Papers, Slides, & Handouts

Our classes aren't the only things going on in our students' lives, and their other obligations are not always perfectly compatible with the structure, the schedule, and the hidden curriculum of higher education. However, the routine of students asking for exceptions and extensions can wear down both student and instructor. A system which removes this stressor would certainly be welcomed by many, but comes with many questions. How can you incentivize students to keep up with the material without enforcing deadlines? How can difficult assignments be made less stressful without sacrificing learning? Is there a way to let students get a second chance at a concept if they couldn’t complete their first attempt, especially if their progress has been disrupted by things out of their control? This talk will discuss some emerging alternative grading systems which tackle some of these questions in the context of core Linguistics courses (e.g., semantics, syntax). We will focus primarily on two variations of Skills-Based Grading – a system which centers the course and the final grade around students mastering the “skills” taught in the course. Importantly, students are allowed to demonstrate mastery over those skills any time before the end of the course, and every assignment merely offers an opportunity to demonstrate one or more skills. There is no penalty for missing or underperforming on assignments due to external setbacks, such as illness, grief, global pandemics, or familial obligations, because, until the final exam, there is always another chance to try again. This leniency is particularly beneficial to students facing structural and institutional disadvantages, such as first generation students, transfer students, students with disabilities, and minority groups, who may be more likely to experience external stressors that lead to temporary setbacks.

It has been well established that example sentences in linguistics use a remarkably non-diverse set of proper names in terms of gender, culture, and ethnicity (e.g., Macaulay & Brice 1997, Cépeda et al. 2021, and Kotek et al. 2021). Here, we introduce a new resource, the Diverse Names Generator (DNG), which provides randomly selected proper names with IPA transcriptions from a user-contributed, linguist-curated database of names from a wide range of languages and cultures. Generating names randomly helps users to overcome unconscious bias that may lead them to default to using Anglophone, male-gendered names. The DNG can be accessed both through a website interface and through a downloadable app, both with offline capabilities. This novel resource is the first of its kind and can be used both while preparing examples ahead of time and while generating examples live in the classroom. 

This paper discusses how to implement ‘Skills-Based Grading’ (SBG) in linguistics, with a university-level course in formal semantics as a case study. Particular focus is given to transitioning to SBG from traditional grading. Regardless, of the grading system, all courses start out from a set of desired learning objectives, with final grades intended to reflect the extent to which these objectives have been achieved. Traditionally, grades are determined by a series of mandatory assessments and tests, for which somewhere between partial and full credit is awarded. In this paper, we illustrate how mastery of learning outcomes can be more directly measured by re-packaging them as “skills”. In SBG, students are given multiple opportunities to demonstrate full mastery of each skill. However, grading is all-or-nothing, with no partial credit awarded. SBG has been shown to improve student learning, encourage effective study, lower student stress, and achieve more equitable outcomes, and has been successfully adapted for linguistics courses in phonology (Zuraw et al. 2019) and semantics (O’Leary & Stockwell 2021). Here, we offer step-by-step instructions for creating an SBG course, covering skill types, skill groupings, opportunities, grading, and assessments. 

In this invited talk, I present the benefits of skills-based grading (SBG) as an alternative to traditional grading systems. Based on work with Richard Stockwell (Christ Church, Oxford) and Douglas Ezra Morrison (UCLA), I present both qualitative and quantitative data (collected from surveys and student grades) that suggest that SBG both lowers student stress and offers a more equitable system of evaluation. While maintaining rigorous standards, SBG is more lenient than traditional grading, in that it allows students multiple opportunities to demonstrate and gain credit for their learning. Among other benefits, the availability of multiple opportunities to earn the same credit ensures that there are not penalties for temporary setbacks; this is, in particular, fairer to students facing structural and institutional disadvantages, who are more likely to experience external stressors that lead to missed assignments.

This work, which summarizes the bulk of my  dissertation, seeks to add to our knowledge of the constraints which govern nominal property times, with the goal of a model that can predict all available nominal property times for a given noun in a given sentence. Backed by novel empirical observations, I propose three additions to the groundwork laid by Enç, Musan, Tonhauser, and Keshet (among others), namely:

(i) Nouns exhibit lexical aspect. Importantly, stage nouns in English naturally separate into two classes, which exhibit different aspectual behavior

(ii) The time arguments of nouns used to introduce novel referents are subject to a locality constraint; they are bound by the nearest scoping time abstraction--either the utterance time or the time introduced by a sentential tense.

