Colloquium and Events Schedule (2014-2015)
All colloquium talks are in B342 Wells Hall at 4:30pm unless otherwise noted.
Past Talks (2014-2015)
The relation between production and perception in individual language users: Afrikaans devoicing as a case study
Andries Coetzee and Patrice Beddor (University of Michigan)
Date: April 23rd
The relation between speech perception and speech production is a foundational issue in the study of the phonetic bases of sound change. Contemporary experimental approaches to sound change investigate how the phonetic variants in the ambient language might serve as a source of new sound patterns that spread through a speech community. In these approaches, when the listeners' systematic perceptual biases are argued to be the source of the change, it is (often tacitly) assumed that listeners-turned-speakers will manifest those biases in their own productions. But how closely yoked are production and perception? Our current research addresses this broad issue through the more specific question of whether language users who produce more innovative variants also weight the innovative property more heavily in perception.
As a case study, we report production and perception data for older and younger speakers of Afrikaans, which is undergoing a change by which an earlier voicing contrast for obstruents is in the process of being replaced by a tonal contrast. The production measures show that devoicing (production of /b d/ as [p t]) is more common for younger speakers than for older speakers. However, there is not an age difference for production of post-stop fundamental frequency (f0): for all speakers, regardless of phonetic voicing, vocalic f0 is systematically higher following phonologically voiced /b d/ than phonologically voiceless /p t/. The perception findings for these same participants parallel their production data: although all listeners use both f0 and voicing to perceptually differentiate /b d/ from /p t/, older participants rely more than younger ones on voicing cues. In addition, production and perception data for individual participants are correlated: participants who produce more voicing also weight voicing more heavily in perception. We will interpret these findings both with respect to the specific change of emergent tonogenesis in Afrikaans and a more general consideration of the contributions of innovative listeners-turned-speakers to sound change.
MSULC Keynote Talk: What's the Difference between a Spartan and a Wolverine? An Interdisciplinary Study of Phonetic Category Learning
Chris Heffner (University of Maryland)
Date: April 17th, 2:15pm
Where: B342 Wells Hall
Models of category learning in phonetics can generally be broken into two different types. Under rule-based models, categories are the result of a list of rules along the lines of a checklist. Under exemplar-based models, meanwhile, categories are the sum total of the experiences with the members of a certain category. In this study, listeners heard phonetic tokens on a continuum of palatal to velar fricatives, which were paired with one of either two or three categories. The pairings of continuum members to categories varied from condition to condition. In the end, though listeners could learn certain pairings of phonetic tokens to categories, they found learning several of the mappings from speech sounds to categories challenging. Their performance indicates that they may instead be imposing rule-like structure on the disorderly categories they attempted to learn. This indicates that listeners may come to the task of categorizing phonetic tokens with strong expectations for how the sounds should be categorized. This contrasts with the predictions of exemplar-based models, under which listeners may only use their direct linguistic experience to assist in categorizing sounds. We instead bring in models from cognitive psychology called dual-system models, which have implications for speech pathology and for the biology of language.
Chris Heffner (University of Maryland)
Date: April 17th, 2:15pm
Where: B342 Wells Hall
Models of category learning in phonetics can generally be broken into two different types. Under rule-based models, categories are the result of a list of rules along the lines of a checklist. Under exemplar-based models, meanwhile, categories are the sum total of the experiences with the members of a certain category. In this study, listeners heard phonetic tokens on a continuum of palatal to velar fricatives, which were paired with one of either two or three categories. The pairings of continuum members to categories varied from condition to condition. In the end, though listeners could learn certain pairings of phonetic tokens to categories, they found learning several of the mappings from speech sounds to categories challenging. Their performance indicates that they may instead be imposing rule-like structure on the disorderly categories they attempted to learn. This indicates that listeners may come to the task of categorizing phonetic tokens with strong expectations for how the sounds should be categorized. This contrasts with the predictions of exemplar-based models, under which listeners may only use their direct linguistic experience to assist in categorizing sounds. We instead bring in models from cognitive psychology called dual-system models, which have implications for speech pathology and for the biology of language.
GLEEFUL Keynote Talk: Object symmetry effects in three Germanic languages
Bill Haddican (CUNY, Queens College)
When: April 18th, 2:00pm
Where: B342 Wells Hall
This talk focuses on the passive symmetry problem how to understand cross-linguistic variation in patterns of passivization out of double object constructions, like the English sentence in (1). Some varieties, "asymmetric passive'' languages/dialects, allow only the higher, goal argument to passivize, as in the standard English example in (2). These differ from "symmetric passive'' varieties, which allow both the theme and goal arguments to passivize. Such is the case for some dialects of Northern and Western England, which allow both (2) and (3).
(1) Tom gave Maria a book.
