I’m a linguist working at the interface between syntax and morphology. I am particularly interested in wh-question formation strategies and the derivation of morphological alternations that appear in the context of A′-movement. Most of my research involves data from Bantu languages, drawn from existing sources as well as original fieldwork with native speakers. Supervisors: Bob Frank and Raffaella Zanuttini
Bantu languages, which are spoken throughout most of sub-Saharan Africa, permit wh-questions to b... more Bantu languages, which are spoken throughout most of sub-Saharan Africa, permit wh-questions to be constructed in multiple ways, including wh-in-situ, full wh-movement, and partial wh-movement. Shona, a Bantu language spoken by about 13 million people in Zimbabwe and Mozambique, allows all three of these types. In this dissertation, I conduct the first in-depth examination of Shona wh-questions, drawing on fifty hours of elicitation with a native speaker consultant to explore the derivational relationships among these strategies.
Wh-in-situ questions have received a wide variety of treatments in the syntactic literature, ranging from covert or disguised movement to postsyntactic binding of the wh-phrase by a silent question operator. In Bantu languages, wh-in-situ questions are often taken to be derived via a non-movement relation (e.g., Carstens 2005 for Kilega, Diercks 2010 for Lubukusu, Muriungi 2003 for Kîîtharaka, Sabel 2000 for Kikuyu and Duala, Sabel & Zeller 2006 for Zulu, Schneider-Zioga 2007 for Kinande), but alternatives have rarely been considered. I demonstrate how movement-based analyses that have been proposed for wh-in-situ in non-Bantu languages make the wrong predictions for Shona wh-in-situ, which lacks word order permutation, extraction marking, island effects, and intervention effects. These properties provide support for the traditional Bantuist view that the relation between the pronunciation site of an in-situ wh-phrase and its scopal position in the left periphery is not movement; I claim that in Shona it is unselective binding.
Many Bantu languages, including Shona, prohibit wh-phrases from appearing in the canonical preverbal subject position. Wasike (2007) demonstrates that this restriction applies to topicalized non-subjects as well as preverbal subjects. I replicate these results for Shona and argue that they cast doubt on Sabel & Zeller's (2006) attempt to characterize the ban with an appeal to improper movement. I argue instead that restrictions on the distribution of wh-in-situ in Bantu are tied to restrictions on the domain for focus licensing. This claim is further bolstered by an examination of crosslinguistic variation within Bantu with respect to whether the ban on in-situ preverbal wh-subjects applies in embedded clauses. I observe a previously unnoticed generalization: languages that universally ban in-situ preverbal wh-subjects (like Zulu) have immediately after the verb (IAV) focus effects; languages that do allow in-situ preverbal wh-subjects in embedded clauses (like Shona, Lubukusu, and Kîîtharaka) also lack IAV effects.
Full wh-movement in Shona gives rise to questions that bear a certain similarity to English wh-questions. However, using a range of diagnostics including extraction marking, island effects, reconstruction effects, and the distribution of temporal modifiers, I argue that what appears to be full wh-movement in Shona actually has a cleft structure: the wh-phrase moves to become the head of a relative clause, which is selected by a copula in the matrix clause. Just as in wh-in-situ, an ex-situ wh-phrase is pronounced lower than its scopal position, and the relation between these two positions is established via unselective binding. Additional evidence for this proposal comes from the sensitivity of partial wh-movement to island boundaries below but not above the pronunciation site of the wh-phrase, a pattern that has been predicted by previous analyses (e.g., Abels 2012, Sabel 2000, Sabel & Zeller 2006) but for which empirical support has been lacking until now. I therefore unify full and partial wh-movement under a single analysis for cleft-based wh-in-situ that involves a step of relativization (independently needed for relative clauses) and a step of unselective binding (independently needed for wh-in-situ).
This article explores geographical variation in a range of understudied dative constructions in A... more This article explores geographical variation in a range of understudied dative constructions in American English. It shows that these constructions are found primarily in the South and that they permit numerous syntactic variations and permutations. However, not all sentences and constructions have an equal status. In particular, the authors find that they lie on a continuum of markedness. More marked variants are judged acceptable by fewer speakers and have a more limited geographic distribution. And yet, even the most marked variants cannot be dismissed: the strong geographic nature of their distribution shows that they are a genuine part of the grammar of many speakers. Overall, this research contributes a more detailed picture of dative constructions in American English and a more nuanced picture of syntactic variation in Southern American English; moreover, the authors offer a novel approach to measuring geographical markedness in syntactic variation.
The Yale Grammatical Diversity Project approaches the empirical domain of North American English ... more The Yale Grammatical Diversity Project approaches the empirical domain of North American English from the perspective of generative microcomparative syntax. In addition to eliciting judgments from speakers of particular varieties, we also conduct large-scale surveys, map the results of those surveys geographically, conduct statistical tests taking geography and other social variables into account, and look for theoretically significant linguistic correlations. In all cases, we do this with the primary goal of understanding variation between speakers at the individual level. While our goals and methodologies are informed by our theoretical perspective, we expect that our work and results will be of interest to linguists working in other frameworks and even to the public more generally. This article outlines the goals and methodologies of the project and describes in broad strokes some of the results obtained so far, as well as some of the ways we have shared our findings with others, inside and outside academia.
Selected proceedings of the 44th Annual Conference on African Linguistics (ACAL), 2015
In many languages, A'-movement in the syntax is reflected by alternations in the morphology. Cate... more In many languages, A'-movement in the syntax is reflected by alternations in the morphology. Categorizing these extraction morphology phenomena on the basis of the morphological alternation involved produces three types, all of which are found in Bantu languages: replacement, deletion, and addition. This paper illustrates each of these types of extraction morphology with Bantu examples before demonstrating how the existence of the third type (addition extraction morphology or wh-agreement) in languages such as Shona, Duala, Lubukusu, and Akɔɔse proves problematic for an account of extraction morphology that relies on the deletion of features. Lahne (2009) proposes that extraction morphology is always less specific than canonical morphology; she implements this intuition using a mechanism called probe impoverishment, whereby a feature on the probe is deleted each time movement occurs. Here, cases of Bantu wh-agreement are shown to be counterexamples to Lahne's Generalization and to be more amenable to traditional analyses in which the additional morphology is the realization of the movement-triggering feature.
"Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological ... more "Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological arguments for a derivation of adverbial (subordinate) clauses that involves movement of an (often null) operator (see Haegeman 2010a for a review). This paper provides morphological evidence for this view while arguing for separate extraction sites for the moved elements in temporal and conditional clauses.
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typology); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses.
