Undergraduate Research in Linguistics and Cognitive Science

In the Linguistics and Cognitive Science Department, students are encouraged to undertake research projects, making new explorations into this field. Below are completed Summer Undergraduate Research Program (SURP) projects from our students.


Language Learning in Virtual Reality

Brendan Ly ’22; Advisor: Franny Brogan

Virtual reality (VR) has vast potential for language learning because VR has the ability to simulate an educational and social learning environment without requiring a teacher facilitator. While VR programs focused on second language acquisition (SLA) have been developed, to the best of our knowledge scholars have yet to evaluate their efficacy through the lens of contemporary SLA literature. This project assesses the successes and failures of VR in simulating conditions, tasks, and/or environments that are conducive to language learning, as set forth by contemporary research in SLA (e.g. Myles 2002, Richards 2006). This body of literature provides a foundation for juxtaposing VR-based language learning programs with successful real-life and online learning models, allowing ease of comparison and analysis between existing models and their VR counterparts.

In addition to evaluating the current landscape, this project suggests ways in which VR-based programs can improve and even offer a better alternative to their non-VR counterparts. For example, VR can simulate the cultural context of a language without the need to travel. However, despite its potential, VR technology is still in infantile stages. Although there are immediate improvements that can be made for current programs, the future prospects are unclear. Despite this, studying examining SLA and VR in tandem is paramount in testing these limits and determining these future prospects.

Rethinking the Gender Paradox in El Salvador: Evidence from /s/ Weakening

Deborah Yi ’21; Advisor: Franny Brogan

In many Spanish dialects, syllable-final /s/ is variably weakened to [h] and/or deleted; like other lenition processes, /s/ weakening is stigmatized and considered nonstandard. Previous studies on /s/ weakening in Spanish have consistently supported Labovian (2001) generalizations regarding language variation and gender: women tend to prefer standard linguistic forms, while men favor nonstandard, more stigmatized variants (e.g., Cepeda, 1995; Erker, 2012). However, in Salvadoran Spanish—a dialect that weakens /s/ at high rates across syllable positions and varies between six different allophones, or phonetic realizations—we find that gender-based patterns of /s/ lenition contradict this generalization. For this study, 72 Salvadorans balanced for region, rural vs. urban origin, age, and gender participated in sociolinguistic interviews in El Salvador in 2015. We acoustically analyzed 200 occurrences of /s/ per participant (N = 14,400 tokens) in Praat (Boersma & Weenink, 2018), and each token was coded for a number of factors including prosodic position, syllable stress, preceding and following segments, and allophone. We found that women produce significantly more of the variants that carry the most social stigma, as attested both in sociolinguistic interviews and in previous research on this dialect. We argue that the contemporary sociopolitical situation of El Salvador has resulted in the sociolinguistic isolation of women and a move toward standardization led by young men.

How Anti-Indigenous Bias Manifests in Language: Mayan Kaqchikel and The Non-Use of the "Ixta" Noun Classifier

Shanaya Stephenson ’19; Advisor: Robin Melnick​

This summer I participated in the annual Summer Field School at the University of Maryland, which brings students, researchers, and faculty from all over the world to Patzun, Guatemala for intensive Mayan Kaqchikel language classes combined with research and other community-based projects. While in Patzun, I received a certificate for my work in learning Kaqchikel. I engaged in research to further understand how the intersection of gender and indigeneity informs the use of language (and vice versa) among the different speech communities in Patzun. The results of my study indicate that the use of the ixta noun classifier in Kaqchikel has been adopted into Spanish as a slur (ixtan/ixto) and is used by the Ladino people (people of Spanish decent living in Central American countries) towards both male and female Kaqchikeles. As a result, that noun classifier is not used as often by the Kaqchikeles. Instead it is shortened to "xta" to make the distinction. Additionally, I ran a demographical study of Kaqchikel use on the internet, and contributed to a Kaqchikel corpus for the future study of linguists there.

