529 Murdoch
3420 Forbes Ave
Pittsburgh, PA 15260
412 624 7460
my first name AT pitt DOT edu

Senior Scientist, Learning Research and Development Center, University of Pittsburgh
Professor, Psychology, Linguistics, and Communication Sciences and Disorders Departments, University of Pittsburgh


1996 graduate of Yale College, B.A. in Cognitive Psychology
2001 graduate of MIT, PhD in Cognitive Science
2001-2003 Postdoctoral fellow, University of Massachusetts at Amherst

Current Research

I was trained as both a cognitive scientist and a formal linguist at MIT and UMass, and I study language comprehension, a process that is automatic and usually seems effortless, but is also surprisingly complex. For example, to understand The cats chased a dog, after accessing the words, one must mentally represent the relevant set of cats, decode the syntax to figure out who is doing what to whom, and resolve the conflict between that representation and one's knowledge of cats' and dogs' usual behavior. My work has advanced our understanding of each of these processes. A primary focus of mine has been on understanding the ways that our knowledge of both language and what is likely to happen in the world are brought to bear during language comprehension (Warren & Dickey, 2021). For example, when one encounters the word "dog" in "The cats drank the dog", how much does the ensuing disruption reflect the fact that the object of the verb "drink" is not liquid versus the extreme unlikelihood of this event? How does reliance on event likelihood versus verb-based knowledge change when the comprehension context changes? How does it change for a person with aphasia (a language disorder usually brought on by stroke)? The answers to these questions have both basic and applied import: they provide a window into the cognitive mechanisms supporting the language comprehension system and could improve the design and targeting of aphasia treatments based on strengthening event knowledge (e.g. Edmonds, 2016).

The Representation and Use of World and Linguistic Knowledge in Language Processing As a postdoc, I embarked on a line of work that used the detection of carefully controlled semantic violations during reading as a window into how, when, and under what conditions world and linguistic knowledge influence language comprehension: Warren & McConnell (2007); Warren, McConnell, & Rayner (2008); Patson & Warren (2010); Warren, Milburn, Patson & Dickey (2015). Contrary to Hagoort et al. (2004), we found that violations of verb-based (linguistic) knowledge seemed to be privileged over violations of event-based (world) knowledge. However, we found no evidence of any privilege for verb-based knowledge when we tested these kinds of knowledge's roles in guiding predictions about what's coming next in a sentence (Milburn, Warren, & Dickey, 2016). These findings support the view that the comprehension system flexibly uses whatever knowledge is most informative in a given situation (e.g. Kuperberg, 2016). This view is consistent with the long-standing, but not particularly well fleshed out, hypothesis that people with aphasia (PWA) might compensate for their linguistic impairments by adapting to rely more heavily on world knowledge.

Dr. Mike Dickey and I started a collaboration to evaluate this hypothesis. We developed a new aphasia-friendly test of event knowledge (Dresang, Dickey, & Warren, 2019), so we could relate the relative degree of PWA's verb vs. event knowledge impairment to their use of these sources of knowledge in different comprehension tasks: Colvin, Warren, & Dickey (2019); Dickey & Warren (2015); Warren, Dickey, & Liburd (2017). Our results in this area have been suggestive, but not entirely clear. We simultaneously began to investigate whether PWA predict during comprehension, because error-based learning, which has been hypothesized to drive adaptation, relies on prediction. In Hayes, Dickey & Warren (2016), Warren, Dickey, & Lei (2016), and a still unpublished data set, we found evidence that PWA do predict during comprehension, at least to some extent. Mike and I continue to work in this area with current students.

Plural Processing I have always found referential processing fascinating (e.g., Warren & Gibson, 2002; Warren, 2003), but my biggest contribution in this area came when my PhD student Nikole Patson and I started investigating how we understand plurals like "the cats". In this work, we have developed a number of creative methods (Patson & Warren, 2017) and drawn from an unusually wide range of theoretical perspectives, from formal semantics to embodied cognition (e.g. Patson, George, & Warren, 2014). One question we asked is: do comprehenders always interpret singulars as representing a single entity and plurals as representing more than one? The answer is no. Patson and Warren (2010) found that comprehenders represent singular noun phrases within distributed predicates as conceptually plural; i.e. in a sentence like "Each man carried a cat", one immediately thinks of multiple cats. Patson et al. (2014) found that in simple sentences, comprehenders processed singulars ("a crayon") as individuals, and duals ("two crayons") as dyads, but definite plurals ("the crayons") didn't seem to be processed as necessarily plural at all. (Nikole has since followed this up in work she's done as a professor at Ohio State- Marion.) Another question we asked is: under what conditions do we mentally represent plurals as undifferentiated groups versus as multiple individuals? Patson and Warren (2011) addressed this question for plural noun phrases, whereas Patson and Warren (2015) addressed it for inherently distributed verbs (i.e. verbs like "sleep", which can only be done by individuals, so if a group of cats slept, every individual cat must have slept). When I was on sabbatical in Germany in 2018-2019, Nikole and I began a new series of experiments in this vein in collaboration with my host Barbara Kaup. This new work uses the spatial-numerical association of response codes (SNARC) effect (an effect such that smaller magnitudes are responded to more quickly when presented on the left side of the body and larger magnitudes on the right) to investigate whether collective nouns (words like "choir", which are conceptually plural but grammatically singular) are conceptually processed more like singulars or plurals. Our results suggest that when the judgment to be made is grammatical (i.e. would you use "ist" (is) or "sind" (are) with this noun-- critically in German, collectives are always grammatically singular and therefore participants are making identical responses to singulars and collectives), collectives pattern with the plurals with respect to the SNARC effect (Patson, Warren, Hurler & Kaup, under review). This is an exciting finding and opens the possibility of using this paradigm to answer other interesting questions about numerosity in our language representations.

Syntactic Processing A third thread of my work investigates syntactic processing and syntactic complexity. This started with my work as a student and postdoc investigating the way the processing demands of syntactic structure building and referential processing might interact (Warren & Gibson, 2002, 2005). This line of work has since extended into the following questions and partial answers: (a) When do we build syntax?-- probably whenever we can. My work suggests that unimpaired college students and even adults with aphasia might build syntax predictively- even before the relevant bottom-up input (Hayes et al., 2016; Warren & Dickey, 2011; Warren et al., 2016). But I've also found that syntactic processing might contribute to readers pausing at the ends of clauses (Warren, White, & Reichle, 2009); (b) What drives our syntactic choices? -- I've researched factors that influence whether potentially reciprocal verbs are interpreted reciprocally or transitively (Patson & Warren, 2014); (c) How does syntactic difficulty influence eye movements in reading?-- I provided a potential answer to this when Erik Reichle and I modified the EZ Reader model of eye-movement control in reading to allow it to capture effects of syntactic and other higher-level language processing difficulty (Reichle, Warren, & McConnell, 2009). I've also investigated the relative timing of lexical vs. syntactic processing effects in eye movements (Warren, Reichle, & Patson, 2011); (d) Do we transfer our syntactic processing routines in the early stages of learning a new language?-- My work with Natasha Tokowicz suggests that yes, language learners rely on syntactic cues from their native language when processing a new language (Tokowicz & Warren, 2010; Tuninetti, Warren, & Tokowicz, 2015). We have ongoing experiments in this vein testing structures in Arabic and Slovak.

I have also started a new collaboration with Duane Watson and his group at Vanderbilt aimed at investigating the effects of reading experience on syntactic processing during reading in large and diverse samples of adults and adolescents.