Generative Semantics set out to unearth the intricacies of paradigms by applying the same computational devices that helped generative grammar account for syntagmatic dependencies. The proposal failed on empirical grounds, as paradigmatic relations lack the productivity, transparency and systematicity of syntagmatic ones, which the computational approach is ideal at capturing. That said, Ray Jackendoff observed in the late eighties how one significant fact that the demised paradigm was designed to capture continued to be puzzling: logically possible though syntactically impossible gaps in lexical paradigms. The classic instance, due to Paul Postal, emphasizes the absence of verbs in any language whose meaning is, say, “meet x and relatives”, a construction whose description in Generative Semantics terms would violate the Coordinate Structure Constraint (other such impossible verbs can be constructed on the basis of any other syntactic principle). This note discusses how Jackendoff’s puzzle has not been resolved, and speculates that an Evo-Devo treatment of the puzzle might be possible if we think of the problem in two stages: at a language acquisition phase that lasts up to roughly puberty and at a consolidation phase that ensues thereafter.
This chapter pursues the issue of phonetically silent lexical items (cf. Van Riemsdijk 2002). Two more empirical domains are examined, and in both cases I tentatively conclude that a silent verb is involved. The first of these concerns the copula/auxiliary be when combined with a directional PP in Dutch and Swiss German. The second has to do with what I call silent particle verbs in Dutch. In a last section I discuss the more general issue of how widespread the phenomenon of silent lexical verbs is? And if the impression that this is a relatively rare phenomenon is correct, why do languages make such a limited use of what looks like a very economical device?
This chapter explores new evidence from lexical change in connection with the debate concerning the nature of the Lexicon: whether it is root-based or word-based. A principled distinction is drawn between (genuine) lexical change, which affects roots, and grammaticalization, which affects selected words in sentence contexts and involves copying the targeted words. The data is mostly drawn from various Arabic dialects with implications for other similar Semitic languages such as Hebrew, where roots are purely consonantal and where words consist of consonantal roots and vocalic melodies in a non-concatinative arrangement.
This chapter applies to Dylan Thomas’s metrical poetry the theory of meter developed in collaboration with Carlos Piera in Fabb and Halle (2008). We explain the various properties of the strict and loose meters used by Thomas. We also discuss some poems which can be analyzed as simultaneously in two meters at the same time: a strict syllable-counting meter and a loose stress-based meter. These polymetrical poems may reflect the influence, possibly mediated through Gerard Manley Hopkins, of the Welsh poetic tradition.
This paper attempts to explain the metrical practice and theorizing of the American poet William Carlos Williams. Observing that in many of his short poems, the total number of syllables is exactly, or nearly exactly, equal to the total number of positions in an abstract metrical representation containing a regular line and stanza structure, it is argued that Williams invented a remarkably successful new type of prosody, termed ‘syllabic-rhythmic’ verse, in which relative ‘speed’ or ‘tempo’ is computed by comparing the actual number of syllables in a line to the number of positions in the metrical schema. Williams’ later theory of the ‘variable foot’ and the ‘triadic line’ is further accounted for by incorporating the unit ‘foot’ in metrical schemata.
In this chapter, we discuss some recent results from studies of word order processing in Basque that show that despite the apparently unconstrained freedom the language displays in linearizing major phrasal constituents in the sentence, native speakers’ processing strategies reveal a clear advantage favoring the linearization that corresponds to minimal syntactic computation. This is reflected in the time speakers employ to read sentences, and also in the electrophysiological signals of neural processing measured by means of event related potentials (ERPs). Moreover, when confronted with chains that are syntactically ambiguous, speakers process them as if they were unambiguous, choosing the minimal syntactic computation (SOV) given their head final grammar. We also suggest that the claim that processing preferences universally favour subjects might need to be revisited under the light of ergativity. We suggest that accounts of processing mechanisms and strategies based on notions like “subject of” or “object of” as explanatory primitives are likely to fall short, and we argue that accounts that take into consideration the impact of morphological variability in processing yield a more accurate view of the interplay of variant and invariant processing mechanisms of language.
The present chapter proposes a novel account of the existence of fronted non-contrastive topics in German and its absence in Dutch. The main claim is that the relevant difference between the two languages reduces to the type of scrambling they have: only Dutch scrambling is triggered by an uninterpretable phi(person)-feature on v* (M. Richards 2008). The proposal, fully compatible with Chomsky’s (2000 and subsequent work) model of cyclic Spell-Out, is extended to cover the more restrictive pattern of non-contrastive topicalization in Swedish, absent in Danish and Norwegian.