(2) Maria was given a book. (Goal passivization)
(3) A book was given Maria. (Theme passivization, ok in some UK dialects)
The consensus view in contemporary syntax to express this variation in terms of locality: in asymmetric passive varieties, passivization of the theme is blocked by the presence of the goal argument. Symmetric passive varieties differ in having some extra movement step that obviates this locality effect, making theme-passivization possible. An alternative view is that the variation between symmetric and asymmetric varieties reflects differences in the way that case is assigned to the two objects in these languages.
In this talk, we discuss evidence from a set of judgment experiments with native speakers of British English, Swedish and Norwegian, all of which have cross-speaker variation in the availability of theme-passivization. The results suggest that object symmetry is not a unified phenomenon but rather that there are several different way that locality can be circumvented. We argue that neither the case approach and the locality approach are exclusively correct, but rather that both are needed to model the relevant facts across Germanic varieties.
Bill Haddican (CUNY, Queens College)
When: April 18th, 2:00pm
Where: B342 Wells Hall
This talk focuses on the passive symmetry problem how to understand cross-linguistic variation in patterns of passivization out of double object constructions, like the English sentence in (1). Some varieties, "asymmetric passive'' languages/dialects, allow only the higher, goal argument to passivize, as in the standard English example in (2). These differ from "symmetric passive'' varieties, which allow both the theme and goal arguments to passivize. Such is the case for some dialects of Northern and Western England, which allow both (2) and (3).
(1) Tom gave Maria a book.
(2) Maria was given a book. (Goal passivization)
(3) A book was given Maria. (Theme passivization, ok in some UK dialects)
The consensus view in contemporary syntax to express this variation in terms of locality: in asymmetric passive varieties, passivization of the theme is blocked by the presence of the goal argument. Symmetric passive varieties differ in having some extra movement step that obviates this locality effect, making theme-passivization possible. An alternative view is that the variation between symmetric and asymmetric varieties reflects differences in the way that case is assigned to the two objects in these languages.
In this talk, we discuss evidence from a set of judgment experiments with native speakers of British English, Swedish and Norwegian, all of which have cross-speaker variation in the availability of theme-passivization. The results suggest that object symmetry is not a unified phenomenon but rather that there are several different way that locality can be circumvented. We argue that neither the case approach and the locality approach are exclusively correct, but rather that both are needed to model the relevant facts across Germanic varieties.
Meaning alternations and kind-reference in modified nouns
Timothy Leffel (University of Chicago)
Date: April 9th
Where: B243 Wells Hall
A number of researchers have observed that the type/token distinction in nominal reference somehow corresponds to the structural layers of N versus D (Vergnaud & Zubizarreta 1992; Longobardi 1994; Carlson 2003). It has also been noted (approximately) that "inner" noun modifiers often have a "generic" interpretation where the same modifiers in "outer" positions do not (Larson 1998,1999). While supported by clear empirical patterns, neither observation has been explained from basic principles in a satisfying way. In this talk, I argue that the type-N, token-D and the inner-generic, outer-not correspondences are actually two sides of the same coin. More specifically, I formulate a theory of nominal roots as predicates of subkinds according to which the type-to-token conversion is encoded in inflectional morphology. The correspondences stated above follow from this approach as a single byproduct of the basic compositional semantics. Combined with a simple syntax for nominal modification, the approach also explains some basic distributional facts about notoriously puzzling "Bolinger contrasts" (e.g. 'the visible stars' vs 'the stars visible'). I discuss some tentative extensions of the theory presented here, as well as some connections to broader theoretical issues about the relationship between word-order and meaning alternations.
Timothy Leffel (University of Chicago)
Date: April 9th
Where: B243 Wells Hall
A number of researchers have observed that the type/token distinction in nominal reference somehow corresponds to the structural layers of N versus D (Vergnaud & Zubizarreta 1992; Longobardi 1994; Carlson 2003). It has also been noted (approximately) that "inner" noun modifiers often have a "generic" interpretation where the same modifiers in "outer" positions do not (Larson 1998,1999). While supported by clear empirical patterns, neither observation has been explained from basic principles in a satisfying way. In this talk, I argue that the type-N, token-D and the inner-generic, outer-not correspondences are actually two sides of the same coin. More specifically, I formulate a theory of nominal roots as predicates of subkinds according to which the type-to-token conversion is encoded in inflectional morphology. The correspondences stated above follow from this approach as a single byproduct of the basic compositional semantics. Combined with a simple syntax for nominal modification, the approach also explains some basic distributional facts about notoriously puzzling "Bolinger contrasts" (e.g. 'the visible stars' vs 'the stars visible'). I discuss some tentative extensions of the theory presented here, as well as some connections to broader theoretical issues about the relationship between word-order and meaning alternations.