(iii) Nominal predicates in non-novel DPs (effectively) have the evaluation time of the initial, novel use of that predicate in the discourse. Nominal lexical aspect and the temporal locality constraint only apply to predicates picking out entities novel to a discourse. In a novel DP, the nominal predicate's time argument is used to pick out the correct group of entities along a temporal dimension. However, a non-novel DP is merely anaphoric to the previously established set of entities, and therefore any use of a nominal predicate in a non-novel DP does not need to consider a time argument in order to denote the correct referents. (In this work in progress,) I argue that the tools we need to model this data require more than simple indices on DPs--we need a theory of dynamic semantics which stores variables and the predicates which describe them, so that the use of a particular nominal predicate in a non-novel DP can be used to identify which variables the DP should be coindexed with.

(Winner of the Linguistic Society of America's 2021 Student Abstract Award)

Skills-based grading (SBG) provides students multiple opportunities to acquire and demonstrate mastery of course material. Following Zuraw et al. (2019), who report the first use of SBG in linguistics for teaching phonetics and phonology, we implement SBG in a formal semantics class for the first time. SBG is known to work well for teaching skills which require algorithmic approaches to arrive at an inarguably correct answer. In our application to semantics, we show that the benefits of SBG are transferable to courses skills approaching the philosophical, where there may not be a “right” answer. 

Richard Stockwell and I show how semantics instructors can implement skills-based grading, a modern evaluation system shown to lower student stress and provide a lasting understanding of the material. We provide grade data and survey responses from students who were evaluated using skills-based grading in our semantics course.

Click here to learn more and access our materials.

(See our slides - below - for  statistical analysis of and visual representation of grade data throughout the course. Many thanks to our statistical consultant Douglas Ezra Morrison.)

Hän, a Dene language spoken in Alaska and the Yukon Territory, only has six remaining speakers. While we work towards revitalization as an L2, we are faced with the necessary task of rapidly documenting the details of an understudied language which is not used in the daily life of any of the remaining speakers. This talk discusses ongoing theoretical work on the pronominal prefix system used in Hän for direct object verbal prefixes, indirect object postposition prefix, and possessor-marking nominal prefixes. I propose that Hän's pronominal system has a distinction between reflexive and disjoint anaphors, and a distinction between definite and indefinite pronouns, with a single phonetic form representing the disjoint anaphor and the definite pronoun. I additionally discuss the challenges of conducting fieldwork on a dying language that is seldom used, especially those that affect this particular theoretical question. 

Many Athabaskan languages have two third person singular object pronouns which occur only in conjunction with third person subjects. The distribution of these pronouns is claimed to be semantically motivated; proposed theories include inverse voice (Willie 2000), topic/focus distinction (Hale, Jelinek, and Willie 2003, Jelinek and Willie 1996, Thompson 1989, 1996, Willie 1991, 2000, Willie and Jelinek 2014, a.o.), and obviation (Thompson 1989, 1996, Aissen 2000, Rice and Saxon 2001). In stark contrast to the data from other Athabaskan languages, we present here the two analogous third person object pronouns in Hän, which are fully predictable based on syntactic traits. In Hän, the two third person object pronouns, yë- and wë-, are used to fill an obligatory object position within the VP when it is not saturated by an overt object, even when the overt object appears elsewhere in the sentence. The use of one pronoun over the other is determined by the person of the subject. Yë- is used when the subject is third person and wë- when the subject is first or second person. We argue that this distinction comes from a person agreement morpheme that occurs directly before the object pronoun.

Kazan Tatar, like other Turkic languages, has a system of verbal suffixes which are generally considered to encode both tense and evidentiality. In this paper, I discuss several embedded environments where only a subset of these suffixes are allowed. I argue that the morphemes that can be used in these contexts encode tense but not evidentiality. In more conventional environments, these morphemes are compatible with evidential features, but the morphemes themselves do not carry evidential information.