Adjectives in many Indo-European languages morphologically agree with the nouns they modify in gender, number and case (if the language’s nouns can differ in case). These agreeing items seem to undermine several otherwise broad generalizations about morphology, word order and phrasal stress. Mysteriously, agreement in Germanic languages is limited to pre-nominal attributes, while all adjectives agree in Romance and Slavic languages. This essay proposes to analyze adjectival agreement in terms of a “Derived Nominal Hypothesis,” which assigns agreeing adjectives a word-internal nominal structure whose head is the agreement suffix itself. Consequently, these “adjectives” are actually Nouns (by Lieber’s Right Hand Head Rule), and so qualify as unexceptional heads of NPs. This supports Abney’s controversial conjecture for pre-nominal attributive adjectives (that they are heads of NPs). The Derived Nominal Hypothesis additionally succeeds in making several traditional observations on the behavior of agreeing adjectives fully compatible with current explanatory grammatical theory. It also accounts for many previously puzzling morphological properties of agreement and its syntactic distribution in those languages on which the study primarily focuses, namely Dutch, German, Latin and Czech. To a lesser extent, the essay touches on English adjectival word order and on common properties of Spanish and Latin agreement.
This chapter describes the syntactic behavior of constructions like the prospects are for peace. The construction is argued to be a sub-case of equational (or ‘specificational’) copular constructions. It is shown that the construction exhibits a definiteness effect on the subject of the copula and an indefiniteness effect on the complement of for, not shared by similar equational sentences. An explanation is provided that derives the descriptive pattern from general properties holding at the interface of the syntactic and semantic components in conjunction with local deletion operations operating at the interface with the phonological component.
In this paper I focus on the odd distribution of Spanish prepositional possessives inside nominals, and offer an explanation of those facts on the basis of a “Phrasal Spell-Out” approach to lexical insertion (cf. Caha 2009; Fábregas 2007a, 2007b, 2010). The core of the proposal is that certain contrasts much emphasized in previous accounts, like the one between weak and strong pronouns, are irrelevant for the problem at issue. Instead, I derive the distributional pattern of Spanish possessives by relying on feature contrasts like Oblique/Genitive, [±participant] and [±number], which provide the right natural classes for those elements.
Taking Folli and Harley’s (2007) analysis of Italian fare-causatives as a starting point, and focusing on Spanish, I examine variation in the distribution of the subject of the embedded infinitive in so-called faire-causatives, and I suggest that there is a robust correlation between the size of the embedded complement and the licensing of particular arguments. I reach this conclusion by investigating syntactic complexity in the domain of hacer-causatives, showing that richer structures obtain when Case factors associated with dative-case-marked arguments are considered. I further show that the specific conditions imposed by these arguments are language particular and arise in the language independently of analytical causatives.
Generative Semantics set out to unearth the intricacies of paradigms by applying the same computational devices that helped generative grammar account for syntagmatic dependencies. The proposal failed on empirical grounds, as paradigmatic relations lack the productivity, transparency and systematicity of syntagmatic ones, which the computational approach is ideal at capturing. That said, Ray Jackendoff observed in the late eighties how one significant fact that the demised paradigm was designed to capture continued to be puzzling: logically possible though syntactically impossible gaps in lexical paradigms. The classic instance, due to Paul Postal, emphasizes the absence of verbs in any language whose meaning is, say, “meet x and relatives”, a construction whose description in Generative Semantics terms would violate the Coordinate Structure Constraint (other such impossible verbs can be constructed on the basis of any other syntactic principle). This note discusses how Jackendoff’s puzzle has not been resolved, and speculates that an Evo-Devo treatment of the puzzle might be possible if we think of the problem in two stages: at a language acquisition phase that lasts up to roughly puberty and at a consolidation phase that ensues thereafter.
This chapter pursues the issue of phonetically silent lexical items (cf. Van Riemsdijk 2002). Two more empirical domains are examined, and in both cases I tentatively conclude that a silent verb is involved. The first of these concerns the copula/auxiliary be when combined with a directional PP in Dutch and Swiss German. The second has to do with what I call silent particle verbs in Dutch. In a last section I discuss the more general issue of how widespread the phenomenon of silent lexical verbs is? And if the impression that this is a relatively rare phenomenon is correct, why do languages make such a limited use of what looks like a very economical device?
This chapter explores new evidence from lexical change in connection with the debate concerning the nature of the Lexicon: whether it is root-based or word-based. A principled distinction is drawn between (genuine) lexical change, which affects roots, and grammaticalization, which affects selected words in sentence contexts and involves copying the targeted words. The data is mostly drawn from various Arabic dialects with implications for other similar Semitic languages such as Hebrew, where roots are purely consonantal and where words consist of consonantal roots and vocalic melodies in a non-concatinative arrangement.
This chapter applies to Dylan Thomas’s metrical poetry the theory of meter developed in collaboration with Carlos Piera in Fabb and Halle (2008). We explain the various properties of the strict and loose meters used by Thomas. We also discuss some poems which can be analyzed as simultaneously in two meters at the same time: a strict syllable-counting meter and a loose stress-based meter. These polymetrical poems may reflect the influence, possibly mediated through Gerard Manley Hopkins, of the Welsh poetic tradition.
