This study was designed as an extension of previous research in which pronunciation of a foreign language was experimentally manipulated using alcohol or hypnosis, t o confirm a relationship between permeability of language ego boundaries and pronunciation. The present study used Benzodiazepine (Valium) to manipulate pronunciation in Thai. Seventy-five subjects were assigned to one of four treatment conditions: Placebo, 2 mg., 5 mg., or 10 mg., and subsequently tested on the Standard Thai Procedure (STP) and the Digit Symbol Test. Results seem to suggest that Benzodiazepine (Valium) facilitates the empathic sensitivity of the subjects to the tester rather than to the voice on the tape. The combined findings of the Valium and alcohol studies are interpreted as supporting the theoretical connection between language ego boundaries and ego boundaries in general and illustrate the extraordinary sensitivity of the test to fluctuations in the state of the subject's ego.In successive publications (Guiora 1967, Guiora 1970, Guiora 1972, Guiora et al. 1968, Taylor et al. 1969 we developed the hypothesis that individual variation in the ability to approximate nativelike pronunciation in a second language is, in part, determined by certain psychological variables best subsumed under the construct empathy or, more broadly, the concept of permeability of ego boundaries. These variables have been investigated in three manifestations: (1) as stable characteristics of adults, (2) as milestones in the growth curve in the course of development, and (3) as experimentally induced behaviors.In our program dealing with the third of these propositions, i.e., the experimental induction of behaviors, we had planned to study the effects of different agents (e.g., alcohol, drugs, hypnosis) on the target behavior (i.e., ' We gratefully acknowledge the invaluable counsel of J.E.K. Smith-so generously extended-in matters of data analysis. We are indebted to Allison Edwards, Dobroslaw Lachowicz, James Runner, and Douglas Woken for their assistance in running the experiment. Stephen Tyma has read an earlier version of this paper, and we wish to thank him for his comments.