Crucially, Akɔɔse wh-agreement encodes whether the extracted element originated above or below v. The distribution of wh-agreement morphology shows that the operator in central conditional clauses is extracted from a position above v (supporting Haegeman's (2010b) claim), the moved element
in central temporal clauses originates below v (supporting Larson's (1987, 1990) position), and there is no extraction in peripheral adverbial clauses (as argued in Haegeman 2007).
Wh-agreement provides compelling evidence for a movement analysis for both temporal and conditional central adverbial clauses. Due to its sensitivity to height of extraction, Akɔɔse lends insight into the question of where the moved elements originate, unlike languages like Irish (McCloskey 2001) where wh-agreement only registers the presence of movement."
This paper examines phonological development and its relationship to input statistics. Using nove... more This paper examines phonological development and its relationship to input statistics. Using novel data from a longitudinal corpus of spontaneous child speech in Polish, we evaluate and compare the predictions of a variety of input-based phonotactic models for syllable structure acquisition. We find that many commonly examined input statistics can make dramatically different predictions, as do different assumptions about the representational units over which statistics are calculated. We find that development is sensitive to multiple abstract units of phonological representation, supporting a crucial role for feature-based generalization. We also identify departures between the predictions of the best phonotactic models and children’s production patterns that indicate input sensitivity alone cannot fully explain the developmental patterns. We discuss the role of universal markedness and phonetic difficulty and argue that a full explanation requires reference to these biases.
While the numeral systems of Australian languages are small, they are not uniform. In this articl... more While the numeral systems of Australian languages are small, they are not uniform. In this article we shed light on the extent of variation in small numeral systems by systematically surveying 189 languages from Pama- Nyungan and non-Pama-Nyungan families. We show that, contra previous assumptions, Australian languages vary extensively in the limits of their numeral system, in the ways in which smaller numerals may be combined to form larger ones, and in their ability to denote vague quantities like ‘some’ or ‘few’. We also investigate the etymological sources of words for numerals and find that few terms are reconstructible.
Numerals vary extensively across the world’s languages, ranging from no precise numeral terms to ... more Numerals vary extensively across the world’s languages, ranging from no precise numeral terms to practically infinite limits. Particularly of interest is the category of “small” or low-limit numeral systems; these are often associated with hunter-gatherer groups, but this connection has not yet been demonstrated by a systematic study. Here we present the results of a wide-scale survey of hunter-gatherer numerals. We compare these to agriculturalist languages in the same regions, and consider them against the broader typological backdrop of contemporary numeral systems in the world’s languages. We find that correlations with subsistence pattern are relatively weak, but that numeral trends are clearly areal.
In recent years, linguists have begun to increasingly rely on quantitative phylogenetic approache... more In recent years, linguists have begun to increasingly rely on quantitative phylogenetic approaches to examine language evolution. Some linguists have questioned the suitability of phylogenetic approaches on the grounds that linguistic evolution is largely reticulate due to extensive lateral transmission, or borrowing, among languages. The problem may be particularly pronounced in hunter-gatherer languages, where the conventional wisdom among many linguists is that lexical borrowing rates are so high that tree building approaches cannot provide meaningful insights into evolutionary processes. However, this claim has never been systematically evaluated, in large part because suitable data were unavailable. In addition, little is known about the subsistence, demographic, ecological, and social factors that might mediate variation in rates of borrowing among languages. Here, we evaluate these claims with a large sample of hunter-gatherer languages from three regions around the world. In this study, a list of 204 basic vocabulary items was collected for 122 hunter-gatherer and small-scale cultivator languages from three ecologically diverse case study areas: northern Australia, northwest Amazonia, and California and the Great Basin. Words were rigorously coded for etymological (inheritance) status, and loan rates were calculated. Loan rate variability was examined with respect to language area, subsistence mode, and population size, density, and mobility; these results were then compared to the sample of 41 primarily agriculturalist languages in [1]. Though loan levels varied both within and among regions, they were generally low in all regions (mean 5.06%, median 2.49%, and SD 7.56), despite substantial demographic, ecological, and social variation. Amazonian levels were uniformly very low, with no language exhibiting more than 4%. Rates were low but more variable in the other two study regions, in part because of several outlier languages where rates of borrowing were especially high. High mobility, prestige asymmetries, and language shift may contribute to the high rates in these outliers. No support was found for claims that hunter-gatherer languages borrow more than agriculturalist languages. These results debunk the myth of high borrowing in hunter-gatherer languages and suggest that the evolution of these languages is governed by the same type of rules as those operating in large-scale agriculturalist speech communities. The results also show that local factors are likely to be more critical than general processes in determining high (or low) loan rates.
Foreword to the Yale Law School's Pronouncing Dictionary of the Supreme Court of the United State... more Foreword to the Yale Law School's Pronouncing Dictionary of the Supreme Court of the United States, found at http://documents.law.yale.edu/pronouncing-dictionary.
This paper addresses the superficially similar focus constructions in Shona ([sna], Bantu, Zimbab... more This paper addresses the superficially similar focus constructions in Shona ([sna], Bantu, Zimbabwe) and Kîîtharaka ([thk], Bantu, Kenya). Using a topicalization test from Abels & Muriungi (2008), I argue that Shona has a biclausal cleft construction, while the Kîîtharaka construction is monoclausal. This asymmetry has consequences for the availability of the "intermediate" strategy for wh-question formation (Muriungi 2003).
This paper addresses partial wh-movement in Shona ([sna], Bantu, Zimbabwe), which is sensitive to... more This paper addresses partial wh-movement in Shona ([sna], Bantu, Zimbabwe), which is sensitive to islands below but not above the pronunciation site of the wh-phrase. This contrasts with partial wh-movement in languages such as Kîîtharaka (Muriungi 2003, Abels 2012) and Singaporean Malay (Cole & Hermon 1998), which is sensitive to islands both below and above the pronunciation site. I argue for a composite derivation of the Shona phenomenon: the wh-phrase moves overtly to its pronunciation site at an intermediate clause boundary, where it is unselectively bound by a null operator in the scopal position.
I show that in Shona partial wh-movement, the relation between the wh-phrase's base position and its pronunciation site patterns like full wh-movement in terms of island sensitivity, cleft structure, extraction marking, and reconstruction effects. On the other hand, the relation between the wh-phrase's pronunciation site and its scopal position displays the same lack of island sensitivity and lack of extraction marking exhibited by wh-in-situ in this language. Thus, Shona partial wh-movement can be reduced to a hybrid of full wh-movement and wh-in-situ. This composite derivation has been predicted to be possible (Sabel 2000, Abels 2012), but clear empirical support for it has been lacking until now.