“Don’t Take That Tone With Me”: Linguistic Variation and Disciplinary Action on African American and Latinx Children in Educational Settings

Emily Vomas ’22 and Jade Hill ‘20; Advisor: Nicole Holliday

African American and Latinx children are disproportionately likely to be suspended or expelled from school for “subjective behavioral infractions” such as being “loud, unruly, or unmanageable” (Crenshaw et. al 2015). This research project aims to understand how and why black and Latinx students are subject to such linguistic discrimination that leads to discipline by addressing the following:

1. What patterns of ethnolinguistic feature use exist for a community of 14-18 year old black and Latinx high school students? In what ways do patterns of use of several AAL and Chicano English (CE) features vary systematically?

2. What is the relationship between use of such features and experiences in school disciplinary systems?

In order to obtain data for exploring both production and perception, we collected data from 28 high school students who participated in the PAYS program at Pomona. Participants were recorded in casual situations as well as in reading tasks, and their salient ethnolinguistic features were identified. They were interviewed about their attitudes towards school and experiences with discipline inside and outside the classroom. This data was then qualitatively coded to examine student experiences of linguistic discrimination, then quantitatively coded for features of AAL and CE. Preliminary results indicate a relationship between use of such features and experiences of linguistic discrimination in school.

Phonological Components in Facilitating Tip-of-the-Tongue Resolution in Native Mandarin Speakers

Kristine Chang ’21; Advisor: Lise Abrams

The tip-of-the-tongue (TOT) phenomenon is a failure to recall a word one knows. For English speakers, encountering a word’s phonology (sounds), specifically its first syllable, is critical to resolving TOT states, more than the initial sound or other syllable of the word. While English words contain sound-based syllables, Mandarin words consist of characters not based on alphabets. Rather, each character is a visual representation that doesn’t spell out how it should be pronounced. Therefore, the present experiment explored whether phonology can facilitate TOT resolution in a language that does not emphasize phonological representations. 54 Mandarin speakers saw descriptions of cheng-yu, 4-character Chinese idioms. Participants said if they knew, didn’t know, or were experiencing a TOT for the cheng-yu being described. They then saw a list of five Mandarin words, where one was manipulated in three ways. Its first character was always visually different from the cheng-yu's first character but contained (a) the cheng-yu’s first phoneme, (b) the sounds of the cheng-yu’s first character, or (c) no sound overlap. The cheng-yu’s description was presented again, and participants attempted to retrieve the answer. If shared phonology helps to resolve TOT states in Mandarin, presenting the sounds of the entire character should help word retrieval the most. The results shed light on whether TOTs are caused by deficits in retrieving a word's sounds even within visual language systems.


"I've Code-Switched on Behalf of the Black Student Population": Linguistic Insecurity Among Black Students at HWIs

Lemuel Lan ’20
Advisor(s): Missing Advisor

At historically white institutions (HWI), Black students are burdened with social/emotional labor (Wallace and Bell 1999; Craig 2014; Harper 2013) as they are stereotyped from negative association with African American English. This study explores how language relates to other elements of black students’ experiences in HWIs and linguistic diversity in higher education. We interviewed 15 self-identified Black students from a university in the Midwest. The interviews lasted an hour each with 20 open-ended questions that had participants reflect on four broad topics: linguistic background, language in the classroom, language elsewhere on campus and navigating the university. The students consistently told stories around the following themes:

  1. Stereotype threat: Students perceived a risk of appearing lazy/uneducated if people heard them using AAE
  2. Perceived incompetence: Students felt the need to prove their intellectual merit; both peers and faculty questioned their belonging
  3. Bifurcated sociolinguistic identities: Students reported having “black spaces” on campus where they could “relax”/use AAE; they monitored themselves and used Standard English elsewhere

These accounts reveal the daily linguistic labor these students perform as they navigate a campus environment that fosters sociolinguistic prejudice. Our study highlights complex interactions between racism and linguistic discrimination, providing a starting point for addressing linguistic inequality at universities

