà ‘go’ fulfil different functionsaccording to (1) the semantics of the verb they modify, (2) the way in whichevents are represented in the clause/sentence, and (3) the context of occurrencein which -la ‘come’, and thwà ‘go’ appear. With motion verbs, they expressdirectionality of motion of the participants in the speech act. With non-motionverbs, they function as inchoative markers, i.e. as markers of a change in theentity’s state. In addition, thwà ‘go’ is used in clauses/sentences to mark the decreasedsalience of the Agent and the correspondent increased salience of the Undergoer.This function of thwà ‘go’ will be compared with the function of -lai ‘follow’, derived from the motion verb lai ‘follow’, which is used to mark a high degree oftransitivity of the clause.
The linguistic category of irrealis does not show stable semantics across languages. This makes it difficult to formulate general statements about this category, and it has led some researchers to reject irrealis as a cross-linguistically valid category. In this paper we look at the semantics of the irrealis category of Yurakaré, an unclassified language spoken in central Bolivia, and compare it to irrealis semantics of a number of other languages. Languages differ with respect to the subcategories they subsume under the heading of irrealis. The variable subcategories are future tense, imperatives, negatives, and habitual aspect. We argue that the cross-linguistic variation is not random, and can be stated in terms of an implicational scale.
In this paper, the selection of indicative or subjunctive for complement clauses in Romance languages is considered, the proposal being made that the selection of one or another mood is related to the kind of attitude expressed by the main predicate. Specifically, indicative is selected when the expressed attitude is one of knowledge or belief, otherwise subjunctive being selected. Hence, a relation is established between epistemic or doxastic modality and the selection of indicative, while subjunctive is not linked to any particular kind of modal value.
This paper provides a unified semantic analysis of the so-called ‘out-of-control’ circumfix ka-…-a in St’át’imcets (Lillooet Salish). ka-…-a expresses an initially puzzling range of meanings, including “be able to”, “manage to”, “suddenly”, “accidentally”, and “non-controllable”. We propose that ka-…-a encodes circumstantial modality; we show that its various meanings all reduce to either an existential (ability) or universal (involuntary action) interpretation. Our analysis provides further support for a striking difference between St’át’imcets and English. In English, modals lexically encode quantificational strength, but do not encode distinctions between epistemic, deontic and circumstantial interpretations. St’át’imcets modals display exactly the inverse pattern (Rullmann et al. 2008). In line with this, ka-…-a lexically encodes circumstantial modality, but does not encode quantificational strength. The parallel between ka-…-a and other St’át’imcets modal elements provides support for our analysis, in contrast to previous accounts (e.g., Demirdache 1997), which treat ka-…-a as primarily aspectual in nature.
This paper takes a closer look at the ‘geometric’ structure of the semantic map of modality (Van der Auwera & Plungian 1998). By analyzing the different modalities into more basic modal features, we can get a better view on how the map is organized along different dimensions around a neutral middle modality, how the deontic modality fits on the map and what role connectivity plays in defining polyfunctionality. Drawing on data from Dutch, we argue that a basic distinction on the map corresponds to the grammatical raising/control distinction.
The paper explores the fact that ‘get’ etymons may acquire modal meanings. It tries to fit this fact into the modal map proposal of van der Auwera & Plungian (1998) and concludes that the map has to be revised, in part because the lexical input, the predicates meaning ‘get/acquire’ allow more than one reading, an agentive and a receptive one. The paper focuses on the two areas in which so-called ‘acquisitiv modality’ is very prominent, viz. Northern Europe and South(east) Asia.
The Dutch modal auxiliaries kunnen ‘can’ and moeten ‘must’ can be interpreted in different ways: ‘participant-internal, ‘participant-external’, and epistemic. For each of the verbs, we assume a basic, default interpretation: ‘participant-internal’ for kunnen, ‘participant-external’ for moeten. In sentences with a ‘neutral’ main verb like zwemmen ‘to swim’, and with a third person subject, the hearer chooses this basic interpretation. We subsequently show that other elements in the sentence can induce a non-basic interpretation. In particular the controllability of the activity expressed in the main verb (‘to swim’ versus ‘to pee’), progressive aspect, and person of the subject (in particular second person subject) are relevant factors influencing the interpretation of the modal verb. We model the factors influencing the interpretation as violable constraints in an optimal theoretic analysis, which leads to tableaux with a sentence as input and an optimal interpretation of that sentence as output.
This paper argues that the analysis of modality in terms of generalized quantification falls short on three issues. First it is shown that such an analysis encounters serious problems when it come to deontic modality. Second I will show that the standard analysis makes false predictions by allowing unwanted combinations of modal items. Third I will argue that the data from Lillooet challenges the position that modality should be analyzed across the board as the interaction of a neutral operator with an externally provided intensional context. Finally I will sketch a solution to those problems within the framework of update semantics. I will propose a polysemous treatment of modality where each modality type (epistemic, deontic and circumstantial) has a distinct meaning.