Vocal Effort and Comfort in Speech: Accommodation to Room Acoustic Conditions
Simone Graetzer and Pasquale Bottalico (Michigan State University)
Date: March 19th
Vocal effort is a physiological entity that accounts for variation in voice production as loading increases, measured as Sound Pressure Level (SPL). A number of studies have investigated vocal effort and load, but few have considered the role of room acoustics. In this experiment, 20 subjects performed vocal tasks in the presence of babble noise in anechoic, semi-reverberant and reverberant environments. The room acoustics in each environment were modified by introducing two reflective panels placed at 45⁰ from the axis mouth at 0.5m from the subject. Tasks were performed at normal and loud volumes. After each task, the subject answered questions addressing their perception of vocal effort, comfort, control, and the clarity of their own voice. SPL and fundamental frequency (F0) were measured. Vocal effort (SPL) decreased when the panels were present. An assessment of F0 and subjective measures is also reported. The results indicate that even when keeping reverberation time constant, reflective surfaces may be optimized to reduce vocal effort.
Simone Graetzer and Pasquale Bottalico (Michigan State University)
Date: March 19th
Vocal effort is a physiological entity that accounts for variation in voice production as loading increases, measured as Sound Pressure Level (SPL). A number of studies have investigated vocal effort and load, but few have considered the role of room acoustics. In this experiment, 20 subjects performed vocal tasks in the presence of babble noise in anechoic, semi-reverberant and reverberant environments. The room acoustics in each environment were modified by introducing two reflective panels placed at 45⁰ from the axis mouth at 0.5m from the subject. Tasks were performed at normal and loud volumes. After each task, the subject answered questions addressing their perception of vocal effort, comfort, control, and the clarity of their own voice. SPL and fundamental frequency (F0) were measured. Vocal effort (SPL) decreased when the panels were present. An assessment of F0 and subjective measures is also reported. The results indicate that even when keeping reverberation time constant, reflective surfaces may be optimized to reduce vocal effort.
Interface Delay and Developmental Semantics
John Grinstead (The Ohio State University)
Date: February 26th
Interface Delay is a generalization that attempts to explain phenomena characterized by relatively slow developmental trajectories of particular linguistic constructions that seem to require information from multiple cognitive domains. In earlier work, I argue that this explains delayed development in children’s ability to count using the conventional count routine in English (Grinstead, MacSwan, Curtiss & Gelman 1998) and the pattern of delayed and non-delayed morphosyntactic development (Grinstead, Baron, Vega-Mendoza, De la Mora, Cantú-Sánchez and Flores 2013) in child Spanish-speaking children with specific language impairment (SLI). For example, while the morphophonological dimensions of plural marking on nouns in English and Spanish seem to develop relatively quickly, in both typical and atypical children, tense marking, which requires a broad array of interfaces between morphophonology, syntax, semantics and discourse pragmatics, seems to develop more slowly. In this talk, I will explore a parallel Interface Delay account in the domain of semantics and pragmatics. First, in the development of pragmatic implicatures in child English, we observe the critical role played by phonology in marking the “some, but not all” scalar implicature associated with the existential quantifier some: it is typically expressed with a full vowel and a L+H* pitch accent. When this quantifier is expressed as vowel-less sm, in contrast, it rarely expresses the pragmatically-enriched meaning and, instead, tends to be interpreted as “some, and possibly all”. The full vowel, but unaccented version is most able to switch between meanings. Child English-speakers do not seem to grasp this range of phonologically indicated meanings until they are school aged. Adult Spanish, on the other hand, computes these meanings with distinct lexical items (unos and algunos), and depends less on prosody and vowel length. It seems that, given this relatively smaller contribution of cross-domain coordination, Spanish-speaking children appear to grasp the distinction in meaning while still in the preschool age range. Second, I consider other consequences of cross-domain coordination of linguistic information for semantics, including what I refer to as Interface Deficit for Spanish-speaking children with SLI, who appear to have a harder time making prototypical connections between lexical aspect (telicity, in particular), grammatical aspect and tense, in contrast to typically-developing Spanish-speaking children. A final example of Interface Delay is between the conceptual system and the lexicon. In the case of distributive and collective interpretations, a plausible cognitive precursor to distributivity and collectivity, individuation, appears to be conceptually available to children from infancy, but Spanish-speaking children take until 10 or 11 to develop adult-like compositional representations of sentences that require distributional or collective interpretations in adult Spanish.