This paper attempts to explain the metrical practice and theorizing of the American poet William Carlos Williams. Observing that in many of his short poems, the total number of syllables is exactly, or nearly exactly, equal to the total number of positions in an abstract metrical representation containing a regular line and stanza structure, it is argued that Williams invented a remarkably successful new type of prosody, termed ‘syllabic-rhythmic’ verse, in which relative ‘speed’ or ‘tempo’ is computed by comparing the actual number of syllables in a line to the number of positions in the metrical schema. Williams’ later theory of the ‘variable foot’ and the ‘triadic line’ is further accounted for by incorporating the unit ‘foot’ in metrical schemata.
In this chapter, we discuss some recent results from studies of word order processing in Basque that show that despite the apparently unconstrained freedom the language displays in linearizing major phrasal constituents in the sentence, native speakers’ processing strategies reveal a clear advantage favoring the linearization that corresponds to minimal syntactic computation. This is reflected in the time speakers employ to read sentences, and also in the electrophysiological signals of neural processing measured by means of event related potentials (ERPs). Moreover, when confronted with chains that are syntactically ambiguous, speakers process them as if they were unambiguous, choosing the minimal syntactic computation (SOV) given their head final grammar. We also suggest that the claim that processing preferences universally favour subjects might need to be revisited under the light of ergativity. We suggest that accounts of processing mechanisms and strategies based on notions like “subject of” or “object of” as explanatory primitives are likely to fall short, and we argue that accounts that take into consideration the impact of morphological variability in processing yield a more accurate view of the interplay of variant and invariant processing mechanisms of language.
The present chapter proposes a novel account of the existence of fronted non-contrastive topics in German and its absence in Dutch. The main claim is that the relevant difference between the two languages reduces to the type of scrambling they have: only Dutch scrambling is triggered by an uninterpretable phi(person)-feature on v* (M. Richards 2008). The proposal, fully compatible with Chomsky’s (2000 and subsequent work) model of cyclic Spell-Out, is extended to cover the more restrictive pattern of non-contrastive topicalization in Swedish, absent in Danish and Norwegian.
Adjectives in many Indo-European languages morphologically agree with the nouns they modify in gender, number and case (if the language’s nouns can differ in case). These agreeing items seem to undermine several otherwise broad generalizations about morphology, word order and phrasal stress. Mysteriously, agreement in Germanic languages is limited to pre-nominal attributes, while all adjectives agree in Romance and Slavic languages. This essay proposes to analyze adjectival agreement in terms of a “Derived Nominal Hypothesis,” which assigns agreeing adjectives a word-internal nominal structure whose head is the agreement suffix itself. Consequently, these “adjectives” are actually Nouns (by Lieber’s Right Hand Head Rule), and so qualify as unexceptional heads of NPs. This supports Abney’s controversial conjecture for pre-nominal attributive adjectives (that they are heads of NPs). The Derived Nominal Hypothesis additionally succeeds in making several traditional observations on the behavior of agreeing adjectives fully compatible with current explanatory grammatical theory. It also accounts for many previously puzzling morphological properties of agreement and its syntactic distribution in those languages on which the study primarily focuses, namely Dutch, German, Latin and Czech. To a lesser extent, the essay touches on English adjectival word order and on common properties of Spanish and Latin agreement.
This chapter describes the syntactic behavior of constructions like the prospects are for peace. The construction is argued to be a sub-case of equational (or ‘specificational’) copular constructions. It is shown that the construction exhibits a definiteness effect on the subject of the copula and an indefiniteness effect on the complement of for, not shared by similar equational sentences. An explanation is provided that derives the descriptive pattern from general properties holding at the interface of the syntactic and semantic components in conjunction with local deletion operations operating at the interface with the phonological component.
In this paper I focus on the odd distribution of Spanish prepositional possessives inside nominals, and offer an explanation of those facts on the basis of a “Phrasal Spell-Out” approach to lexical insertion (cf. Caha 2009; Fábregas 2007a, 2007b, 2010). The core of the proposal is that certain contrasts much emphasized in previous accounts, like the one between weak and strong pronouns, are irrelevant for the problem at issue. Instead, I derive the distributional pattern of Spanish possessives by relying on feature contrasts like Oblique/Genitive, [±participant] and [±number], which provide the right natural classes for those elements.
Taking Folli and Harley’s (2007) analysis of Italian fare-causatives as a starting point, and focusing on Spanish, I examine variation in the distribution of the subject of the embedded infinitive in so-called faire-causatives, and I suggest that there is a robust correlation between the size of the embedded complement and the licensing of particular arguments. I reach this conclusion by investigating syntactic complexity in the domain of hacer-causatives, showing that richer structures obtain when Case factors associated with dative-case-marked arguments are considered. I further show that the specific conditions imposed by these arguments are language particular and arise in the language independently of analytical causatives.