university of california, berkeley


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


Due to the COVID-19 mitigation efforts, meetings will be held remotely over zoom. Recurring link.


Maddy Bossi

Tyler Lemon


University of California, Berkeley
Department of Linguistics

Spring 2021


3 may
SALT practice talks
Note the Monday date! The meeting time will be 2pm.

Rebecca Jarvis (UC Berkeley)
English non-manner how-clauses as answers to deficient questions

The abstract for this talk can be found here.

30 april
Maksymilian Dąbkowski (UC Berkeley)
Pied-piping by Cyclic Agree. In defense of feature percolation

The abstract for this talk can be found here.

23 april
Vera Hohaus (Manchester)
Composing meanings with alternatives: A case of no variation?

The abstract for this talk can be found here.

16 april
Note the special meeting time of 3:30pm-4:30pm!

Workshopping a manuscript by Emily Clem (UCSD):
Toward a unified account of inverse marking and the Person-Case Constraint

Please email Tyler Lemon or Maddy Bossi to get a copy of the manuscript!

9 april

2 april

WCCFL practice talks (round 2)
Wesley dos Santos (UC Berkeley): Long head movement is A-bar movement: the case study of Kawahíva
Emily Drummond (UC Berkeley): Maintaining syntactic identity under sluicing: Pseudoclefts and voice (mis)matches
Rebecca Jarvis (UC Berkeley): Presuppositionality and syntactic nominalization in finite clausal complements

25 march
Note the Thursday date and special meeting time of 4pm!

WCCFL practice talks (round 1)
Maksymilian Dąbkowski (UC Berkeley): Laryngeal feet in A’ingae: Implications for metrical theory
Tyler Lemon (UC Berkeley): Low nominative agreement in Uab Meto

19 march
Round Robin!

12 march
Rebecca Jarvis (UC Berkeley)
Presuppositionality and syntactic category in finite complement clauses

Though a traditional analysis of finite complement clauses treats them as CPs, numerous accounts have posited at least some nominal structure above the CP. One recent account along these lines comes from Kastner (2015), who argues for a bijection between presuppositionality and syntactic category. On this account, all and only presuppositional clauses, across languages, are DPs. By contrast, non-presuppositional clauses are analyzed as CPs with no additional nominal structure.

In this talk, I argue against such a bijection. Empirically, I suggest that our view of English finite complement clauses should be expanded beyond just that-clauses. With this broadened view of English complement clauses, I incorporate evidence both from English and cross-linguistically to argue that presuppositionality is neither necessary nor sufficient for nominalization in finite complement clauses.

5 march
Michael Diercks (Pomona)
Bukusu object marking: at the interface of pragmatics and syntax
(joint work with Justine Sikuku (Moi University))

This talk describes a complex system of object marking (i.e. object clitics) in Bukusu that interacts with pragmatics/discourse semantics (of focus/givenness). We show that object markers can co-occur with (i.e. double) in situ lexical objects, but that this possibility is linked with 1) a discourse-given interpretation of the object and, necessarily, 2) the presence of a focused element inside the verb phrase. We propose that Bukusu (doubling) object markers arise via Agree relations generated by phi-features on a functional projection at the edge of vP, but that this focus projection also contains a Roothian-style focus operator that introduces focus presuppositions about its complement (i.e. vP). Therefore, rather than object marker (OM) doubling being driven by properties such as case, specificity, or linked with object shift as has been claimed for many clitic doubling languages, instead it appears that the closest empirical correlate of Bukusu OM-doubling is the semantics of focus and givenness in (for example) English intonation. This is especially evident in instances where OM-doubling can target a structurally lower object in a multiple object construction: such non-locality is possible, but restricts the focus domain of the sentence to that lower object itself, in rather stark contrast to the typical situation where the entire vP is the focus domain. The result will hopefully be an interesting empirical look at a novel system of OM-doubling, as well as raising important theoretical questions about what kinds of pragmatic information can be represented grammatically in the syntax.

