An added complication is that language is not just grounded in such complex multimodal representations, it also helps us to abstract across them. Language provides a heuristic shortcut that allows us to process and share our experiences, largely independent of time and space. This makes language an incredible cognitive tool but also a deceptive cognitive despot, creating the illusion of stable representations that objectively “name” reality, as claimed in the Sapir-Whorf hypothesis (Whorf, 1956/2011). For example, it has been claimed that there are 11 basic colours, which are experienced similarly across all cultures (Berlin & Kay, 1969). This seems fine when working from the English language (which has 11 basic colour words), but is quite rightly challenged by, among others, speakers of Russian. Their language distinguishes 12 basic colours, due to a compulsory distinction between an area of the colour spectrum English-language speakers would label as light blue and dark blue (see Pavlenko,2014, for a discussion of this and other effects of language on thought). Language co-constructs our conceptual knowledge, including, if Lisa Barrett (2017; see the May, 2018 Think Tank) is right, our emotions. The language(s) we speak alter how we perceive reality, such as whether we see an object as a cup, a mug, or a glass, although we are typically not aware of this.

In sum, if language is embodied, not purely symbolic, when we teach our students that “cow” means “牛” (ushi), we should not assume that we all have exactly the same understanding. When discussing a word like “cow,” the conceptual difference (or distance) might be greater between rural folks like myself and urbanites than between speakers of two different native languages. I’m not suggesting that the language a person speaks dictates the whole content of their concepts. As I said earlier, it is our experiences that dictate our concepts, and language is a very important part of human experience, but it is only a part. The point is that although the conceptual knowledge underlying our meaning for any linguistic label shares commonalities, it probably also differs more than we realise, at any one time. And the linguaculture we grew up in can be a big differentiator. When I taught Japanese students in the U.K., their previous teachers had clearly tried to teach them the pronunciation difference between rice and lice, but it was a real shame nobody had thought to teach them that British people mainly eat long-grain, dry rice, so these learners wouldn’t have been disappointed when ordering it. Intercultural miscommunication is not just a language issue, but also a multimodal conceptual one.

Overall, the embodiment of language suggests why deep learning of a foreign language may be particularly problematic in the classroom learning context (Shaules, 2019). One hypothesis is that learners initially link foreign language words to their nearest native language translation equivalents and thus to conceptual knowledge. As proficiency develops, direct links between foreign words and the concept they refer to become stronger (Kroll & Stewart, 1994). But if all we give students in classrooms is words and phrases, without pictures or other stimuli or multimodal mental imagery, let alone embedding the language in the rich context of situational use, how can we expect them to create an embodied foreign language? Maybe, just maybe, teachers need to add embodiment of the language to their list of things to worry about students experiencing, beyond vocabulary and grammar and pragmatics (there is evidence that gesture may improve vocabulary acquisition; for a review, see Macedonia & von Kriegstein, 2012). Or maybe we need to factor in the poverty of the stimulus of the classroom when forming our expectations of what students can achieve.