John Grinstead (The Ohio State University)
Date: February 26th
Interface Delay is a generalization that attempts to explain phenomena characterized by relatively slow developmental trajectories of particular linguistic constructions that seem to require information from multiple cognitive domains. In earlier work, I argue that this explains delayed development in children’s ability to count using the conventional count routine in English (Grinstead, MacSwan, Curtiss & Gelman 1998) and the pattern of delayed and non-delayed morphosyntactic development (Grinstead, Baron, Vega-Mendoza, De la Mora, Cantú-Sánchez and Flores 2013) in child Spanish-speaking children with specific language impairment (SLI). For example, while the morphophonological dimensions of plural marking on nouns in English and Spanish seem to develop relatively quickly, in both typical and atypical children, tense marking, which requires a broad array of interfaces between morphophonology, syntax, semantics and discourse pragmatics, seems to develop more slowly. In this talk, I will explore a parallel Interface Delay account in the domain of semantics and pragmatics. First, in the development of pragmatic implicatures in child English, we observe the critical role played by phonology in marking the “some, but not all” scalar implicature associated with the existential quantifier some: it is typically expressed with a full vowel and a L+H* pitch accent. When this quantifier is expressed as vowel-less sm, in contrast, it rarely expresses the pragmatically-enriched meaning and, instead, tends to be interpreted as “some, and possibly all”. The full vowel, but unaccented version is most able to switch between meanings. Child English-speakers do not seem to grasp this range of phonologically indicated meanings until they are school aged. Adult Spanish, on the other hand, computes these meanings with distinct lexical items (unos and algunos), and depends less on prosody and vowel length. It seems that, given this relatively smaller contribution of cross-domain coordination, Spanish-speaking children appear to grasp the distinction in meaning while still in the preschool age range. Second, I consider other consequences of cross-domain coordination of linguistic information for semantics, including what I refer to as Interface Deficit for Spanish-speaking children with SLI, who appear to have a harder time making prototypical connections between lexical aspect (telicity, in particular), grammatical aspect and tense, in contrast to typically-developing Spanish-speaking children. A final example of Interface Delay is between the conceptual system and the lexicon. In the case of distributive and collective interpretations, a plausible cognitive precursor to distributivity and collectivity, individuation, appears to be conceptually available to children from infancy, but Spanish-speaking children take until 10 or 11 to develop adult-like compositional representations of sentences that require distributional or collective interpretations in adult Spanish.
Be free! (of construction-specific grammar)
Heather Taylor (Michigan State University)
Date: January 29th
Ever since transformational grammar began morphing into a model of grammar that eliminated construction-specific rules (Chomsky 1977, 1986), there has been lively discussion over a loss of descriptive adequacy. A central argument against eliminating sentential and phrasal constructions is that most (if not all) linguistic data cannot be derived from the parts alone. According to this argument, we are frequently faced with cases where an entire expression is described as 'more than the sum of its parts.' Peripheral (i.e., unusual, infrequent, or previously unanalyzed) data are frequently invoked as particularly strong evidence for this claim.
This talk will primarily deal with my proposed analysis for a specific case of such 'peripheral' data, comparative correlatives (an example from English is in (1)). The analysis demonstrates that, despite the unusual appearance of these expressions, they can be derived in a construction-free syntax by appealing to well-attested operations of the computational system, like successive cyclic A'-movement and a richly structured CP-domain (Rizzi 1997, 2004).
(1) The longer the storm lasts, the worse the damage is
Further, my treatment of these data demonstrate that they can be derived cross linguistically without appealing to constructions. As such, this analysis can be taken as a case study of how constructions are not needed in a grammar, not even for the derivation of 'peripheral' data. Beyond this, I argue that introducing constructions into a grammar creates an burden on the computational system by forcing choices at every step in a derivation. Thus, if they are not essential to the computational system, they should necessarily be eliminated in favor of a simpler system.
Heather Taylor (Michigan State University)
Date: January 29th
Ever since transformational grammar began morphing into a model of grammar that eliminated construction-specific rules (Chomsky 1977, 1986), there has been lively discussion over a loss of descriptive adequacy. A central argument against eliminating sentential and phrasal constructions is that most (if not all) linguistic data cannot be derived from the parts alone. According to this argument, we are frequently faced with cases where an entire expression is described as 'more than the sum of its parts.' Peripheral (i.e., unusual, infrequent, or previously unanalyzed) data are frequently invoked as particularly strong evidence for this claim.
This talk will primarily deal with my proposed analysis for a specific case of such 'peripheral' data, comparative correlatives (an example from English is in (1)). The analysis demonstrates that, despite the unusual appearance of these expressions, they can be derived in a construction-free syntax by appealing to well-attested operations of the computational system, like successive cyclic A'-movement and a richly structured CP-domain (Rizzi 1997, 2004).
(1) The longer the storm lasts, the worse the damage is
Further, my treatment of these data demonstrate that they can be derived cross linguistically without appealing to constructions. As such, this analysis can be taken as a case study of how constructions are not needed in a grammar, not even for the derivation of 'peripheral' data. Beyond this, I argue that introducing constructions into a grammar creates an burden on the computational system by forcing choices at every step in a derivation. Thus, if they are not essential to the computational system, they should necessarily be eliminated in favor of a simpler system.
Polar Questions and Disjunction: clues from Hindi-Urdu `kyaa'
Rajesh Bhatt (University of Massachusetts, Amherst)
Date: January 15th
Hindi-Urdu has an optional marker `kyaa' that appears in polar and alternative questions. We delineate its properties distinguishing from the homophonous thematic `kyaa' (what); in particular we locate it in ForceP. Then we use `kyaa' to argue that projection of alternatives (as in Alternative and Inquisitive Semantics) is constrained by the syntax. In particular, A-bar movements lead to `closure' of alternatives, making them inaccessible. Consequently we expect a bleeding relationship between such movements and operations that depend upon alternatives such as alternative questions. Fina lly we also explore interactions between the intonational marking of Y/N questions and syntax.