This paper addresses partial wh-movement in Shona ([sna], Bantu, Zimbabwe), which is sensitive to... more This paper addresses partial wh-movement in Shona ([sna], Bantu, Zimbabwe), which is sensitive to islands below but not above the pronunciation site of the wh-word. I argue for a composite derivation of this phenomenon: the wh-word moves overtly to its pronunciation site at an intermediate clause boundary, where it is unselectively bound by a null operator in the scopal position. Thus, Shona partial wh-movement can be reduced to a hybrid of full wh-movement and wh-in-situ. This composite derivation has been predicted to be possible (Sabel 2000, Abels 2012), but clear empirical support for it has been lacking until now.
"This poster provides empirical support for the inheritance of non-φ-features by T from C. In the... more "This poster provides empirical support for the inheritance of non-φ-features by T from C. In the Bantu language Duala (Epée 1975, 1976a,b), A′-movement (focus movement, relativization, or wh-movement) of an element other than the matrix subject requires the presence of an invariant particle no following the highest verb in the clause where the A′-operator takes scope. The analysis presented here demonstrates that no is the morphological reflex on T of a feature associated with an A′-operator that takes scope in the CP domain. This featural link between C and T is straightforwardly derived under feature inheritance (Chomsky 2007, 2008)."
Many languages exhibit morphological reflexes of syntactic movement. Within Bantu, there are at l... more Many languages exhibit morphological reflexes of syntactic movement. Within Bantu, there are at least three types of this extraction morphology. In “anti-agreement,” extraction of a class 1 subject blocks canonical subject/verb agreement; in its place appears a featurally deficient agreement marker. The term "wh-agreement” covers two cases where verbal agreement remains intact but other morphological changes are triggered by syntactic movement. In the first, canonical morphophonology fails to appear in extraction contexts. In the second, canonical morphology remains constant across extraction and non-extraction contexts, but extraction contexts require the presence of additional morphology.
Lahne’s (2008) analysis of extraction morphology relies on the mechanism of “probe impoverishment,” where movement to the edge of a phase results in the deletion of one of the probe’s features. Probe impoverishment together with the Subset Principle (Halle 1997) predicts that “whenever a language shows different exponents in movement and non-movement contexts, when the marker appearing in movement contexts is less specific than the marker in non-movement contexts” (Lahne 2008: 9). This generalization is consistent with the facts of Bantu anti-agreement and the first type of wh-agreement.
However, here I illustrate the second type of wh-agreement with data from Kinande, Duala, and Akɔɔse, where extraction morphology does not compete with canonical morphology and is not found elsewhere in the language. In these cases, the extraction morphology cannot be the realization of an impoverished probe but rather the realization of the feature(s) driving movement.
Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological a... more Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological arguments for analyzing adverbial clauses as free relatives, involving movement of an (often null) operator (see Haegeman 2010 for a review). This paper provides morphological evidence for this view while arguing for separate extraction sites for the moved elements in temporal and conditional clauses.
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typological profile); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. The extraction of subjects is marked differently on the verb than the extraction of non-subjects. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses, suggesting that a free relative analysis of these clauses is warranted. Verbs in central temporal clauses take the non-subject extraction morphology, while verbs in central conditional clauses show the morphology found in subject extraction contexts. As would be expected from Haegeman's (2007) claim that peripheral adverbial clauses do not involve movement, these clauses show no extraction morphology in Akɔɔse.
In order to probe the question of where the relativizing operators in central temporal and conditional clauses originate, this paper offers a novel syntactic account for the morphological patterns found in Akɔɔse wh-agreement. According to this account, the temporal operator must originate within the VP layer, while the locus of extraction for the operator involved in central conditional clauses must be higher.
Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological a... more Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological arguments for a derivation of adverbial (subordinate) clauses that involves movement of an (often null) operator (see Haegeman 2010a for a review). This paper provides morphological evidence for this view while arguing for separate extraction sites for the moved elements in temporal and conditional clauses.
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typology); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. The extraction of subjects is marked differently on the verb than the extraction of non-subjects. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses. Verbs in central temporal clauses take the non-subject extraction morphology, while verbs in central conditional clauses show the morphology found in subject extraction contexts.
This paper offers a syntactic account for the patterns in Akɔɔse wh-agreement morphology and uses this account to refine our knowledge of the extraction sites for the operators in central temporal and conditional clauses, as well as to argue that peripheral adverbial clauses do not involve extraction (supporting Haegeman 2007).
Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological a... more Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological arguments for a derivation of adverbial (subordinate) clauses that involves movement of an (often null) operator (see Haegeman 2010a for a review). This paper provides morphological evidence for this view while arguing for separate extraction sites for the moved elements in temporal and conditional clauses.
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typology); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses.
Crucially, Akɔɔse wh-agreement encodes whether the extracted element originated above or below v. The distribution of wh-agreement morphology shows that the operator in central conditional clauses is extracted from a position above v (supporting Haegeman's (2010b) claim), the moved element in central temporal clauses originates below v (contra suggestions by Demirdache & Uribe-Extebarria (2004) and Haegeman (2007)), and there is no extraction in peripheral adverbial clauses (as argued in Haegeman 2007).
Wh-agreement provides compelling evidence for a movement analysis for both temporal and conditional central adverbial clauses. Due to its sensitivity to height of extraction, Akɔɔse lends insight into the question of where the moved elements originate, unlike languages like Irish (McCloskey 2001) where wh-agreement only registers the presence of movement.
Bantu languages, which are spoken throughout most of sub-Saharan Africa, permit wh-questions to b... more Bantu languages, which are spoken throughout most of sub-Saharan Africa, permit wh-questions to be constructed in multiple ways, including wh-in-situ, full wh-movement, and partial wh-movement. Shona, a Bantu language spoken by about 13 million people in Zimbabwe and Mozambique, allows all three of these types. In this dissertation, I conduct the first in-depth examination of Shona wh-questions, drawing on fifty hours of elicitation with a native speaker consultant to explore the derivational relationships among these strategies.
Wh-in-situ questions have received a wide variety of treatments in the syntactic literature, ranging from covert or disguised movement to postsyntactic binding of the wh-phrase by a silent question operator. In Bantu languages, wh-in-situ questions are often taken to be derived via a non-movement relation (e.g., Carstens 2005 for Kilega, Diercks 2010 for Lubukusu, Muriungi 2003 for Kîîtharaka, Sabel 2000 for Kikuyu and Duala, Sabel & Zeller 2006 for Zulu, Schneider-Zioga 2007 for Kinande), but alternatives have rarely been considered. I demonstrate how movement-based analyses that have been proposed for wh-in-situ in non-Bantu languages make the wrong predictions for Shona wh-in-situ, which lacks word order permutation, extraction marking, island effects, and intervention effects. These properties provide support for the traditional Bantuist view that the relation between the pronunciation site of an in-situ wh-phrase and its scopal position in the left periphery is not movement; I claim that in Shona it is unselective binding.