Funding Provided By: Mr. & Mrs. Richard N. Frank

Cognitive Processing of Heavy Noun-Phrase Shift

Sadaf Khan ’20
Advisor(s): Laura Johnson

Time perception studies find that processing of complex stimuli diverts attention from the passage of time, resulting in reduced estimates for the duration of a given time period. Subjective time perception can thus be utilized as a measure of processing complexity of a stimulus. Heavy Noun-Phrase Shift (HNPS) is a linguistic phenomenon wherein longer, more syntactically complex noun phrases move to a later position in a sentence after a prepositional phrase. Previous research suggests that this occurs due to production demands on working memory as opposed to core grammar requirements. Although the effects of HNPS on listeners’ attentional resources is far less studied, we hypothesized that participants would produce shorter time estimates for non-shifted than shifted sentences due to the increased processing complexity. Twenty-three participants read noun-phrase shifted and non-shifted versions of the same sentences, and then attempted to reproduce the reading time period for each sentence. The results revealed that participants consistently spent longer reading non-shifted than shifted sentences, and that their reproduced time periods were significantly shorter compared to actual reading times for non-shifted than for shifted sentences. Both of these findings suggest that non-shifted sentences increased processing demands and reduced time estimates; as such, the findings lend support to the theory that parsing demands are also eased by HNPS.

Documenting Wanga Syntax

Felicity Walston ’20
Advisor(s): Michael Diercks

Wanga is a Luyia-Bantu language spoken by approximately 309,000 people in Western Kenya. Our work this summer builds upon previous SURP projects and Pomona College student theses, all related to a broader project funded by a National Science Foundation grant with two other institutions. The purpose is to document the lexicon and grammar of four Luyia-Bantu languages, including Wanga. This summer, we conducted research on grammatical topics by reading existing literature and transcripts of interviews and narratives in Wanga. We held elicitation sessions with a language consultant who produced forms and confirmed our suspicions of underlying phenomena in the language. We composed chapters outlining twelve specific grammatical features that we investigated in Wanga; those include negation, anti-agreement effects, clause-chaining, copulas, discourse particles, locatives, relative clauses, auxiliary verb constructions, interrogatives, nasal prefixes, conjunction and disjunction. Further work will continue to confirm our findings with native speakers and uncover more data in spoken narratives or interviews.

Funding Provided By: Mr. & Mrs. Richard N. Frank

Investigating the Affective Reactions Elicited Through Inconsistency Processing

Kerem Oktar ’20
Advisor(s): Shlomo Sher

There is significant inconsistency in the affective reactions that scholars have postulated to accompany the cognitive processing of inconsistencies. The disparity in these notions is driven partly by the field’s current lack of an efficient experimental paradigm through which the affective reactions accompanying the processing of different types of inconsistencies can be investigated. This research project is therefore an attempt at developing such a paradigm: one that has the potential to consolidate findings across multiple fields (notably the literatures on cognitive dissonance and curiosity) and provide insight into how we update our models of the world in light of new information. The project further aims to explore the veracity of our hypothesis that belief-conflicting information has opposite effects depending on its self-relevance with neutral conflicts leading to positive affect and increased information search, and self-relevant conflict having opposite effects. In order to investigate these notions, we have developed the affective spillover paradigm: a Qualtrics survey comprised of two parts. In the first part, various household object and statement pairings are presented together. In the second part, the affective reactions elicited by the neutral and conflicting information are measured through the subsequent valuation of the objects by the participants.


Documenting Tiriki Syntax

Kristen Hernandez ’18; Advisor: Michael Diercks; Collaborators: Zoe Bauer ’20, Kang (Franco) Liu ’20

This project documents various aspects of the grammar of Tiriki, an understudied member of the Luyia subgroup of the Bantu language family spoken in Western Kenya. Our work is part of an ongoing NSF grant to document Tiriki along with three related Luyia languages. We worked with texts collected from interviews with native Tiriki speakers, adding morphological glosses into a computer database called FLEx, which we used to share the texts with researchers from other institutions. We also collected samples of theoretically relevant grammatical features of the language to begin the foundations of a grammar sketch; these included auxiliary verb constructions, anti-agreement effects, clause chaining, focalization, and relative clauses. In addition, we conducted elicitation interviews with our Tiriki consultant to answer questions about the constructions we found in the texts, as well as to analyze the phenomenon of hyper-raising in particular. We used this data to write a manuscript describing the Tiriki hyper-raising facts, a possible analysis, and the theoretical implications for the field. Overall, this project will contribute to both a broader understanding of Tiriki morphosyntax, as well as to a deeper understanding of the specific phenomenon of hyper-raising. This will offer a foundation for future research on Tiriki grammar as well as the grammar of other related languages.
Funding Provided By: Fletcher Jones Foundation

