Predictive Processing and Linguaculture Learning
By: Joseph Shaules
Ayako in Oz
Ayako landed in Los Angeles on her first trip abroad. She passed through immigration and customs, excited to hear English around her. Before heading for the exit, however, she walked into the women’s restroom. She swung open the door but stopped in her tracks. In front of her was a line of toilet stalls, but unlike in Japan the doors did not go all the way to the floor. In fact, she could see someone’s legs inside.
She stood for a moment, embarrassed to imagine herself being so exposed. Her mind raced . . . Should I look for another restroom? . . . Aren’t people ashamed? . . . What kind of country is this? . . . Then, suppressing her discomfort she decided she had no choice and proceeded. Welcome to America!
Ayako told me this story years after it happened. Yet she recalled the smallest details . . . the legs inside the stall . . . her feelings of embarrassment . . . her train of thoughts. It was a small experience but it left a big impression. Such experiences are called Oz Moments–named after a scene in the movie The Wizard of Oz. The main character, Dorothy, is in her house when it gets carried away by a tornado. It lands with a crash far away, in the magical kingdom of Oz. When she opens the door, she steps out into a wonderland unlike anything she could have imagined. Then, famously, she says to her dog . . . I’ve a feeling we’re not in Kansas anymore, Toto! This is the Oz moment–the feeling of being in a foreign world.
Oz moments and predictive processing
Oz moments have a lot to teach us. They give us a window into the central role that predictive processing plays in language and culture learning. To navigate everyday life, our brain acts as an inference machine—an “organ that actively generates predictions of its sensory inputs using an internal or generative model” (Friston, 2011, p. 248). These models relate not only to our perceptions of the physical world, but also the cultural and linguistic worlds we inhabit (Gastaldon et al., 2024).
Ayako stopped short because the mental model she carried within her for “restroom” did not match what she saw. Figure 1 is a representation of the cognitive processes triggered by Oz moments (Shaules, 2014). In the upper left corner we see the initial trigger of Oz moments–a mismatch between the environment and the mental models we use to navigate our experiences. We have an emotional response, do a reflective sorting and analysis, respond intuitively to the situation, and then are affected by the experience. This hints at the complexity of the mental models that predictive processing relies on, and helps us understand why Ayako remembered her Oz moment years later. Oz moments stick with us because they are part of a process of deep learning–one in which we do more than add to our store of knowledge; we are adjusting the mental models we use to navigate in the world.
Predictive processing and embodied simulation
Predictive processing is central to both language and culture learning. Learning a foreign language involves much more than inputting a new information code. We must create complex experiential models in order to navigate in the world using that language. Larsen-Freeman (2011) points out that constructing such models is a highly complex and embodied process. It is “not just about adding knowledge to an unchanging system. It is about changing the system” (p. 57).
There are competing theories about the predictive processes involved in language use (Gastaldon et al., 2024). One thing that’s increasingly clear, however, is that these predictive models involve much more than a manipulation of mental symbols. According to embodied simulation theory, language use involves a simulation of lived experience (Bergen, 2012). Our personal and cultural experiences, emotions, and bodily states are all intimately tied to language use. When you hear the word dog, for example, the dog that comes to mind is of a particular breed–perhaps the one you had as a child. When you hear the words wet dog you may remember the smell of wet fur, or the sight of a dog shaking itself off. The linguistic models we rely on are grounded in lived experience.
The linguaculture tree
From the perspective of embodied simulation and predictive processing, there is no clear line separating language from culture. The linguaculture tree (Fig. 2) can help us visualize this connection. Language is represented by the trunk, branches, and leaves, while culture is represented by the roots. This emphasizes that language and culture are two complementary parts of a single, complexly interacting system (Risager, 2015).