Rajesh Bhatt (University of Massachusetts, Amherst)
Date: January 15th
Hindi-Urdu has an optional marker `kyaa' that appears in polar and alternative questions. We delineate its properties distinguishing from the homophonous thematic `kyaa' (what); in particular we locate it in ForceP. Then we use `kyaa' to argue that projection of alternatives (as in Alternative and Inquisitive Semantics) is constrained by the syntax. In particular, A-bar movements lead to `closure' of alternatives, making them inaccessible. Consequently we expect a bleeding relationship between such movements and operations that depend upon alternatives such as alternative questions. Fina lly we also explore interactions between the intonational marking of Y/N questions and syntax.
An “easy questions” approach to hard problems in prosodic analysis
Jennifer Cole (University of Illinois at Urbana-Champaign)
Date: December 4th, 2014
This talk demonstrates the merits of the “easy questions” approach to hard scientific problems concerning the phonetic correlates of prosodic structure and their representation in cognitive encoding. Problems in prosodic analysis, which are traditionally investigated with prosodic annotations from expert transcribers, are factored into a series of easy (binary) problems that can be answered by non-experts. We apply this method using crowdsourcing to achieve efficiency and reliability in prosodic annotation with large speech datasets. This approach allows us to investigate prosody perception for diverse languages, and with data from the broader speech community. It is provably successful because each crowdsourcer only needs to be partially correct (Hasegawa-Johnson et al., 2014). I present an overview of our research using crowdsourcing and Rapid Prosody Transcription to investigate prosody in American English, focusing here on prosodic prominence. Our findings show that despite the apparent uncertainty of individual transcribers, non-expert listeners perceive categorical prominence distinctions in American English conversational speech, based on a composite of acoustic cues, and these prosodic distinctions are strongly predictive of the more complex prosodic labels assigned by experts using the ToBI annotation system. We observe that patterns of perceived prominence are weakly asymmetric within the prosodic phrase, with the strongest prominence response typically for words in the nuclear (rightmost) position of the phrase, lending qualified support to earlier descriptions of English as a head-final language in its phonological structure. Our findings also show that, with explicit instructions, listeners can attend more to acoustic cues or (under different instructions) to lexico-syntactic factors related to the meaning functions of prosody, but prominence labels strongly converge under the two task conditions. These findings for American English are briefly compared with findings from our parallel study of prosody perception in Spanish which reveal significant differences between the languages in the acoustic cues to prominence and their relationship to meaning functions. Our research demonstrates that “easy questions” can reveal the complexity of prominence in everyday spoken language, and shed light on how languages differ in the grammatical use of prominence.
Jennifer Cole (University of Illinois at Urbana-Champaign)
Date: December 4th, 2014
This talk demonstrates the merits of the “easy questions” approach to hard scientific problems concerning the phonetic correlates of prosodic structure and their representation in cognitive encoding. Problems in prosodic analysis, which are traditionally investigated with prosodic annotations from expert transcribers, are factored into a series of easy (binary) problems that can be answered by non-experts. We apply this method using crowdsourcing to achieve efficiency and reliability in prosodic annotation with large speech datasets. This approach allows us to investigate prosody perception for diverse languages, and with data from the broader speech community. It is provably successful because each crowdsourcer only needs to be partially correct (Hasegawa-Johnson et al., 2014). I present an overview of our research using crowdsourcing and Rapid Prosody Transcription to investigate prosody in American English, focusing here on prosodic prominence. Our findings show that despite the apparent uncertainty of individual transcribers, non-expert listeners perceive categorical prominence distinctions in American English conversational speech, based on a composite of acoustic cues, and these prosodic distinctions are strongly predictive of the more complex prosodic labels assigned by experts using the ToBI annotation system. We observe that patterns of perceived prominence are weakly asymmetric within the prosodic phrase, with the strongest prominence response typically for words in the nuclear (rightmost) position of the phrase, lending qualified support to earlier descriptions of English as a head-final language in its phonological structure. Our findings also show that, with explicit instructions, listeners can attend more to acoustic cues or (under different instructions) to lexico-syntactic factors related to the meaning functions of prosody, but prominence labels strongly converge under the two task conditions. These findings for American English are briefly compared with findings from our parallel study of prosody perception in Spanish which reveal significant differences between the languages in the acoustic cues to prominence and their relationship to meaning functions. Our research demonstrates that “easy questions” can reveal the complexity of prominence in everyday spoken language, and shed light on how languages differ in the grammatical use of prominence.
