The Circle is a weekly forum dedicated to discussion of the descriptive, experimental, and theoretical study of syntax and semantics, featuring presentations of ongoing research by members of the Berkeley Linguistics Department and other departments, as well as discussion of previously published works.


Fridays, 3:00-4:30pm


1303 Dwinelle Hall


Virginia Dawson

Erik Hans Maier


University of California, Berkeley
Department of Linguistics


16 december
No Meeting (Finals)


9 december
No Meeting (RRR Week)

2 december
Lev Michael (UC Berkeley)
New insights into the diachrony of negation from the evolution of Arawakan privatives

Since the discovery of Jespersen's cycles (Jespersen 1917, Meillet 1911), the study of the sources of standard negation markers and their diachronic development has been a central topic in the field of historical morphosyntax (van der Auwera 2009, Croft 1991, Givón 1978, Horn 1989). In this talk I employ comparative data from 28 Arawakan languages to argue for a novel source for standard negation, namely privative derivational morphology, and describe the two main trajectories by which Arawakan privatives extended their function from privative derivation towards standard negation.

In its original function (i.e. in Proto-Arawakan), the privative prefix derived negative stative verbs stems from nouns (‘child’ > ‘be childless’). Independently in different branches of the family, however, the distribution of the privative broadened in two different ways. In the first trajectory, the distribution of privative extended from underived nouns to deverbal nominalizations, becoming the normal means of negating the nominalized verbs of subordinate clauses. In a small number of languages, these subordinate clause constructions underwent insubordination (Evans 2007), with the result that the erstwhile privative became normal main clause negation, i.e. became standard negation. Not only is this specific source for standard negation novel, but as far as I am aware, this is the first substantiated case of derivational morphology serving as the source of standard negation.

In the second trajectory, the distribution of the privative extended not via subordinate clauses, but via main clauses. The privative spread from its purely denominal function to also negating stative verbs, maintaining the restriction that the negated form exhibit stative lexical aspect, but not requiring the root to which the privative attaches be nominal. In a small number of languages, this logic extended to active verb roots, such that the privative came to express stative/habitual negation in main clauses for verb roots of all lexical aspects (e.g. ‘He does not sing.’ (as a matter of custom or habit)). Via this trajectory, the privative did not quite break through to becoming standard negation (because of its habitual semantics) but does serve as a means of main clause negation.

I conclude with with a brief overview of evidence in other languages families for similar trajectories, including Tupí-Guaraní family of South America and languages of Australia, which suggest that these process may not be unusual in language families that exhibit verbalizing privatives.

25 november
No Meeting (Thanksgiving)

18 november
Rose-Marie Déchaine (University of British Columbia) and H.C. Wolfart (University of Manitoba)
What Plains Cree teaches us about negation

Plains Cree (PC, central Algonquian) has two forms of negation, nama and êkâ, which occur at the left edge of the clause, and whose distribution is conditioned by a complex interplay of syntactic and semantic factors. While êkâ NEG occurs in averidical root clauses and embedded clauses, nama NEG is the elsewhere case, occurring with veridical root clauses, constituent negation, negative predicates, and negative answers. This converges with Lacombe’s (1874) findings, but challenges Bloomfield’s (1928) treatment. More broadly, the PC data bears on recent proposals concerning the syntax, semantics, pragmatics, and formal typology of negation. Relative to the syntax of NEG, the placement of PC NEG at the left-edge of the CP challenges Horn’s (1989) claim that natural language negation is always and only a predicate operator. Accepting the postulate that scope reduces to c-command (Szabolcsi 1997), the PC data challenges proposals that dispense with c-command (Barker 2012, Bruening 2014); in PC, the status of NEG as a c-commanding operator is confirmed by tense-marking, scopal interaction with indefinites, cross-clausal scope, and CP-ellipsis. Relative to the semantics of NEG, PC supports the view that NEG is conditioned by veridicality (Giannikidou 2011), as confirmed by the interaction of NEG with clause-typing. (NEG also diagnoses a previously un-noticed contrast between HUMAN versus NON- HUMAN indefinites in PC.) Relative to the pragmatics of NEG, the PC data supports the view that the contribution of NEG is conditioned by the question-under-discussion (QUD, Tian et al. 2016). This correctly predicts a partition between presentative and clefted clauses, and converges with Seeliger’s (2016) analysis of NEG as a FALSUM (Repp 2013) operator. Relative to the formal typology of NEG, the PC data shows NEG is positioned high in the clausal domain, converging with the findings of Christiansen (1986) and Brandtler (2012) on Swedish, but challenging analyses that position NEG low (De Clercq 2013, Collins & Postal 2014). Sells (2000) argues that high NEG yields “double negation”, while low NEG yields “negative concord”, and the PC data supports this view.