It’s possible, of course, to look at language in disembodied terms–as a collection of lexical items or a set of grammatical structures. Many language classes are taught in this way. But it’s extremely difficult to learn a language purely as a mental code. To test this, try mastering Latin! Language is alive with a dynamic complexity that emerges from the interaction of its family of speakers. As members of a linguistic community interact, their shared experiences give rise to linguistic and cultural patterns. Linguistic meaning does not come from a dictionary–it is grounded in the shared experience of linguaculture communities.
Let’s take a simple example. A Japanese student of English may learn that have can be translated as 持つ (motsu). Yet a conceptual translation does little to inform the predictive models we use to communicate. If you see someone sneeze, then hear them say I have . . . you can anticipate the next word . . . a cold. If someone points at their watch and says I have . . . you can anticipate a time expression (e.g., I have only a few minutes). And when you hear the name Martin Luther King, Jr, you may anticipate another use of this word (I have a dream!) The word have is not simply a concept–it is part of a complexly embodied system of meaning grounded in the experiential worlds of English speakers.
It’s a mistake to think of language as an information code that is divorced from culture. It’s true, of course, that there is no singular “English culture.” Americans and Brits do not own English! This misses the point, however, that the mental models that we rely on are grounded in lived experience. For a Japanese person, learning to use English is a foreign experience. Whether they are speaking with an Australian, Argentinian, or Armenian, learning involves the construction of a new (and foreign) embodied predictive linguaculture model.
The psychology of linguaculture learning
The term predictive processing can be misleading. It can give the impression that the brain processes information like a computer. But our mental models are grounded in lived experiences . . . sights, sounds, impressions, emotions, and intuitions. As we gain fluency, we get a feel for a foreign language. We feel more ourselves and get comfortable with interacting in new ways. This is not a psychologically neutral process. It is disruptive to our cognitive habits. Because our mental models are integral to who we are, we easily feel stressed, awkward, or stupid when trying to use a foreign language.
If you ask Ayako about her first experience in the US, she will tell you it sparked her interest in travel and made her want to improve her language skills. This was not simply because she had a chance to “use English” with Americans. Her foreign experiences challenged her personally, even as they enriched the mental models she uses to navigate the world in English. Language and cultural learning is a highly psychological process.
So where does this leave us as classroom teachers? An understanding of predictive processing teaches us that the psychological impact of language learning is similar to that of adapting to a new culture (Shaules, 2019). To that end, we need to provide a secure environment that allows for trial and error learning. Learners need emotional engagement, meaningful interaction, and the freedom to experiment and make mistakes. This will help them create a new predictive model. Like Ayako and Dorothy, they are entering into a new world of experience and self. Our job is to help them on their journey.
References
Bergen, B. K. (2012). Louder than words: The new science of how the mind makes meaning. Basic Books.
Friston, K. (2012). Prediction, perception, and agency. International Journal of Psychophysiology, 83(2), 248–252. https://doi.org/10.1016/j.ijpsycho.2011.11.014
Gastaldon, S., Bonfiglio, N., Vespignani, F., & Peressotti, F. (2024). Predictive language processing: Integrating comprehension and production, and what atypical populations can tell us. Frontiers in Psychology, 15, 1369177. https://doi.org/10.3389/fpsyg.2024.1369177
Larsen-Freeman, D. (2011). A complexity theory approach to second language development/acquisition. In D. Atkinson (Ed.), Alternative approaches to second language acquisition (pp. 48–72). Routledge.
Risager, K. (2015). Linguaculture: The language-culture nexus in transnational perspective. In F. Sharifian (Ed.), The Routledge handbook of language and culture (pp. 87–99). Routledge.
Shaules, J. (2014). The intercultural mind: Connecting culture and cognition. Intercultural Press.
- Shaules, J. (2019). Language, culture, and the embodied mind: A developmental model of linguaculture learning. Springer.
Joseph Shaules has worked in intercultural education for more than 25 years. He is a Specially Appointed Professor at Keio University. He is the director of the Japan Intercultural Institute. He teaches in the Tsuda University Graduate Program in TESOL. He is widely published. He hosts The Deep Culture Podcast.