If there is time, we can explore an additional empirical goody: Bukusu OM-doubling appears to be largely dependent on focus, but focus is not sufficient to trigger OM-doubling. Instead we propose that Bukusu OM-doubling is in fact a mirative focus construction, relying on recent analyses of Mirative Fronting in Romance (e.g. Bianchi et al 2016, Cruschina 2019).

26 february
Hadas Kotek (MIT)
Top-down derivations: Flipping syntax on its head
(joint work with Bob Frank (Yale))

The past few decades of Minimalist syntax have seen the development of grammatical theories that build syntactic structure in a bottom-up fashion. Such directionality derives from the formulation of the structure building operation, Merge, which combines syntactic objects into a new one that contains them. As sometimes noted, derivations could proceed in a top-down manner instead. It has, however, proven difficult to argue empirically for one approach over the other (but cf. Phillips 2003; den Dikken 2018).

In this talk, we survey a class of phenomena which point to a top-down approach. We present phenomena where properties of a higher structure condition elements or operations in a lower domain. The opposite pattern is, we claim, much rarer and more restricted. This follows from a top-down approach, given the relative time at which elements are integrated into the structure. Bottom-up derivations, in contrast, naturally lead to opposite predictions. In all cases, our goal is not to show that bottom-up theories cannot explain the facts, but rather that a top-down account is more natural and less stipulative.

19 february
Andrew McKenzie (Kansas)
Interpreting verb incorporation in Kiowa

This presentation dives into the semantics of verb incorporation (VI), which has largely been set aside in favor of understanding noun incorporation. The semantics of VI is complicated by three major factors: The variety of verb-incorporation structures, the weak compositionality of word-building structures in general, and the decomposition of verbal roots. Parallels can be drawn to Noun Incorporation, especially focusing on the Kiowa language of Oklahoma.

Verb stems combine with verbs in a number of ways, from restructuring to light verbs to morphologically obvious verb incorporation. In Kiowa, I list (2019) fourteen semantically distinct mechanisms for two verbs to combine, from depictive conjunction to tough-constructions to restructuring contexts. Focusing on a few of these we see several parallels to the semantics of noun incorporation (NI). NI semantics has to account for three components of the noun’s interpretation: Its entity argument’s saturation (i.e. its determiner), its possible world, and its thematic relation to the verbal predicate. Different accounts assign these components to different elements of structure, while my own work proposes mediating relations between noun and verb, or above the verb.

VI semantics has to account for the event argument’s saturation (i.e. its aspect), its possible world, and some kind of relation to the actual event. We can propose similar relations, but they have do a lot of work while being morphologically inert, raising questions about the match between the morphosyntax and the semantics, the nature of roots, and perhaps the nature of composition itself.

We can also ask about a universally-limited set of possible relations between events that grammars can encode. With NI, non-themes are limited to thematic roles that are usually introduced by an adposition, but not those introduced by heads in the verbal spine. With VI, relations with the main verb are usually introduced by a conjunction or adposition, and cross-linguistically, those meanings have effects in a number of parts of grammar. Are there similar limits to what we see in NI?

We also look at the structure of the verbs themselves, because ‘root’ verbal meaning is not reducible to a single head. Classic accounts like Baker (1988, 1996) propose simple movement of one V head to another. More recently, Keine & Bhatt (2016)’s analysis of the German long passive relies on the lack of higher projection. However it does not generalize to the wide variety of restructuring phenomena shown by Wurmbrand (and see Homer & Bhatt 2020). Moreover, I will show facts from Kiowa argument structure that require a vP in at least some instances of VI. How much structure does an incorporated/restructuring verb have? Why is it always non-finite, and usually modal?

12 february
Suzana Fong (MIT)
A dependent case analysis of pseudo noun incorporation in Wolof