A compositional account of scope-marking via embedded topics
Kyle Rawlins (Johns Hopkins University)
Date: November 20th, 2014
Scope marking is a multiple-wh phenomenon where, descriptively, a (semantically minimal) root-clause interrogative pronoun indicates that the sentence is a question, and an embedded interrogative pronoun indicates what clause is actually questioned. This pattern in broad terms is cross-linguistically robust and has led to a large literature; a classic example from Hindi, due to Dayal (1994), is shown in (1).
(1) juan kyaa soctaa hai meri kis-se baat karegii?
John what thinks PRES Mary who-with talk will
'Who does John think Mary will talk to?'
Intuitively, (1) can be thought of as pragmatically asking the embedded question ('who will Mary talk to?'), with some extra caveats about the desired source or evidence for the answer. This species of data has been the center of substantial discussion and debate, and a range of analyses for scope marking are therefore on the market. These include (‘direct dependency') syntactic approaches involving partial wh-movement or other such mechanisms, and semantic (‘indirect dependency’; Dayal 1994 et seq) approaches involving higher-order restricted questions. In this talk I focus on a seldom-discussed scope marking pattern in English, where the embedded clause is marked with a preposition, typically ‘about’; this is illustrated in (2). Descriptively, this configuration acts just like scope-marking constructions in other languages (and in fact is sometimes used as a paraphrase in such discussions).
(2) What did Alfonso say about who is going to the party?
`Who is going to the party, according to what Alfonso said?'
I develop a new compositional semantic ('indirect') account of this pattern, building on Rawlins (2013): "about"-phrases are (semantically) modifiers to clause-embedding predicates, with a broad distribution, and provide (roughly) a topic for the 'content' of the attitude or communication. I implement this notion of topic via David Lewis's formalization of relevance/aboutness, combined with a neo-Davidsonian account of clause-embedding predicates. In a scope-marking configuration the root-clause wh-item abstracts over the direct object to the embedding predicate, and so (2) asks, roughly: what was the content of Alfonso's utterance(s), restricting ourselves to utterances whose content was 'about' what Alfonso said? On this account, (2) has a standard and in fact boring syntax, with just two cases of local, ordinary wh-movement. I show that this account makes a range of desirable predictions about English, for example predicting exactly the set of English verbs that participate in this form of scope marking. Finally, this account can extend to some, but not all, of the scope-marking patterns discussed in the cross-linguistic literature, contributing to the idea that the direct/indirect analytical distinction forms a spectrum rather than a forced choice.
Kyle Rawlins (Johns Hopkins University)
Date: November 20th, 2014
Scope marking is a multiple-wh phenomenon where, descriptively, a (semantically minimal) root-clause interrogative pronoun indicates that the sentence is a question, and an embedded interrogative pronoun indicates what clause is actually questioned. This pattern in broad terms is cross-linguistically robust and has led to a large literature; a classic example from Hindi, due to Dayal (1994), is shown in (1).
(1) juan kyaa soctaa hai meri kis-se baat karegii?
John what thinks PRES Mary who-with talk will
'Who does John think Mary will talk to?'
Intuitively, (1) can be thought of as pragmatically asking the embedded question ('who will Mary talk to?'), with some extra caveats about the desired source or evidence for the answer. This species of data has been the center of substantial discussion and debate, and a range of analyses for scope marking are therefore on the market. These include (‘direct dependency') syntactic approaches involving partial wh-movement or other such mechanisms, and semantic (‘indirect dependency’; Dayal 1994 et seq) approaches involving higher-order restricted questions. In this talk I focus on a seldom-discussed scope marking pattern in English, where the embedded clause is marked with a preposition, typically ‘about’; this is illustrated in (2). Descriptively, this configuration acts just like scope-marking constructions in other languages (and in fact is sometimes used as a paraphrase in such discussions).
(2) What did Alfonso say about who is going to the party?
`Who is going to the party, according to what Alfonso said?'
I develop a new compositional semantic ('indirect') account of this pattern, building on Rawlins (2013): "about"-phrases are (semantically) modifiers to clause-embedding predicates, with a broad distribution, and provide (roughly) a topic for the 'content' of the attitude or communication. I implement this notion of topic via David Lewis's formalization of relevance/aboutness, combined with a neo-Davidsonian account of clause-embedding predicates. In a scope-marking configuration the root-clause wh-item abstracts over the direct object to the embedding predicate, and so (2) asks, roughly: what was the content of Alfonso's utterance(s), restricting ourselves to utterances whose content was 'about' what Alfonso said? On this account, (2) has a standard and in fact boring syntax, with just two cases of local, ordinary wh-movement. I show that this account makes a range of desirable predictions about English, for example predicting exactly the set of English verbs that participate in this form of scope marking. Finally, this account can extend to some, but not all, of the scope-marking patterns discussed in the cross-linguistic literature, contributing to the idea that the direct/indirect analytical distinction forms a spectrum rather than a forced choice.
