To test the thinking-for-speaking hypothesis, cross-linguistic research has been conducted in the domain of motion events ― movements of entities through space ― in a number of different languages. On the basis of where a language encodes path, Talmy (1985, 1991, 2000) has classified languages into two types: verb-framed and satellite-framed. In verb-framed languages (Romance, Semitic, and Japanese), path or directionality is encoded on the verb, while in satellite-framed languages (Indo-European except Romance, Finno-Ugric and Chinese) it is encoded on a satellite, a particle.

Spanish and English exemplify these two typologically different languages (Talmy, 1991). In Spanish, a verb-framed language, motion and path are indicated by the verb, and manner if present in speech is indicated outside the verb by an adjunct, an adverbial such as a gerund or a phrase. In contrast, in English, a satellite-framed language, motion and manner are indicated by the verb, and path is indicated by a satellite, a particle. Research on Spanish and English speakers’ narrations of motion events show that Spanish speakers tend to describe states and to elaborate descriptions of settings and their path gestures tend to fall on the verb, whereas English speakers tend to describe processes and to accumulate path components and their path gestures tend to fall on the satellite (Slobin, 1991, 1996a, 1996b, 2003, 2004; Berman and Slobin 1994; McNeill and Duncan, 2000).

Slobin has hypothesized that many language patterns that are acquired in childhood are “resistant to restructuring in adult second language acquisition” (Slobin, 1996a, 89). Here, the differences in typology between languages are important. If different patterns of thinking for speaking exist in the first language (L1) and the second language (L2), then learners must learn another pattern of thinking for speaking in order to be proficient speakers in their L2 (Stam, 1998). At issue is how to ascertain what language learners are thinking for speaking in when speaking their second language: their L1, their L2, or a combination of the two.

Traditionally, learners’ performance and production errors have been used as a basis for assessing their level of competency in their L2 (Ellis, 1986). Although this provides us with some important information about learners’ ability, it does not provide us with a full picture of their mastery of the language. Stam (2006a, 2006b, 2008, in press) has demonstrated that looking not only at learners’ speech but also at their accompanying gestures gives us a more complete picture of their progress in their L2.

The gestures that are referred to in this talk are spontaneous movements of hands and arms that accompany speech. They are external manifestations of speakers' on-line thinking-for-speaking and express the same meanings as well as perform the same pragmatic functions as speech (Kendon, 1980; McNeill, 1992, 2000). Speech and gesture can represent the same entities, or they can complement each other.

In this talk, I will demonstrate the different patterns of thinking for speaking about motion that Spanish speakers and English speakers have. Then I will discuss two studies: (1) a cross-sectional study that compared the narrations of intermediate and advanced learners of English as a second language with those of monolingual Spanish speakers and native English speakers and (2) a longitudinal study that examined how patterns of thinking for speaking about motion changed linguistically and gesturally for an advanced English learner from 1997 to 2006 in both her L1 and L2.