Many Bantu languages, including Shona, prohibit wh-phrases from appearing in the canonical preverbal subject position. Wasike (2007) demonstrates that this restriction applies to topicalized non-subjects as well as preverbal subjects. I replicate these results for Shona and argue that they cast doubt on Sabel & Zeller's (2006) attempt to characterize the ban with an appeal to improper movement. I argue instead that restrictions on the distribution of wh-in-situ in Bantu are tied to restrictions on the domain for focus licensing. This claim is further bolstered by an examination of crosslinguistic variation within Bantu with respect to whether the ban on in-situ preverbal wh-subjects applies in embedded clauses. I observe a previously unnoticed generalization: languages that universally ban in-situ preverbal wh-subjects (like Zulu) have immediately after the verb (IAV) focus effects; languages that do allow in-situ preverbal wh-subjects in embedded clauses (like Shona, Lubukusu, and Kîîtharaka) also lack IAV effects.
Full wh-movement in Shona gives rise to questions that bear a certain similarity to English wh-questions. However, using a range of diagnostics including extraction marking, island effects, reconstruction effects, and the distribution of temporal modifiers, I argue that what appears to be full wh-movement in Shona actually has a cleft structure: the wh-phrase moves to become the head of a relative clause, which is selected by a copula in the matrix clause. Just as in wh-in-situ, an ex-situ wh-phrase is pronounced lower than its scopal position, and the relation between these two positions is established via unselective binding. Additional evidence for this proposal comes from the sensitivity of partial wh-movement to island boundaries below but not above the pronunciation site of the wh-phrase, a pattern that has been predicted by previous analyses (e.g., Abels 2012, Sabel 2000, Sabel & Zeller 2006) but for which empirical support has been lacking until now. I therefore unify full and partial wh-movement under a single analysis for cleft-based wh-in-situ that involves a step of relativization (independently needed for relative clauses) and a step of unselective binding (independently needed for wh-in-situ).
This article explores geographical variation in a range of understudied dative constructions in A... more This article explores geographical variation in a range of understudied dative constructions in American English. It shows that these constructions are found primarily in the South and that they permit numerous syntactic variations and permutations. However, not all sentences and constructions have an equal status. In particular, the authors find that they lie on a continuum of markedness. More marked variants are judged acceptable by fewer speakers and have a more limited geographic distribution. And yet, even the most marked variants cannot be dismissed: the strong geographic nature of their distribution shows that they are a genuine part of the grammar of many speakers. Overall, this research contributes a more detailed picture of dative constructions in American English and a more nuanced picture of syntactic variation in Southern American English; moreover, the authors offer a novel approach to measuring geographical markedness in syntactic variation.
The Yale Grammatical Diversity Project approaches the empirical domain of North American English ... more The Yale Grammatical Diversity Project approaches the empirical domain of North American English from the perspective of generative microcomparative syntax. In addition to eliciting judgments from speakers of particular varieties, we also conduct large-scale surveys, map the results of those surveys geographically, conduct statistical tests taking geography and other social variables into account, and look for theoretically significant linguistic correlations. In all cases, we do this with the primary goal of understanding variation between speakers at the individual level. While our goals and methodologies are informed by our theoretical perspective, we expect that our work and results will be of interest to linguists working in other frameworks and even to the public more generally. This article outlines the goals and methodologies of the project and describes in broad strokes some of the results obtained so far, as well as some of the ways we have shared our findings with others, inside and outside academia.
Selected proceedings of the 44th Annual Conference on African Linguistics (ACAL), 2015
In many languages, A'-movement in the syntax is reflected by alternations in the morphology. Cate... more In many languages, A'-movement in the syntax is reflected by alternations in the morphology. Categorizing these extraction morphology phenomena on the basis of the morphological alternation involved produces three types, all of which are found in Bantu languages: replacement, deletion, and addition. This paper illustrates each of these types of extraction morphology with Bantu examples before demonstrating how the existence of the third type (addition extraction morphology or wh-agreement) in languages such as Shona, Duala, Lubukusu, and Akɔɔse proves problematic for an account of extraction morphology that relies on the deletion of features. Lahne (2009) proposes that extraction morphology is always less specific than canonical morphology; she implements this intuition using a mechanism called probe impoverishment, whereby a feature on the probe is deleted each time movement occurs. Here, cases of Bantu wh-agreement are shown to be counterexamples to Lahne's Generalization and to be more amenable to traditional analyses in which the additional morphology is the realization of the movement-triggering feature.
"Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological ... more "Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological arguments for a derivation of adverbial (subordinate) clauses that involves movement of an (often null) operator (see Haegeman 2010a for a review). This paper provides morphological evidence for this view while arguing for separate extraction sites for the moved elements in temporal and conditional clauses.
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typology); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses.
Crucially, Akɔɔse wh-agreement encodes whether the extracted element originated above or below v. The distribution of wh-agreement morphology shows that the operator in central conditional clauses is extracted from a position above v (supporting Haegeman's (2010b) claim), the moved element
in central temporal clauses originates below v (supporting Larson's (1987, 1990) position), and there is no extraction in peripheral adverbial clauses (as argued in Haegeman 2007).
Wh-agreement provides compelling evidence for a movement analysis for both temporal and conditional central adverbial clauses. Due to its sensitivity to height of extraction, Akɔɔse lends insight into the question of where the moved elements originate, unlike languages like Irish (McCloskey 2001) where wh-agreement only registers the presence of movement."
This paper examines phonological development and its relationship to input statistics. Using nove... more This paper examines phonological development and its relationship to input statistics. Using novel data from a longitudinal corpus of spontaneous child speech in Polish, we evaluate and compare the predictions of a variety of input-based phonotactic models for syllable structure acquisition. We find that many commonly examined input statistics can make dramatically different predictions, as do different assumptions about the representational units over which statistics are calculated. We find that development is sensitive to multiple abstract units of phonological representation, supporting a crucial role for feature-based generalization. We also identify departures between the predictions of the best phonotactic models and children’s production patterns that indicate input sensitivity alone cannot fully explain the developmental patterns. We discuss the role of universal markedness and phonetic difficulty and argue that a full explanation requires reference to these biases.
While the numeral systems of Australian languages are small, they are not uniform. In this articl... more While the numeral systems of Australian languages are small, they are not uniform. In this article we shed light on the extent of variation in small numeral systems by systematically surveying 189 languages from Pama- Nyungan and non-Pama-Nyungan families. We show that, contra previous assumptions, Australian languages vary extensively in the limits of their numeral system, in the ways in which smaller numerals may be combined to form larger ones, and in their ability to denote vague quantities like ‘some’ or ‘few’. We also investigate the etymological sources of words for numerals and find that few terms are reconstructible.