Natural sequences in second language acquisition
Patti Spinner (Michigan State University)
Date: September 11th, 2014
One of the great questions in second language acquisition research is whether second languages are fundamentally the same as or different from first languages (The Fundamental Difference Hypothesis, Bley-Vroman, 1989, 2009). One of the ways researchers have examined this question is to compare natural sequences of grammar acquisition in first (L1) and second (L2) languages. At first, morpheme order studies (e.g., Brown, 1973; Dulay & Burt, 1974) seemed to offer promising insights into this issue, but this research has fallen out of favor due to methodological and theoretical concerns.
However, two more recent proposals, Organic Grammar (OG) (Vainikka & Young-Scholten, 1996, 2006) and Processability Theory (PT) (Pienemann, 1998, 2005), explore the possibility of natural L2 orders of acquisition for a variety of morphosyntactic forms and structures. Both OG and PT, theoretically grounded in Generative grammar and Lexical Functional grammar respectively, propose that L2 acquisition proceeds in a parallel fashion to L1 acquisition, with differences mainly accounted for by the initial state of L2s versus L1s.
In this talk, I review recent research that has examined the claims of both OG and PT. Strengths and weaknesses of each are explored. Practical applications as well as implications for the Fundamental Difference Hypothesis and second language production, reception, and representation are discussed.
Patti Spinner (Michigan State University)
Date: September 11th, 2014
One of the great questions in second language acquisition research is whether second languages are fundamentally the same as or different from first languages (The Fundamental Difference Hypothesis, Bley-Vroman, 1989, 2009). One of the ways researchers have examined this question is to compare natural sequences of grammar acquisition in first (L1) and second (L2) languages. At first, morpheme order studies (e.g., Brown, 1973; Dulay & Burt, 1974) seemed to offer promising insights into this issue, but this research has fallen out of favor due to methodological and theoretical concerns.
However, two more recent proposals, Organic Grammar (OG) (Vainikka & Young-Scholten, 1996, 2006) and Processability Theory (PT) (Pienemann, 1998, 2005), explore the possibility of natural L2 orders of acquisition for a variety of morphosyntactic forms and structures. Both OG and PT, theoretically grounded in Generative grammar and Lexical Functional grammar respectively, propose that L2 acquisition proceeds in a parallel fashion to L1 acquisition, with differences mainly accounted for by the initial state of L2s versus L1s.
In this talk, I review recent research that has examined the claims of both OG and PT. Strengths and weaknesses of each are explored. Practical applications as well as implications for the Fundamental Difference Hypothesis and second language production, reception, and representation are discussed.
Restrictive Phi in a Partial Typology of Noncanonical Passives
Julie Legate (University of Pennsylvania)
Date: October 2nd, 2014
In this talk, I investigate the syntactic structure of noncanonical passives, focusing on the role played by phi-features that restrict rather than saturate the external argument position. Building on previous work by myself and others, I show that voice is encoded in a functional projection, VoiceP, which is distinct from, and higher than, vP. I demonstrate that microvariations in the properties of VoiceP and in the location of restrictive phi-features explain a wide range of noncanonical passives, including agent-agreeing passives, restricted agent passives, accusative object passives, impersonals, and object voice. The analysis draws on data from a typologically diverse set of languages.
Julie Legate (University of Pennsylvania)
Date: October 2nd, 2014
In this talk, I investigate the syntactic structure of noncanonical passives, focusing on the role played by phi-features that restrict rather than saturate the external argument position. Building on previous work by myself and others, I show that voice is encoded in a functional projection, VoiceP, which is distinct from, and higher than, vP. I demonstrate that microvariations in the properties of VoiceP and in the location of restrictive phi-features explain a wide range of noncanonical passives, including agent-agreeing passives, restricted agent passives, accusative object passives, impersonals, and object voice. The analysis draws on data from a typologically diverse set of languages.
Non-thematic subjects in Brazilian Portuguese
Aroldo Andrade (University of Quebec at Montreal)
Date: October 16th, 2014
Where: B243 Wells Hall
Brazilian Portuguese shows genitive and locative elements that behave as subjects. The presentation offers a unified account of the distribution of these two elements along different classes of unaccusative verbs that share some type of secondary predicate. Finally, a comparison with European Portuguese, which is devoid of such a construction, is offered.
Aroldo Andrade (University of Quebec at Montreal)
Date: October 16th, 2014
Where: B243 Wells Hall
Brazilian Portuguese shows genitive and locative elements that behave as subjects. The presentation offers a unified account of the distribution of these two elements along different classes of unaccusative verbs that share some type of secondary predicate. Finally, a comparison with European Portuguese, which is devoid of such a construction, is offered.
Innit and its variable contexts: Discourse-pragmatic innovations in Multicultural London English
Heike Pichler (Newcastle University)
Date: October 21st, 2014 (note: this is a Tuesday)
Previous studies of the tag form innit, illustrated in (1)-(3), have examined its derivation from isn’t it, its spread across social and geographical space, and its gradual levelling across the inflectional paradigm (Andersen 2001; Cheshire et al. 2005; Pichler 2013). The common consensus across these studies has been that the use of innit is limited to clause-final positioning and that it shares the functional profile of canonical negative polarity question tags (neg-tags) (e.g. isn’t it, haven’t you, wouldn’t we).
