Title: In Honor of Michael K. Tanenhaus for Receiving the 2018 Rumelhart Prize
Abstract: It is a great honor to provide this brief introduction to the Special Issue of topiCS honoring Michael K. Tanenhaus and his receipt of the 2018 David E. Rumelhart Prize of the Cognitive Science Society. Michael Tanenhaus is one of the most influential researchers today in cognitive science, especially within the subfield of human language processing. He has made numerous contributions to the study of language over his impressive career. His work has done more than simply advance our understanding of key facets of the language comprehension system; it has transformed the field of language processing. Indeed, one must look back to the seminal work of Noam Chomsky, George Miller, Thomas Bever, and Jerry Fodor, which gave birth to modern psycholinguistics in the early 1960s, to find research that has reshaped the field as dramatically as the work of Tanenhaus. Through ingenious theory development and experimentation, he has marshaled evidence for the highly interactive nature of language comprehension; he has demonstrated how the study of language processing both informs and is informed by other aspects of cognition, including visual and auditory perception, attention, representation, and social-pragmatic interaction. In fact, Tanenhaus's most significant contribution to the field of language processing may very well be that he made the field relevant again to those outside its core members. Many of Tanenhaus's accomplishments are tied to his development of what he dubbed the Visual World (VW) paradigm (Tanenhaus, Spivey-Knowlton, Eberhard, & Sedivy, 1995, 1996). In the VW paradigm, participants’ eye movements are recorded as they use language in natural ways while interacting with the physical world and with conversational partners. By time locking auditory linguistic events to the eye movements that language users make to objects, Tanenhaus revealed the exquisitely timed process of how humans connect speech to the referent world on a moment-by-moment basis (e.g., Allopenna, Magnuson, & Tanenhaus, 1998; Spivey, Tanenhaus, Eberhard, & Sedivy, 2002). And most importantly, through ingenious and creative application of the VW paradigm, Tanenhaus showed us how to study language as an integrated system, from speech processing (e.g., McMurray, Tanenhaus, & Aslin, 2002), to syntactic parsing (e.g., Spivey et al., 2002), to referential processing (e.g., Sedivy, Tanenhaus, Chambers, & Carlson, 1999), to semantic and pragmatic processing (e.g., Grodner, Klein, Carbary, & Tanenhaus, 2010), to interactive conversation (e.g., Brown-Schmidt & Tanenhaus, 2008). In all of these domains, he has left a lasting influence both methodologically and theoretically. Indeed, the breadth of Tanenhaus's thinking, combined with his uncanny experimental skills, inspired a generation of researchers to take a fresh look at what it means to communicate by language. His influence goes well beyond psychology, as he has engaged in a significant way the fields of formal and computational linguistics—through his work on thematic roles, phonemic processing, prosody, and pragmatics. A person with Tanenhaus's depth and breadth of thinking is well deserving of an award that honors David E. Rumelhart, who, like Tanenhaus, showed great skill in identifying issues worthy of research and valued intellectual engagement with other disciplines. One of Mike's most well-known attributes is his striking ability to mentor and develop success in others. The field of language processing is populated by many who went through his lab at the University of Rochester as a mentee—hundreds of undergraduate students, graduate students, postdoctoral researchers, and faculty colleagues. Perhaps then it is surprising that this Special Issue of topiCS has just three articles, rather than a long volume by these many (truly grateful) mentees. The reason for this state of affairs brings up two other well-known attributes of Mike: his quirkiness and humbleness. Mike seldom takes the obvious path or seeks the limelight. As a result, when asked to select those who might present at the Cognitive Science conference in the Rumelhart Prize Symposium in his honor, and by tradition offer published papers in topiCS, Mike decided against the usual path of selecting among the army of students who would have happily praised him and expressed his influence; he instead decided to select those who have influenced him, purposely avoiding the praise (and no doubt the awkwardness of excluding some students from giving a talk!). So, for the Rumelhart Prize Symposium, he chose senior and junior researchers who were never his mentees; instead, these people had influenced him in some interesting and important way. Two of these researchers, Dana Ballard and Herbert Clark, were selected because Mike wanted to acknowledge and pay homage to significant research programs in and of themselves, which inspired and guided the development and exploration of the VW