Systematicity and Variation in Word Structure Processing Across Languages: a Neuro-Typology approach (SAVANT)

In general, research on how the human brain processes language has mostly focused on a very small set of familiar, related European languages like English, Dutch, German, Spanish and French. We know almost nothing about how the brains of speakers of most of the world's languages respond to even simple tasks like processing a single word. Not only does this limit our scientific knowledge, it also has the effect of making only a few languages seem worthy of scientific study.
Speakers of languages know many things about their language that they have never been taught, and probably do not even know that they know. For example, speakers of English know that the prefix re- attaches to verbs (eg. refill, repaint). Attaching re- to nouns creates impossible words: reidea, resofa. Speakers of English also know that re- can only attach to specific kinds of verbs like 'fill' and 'open' that describe an event which causes a change of state (fill = cause to become filled). If we attach re- to verbs which do not have the right meaning, the result is not a possible word of English: reknow, resmile. Every human language has rules like this which place limits on the way pieces of the language can be combined to make new words. A combination that is impossible in English, like relaugh, is perfectly fine in French (resourir) or Greek (xanagelo).
We can test people's unconscious knowledge of these kinds of restrictions with a very simple experiment, in which we show speakers real English words, and impossible words like reknow, one word at a time, and ask them to say judge how good a word it is. As people are doing this experiment, we use neuroimaging equipment to record their brain activity to find out how their brains process these impossible words. We have found that both English and Greek speakers' brains produce the same patterns of activity. When speakers of these two languages read a word that combines a prefix or suffix with a stem of the wrong category (eg. reidea), their brains produce more activity in a region in the left temporal lobe about 200ms after they first see the word. But Greek and English speakers produce a different response when they read a word that combines a prefix or suffix with a stem that is the right category, but has the wrong semantics (eg. reknow): in this case, their brains produce more activity about 450ms after they first see the word, and the response comes from the frontal lobe. Just from these two languages, it seems that human brains have the same kinds of responses in the same locations, and at the same times, to similar kinds of linguistic information across different languages. But Greek and English are only two of thousands of languages.
Our project will test speakers from a wider range of different languages, which have a range of different word formation rules and processes. These languages will include other Indo-European languages like Slovenian and Bosnian/Croatian/Serbian, in which verbs usually have four or five separate pieces (morphemes), to mark things like tense, aspect, and the person, number and gender of the subject; and Bangla, a language in which verbs often change their pronunciation to mark grammatical features (like English sing~sang or write~written). We will also include Arabic, in which words are made by combining a 3 consonant root like KTB with different vowel 'melodies' (eg. kitab = book, kaatib = writer, katab = 'to read'), and Tagalog, a language in which words can be made by doubling part of the word (eg. 'halo' = 'a mix', 'hahalo' = 'to mix'), or by inserting an affix into the middle of the word (eg. 'h-in-alo' = 'it was mixed'). We'll use the same simple experiment to show speakers of these languages words which break either a category rule (reidea) or a semantic rule (unsmile) to see whether the brain responses we observed for English and Greek are truly universal, and how different word-formation processes might use the same basic language network in different ways.

This project investigates how speakers of a diverse range of languages solve the basic problem of detecting, recognising and interpreting constituent pieces of complex words, by recording their brain activity while they read and judge the wellformedness of familiar and novel words in their language. By employing a very simple paradigm, that can be replicated across all the languages in our sample, we can both better understand the shared neurocognitive bases for the human language capacity, while also uncovering the neurobiological basis for the distribution of different linguistic patterns across the languages of the world. The systematic comparison of the responses evoked by the same manipulations across a range of languages will lead to new discoveries and to the refinement of existing models of how word-internal linguistic structure is parsed.

In addition to addressing both big picture and fine-grained questions about the representation and processing of complex words, our project has a fundamental capacity building goal. We can only enrich the set of languages for which there is neurolinguistic data by a small number in this project. But we can enable a much larger set of languages and languages speakers to be investigated, (a) by training our postdocs, postgraduate and undergraduate research assistants, and, through a summer school program, local secondary school students, all of whom will be speakers of under-investigated languages, and (b) by developing the kinds of databases that are essential for language science research and clinical applications, and training others to build their own. The program to give UG and secondary school students the skills and motivation to investigate the languages in their lives critically includes public showcase events, co-organised with the advice and advocacy organisation Bilingualism Matters (in London) and the science communication space HiSa Experimentov (House of Experiments) (in Ljubljana).

Significantly widening the range of languages investigated by neuroscience in this way not only ensures that our theory building and testing is much more representative of the reality of linguistic experience across the globe, it also firmly situates the native speaker researchers and participants as stakeholders with a vested interest in research that replaces cultural insecurity and marginalisation with a deep sense of pride and confidence, enhances the prestige of their languages and celebrates their linguistic capacities and the value of multilinguistic communities.