I study language and memory.

I study memory and language in the Psychological Sciences department at Rice University. My research is supervised by Dr. Randi Martin. I use a combination of behavioral, neuroimaging, and neuropsychological techniques to understand the theoretical and neural basis of language and memory. The goal of my research is to produce converging evidence from multiple approaches that illuminates the relationship between language processing and memory. I am especially interested in how working memory, the cognitive system that allows us to remember information over very short time periods, is related to language production and comprehension. These are a few projects that I am currently working on:

White matter correlates of phonological and semantic working memory

Working memory (WM) is the cognitive system that maintains information over short time periods. WM is important for successful language production and comprehension. Evidence from people who have WM deficits suggests that there are separable WM buffers for maintaining different types of representations, including phonological (i.e., speech sound) and semantic (i.e., meaning) information. Phonological and semantic WM buffers make distinct contributions to language processing, and recent evidence suggests that the two buffers are also located in different brain regions. The proposed location of the phonological WM buffer is the supramarginal gyrus, while the semantic WM buffer may be located in the inferior frontal gyrus.


However, understanding which gray matter regions support WM is not enough for a full picture of the neural basis of WM. It is also important to understand how white matter tracts that connect gray matter regions support WM. I use diffusion tensor imaging (DTI) to visualize the white matter tracts of people who have experienced left hemisphere stroke. DTI is a neuroimaging analysis method used to infer the presence and integrity of the white matter tracts which connect gray matter regions. 

My goal is to help us better understand how damage to specific white matter tracts affects WM and, thus, language processing.


The white matter tracts in the left and right hemispheres of the same individual’s brain would typically look similar. However, as a result of brain damage in the left hemisphere, there are far fewer white matter connections remaining between brain areas (a) compared to the same individual’s undamaged right hemisphere (b).  Damage to some tracts may lead to specific kinds of WM deficits. For example, reduced volume in the arcuate fasciculus may be related to phonological WM performance (c). On the other hand, reduced volume in the inferior fronto-occipital fasciculus may be related to semantic WM (d).

Sentence comprehension and working memory

Sentence comprehension involves maintaining, processing, and integrating linguistic information actively and continuously. Sentence comprehension is supported by many cognitive systems, one of which is working memory (WM). I am interested in how different types of WM capacities, especially phonological and semantic WM, support comprehension. Past work using small groups of people with left hemisphere brain damage has suggested that semantic, but not phonological, WM is critical for successful sentence comprehension. 


As an extension of this work, I analyzed a dataset that includes the sentence processing and working memory performance of over 40 people with left hemisphere brain damage. As a result of their brain damage, each person included in this dataset had a unique combination of language and working memory difficulties. My findings (currently under review) help us understand the role that phonological and semantic WM have on sentence comprehension.


Depiction of the domain-specific working memory model which includes separate working memory capacities for phonological and semantic information (figure adapted from Martin, Lesch, and Bartha, 1999).

Single word production and working memory

Fluent speech production requires interaction between the language processing system and many other cognitive systems, including working memory. The language production system engages working memory to plan speech in advance of articulation. Past neuropsychological evidence from chronic stroke patients has shown that semantic working memory is critical for language production, but phonological working memory is not. While it is unlikely that phonological working memory capacity has a large influence on speech production, there may still be some processes, such as phonological retrieval, that are shared between the two. In three individual differences studies, I am testing the prediction that a phonological retrieval process is shared by both single word production and rehearsal in phonological working memory.