The Russian Genitive of Negation construction involves alternations of genitive with nominative or accusative case under negation; typically a genitive NP is interpreted as “weaker” than a corresponding nominative or accusative, having narrow scope with respect to negation and as lacking any existence presupposition. A similar alternation is found with some intensional verbs, with genitive (sometimes) used for the “opaque” reading of the direct object. The similarity among these uses of genitive for ‘less referential’ objects was observed by Neidle (1982). One challenge for compositionality is the apparent nonuniformity of the semantics: the case alternations often have semantic correlates, but not always; when they do, the semantic correlates are not self-evidently the same in each case. We believe that a solution might be approached by combining the Russian “verb-centered” view of the phenomena with the western “compositionality-centered” view. As a prolegomenon to a fuller study of shifts in semantics and in fine-grained argument structure of verbs under negation and under the influence of intensionality, modality, and imperfective aspect, in this paper we examine the relationships between negation and intensionality and between partitivity and imperfectivity.
This article shows that the Estonian partitive evidential marks predicates in sentences that express incomplete evidence. Partitive occurs in the categories of aspect, epistemic modality, and evidentiality, marking objects and present participles. Despite the difference in syntax, the semantics of these categories is based on parallel relationships. More specifically, the aspectual partitive marks objects in sentences describing incomplete events, and the partitive evidential appears in sentences that encode incomplete evidence compared to the expectation of complete evidence.
In this paper I propose a general approach to the study of constraints on cooccurrence of grammatical categories and present one case study of a functionally infelicitous combination from the domain of TAM categories, the present perfective. It is argued that constraints on co-occurrence of particular categories can be accounted for in terms of local markedness and markedness hierarchies. This approach lends itself naturally for formalization in Optimality Theoretic terms. It was further shown that both production optimization (OT syntax) and comprehension optimization (OT semantics) is needed to model syntagmatic interaction of grammatical categories.
I defend an aspectual operator approach of the perfective/imperfective distinction against a coercion approach, as, for example, proposed for French by de Swart (1998). I propose an analysis that follows de Swart on many points, but keeps temporal and aspectual contributions separate. I argue that such an analysis has a larger cross-linguistic coverage than one that combines the two in a single operator. The argumentation is based on the aspectual system of Ancient Greek, but holds for any language in which temporal and aspectual information are encoded in separate morphemes, and in which the opposition perfective/ imperfective is not restricted to the past tense. In addition, I show that a coercion analysis is problematic for French as well.
This paper deals with different components of aspectual interpretation in Adyghe, a polysynthetic North-West Caucasian language, and hierarchical relations among them. Following Tatevosov (2002), I propose a classification of Adyghe predicates into actional classes, and then show how this classification can account for the distribution of temporal adverbials. I argue that temporal adverbials in Adyghe are able to shift the lexically specified actional characteristic of the predicate (coercion in the sense of de Swart 1998) and thus constitute a separate level of aspectually relevant operators intermediate between lexical and grammatical aspect.
In this paper, we examine failed attempt and partial success interpretations of accomplishment verbs cross-linguistically. We observe that accomplishments differ systematically as to which of these readings they can produce. Relying on Rothstein’s (2004) theory of accomplishments, we propose that this diversity can be accounted for through properties of the relation between subevents in the accomplishment event structure.
This paper describes the grammatical uses of the Burmese motion verbs la ‘come’ and thwà ‘go’. The verbal markers -la ‘come’ and thwà ‘go’ fulfil different functionsaccording to (1) the semantics of the verb they modify, (2) the way in whichevents are represented in the clause/sentence, and (3) the context of occurrencein which -la ‘come’, and thwà ‘go’ appear. With motion verbs, they expressdirectionality of motion of the participants in the speech act. With non-motionverbs, they function as inchoative markers, i.e. as markers of a change in theentity’s state. In addition, thwà ‘go’ is used in clauses/sentences to mark the decreasedsalience of the Agent and the correspondent increased salience of the Undergoer.This function of thwà ‘go’ will be compared with the function of -lai ‘follow’, derived from the motion verb lai ‘follow’, which is used to mark a high degree oftransitivity of the clause.
The linguistic category of irrealis does not show stable semantics across languages. This makes it difficult to formulate general statements about this category, and it has led some researchers to reject irrealis as a cross-linguistically valid category. In this paper we look at the semantics of the irrealis category of Yurakaré, an unclassified language spoken in central Bolivia, and compare it to irrealis semantics of a number of other languages. Languages differ with respect to the subcategories they subsume under the heading of irrealis. The variable subcategories are future tense, imperatives, negatives, and habitual aspect. We argue that the cross-linguistic variation is not random, and can be stated in terms of an implicational scale.