Numerals vary extensively across the world’s languages, ranging from no precise numeral terms to ... more Numerals vary extensively across the world’s languages, ranging from no precise numeral terms to practically infinite limits. Particularly of interest is the category of “small” or low-limit numeral systems; these are often associated with hunter-gatherer groups, but this connection has not yet been demonstrated by a systematic study. Here we present the results of a wide-scale survey of hunter-gatherer numerals. We compare these to agriculturalist languages in the same regions, and consider them against the broader typological backdrop of contemporary numeral systems in the world’s languages. We find that correlations with subsistence pattern are relatively weak, but that numeral trends are clearly areal.
In recent years, linguists have begun to increasingly rely on quantitative phylogenetic approache... more In recent years, linguists have begun to increasingly rely on quantitative phylogenetic approaches to examine language evolution. Some linguists have questioned the suitability of phylogenetic approaches on the grounds that linguistic evolution is largely reticulate due to extensive lateral transmission, or borrowing, among languages. The problem may be particularly pronounced in hunter-gatherer languages, where the conventional wisdom among many linguists is that lexical borrowing rates are so high that tree building approaches cannot provide meaningful insights into evolutionary processes. However, this claim has never been systematically evaluated, in large part because suitable data were unavailable. In addition, little is known about the subsistence, demographic, ecological, and social factors that might mediate variation in rates of borrowing among languages. Here, we evaluate these claims with a large sample of hunter-gatherer languages from three regions around the world. In this study, a list of 204 basic vocabulary items was collected for 122 hunter-gatherer and small-scale cultivator languages from three ecologically diverse case study areas: northern Australia, northwest Amazonia, and California and the Great Basin. Words were rigorously coded for etymological (inheritance) status, and loan rates were calculated. Loan rate variability was examined with respect to language area, subsistence mode, and population size, density, and mobility; these results were then compared to the sample of 41 primarily agriculturalist languages in [1]. Though loan levels varied both within and among regions, they were generally low in all regions (mean 5.06%, median 2.49%, and SD 7.56), despite substantial demographic, ecological, and social variation. Amazonian levels were uniformly very low, with no language exhibiting more than 4%. Rates were low but more variable in the other two study regions, in part because of several outlier languages where rates of borrowing were especially high. High mobility, prestige asymmetries, and language shift may contribute to the high rates in these outliers. No support was found for claims that hunter-gatherer languages borrow more than agriculturalist languages. These results debunk the myth of high borrowing in hunter-gatherer languages and suggest that the evolution of these languages is governed by the same type of rules as those operating in large-scale agriculturalist speech communities. The results also show that local factors are likely to be more critical than general processes in determining high (or low) loan rates.
Foreword to the Yale Law School's Pronouncing Dictionary of the Supreme Court of the United State... more Foreword to the Yale Law School's Pronouncing Dictionary of the Supreme Court of the United States, found at http://documents.law.yale.edu/pronouncing-dictionary.
This paper addresses the superficially similar focus constructions in Shona ([sna], Bantu, Zimbab... more This paper addresses the superficially similar focus constructions in Shona ([sna], Bantu, Zimbabwe) and Kîîtharaka ([thk], Bantu, Kenya). Using a topicalization test from Abels & Muriungi (2008), I argue that Shona has a biclausal cleft construction, while the Kîîtharaka construction is monoclausal. This asymmetry has consequences for the availability of the "intermediate" strategy for wh-question formation (Muriungi 2003).
This paper addresses partial wh-movement in Shona ([sna], Bantu, Zimbabwe), which is sensitive to... more This paper addresses partial wh-movement in Shona ([sna], Bantu, Zimbabwe), which is sensitive to islands below but not above the pronunciation site of the wh-phrase. This contrasts with partial wh-movement in languages such as Kîîtharaka (Muriungi 2003, Abels 2012) and Singaporean Malay (Cole & Hermon 1998), which is sensitive to islands both below and above the pronunciation site. I argue for a composite derivation of the Shona phenomenon: the wh-phrase moves overtly to its pronunciation site at an intermediate clause boundary, where it is unselectively bound by a null operator in the scopal position.
I show that in Shona partial wh-movement, the relation between the wh-phrase's base position and its pronunciation site patterns like full wh-movement in terms of island sensitivity, cleft structure, extraction marking, and reconstruction effects. On the other hand, the relation between the wh-phrase's pronunciation site and its scopal position displays the same lack of island sensitivity and lack of extraction marking exhibited by wh-in-situ in this language. Thus, Shona partial wh-movement can be reduced to a hybrid of full wh-movement and wh-in-situ. This composite derivation has been predicted to be possible (Sabel 2000, Abels 2012), but clear empirical support for it has been lacking until now.
This paper addresses partial wh-movement in Shona ([sna], Bantu, Zimbabwe), which is sensitive to... more This paper addresses partial wh-movement in Shona ([sna], Bantu, Zimbabwe), which is sensitive to islands below but not above the pronunciation site of the wh-word. I argue for a composite derivation of this phenomenon: the wh-word moves overtly to its pronunciation site at an intermediate clause boundary, where it is unselectively bound by a null operator in the scopal position. Thus, Shona partial wh-movement can be reduced to a hybrid of full wh-movement and wh-in-situ. This composite derivation has been predicted to be possible (Sabel 2000, Abels 2012), but clear empirical support for it has been lacking until now.
"This poster provides empirical support for the inheritance of non-φ-features by T from C. In the... more "This poster provides empirical support for the inheritance of non-φ-features by T from C. In the Bantu language Duala (Epée 1975, 1976a,b), A′-movement (focus movement, relativization, or wh-movement) of an element other than the matrix subject requires the presence of an invariant particle no following the highest verb in the clause where the A′-operator takes scope. The analysis presented here demonstrates that no is the morphological reflex on T of a feature associated with an A′-operator that takes scope in the CP domain. This featural link between C and T is straightforwardly derived under feature inheritance (Chomsky 2007, 2008)."
Many languages exhibit morphological reflexes of syntactic movement. Within Bantu, there are at l... more Many languages exhibit morphological reflexes of syntactic movement. Within Bantu, there are at least three types of this extraction morphology. In “anti-agreement,” extraction of a class 1 subject blocks canonical subject/verb agreement; in its place appears a featurally deficient agreement marker. The term "wh-agreement” covers two cases where verbal agreement remains intact but other morphological changes are triggered by syntactic movement. In the first, canonical morphophonology fails to appear in extraction contexts. In the second, canonical morphology remains constant across extraction and non-extraction contexts, but extraction contexts require the presence of additional morphology.
Lahne’s (2008) analysis of extraction morphology relies on the mechanism of “probe impoverishment,” where movement to the edge of a phase results in the deletion of one of the probe’s features. Probe impoverishment together with the Subset Principle (Halle 1997) predicts that “whenever a language shows different exponents in movement and non-movement contexts, when the marker appearing in movement contexts is less specific than the marker in non-movement contexts” (Lahne 2008: 9). This generalization is consistent with the facts of Bantu anti-agreement and the first type of wh-agreement.