paradigm. Dana Ballard is a leader in the field of computer vision; his work focuses on computational theories of how the brain implements vision and motor control. Dana was a longtime colleague of Mike's at the University of Rochester. And as Mike has noted to me, one of the most important influences on the development of the VW research was Ballard's work on animate vision (Ballard, 1991), coupled with his work with Mary Hayhoe (e.g., Ballard, Hayhoe, & Pelz, 1995; Hayhoe & Ballard, 2014), which represented some of the earliest applications of video-based head-mounted eye-trackers to studying vision in natural tasks—an approach that is necessary if one is to study perception embedded within goals and actions. Moreover, the first VW paradigm experiments by Mike and his students were conducted in Dana and Mary's lab. In his article (Ballard, 2021), Dana offers an insightful and engaging review of his development of the computational approach to active vision, tracing the development of these ideas from Marr to today. “It was the content of the conversation that left a lasting impact, but also the fact that Herb, as he has continued throughout his career, took the time to seriously engage and treat a graduate student as an intellectual equal. This is something I've tried to emulate; some of the most satisfying moments of my own career come when someone in the field (to my surprise) tells me about how something I said in a conversation at a conference early in their career impacted their work.” Second, and less personally, it was Mike's appreciation of Herb Clark's research program that approached language as joint action that highlighted the importance of embedding real-time language processing in grounded language use. Indeed, Mike observed a deep parallel between Dana's research program and Herb's, though they address very different domains and one will rarely see them both cited in the same paper. In his contribution to this special issue (Clark, 2021), Herb offers a commentary on the importance of anchoring in language, that is, the fact that utterances are tied to the detailed representation of the situation of use including representations of “the speaker, addressees, place, time, display, and purpose of that particular utterance.” Herb points out the challenges of taking into account utterance anchoring when studying language use, but he also reveals the significant pitfalls for researchers who choose to ignore it. His article is likely to become an important read for any psycholinguist who wishes to develop further ideas about how language processing is shaped by its many situations of use. Last but not least, Laura Dilley, with Mike as a coauthor, offers the final contribution to this special issue. Dilley is a well-known speech and hearing researcher, who has made significant advancements to the study of speech and prosody. Dilley's work, and the article found herein and co-authored with Meredith Brown, highlights a forward-looking aspect of the VW approach. As Mike has often told me, some of the most exciting work enabled by the VW paradigm is in speech processing because speech perception can be studied both in continuous speech and in natural contexts, allowing for an appreciation of the importance of understanding the speech signal, even when studying higher level processing. Brown, Tanenhaus, and Dilley's (2021) article highlights this point but also shows how thinking about the speech signal in real-time processing can help inform work at multiple levels of the system. More broadly, Dilley's research program often reexamines known “low-level” speech phenomena (such as the distal rate effect) that, when coupled with VW experimentation, inspires a novel approach to understanding the interplay between high-level and low-level processing of language, including an understanding of its neural processing. Indeed, Dilley is somewhat unique among those studying speech perception in that she frequently examines experimentally so-called casual speech. The nature of the problem of speech perception changes when one realizes that in naturally occurring casual speech cues to syllabic boundaries and word boundaries are often highly ambiguous. So, just as Ballard's work is informed by thinking about active vision and Clark's work is driven by interactive conversation—vision and language use in its natural context, as it were—Dilley's work is informed by studying naturally occurring speech. In closing, this volume is a tribute to Michael Tanenhaus's ability to see connections across topics and disciplines. The three very different articles in this selection have deep connections that he observed and appreciated like no other. This talent is no doubt one of the reasons that Mike's own contributions to the fields of language and language processing have been significant and far reaching. Not only has he connected topics and fields of study in interesting, important and creative ways, he has reshaped how we think about what it means to say that we understand language.