Bare nominals in Wolof can occur in the object position, though not in the subject position of a finite verb. Furthermore, they must be adjacent to the transitive verb that subcategorizes for them. These are properties usually attributed to Pseudo Noun Incorporation. However, there are two circumstances under which the requirement to be adjacent to the verb can be obviated: either an argument is introduced between the subject and the PNI-ed object or the latter is A'-moved. While the introduction of an additional argument and A'-movement are disparate phenomena, a dependent case analysis of nominal licensing (Branan, to appear) can account for why they both allow a PNI-ed object to not be adjacent to the verb in Wolof. Branan argues, following Levin (2015), a.o., that all nominals must be licensed with case, with case assignment being calculated in terms of dependent case (Marantz, 1991). In the impossibility of assigning case to a nominal, a last resort licensing strategy arises, namely, adjacency with the verb. Under the proposal that Branan makes about domains of case assignment and the position of case competitors in the sentential spine, bare nominal objects in Wolof cannot be licensed with case, hence why they must be adjacent to the verb. However, the introduction of an additional argument provides a case competitor to a PNI-ed object, allowing it to do away with licensing via verb adjacency. Likewise, A'-moving a bare nominal object brings it close to the subject, which can transformationally act as a case competitor. I argue thus that a dependent case theory of PNI can provide a uniform analysis of the distribution of bare nominals in Wolof.

5 february
Zach Wellstood (UC Berkeley)
Towards a Free Relative Analysis of Event Existential Constructions in Aklanon

In Aklanon (AKL), an understudied Western Visayan language, the robust correlation between Philippine-type voice and nominal case seems to break down in ‘event existential constructions’ (EECs). In this talk I show that this discrepancy is only apparent and dissolves under the proposed analysis wherein the existential maj selects: (i) a free relative (FR) with a null nominal head as its complement, and (ii) a topic-marked argument which controls a FR-internal PRO. Because violations of case assignment in EECs are only apparent, the robust generalization that case and voice covary can be maintained in Aklanon generally.

29 january
Dan Brodkin (UCSC)
Agent Focus in Austronesian

Many ergative languages appear to require intransitive verbal morphology in clauses where the external argument is extracted. One analysis takes this pattern to reflect a constraint on locality: (i) the external argument can only be extracted when highest in its clause, and (ii) intransitive morphology allows this to occur. Another view treats the pattern as morphological: the appearance of intransitive morphology reflects impoverishment triggered by A’-extraction.

This talk investigates this pattern in Mandar (South Sulawesi, Austronesian). This language shows a split construction in clauses where the external argument is extracted: (i) the verb bears intransitive morphology, but (ii) the internal argument is indexed with absolutive agreement (as in a transitive clause). This construction provides a means to comply with a locality constraint on extraction: it allows the internal argument to stay in the vP and permits the external argument to be highest in its clause. The analysis offers a new perspective on parallel constructions in the morphologically caseless but syntactically ergative languages of Western Indonesia and Mesoamerica.

22 january
Lev Michael (UC Berkeley)
The semantics of graded past tenses in Iquito (Zaparoan, Peru)

Approximately 40% of languages that have been described as expressing past tense make multiple temporal remoteness distinctions (Dahl and Velupillai 2013). These ‘graded’ or ‘metrical’ past tenses have traditionally been analyzed as having non-overlapping temporal reference, effectively dividing the timeline into non-overlapping regions, e.g., into ‘recent past’ and ‘distant past’ regions (see, e.g., Dahl 1985:121-125). Recent research suggests that this picture is incomplete in two ways. First, in some languages, the temporal reference domain of one tense may be wholly contained within the temporal reference domain of another tense, with the apparently non-overlapping referential properties of the tenses arising pragmatically (Hayashi 2011, Cable 2014, Chamorro 2020). And second, in some languages, graded ‘tenses’ do not express relations between utterance time (TU) and topic time (TT), as tenses definitionally do, but rather express relations between utterance time (TU) and situation time (TSit) (Cable 2014), making them ‘temporal remoteness makers’ (TRMs). In this talk I analyze the past tenses of Iquito, a Zaparoan language of Peruvian Amazonia, and show that what have previously described as: 1) a recent past tense suffix; and 2) a remote past tense suffixes, are in fact, respectively: 1) a general past tense suffix; and 2) remote past tense suffixes, where the domain of temporal reference of the latter suffixes are wholly contained within that of the former suffix. In doing so I describe a number of new tests for examining the semantics of graded tenses. I also show that Iquito past tenses are in fact tenses, i.e., that they express TT-TU relations, and are not TRMs. I conclude by placing Iquito in a provisional typology of graded temporal reference systems.