Mary R. Newsome

Dept. of Psychology
Rice University
Houston, TX 77005
tel: (713) 348-2215
fax: (713) 348-5221

I am a post-doc in Dr. Randi Martin's lab at Rice University. In the lab, we investigate working memory and language processing. To do this, we often ask patients who have selective language and working memory deficits to do tasks to help us understand the nature of their deficits. The rationale is that understanding where and how systems break down will help us understand how intact systems work. For example, there are patients who can understand written words but cannot understand spoken words--this suggests that the sounds of written words may not be necessary to process them.

We greatly appreciate the help our patients provide--most notably the patience they have for us…


Language production

Currently, we are working with patients who have difficulty producing speech to gain insight into models of language production. For example, when asked to name pictures (Philadelphia Naming Task, Roach, Schwartz, N. Martin, Grewal, & Brecher, 1996), two patients who are unable to name a helicopter have very different responses. One patient provided a description of a helpicopter, saying "It's not an airplane, but it's one that flies in the air". The other patient called it a "calibusta". What can account for these different responses?

The first patient's response, in addition to separate testing of both patients, suggests that the semantic component of language production in these patients is unimpaired. Both patients' difficulty seems to arise from the phonological level of processing, albeit at different levels. The first patient's difficulty producing words that are low in frequency, and the second patient insertion of extraneous phonemes provides clues. At this point, we are positing that the first patient's difficulty arises from a lexemic (or sound of whole word) level, and the second patient's difficulty arises from a phonemic (or sound of individual letters) level.

Separable semantic and phonological components in working memory

Patients exist who are able to perform well on working memory tasks that require semantic processing (remembering the meaning of words), but are unable to complete tasks that require phonological processing (remembering the sounds of words) (Martin, Shelton, & Yaffee, 1994). We are currently conducting tasks that further test the separability of semantic and phonological working memory.

There are patients who are able to process the phonology of words they say, but are unable to process the phonology of words they hear (Martin, Lesch, & Bartha, 1999). We are further testing the distinction of separable input and output phonological components in working memory by asking patients to view lists of written words and afterward identifying whether a probe word was in the list or not. In order to do this task, people have to say the words to themselves, thereby employing output phonology. Patients who have impaired output phonology should be less accurate on this task than patients and controls who do not. We can test input phonology by presenting the words auditorily. Patients who perform worse when the words are spoken than when they are written would have impaired input phonology.

Semantic dementia

We are currently investigating the relationship between phonology and semantics in working memory with a patient who has semantic dementia. Some researchers (e.g., McCarthy and Warrington, 1987) have found that when having to repeat lists of words, semantic dementia patients show no difference in their ability to repeat words that they still know, compared to words that they no longer know. (Known and unknown words are determined by a word definition task, among others). Such a finding suggests that semantics and phonology are functionally separate. However, other researchers (e.g., Knott, Patterson, and Hodges, 1997) have found better serial recall for known words and have put forth a semantic-lexical-phonological interactive account of working memory.

Randi Martin, Cris Hamilton, and I are conducting working memory research with a semantic dementia patient, BA, who, in one testing session, showed the ability to recall lists of words that she still knows, but was unable to recall words she does not know, suggesting interaction in a semantic-lexical-phonological working memory network. However, at a later testing session no difference was found. Replicating previous research, BA recalls unknown words better than she recalls nonwords, suggesting an intact lexical contribution during serial recall. We are currently investigating why some researchers have found differences in known and unknown word recall, and others have not.

Sentence processing

Another study that is in the works investigates the effects of semantic constraints on thematic role assignment. We'll let you know!

Email me

Click here for Randi Martin's homepage.

View the Rice University Art gallery, fortunately enough located in the same building as the Psychology Department.

Click here to donate free food, paid for by advertisers on the site (

Here to send a letter inquiring into the safety of a political prisoner suspected of being tortured. Amnesty International writes the letter for you.
Click here for a pretty good cartoon about the recent election outcome.