In this paper, the selection of indicative or subjunctive for complement clauses in Romance languages is considered, the proposal being made that the selection of one or another mood is related to the kind of attitude expressed by the main predicate. Specifically, indicative is selected when the expressed attitude is one of knowledge or belief, otherwise subjunctive being selected. Hence, a relation is established between epistemic or doxastic modality and the selection of indicative, while subjunctive is not linked to any particular kind of modal value.
This paper provides a unified semantic analysis of the so-called ‘out-of-control’ circumfix ka-…-a in St’át’imcets (Lillooet Salish). ka-…-a expresses an initially puzzling range of meanings, including “be able to”, “manage to”, “suddenly”, “accidentally”, and “non-controllable”. We propose that ka-…-a encodes circumstantial modality; we show that its various meanings all reduce to either an existential (ability) or universal (involuntary action) interpretation. Our analysis provides further support for a striking difference between St’át’imcets and English. In English, modals lexically encode quantificational strength, but do not encode distinctions between epistemic, deontic and circumstantial interpretations. St’át’imcets modals display exactly the inverse pattern (Rullmann et al. 2008). In line with this, ka-…-a lexically encodes circumstantial modality, but does not encode quantificational strength. The parallel between ka-…-a and other St’át’imcets modal elements provides support for our analysis, in contrast to previous accounts (e.g., Demirdache 1997), which treat ka-…-a as primarily aspectual in nature.
This paper takes a closer look at the ‘geometric’ structure of the semantic map of modality (Van der Auwera & Plungian 1998). By analyzing the different modalities into more basic modal features, we can get a better view on how the map is organized along different dimensions around a neutral middle modality, how the deontic modality fits on the map and what role connectivity plays in defining polyfunctionality. Drawing on data from Dutch, we argue that a basic distinction on the map corresponds to the grammatical raising/control distinction.
The paper explores the fact that ‘get’ etymons may acquire modal meanings. It tries to fit this fact into the modal map proposal of van der Auwera & Plungian (1998) and concludes that the map has to be revised, in part because the lexical input, the predicates meaning ‘get/acquire’ allow more than one reading, an agentive and a receptive one. The paper focuses on the two areas in which so-called ‘acquisitiv modality’ is very prominent, viz. Northern Europe and South(east) Asia.
The Dutch modal auxiliaries kunnen ‘can’ and moeten ‘must’ can be interpreted in different ways: ‘participant-internal, ‘participant-external’, and epistemic. For each of the verbs, we assume a basic, default interpretation: ‘participant-internal’ for kunnen, ‘participant-external’ for moeten. In sentences with a ‘neutral’ main verb like zwemmen ‘to swim’, and with a third person subject, the hearer chooses this basic interpretation. We subsequently show that other elements in the sentence can induce a non-basic interpretation. In particular the controllability of the activity expressed in the main verb (‘to swim’ versus ‘to pee’), progressive aspect, and person of the subject (in particular second person subject) are relevant factors influencing the interpretation of the modal verb. We model the factors influencing the interpretation as violable constraints in an optimal theoretic analysis, which leads to tableaux with a sentence as input and an optimal interpretation of that sentence as output.
This paper argues that the analysis of modality in terms of generalized quantification falls short on three issues. First it is shown that such an analysis encounters serious problems when it come to deontic modality. Second I will show that the standard analysis makes false predictions by allowing unwanted combinations of modal items. Third I will argue that the data from Lillooet challenges the position that modality should be analyzed across the board as the interaction of a neutral operator with an externally provided intensional context. Finally I will sketch a solution to those problems within the framework of update semantics. I will propose a polysemous treatment of modality where each modality type (epistemic, deontic and circumstantial) has a distinct meaning.
The Russian Genitive of Negation construction involves alternations of genitive with nominative or accusative case under negation; typically a genitive NP is interpreted as “weaker” than a corresponding nominative or accusative, having narrow scope with respect to negation and as lacking any existence presupposition. A similar alternation is found with some intensional verbs, with genitive (sometimes) used for the “opaque” reading of the direct object. The similarity among these uses of genitive for ‘less referential’ objects was observed by Neidle (1982). One challenge for compositionality is the apparent nonuniformity of the semantics: the case alternations often have semantic correlates, but not always; when they do, the semantic correlates are not self-evidently the same in each case. We believe that a solution might be approached by combining the Russian “verb-centered” view of the phenomena with the western “compositionality-centered” view. As a prolegomenon to a fuller study of shifts in semantics and in fine-grained argument structure of verbs under negation and under the influence of intensionality, modality, and imperfective aspect, in this paper we examine the relationships between negation and intensionality and between partitivity and imperfectivity.
This article shows that the Estonian partitive evidential marks predicates in sentences that express incomplete evidence. Partitive occurs in the categories of aspect, epistemic modality, and evidentiality, marking objects and present participles. Despite the difference in syntax, the semantics of these categories is based on parallel relationships. More specifically, the aspectual partitive marks objects in sentences describing incomplete events, and the partitive evidential appears in sentences that encode incomplete evidence compared to the expectation of complete evidence.