However, here I illustrate the second type of wh-agreement with data from Kinande, Duala, and Akɔɔse, where extraction morphology does not compete with canonical morphology and is not found elsewhere in the language. In these cases, the extraction morphology cannot be the realization of an impoverished probe but rather the realization of the feature(s) driving movement.
Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological a... more Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological arguments for analyzing adverbial clauses as free relatives, involving movement of an (often null) operator (see Haegeman 2010 for a review). This paper provides morphological evidence for this view while arguing for separate extraction sites for the moved elements in temporal and conditional clauses.
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typological profile); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. The extraction of subjects is marked differently on the verb than the extraction of non-subjects. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses, suggesting that a free relative analysis of these clauses is warranted. Verbs in central temporal clauses take the non-subject extraction morphology, while verbs in central conditional clauses show the morphology found in subject extraction contexts. As would be expected from Haegeman's (2007) claim that peripheral adverbial clauses do not involve movement, these clauses show no extraction morphology in Akɔɔse.
In order to probe the question of where the relativizing operators in central temporal and conditional clauses originate, this paper offers a novel syntactic account for the morphological patterns found in Akɔɔse wh-agreement. According to this account, the temporal operator must originate within the VP layer, while the locus of extraction for the operator involved in central conditional clauses must be higher.
Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological a... more Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological arguments for a derivation of adverbial (subordinate) clauses that involves movement of an (often null) operator (see Haegeman 2010a for a review). This paper provides morphological evidence for this view while arguing for separate extraction sites for the moved elements in temporal and conditional clauses.
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typology); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. The extraction of subjects is marked differently on the verb than the extraction of non-subjects. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses. Verbs in central temporal clauses take the non-subject extraction morphology, while verbs in central conditional clauses show the morphology found in subject extraction contexts.
This paper offers a syntactic account for the patterns in Akɔɔse wh-agreement morphology and uses this account to refine our knowledge of the extraction sites for the operators in central temporal and conditional clauses, as well as to argue that peripheral adverbial clauses do not involve extraction (supporting Haegeman 2007).
Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological a... more Beginning with Geis (1970), several authors have provided syntactic, semantic, and etymological arguments for a derivation of adverbial (subordinate) clauses that involves movement of an (often null) operator (see Haegeman 2010a for a review). This paper provides morphological evidence for this view while arguing for separate extraction sites for the moved elements in temporal and conditional clauses.
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typology); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses.
Crucially, Akɔɔse wh-agreement encodes whether the extracted element originated above or below v. The distribution of wh-agreement morphology shows that the operator in central conditional clauses is extracted from a position above v (supporting Haegeman's (2010b) claim), the moved element in central temporal clauses originates below v (contra suggestions by Demirdache & Uribe-Extebarria (2004) and Haegeman (2007)), and there is no extraction in peripheral adverbial clauses (as argued in Haegeman 2007).
Wh-agreement provides compelling evidence for a movement analysis for both temporal and conditional central adverbial clauses. Due to its sensitivity to height of extraction, Akɔɔse lends insight into the question of where the moved elements originate, unlike languages like Irish (McCloskey 2001) where wh-agreement only registers the presence of movement.
This paper presents an analysis of Warlpiri progressive front vowel harmony using McCarthy's (200... more This paper presents an analysis of Warlpiri progressive front vowel harmony using McCarthy's (2009, 2010) Serial Harmony, a theory of feature assimilation that operates over autosegmental representations within the architecture of Harmonic Serialism, a serial version of Optimality Theory. I show that this Serial Harmony analysis is superior to previous OT analyses of Warlpiri vowel harmony because of its ability to model the myopic spreading that characterizes partial harmony, where the harmonizing feature is successful in spreading until it reaches a blocker. Additionally, I consider the interaction of labial attraction and progressive front harmony and argue that labial attraction may also be implemented in Serial Harmony, eliminating the need for a separate labial attraction constraint. I demonstrate that this approach to labial attraction highlights a structural difference between two types of Pi (labial consonant + [i]) sequences. Under this view, labial blocking of progressive front harmony is not a derived environment blocking effect, but an absolute blocking effect arising from a ban on segments that are both [front] and [labial]. A single constraint, *LabialFront, enforces this ban, placing limits on both labial attraction and progressive front vowel harmony.
We present novel findings on the acquisition of syllable structure in Polish and explore abilitie... more We present novel findings on the acquisition of syllable structure in Polish and explore abilities of various formulations of frequency to account for the acquisition results. We analyze the spontaneous productions of four children, examining accuracy of syllable margins at several levels of representation (CV, sonority, segment). We use mixed-effects modeling to examine differences in accuracy after controlling for effects of word frequency, age, subject, and word length. We also examine differences in acquisition between children. We then compare the observed differences in accuracy to various frequency measures calculated from the child-directed speech to these children. We consider both type and token frequencies calculated at each representational level as well as versions of these measures calculated from speech directed at each individual child. Our results suggest that while frequency can be a good predictor of accuracy, no single frequency measure can fully explain accuracy differences across children or representational levels.
Australian languages tend to have restricted numeral systems, stereotypically “one, two, three, m... more Australian languages tend to have restricted numeral systems, stereotypically “one, two, three, many.” The variability of small systems, however, has been understudied. We examine properties of numeral systems and etymology of numeral forms to shed light on the processes involved in the creation and maintenance of these systems. We find counterexamples to several claims in the literature and discuss data which has been excluded from previous studies. We collected data on numerals and quantifiers from 121 Australian hunter-gatherer languages. We survey limits, composition, base, vagueness, and etymologies of numerals. Secondary counting systems are also discussed. Given the current interest in the implication of small numeral systems for cultural and cognitive constraints on language, a review of such systems is timely. These small systems have an important place in numeral typologies.
Uploads
Wh-in-situ questions have received a wide variety of treatments in the syntactic literature, ranging from covert or disguised movement to postsyntactic binding of the wh-phrase by a silent question operator. In Bantu languages, wh-in-situ questions are often taken to be derived via a non-movement relation (e.g., Carstens 2005 for Kilega, Diercks 2010 for Lubukusu, Muriungi 2003 for Kîîtharaka, Sabel 2000 for Kikuyu and Duala, Sabel & Zeller 2006 for Zulu, Schneider-Zioga 2007 for Kinande), but alternatives have rarely been considered. I demonstrate how movement-based analyses that have been proposed for wh-in-situ in non-Bantu languages make the wrong predictions for Shona wh-in-situ, which lacks word order permutation, extraction marking, island effects, and intervention effects. These properties provide support for the traditional Bantuist view that the relation between the pronunciation site of an in-situ wh-phrase and its scopal position in the left periphery is not movement; I claim that in Shona it is unselective binding.