The Linguistics of a Transnational Cuba

Caitlin Warren ’18; Advisor: Michael Diercks

Miami is heavily shaped by the Cuban exodus and impacted by Cuban culture and language. The goal of this research is to explore the relationship between the different accents and social groups within the Havana and Miami Cuban communities, focusing on defining key characteristics of the Cuban community, more specifically centered on how linguistic markers change with social variables. Data was collected through recorded interviews focusing on where participants have lived, including where in Cuba they grew up or where they’ve lived in Miami. Participants were found through personal and professional networks and interviewed for an average of 10 minutes.  After transcribing and analyzing the interviews, I found that many Miami Cubans disagreed with the Cuban government and often cited their displeasure and inability to voice their concerns in Cuba as reasons for moving to the United States. Their accents shared many characteristics with the Havana Cuban accent, namely a softening of consonants between vowels, and aspiration or deletion of an /s/ at the end of syllables and words. Unsurprisingly, Miami Cubans showed stronger signs of Americanization in their speech, especially women, such as stronger presence of a rhotic /r/ and reduction of /s/ elision.
Funding Provided By: Paul K. Richter and Evalyn E. Cook Richter Memorial Fund

Investigating the Neural Basis of Age Related Deficits in Word Retrieval

Kerem Oktar ’19; Advisor: Deborah Burke

Older adults experience specific difficulties with language production which may be predicted by (and may be predictive of) changes in the underlying neural architecture that supports language processing. Our studies leveraged the dual stream model of language, which implicates dorsal streams in language production and ventral streams in language comprehension. Using neuroimaging techniques, different aspects of these changes can be investigated and compared to predictions made by cognitive models of language networks. Resting state fMRI (rsfMRI), a task-free technique that is used to investigate intrinsic network function, was used in one study. We used rsfMRI and graph theoretical measures to investigate how functional connectivity in the language network in older and younger adults predicts behavioral performance. Preliminary results suggest that, in the left hemisphere, older adults show significantly diminished functional connectivity compared to younger adults. A potential mechanism underlying transmission failures in the language network may be declines in white matter integrity. In order to explore this hypothesis, another study used Diffusion Tensor in order to investigate differences in the integrity of language tracts Preliminary results replicated the finding that dorsal stream tracts have lower integrity than ventral stream tracts. These neuro-anatomical differences were not predictive of accuracy on a picture-naming task.
Funding Provided By: Fletcher Jones Foundation

Does GPS Use Enhance Spatial Navigation Ability?

Laura Zhang ’19; Advisor: Deborah Burke

With the rising popularity of and reliance on navigational aids, it becomes increasingly important to study how these systems affect our spatial knowledge acquisition. Previous studies have explored the influence of different interface technologies (mobile maps, AR, voice) on spatial knowledge acquisition, however, none have established a concrete measure by which to gauge GPS usage and navigational ability (Huang et. al., 2009). The objective of this project is to develop a reliable GPS survey measure that examines the relationship between frequency of GPS use, use of different GPS features, and spatial navigation performance. The method involved performing a factor analysis in SPSS on previously collected pilot survey data, and correlating the resulting factors with the scores for two tests of spatial navigation performance (pointing task and model-building task). Participants were Temple University undergraduate students (N = 66, F = 43). From our preliminary analysis, we did not find any significant factors or correlations between GPS use and navigational ability. However, these results can largely be attributed to a flawed survey and small pool of participants. Future directions include modifying and re-distributing the survey to a larger population, and re-running a factor analysis on the new data.
Funding Provided By: General SURP Fund