Additionally, I discuss the licensing of propositional content in these specific syntactic environments, in order to explain why they are allowed to lack evidential information. Following an analysis of Korean embedded nominal clauses, I propose that propositional content is only licensed in these specific embedded clauses when the embedded proposition has been mentioned previously in the local discourse. The fact that the proposition has previously established, along with evidential material, allows a repetition of the proposition without evidential information.

This work seeks to add further depth to our knowledge of cleft constructions by examining the tense of the matrix copula. There are numerous accounts describing the semantics of clefts, but none of these have integrated the matrix tense, meaning that any current theory of cleft constructions would claim that ‘It was John that died.’ and ‘It is John that died.’ have the same formal meaning, despite the fact that there are situations in which one can be uttered but the other cannot.

This work proposes that the matrix tense overtly marks the topic time of the utterance (see Klein 1994), and that the cleft clause (e.g. ‘that died.’) is interpreted relative to that time. For instance, a past matrix tense communicates that the topic time precedes the utterance time. The property specified in the cleft clause ‘that died’ is therefore true for all individuals who died at some t prior to the topic time, rather than prior to the utterance time. These truth conditions are added to the cleft operator proposed by Büring & Križ (2013).

A longstanding discussion in the study of Athabaskan languages has been the distribution of the two third person object pronouns (often called y- and b- pronouns, from Navajo, see Platero 1982, Rice 2000, Rice and Saxon 2001). These pronouns always occur with third person subjects and either are obligatorily included, as in Apachean languages, or are in complementary distribution with the overt object (Rice and Saxon 2001). The distribution of y- and b- pronouns is claimed to be semantically motivated, although there has been disagreement about the nature of the distinction. Proposed theories have included inverse voice (Willie 2000), a topic/focus distinction (e.g. Hale, Jelinek, and Willie 2003, Jelinek and Willie 1996, Thompson 1989, 1996, Willie 1991, 2000, Willie and Jelinek 2000), and obviation (Thompson 1989, 1996, Aissen 2000, Rice and Saxon 2001), among others. Surprisingly, the two analogous third person object pronouns in Hän, an endangered Athabaskan language spoken in eastern Alaska and the Yukon Territory, Canada, are fully predictable based on syntactic traits.

Musan (1995, a.o.) highlights a number of syntactic environments which could force a noun phrase to be temporally dependent, including weak cardinal determiner phrases and existential there constructions, while other noun phrases are subject to no such constraint and are therefore temporally independent. She concludes that these constructions are used when the noun phrase is new to the hearer, and that hearer-new noun phrases were interpreted as referring to temporal chunks, or stages, of an individual, rather than the whole temporal existence of the individual. She posited that these temporal chunks were then evaluated at the same time as the rest of the sentence.

In this paper, I provided evidence that some nouns inherently describe stages, and are therefore always temporally dependent. Additionally, I showed evidence that temporally independent noun phrases can be evaluated at fewer times than was previously thought and that temporally dependent noun phrases can be evaluated at more times than was previously thought. Temporally independent noun phrases are subject to a familiarity constraint and can only be evaluated at or before the utterance time or the predication time, whichever is later of the two. Temporally dependent nouns are subject to an accessibility constraint and may be evaluated at either the utterance time or the predication time.

Many Italian verbal paradigms have multiple stems associated with a single verb. Single paradigms can be seen to have up to five stems. These stems are defined as a “common formal core” (Aronoff 1994) shared by a group of forms, and are often associated with the “irregular” conjugations. The irregular forms seem to pattern together in noticeably predictable ways: certain verbs have paradigms that are irregular in the same ways and some forms are more likely to be irregular than others. With the noticeable level of predictability in the language, it would seem odd to say that all irregular forms are stored in the lexicon. Instead, I propose that we can minimize the necessary lexical storage space to include only one entry for each unique stem, and a set of rules for how they interact.

Following Bonami & Boye’s (2001) approach to verb paradigms in French, I present a theory of dependency for Italian, which separates Italian verb forms into eight groups. The stem for each of these groups is dependent of the stem of another group. Assuming that the dependency relationships are internalized by the speaker, this greatly reduces the amount of information that needs to be stored in the lexicon.

This presentation covers the basic morphological phenomena found in nominal and prepositional phrases in Italian, aiming to account for the contraction of prepositions and definite articles.