Many Bantu languages, including Shona, prohibit wh-phrases from appearing in the canonical preverbal subject position. Wasike (2007) demonstrates that this restriction applies to topicalized non-subjects as well as preverbal subjects. I replicate these results for Shona and argue that they cast doubt on Sabel & Zeller's (2006) attempt to characterize the ban with an appeal to improper movement. I argue instead that restrictions on the distribution of wh-in-situ in Bantu are tied to restrictions on the domain for focus licensing. This claim is further bolstered by an examination of crosslinguistic variation within Bantu with respect to whether the ban on in-situ preverbal wh-subjects applies in embedded clauses. I observe a previously unnoticed generalization: languages that universally ban in-situ preverbal wh-subjects (like Zulu) have immediately after the verb (IAV) focus effects; languages that do allow in-situ preverbal wh-subjects in embedded clauses (like Shona, Lubukusu, and Kîîtharaka) also lack IAV effects.
Full wh-movement in Shona gives rise to questions that bear a certain similarity to English wh-questions. However, using a range of diagnostics including extraction marking, island effects, reconstruction effects, and the distribution of temporal modifiers, I argue that what appears to be full wh-movement in Shona actually has a cleft structure: the wh-phrase moves to become the head of a relative clause, which is selected by a copula in the matrix clause. Just as in wh-in-situ, an ex-situ wh-phrase is pronounced lower than its scopal position, and the relation between these two positions is established via unselective binding. Additional evidence for this proposal comes from the sensitivity of partial wh-movement to island boundaries below but not above the pronunciation site of the wh-phrase, a pattern that has been predicted by previous analyses (e.g., Abels 2012, Sabel 2000, Sabel & Zeller 2006) but for which empirical support has been lacking until now. I therefore unify full and partial wh-movement under a single analysis for cleft-based wh-in-situ that involves a step of relativization (independently needed for relative clauses) and a step of unselective binding (independently needed for wh-in-situ).
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typology); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses.
Crucially, Akɔɔse wh-agreement encodes whether the extracted element originated above or below v. The distribution of wh-agreement morphology shows that the operator in central conditional clauses is extracted from a position above v (supporting Haegeman's (2010b) claim), the moved element
in central temporal clauses originates below v (supporting Larson's (1987, 1990) position), and there is no extraction in peripheral adverbial clauses (as argued in Haegeman 2007).
Wh-agreement provides compelling evidence for a movement analysis for both temporal and conditional central adverbial clauses. Due to its sensitivity to height of extraction, Akɔɔse lends insight into the question of where the moved elements originate, unlike languages like Irish (McCloskey 2001) where wh-agreement only registers the presence of movement."
I show that in Shona partial wh-movement, the relation between the wh-phrase's base position and its pronunciation site patterns like full wh-movement in terms of island sensitivity, cleft structure, extraction marking, and reconstruction effects. On the other hand, the relation between the wh-phrase's pronunciation site and its scopal position displays the same lack of island sensitivity and lack of extraction marking exhibited by wh-in-situ in this language. Thus, Shona partial wh-movement can be reduced to a hybrid of full wh-movement and wh-in-situ. This composite derivation has been predicted to be possible (Sabel 2000, Abels 2012), but clear empirical support for it has been lacking until now.
Lahne’s (2008) analysis of extraction morphology relies on the mechanism of “probe impoverishment,” where movement to the edge of a phase results in the deletion of one of the probe’s features. Probe impoverishment together with the Subset Principle (Halle 1997) predicts that “whenever a language shows different exponents in movement and non-movement contexts, when the marker appearing in movement contexts is less specific than the marker in non-movement contexts” (Lahne 2008: 9). This generalization is consistent with the facts of Bantu anti-agreement and the first type of wh-agreement.
However, here I illustrate the second type of wh-agreement with data from Kinande, Duala, and Akɔɔse, where extraction morphology does not compete with canonical morphology and is not found elsewhere in the language. In these cases, the extraction morphology cannot be the realization of an impoverished probe but rather the realization of the feature(s) driving movement.
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typological profile); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. The extraction of subjects is marked differently on the verb than the extraction of non-subjects. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses, suggesting that a free relative analysis of these clauses is warranted. Verbs in central temporal clauses take the non-subject extraction morphology, while verbs in central conditional clauses show the morphology found in subject extraction contexts. As would be expected from Haegeman's (2007) claim that peripheral adverbial clauses do not involve movement, these clauses show no extraction morphology in Akɔɔse.
In order to probe the question of where the relativizing operators in central temporal and conditional clauses originate, this paper offers a novel syntactic account for the morphological patterns found in Akɔɔse wh-agreement. According to this account, the temporal operator must originate within the VP layer, while the locus of extraction for the operator involved in central conditional clauses must be higher.
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typology); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. The extraction of subjects is marked differently on the verb than the extraction of non-subjects. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses. Verbs in central temporal clauses take the non-subject extraction morphology, while verbs in central conditional clauses show the morphology found in subject extraction contexts.
This paper offers a syntactic account for the patterns in Akɔɔse wh-agreement morphology and uses this account to refine our knowledge of the extraction sites for the operators in central temporal and conditional clauses, as well as to argue that peripheral adverbial clauses do not involve extraction (supporting Haegeman 2007).
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typology); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses.
Crucially, Akɔɔse wh-agreement encodes whether the extracted element originated above or below v. The distribution of wh-agreement morphology shows that the operator in central conditional clauses is extracted from a position above v (supporting Haegeman's (2010b) claim), the moved element in central temporal clauses originates below v (contra suggestions by Demirdache & Uribe-Extebarria (2004) and Haegeman (2007)), and there is no extraction in peripheral adverbial clauses (as argued in Haegeman 2007).
Wh-agreement provides compelling evidence for a movement analysis for both temporal and conditional central adverbial clauses. Due to its sensitivity to height of extraction, Akɔɔse lends insight into the question of where the moved elements originate, unlike languages like Irish (McCloskey 2001) where wh-agreement only registers the presence of movement.
Wh-in-situ questions have received a wide variety of treatments in the syntactic literature, ranging from covert or disguised movement to postsyntactic binding of the wh-phrase by a silent question operator. In Bantu languages, wh-in-situ questions are often taken to be derived via a non-movement relation (e.g., Carstens 2005 for Kilega, Diercks 2010 for Lubukusu, Muriungi 2003 for Kîîtharaka, Sabel 2000 for Kikuyu and Duala, Sabel & Zeller 2006 for Zulu, Schneider-Zioga 2007 for Kinande), but alternatives have rarely been considered. I demonstrate how movement-based analyses that have been proposed for wh-in-situ in non-Bantu languages make the wrong predictions for Shona wh-in-situ, which lacks word order permutation, extraction marking, island effects, and intervention effects. These properties provide support for the traditional Bantuist view that the relation between the pronunciation site of an in-situ wh-phrase and its scopal position in the left periphery is not movement; I claim that in Shona it is unselective binding.