(1) It’s like that big actually, innit?
(2) We pay them, innit? To rob people, innit?
(3) We were down the west end, innit?
In this talk, I will examine the use of innit and other neg-tag variants in the 1.4-million-word Linguistic Innovators (LIC) corpus collected in 2005-2006 in inner- and outer-city London (Kerswill et al. 2007). Close analysis of the data reveals a range of innovative innit uses not previously discussed in the scholarly literature, including: its use as a clause-final punctuation marker in narrative sequences, as in (4); its turn-taking function in multi-party interactions, as in (5); and its occurrence as a discourse-new marking strategy after left-dislocated and lone NPs in collaborative discourse, as in (6).
(4) Then we took his bike, innit, joyriding it. Then the police came, innit?
(5) Z: I don’t like heights, bruv.
A: Innit! I hate heights, (bruv). I’m shook.
(6) a. The sister, innit, she’s about five times bigger than you, innit, Mark?
b. The girl, innit. There was a girl.
Based on my variationist analysis of the canon of neg-tags in the LIC corpus, I will provide hypotheses about the linguistic (and social) mechanisms that gave rise to these innovative innit uses and explore how they interact with their potential co-variants in these environments. Underpinning my analysis is the view first articulated in Waters (MS): that a comprehensive account of a form’s trajectory of change and sociolinguistic distribution requires a flexible approach to defining the variable context.
Heike Pichler (Newcastle University)
Date: October 21st, 2014 (note: this is a Tuesday)
Previous studies of the tag form innit, illustrated in (1)-(3), have examined its derivation from isn’t it, its spread across social and geographical space, and its gradual levelling across the inflectional paradigm (Andersen 2001; Cheshire et al. 2005; Pichler 2013). The common consensus across these studies has been that the use of innit is limited to clause-final positioning and that it shares the functional profile of canonical negative polarity question tags (neg-tags) (e.g. isn’t it, haven’t you, wouldn’t we).
(1) It’s like that big actually, innit?
(2) We pay them, innit? To rob people, innit?
(3) We were down the west end, innit?
In this talk, I will examine the use of innit and other neg-tag variants in the 1.4-million-word Linguistic Innovators (LIC) corpus collected in 2005-2006 in inner- and outer-city London (Kerswill et al. 2007). Close analysis of the data reveals a range of innovative innit uses not previously discussed in the scholarly literature, including: its use as a clause-final punctuation marker in narrative sequences, as in (4); its turn-taking function in multi-party interactions, as in (5); and its occurrence as a discourse-new marking strategy after left-dislocated and lone NPs in collaborative discourse, as in (6).
(4) Then we took his bike, innit, joyriding it. Then the police came, innit?
(5) Z: I don’t like heights, bruv.
A: Innit! I hate heights, (bruv). I’m shook.
(6) a. The sister, innit, she’s about five times bigger than you, innit, Mark?
b. The girl, innit. There was a girl.
Based on my variationist analysis of the canon of neg-tags in the LIC corpus, I will provide hypotheses about the linguistic (and social) mechanisms that gave rise to these innovative innit uses and explore how they interact with their potential co-variants in these environments. Underpinning my analysis is the view first articulated in Waters (MS): that a comprehensive account of a form’s trajectory of change and sociolinguistic distribution requires a flexible approach to defining the variable context.
The mental status of derivational operations
Tim Hunter (University of Minnesota - Twin Cities)
Date: November 13th, 2014
In this talk I present an approach to integrating generative syntax with information-theoretic sentence complexity metrics, from which it follows that two grammars that are extensionally equivalent --- two grammars which produce the same structures, and differ only in what primitive derivational operations they use to build those structures --- can nonetheless give rise to distinct predictions concerning sentence comprehension difficulty. This provides a linking hypothesis that connects sentence processing observations to subtle questions about the derivational operations that comprise human grammars (merge, move, re-merge, agree, etc.). Competing answers to these sorts of fine-grained questions can therefore be understood as competing claims about cognitive machinery even if they fail to make distinct predictions about acceptability judgement data.
Tim Hunter (University of Minnesota - Twin Cities)
Date: November 13th, 2014
In this talk I present an approach to integrating generative syntax with information-theoretic sentence complexity metrics, from which it follows that two grammars that are extensionally equivalent --- two grammars which produce the same structures, and differ only in what primitive derivational operations they use to build those structures --- can nonetheless give rise to distinct predictions concerning sentence comprehension difficulty. This provides a linking hypothesis that connects sentence processing observations to subtle questions about the derivational operations that comprise human grammars (merge, move, re-merge, agree, etc.). Competing answers to these sorts of fine-grained questions can therefore be understood as competing claims about cognitive machinery even if they fail to make distinct predictions about acceptability judgement data.