11 november
No Meeting (Veteran's Day)

4 november
Peter Jenks
Hyperraising from pseudorelatives in Moro

Certain verbs in Moro (Kordofanian; Sudan) trigger obligatory raising to subject and raising to object from a lower finite clause, an instance of hyperraising (Ura 1994). While hyperraising is attested in many languages, the Moro case is somewhat different in that the finite embedded clause which launches the raised argument is a pseudorelative (Cinque 1992), an apparent relative clause. I show that Moro pseudorelatives are actually structurally reduced complements — FinPs — and suggest that size is an important factor in allowing hyperraising. The other ingredient for hyperraising is case-related: raised nominal must be able to receive case, meaning that the position which is being moved from must be case-defective in a sense I make clear. I show that Moro's status as a dependent case language (Jenks & Sande 2016) means that Moro subjects are always case-defective and hence potential candidates for overt A-movement.

28 october
Emily Clem (UC Berkeley)
Decomposing ergativity: Evidence for the featural complexity of ergative case

The question of how ergative case is assigned has been a topic of debate in the literature with two main views: 1) ergative is an inherent case assigned by a transitive v to an agent, 2) ergative is a dependent case assigned to a DP that c-commands another DP within a case domain. What both of these views have in common is the idea that ergative case is one feature assigned on the basis of one syntactic relationship, be it selectional or configurational. In this talk, I argue that ergative case is not exponence of a single feature but rather of a complex of features occurring on a single DP. I examine data from my fieldwork on Amahuaca, an endangered Panoan language of Peru, in which ergative case marking is sensitive to word order. I demonstrate that in situ subjects are not marked ergative while moved subjects are. Following Deal (2010), I argue that ergative case expones features that are received by a DP that has Agreed with both a v that has already Agreed and with T. This analysis is able to account for the sensitivity of Amahuaca ergative case marking to the syntactic position of the subject, while also yielding insights about the switch reference agreement system of the language and the variable realization of nominative case, both of which support the case-as-a-feature-complex view. Crucially, neither an inherent nor dependent analysis of ergative case can account for the distribution of ergative marked subjects in Amahuaca. Instead, the account I pursue captures the generalization of the inherent case literature that ergativity depends on a relationship with a transitive v and the generalization of the dependent case literature that ergativity depends on the presence of another DP in the clause, while also incorporating the importance of subjecthood and the relationship between the ergative DP and T. This work has implications for how we view ergativity more broadly, as it demonstrates that ergativity is not a unified feature but expresses simultaneously several syntactic relationships, yielding the cross-linguistic diversity that we find in patterns of ergative case marking.

21 october
No Meeting (CUSP 9)

14 october
No Meeting (NELS 47)

7 october
NELS Practice Talks
Hannah Sande (UC Berkeley): Process morphology in a realizational theory
Virginia Dawson (UC Berkeley): Optimal clitic placement in Tiwa
Emily Clem (UC Berkeley): Two types of binding: Evidence from Tswefap pronominals
Peter Jenks and Hannah Sande (UC Berkeley): Case and caselessness in Moro

30 september
Lelia Glass (Stanford)
An underspecified semantics for (non)distributive predication

A predicate is understood distributively if it is understood to be separately true of each member of a plural subject: Alice and Bob smiled is generally understood distributively, in that we can infer from it that Alice smiled and that Bob smiled. A predicate is understood nondistributively ("collectively") if it is not understood to be separately true of each member of a plural subject: Alice and Bob met generally does not allow us to infer that Alice met or that Bob met. Some predicates can be easily understood in both ways: Alice and Bob lifted the table could mean that they each did so, or did so only jointly.