Many Bantu languages, including Shona, prohibit wh-phrases from appearing in the canonical preverbal subject position. Wasike (2007) demonstrates that this restriction applies to topicalized non-subjects as well as preverbal subjects. I replicate these results for Shona and argue that they cast doubt on Sabel & Zeller's (2006) attempt to characterize the ban with an appeal to improper movement. I argue instead that restrictions on the distribution of wh-in-situ in Bantu are tied to restrictions on the domain for focus licensing. This claim is further bolstered by an examination of crosslinguistic variation within Bantu with respect to whether the ban on in-situ preverbal wh-subjects applies in embedded clauses. I observe a previously unnoticed generalization: languages that universally ban in-situ preverbal wh-subjects (like Zulu) have immediately after the verb (IAV) focus effects; languages that do allow in-situ preverbal wh-subjects in embedded clauses (like Shona, Lubukusu, and Kîîtharaka) also lack IAV effects.
Full wh-movement in Shona gives rise to questions that bear a certain similarity to English wh-questions. However, using a range of diagnostics including extraction marking, island effects, reconstruction effects, and the distribution of temporal modifiers, I argue that what appears to be full wh-movement in Shona actually has a cleft structure: the wh-phrase moves to become the head of a relative clause, which is selected by a copula in the matrix clause. Just as in wh-in-situ, an ex-situ wh-phrase is pronounced lower than its scopal position, and the relation between these two positions is established via unselective binding. Additional evidence for this proposal comes from the sensitivity of partial wh-movement to island boundaries below but not above the pronunciation site of the wh-phrase, a pattern that has been predicted by previous analyses (e.g., Abels 2012, Sabel 2000, Sabel & Zeller 2006) but for which empirical support has been lacking until now. I therefore unify full and partial wh-movement under a single analysis for cleft-based wh-in-situ that involves a step of relativization (independently needed for relative clauses) and a step of unselective binding (independently needed for wh-in-situ).
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typology); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses.
Crucially, Akɔɔse wh-agreement encodes whether the extracted element originated above or below v. The distribution of wh-agreement morphology shows that the operator in central conditional clauses is extracted from a position above v (supporting Haegeman's (2010b) claim), the moved element
in central temporal clauses originates below v (supporting Larson's (1987, 1990) position), and there is no extraction in peripheral adverbial clauses (as argued in Haegeman 2007).
Wh-agreement provides compelling evidence for a movement analysis for both temporal and conditional central adverbial clauses. Due to its sensitivity to height of extraction, Akɔɔse lends insight into the question of where the moved elements originate, unlike languages like Irish (McCloskey 2001) where wh-agreement only registers the presence of movement."
I show that in Shona partial wh-movement, the relation between the wh-phrase's base position and its pronunciation site patterns like full wh-movement in terms of island sensitivity, cleft structure, extraction marking, and reconstruction effects. On the other hand, the relation between the wh-phrase's pronunciation site and its scopal position displays the same lack of island sensitivity and lack of extraction marking exhibited by wh-in-situ in this language. Thus, Shona partial wh-movement can be reduced to a hybrid of full wh-movement and wh-in-situ. This composite derivation has been predicted to be possible (Sabel 2000, Abels 2012), but clear empirical support for it has been lacking until now.
Lahne’s (2008) analysis of extraction morphology relies on the mechanism of “probe impoverishment,” where movement to the edge of a phase results in the deletion of one of the probe’s features. Probe impoverishment together with the Subset Principle (Halle 1997) predicts that “whenever a language shows different exponents in movement and non-movement contexts, when the marker appearing in movement contexts is less specific than the marker in non-movement contexts” (Lahne 2008: 9). This generalization is consistent with the facts of Bantu anti-agreement and the first type of wh-agreement.
However, here I illustrate the second type of wh-agreement with data from Kinande, Duala, and Akɔɔse, where extraction morphology does not compete with canonical morphology and is not found elsewhere in the language. In these cases, the extraction morphology cannot be the realization of an impoverished probe but rather the realization of the feature(s) driving movement.
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typological profile); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. The extraction of subjects is marked differently on the verb than the extraction of non-subjects. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses, suggesting that a free relative analysis of these clauses is warranted. Verbs in central temporal clauses take the non-subject extraction morphology, while verbs in central conditional clauses show the morphology found in subject extraction contexts. As would be expected from Haegeman's (2007) claim that peripheral adverbial clauses do not involve movement, these clauses show no extraction morphology in Akɔɔse.
In order to probe the question of where the relativizing operators in central temporal and conditional clauses originate, this paper offers a novel syntactic account for the morphological patterns found in Akɔɔse wh-agreement. According to this account, the temporal operator must originate within the VP layer, while the locus of extraction for the operator involved in central conditional clauses must be higher.
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typology); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. The extraction of subjects is marked differently on the verb than the extraction of non-subjects. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses. Verbs in central temporal clauses take the non-subject extraction morphology, while verbs in central conditional clauses show the morphology found in subject extraction contexts.
This paper offers a syntactic account for the patterns in Akɔɔse wh-agreement morphology and uses this account to refine our knowledge of the extraction sites for the operators in central temporal and conditional clauses, as well as to argue that peripheral adverbial clauses do not involve extraction (supporting Haegeman 2007).
The Bantu language Akɔɔse (Hedinger 2008) exhibits wh-agreement (see Reintges, LeSourd, & Chung 2006 for a typology); that is, it marks its verbs with respect to whether an element has been extracted to the left periphery. This extraction marking occurs not only in canonical wh-movement contexts (Chomsky 1977), such as constituent questions, relative clauses, cleft questions, and topicalization, but also in temporal and conditional adverbial clauses.
Crucially, Akɔɔse wh-agreement encodes whether the extracted element originated above or below v. The distribution of wh-agreement morphology shows that the operator in central conditional clauses is extracted from a position above v (supporting Haegeman's (2010b) claim), the moved element in central temporal clauses originates below v (contra suggestions by Demirdache & Uribe-Extebarria (2004) and Haegeman (2007)), and there is no extraction in peripheral adverbial clauses (as argued in Haegeman 2007).
Wh-agreement provides compelling evidence for a movement analysis for both temporal and conditional central adverbial clauses. Due to its sensitivity to height of extraction, Akɔɔse lends insight into the question of where the moved elements originate, unlike languages like Irish (McCloskey 2001) where wh-agreement only registers the presence of movement.