This talk asks, what should our semantic representation of (non)distributive predication consist of? Equally importantly, what should it not consist of -- what should be explained pragmatically rather than semantically?

Inspired by the principle (Grice 1978) that anything that should be explained pragmatically rather than semantically should be, I propose a highly underspecified semantics for distributive predication, moving many distinctions from the literature out of the semantics. Thanks to an assumption known as lexical cumulativity (Lasersohn 1989, Krifka 1992, Kratzer 2007), I argue that predicates are underspecified between distributive and nondistributive understandings, and that pragmatics comes in to narrow down this underspecification. I argue that this analysis gives world knowledge the starring role it deserves in explaining which predicates can be understood which way(s) and why.

23 september
Filippa Lindahl (University of Gothenburg)
Relative clauses, small clauses, and extraction in Swedish

The mainland Scandinavian languages are known to allow extraction from relative clauses (ERC) (e.g. Erteschik-Shir 1973; Koch Christensen 1982; Engdahl 1997). A typical example from Swedish is given in (1).



`There are many people who speak that language.' (Engdahl 1997, p. 59)

One recent proposal (Kush 2011, Kush, Omaki & Hornstein 2013) argues that the relative clause-like constituents in examples like (1) are not real relative clauses but small clauses. Others (Lindahl 2014, Christensen & Nyvad 2014, Vikner 2016, Nyvad, Christensen & Vikner 2016) argue that examples like (1) involve regular restrictive RCs, and that movement out of RCs in the mainland Scandinavian languages proceeds via an extra specifier in the CP-domain.

In this talk I take a closer look at the RCs in examples like (1) and other ERC sentences, taking as a starting point a collection of 250 extraction sentences from conversation, radio, and written Swedish. Specifically I investigate predictions made by the small clause hypothesis about possible matrix predicates and extraction from non-subject RCs, and provide an argument inspired by Keenan’s (1987) account of existential sentences that the RCs in several ERC sentences attaches inside the DP. This and other observations suggest that the RC in ERC can be a regular restrictive RC.

16 september
No Meeting (Workshop on the Status of Head Movement in Linguistic Theory)

9 september
Round Robin

The summer is over and the Round Robin is returning to roost. Come join us for an informal discussion where any and all can have up to 5-10 minutes to tell us about their summer elicitation or research, ask questions, present new ideas, preliminary analyses, or interesting data, and get valuable answers, feedback, and commiseration from your colleagues. Names will be drawn from a hat to ensure the turns are random.

2 september
Jed Pizarro-Guevara (UC Santa Cruz)
Tagalog voice/agreement morphology and its role in processing filler-gap dependencies

The present study investigates the role of Tagalog voice/agreement morphology (VAM) in real-time sentence processing. Because VAM packages information about the subject's thematic relation and structural position, I hypothesize that it can sharpen the comprehenders' predictions by allowing them to project the structure of vP and thus to guide their interpretation of incoming linguistic material. Using a Stops-Making-Sense task, I tested whether verbs inflected with VAM (i.e., verbs with -um- or -in-) were linked to their arguments any faster than controls without visible inflection (i.e., verbs in the iterative or recent perfective aspect). Results indicated that verbs inflected with -um- facilitated the comprehension of agent wh-questions, but verbs inflected with -in- did not facilitate the comprehension of patient wh-questions. Despite this apparent asymmetry, I maintain that VAM does permit comprehenders to predictively extend their syntactic representations. However, whether it immediately feeds interpretation is mediated by other factors, such as the availability of alternative parses. I conjecture that the syntax of argument wh-questions in Tagalog affects the time-course of parsing, such that they are interpreted less “actively” than comparable constructions in English.

26 august
No Meeting