Theories of speech perception are complex enough that they require computational models to implement their principles and derive predictions by simulation. Yet until recently, our cognitive models could not address the fundamental challenge of speech perception: listeners achieve phonetic constancy despite a many-to-many mapping (lack of invariance) between acoustics and perceptual categories (phonemes, roughly). Phonetic constancy has been out of scope because most cognitive models operate on abstract, simplified inputs rather than real speech. Current deep learning and transformer-based approaches to automatic speech recognition do not offer a viable alternative. While remarkably powerful, they are not understandable: their complexity obscures the mechanisms that allow them to work, limiting theoretical insights into human speech processing. We need a middle ground: models that are sufficiently simple that we can (begin to) understand them, yet sufficiently realistic to work on real speech. I describe our group’s approach, where we develop the simplest neural networks capable of processing real speech, and compare them to core human behavioral and neural benchmarks. I will discuss a recent discovery that has led us to refine our minimalist approach: modest increases in model depth allow a hierarchical division of labor to emerge that maps onto the cortical hierarchy that supports human speech understanding.

Despite language playing a central role in our everyday lives, the fact that language is a subject that can be researched scientifically is not self-evident for members of the general public (Wagner, Patson & Awani, 2022). In a world in which public trust in science and research is decreasing and public funds for science and education are contested, communication of what we do as researchers and also why we do it becomes more and more important. This also holds for the field of linguistics and more specifically, multilingualism research, where we know that a number of myths about what it means to be multilingual still persist in public opinion (Grosjean, 2010). Public discourse that frames linguistic diversity in schools as a problem, in particular, can have negative effects on multilingual children’s language development and wellbeing (Agirdag, 2017).

In this talk, I would like to present two projects in which we communicate research on language and multilingualism to young children and their parents and teachers at Radboud University. The first project, Kletskoppen (engl. ‘Chatterboxes’), is an initiative by language scientists from a range of disciplines at Radboud University and Max Planck Institute of Psycholinguistics (Nijmegen). We organise festivals and lessons about language, linguistics research and multilingualism for primary-school aged children from diverse linguistic and socio-economic backgrounds across the Netherlands. Kletskoppen’s main goals are (1) to increase children’s enjoyment of language, (2) broad their view of science and (3) encourage children to better appreciate their own linguistic repertoire, which for many is multilingual. More specifically, I will present findings from a recent impact study (Unsworth & Knopp, under review) that investigated the short- and longterm effects of participating in Kletskoppen activities on mono- and multilingual children’s attitudes towards multilingualism. The second project I will report on concerns a citizen science project I conducted with colleagues from Tilburg University: Taaldetectives (engl. ‘Language detectives’), in which children from 75 primary schools across the Netherlands collected a data-set of 5.780 examples of the linguistic landscape of the area surrounding their school. I will reflect on what it means to involve children from that age-group in language science, as well as their teachers and parents. I will also present some preliminary findings concerning their awareness of the socio-pragmatic function of language(s) in their environment as well as their attitudes towards languages and linguistic diversity in their environment.

• Agirdag, O. (2017). 8. Schools in the multilingual city. In G. Marini (Ed.), Urban Europe (pp. 67-74). Amsterdam: Amsterdam University Press. https://doi.org/10.1515/9789048535811-009
• Grosjean, F. (2010). Bilingual: Life and Reality. Harvard University Press.
• Wagner, L., Patson, N. D., & Awani, S. (2022). What does the Public Think about Language Science? Language, 98(4), e224–e249. doi.org/10.1353/lan.2022.0029

During language comprehension, adult speakers rapidly use the available grammatical cues to find out about the argument structure, i.e. the who-does-what-to-whom of the sentence. Across languages, the most common cues for assigning thematic roles are word order or case-marking. While adult speakers can efficiently use these cues when comprehending sentences, it’s not clear how and when children learn to do so. Several studies show that word-order is used from the early childhood, but the evidence is not clear for case making: a few studies point to an early acquisition, while other seem to show that it is developed gradually and is only mastered from 7 years onwards. Additionally, there is little evidence on how bilingual children acquire these cues in each of the languages they are acquiring.

In this talk, I will present EEG and behavioural evidence on how adult and children process word order and case marking cues in Basque, with a particular focus on Basque-French bilingual children. I will also discuss the influence of language exposure and prosody for the acquisition of morphosyntactic cues, and I will argue that bilingual children transfer sentence processing strategies between their languages.

Iconicity (a resemblance between form and meaning) in sign languages appears to be much more pervasive and structured compared to spoken languages. Currently, however, we know very little about how iconicity might impact the structure of the lexicon or whether iconic signs are processed differently in the brain. My colleagues and I have been exploring the nature of the distribution of iconic forms in the American Sign Language (ASL) lexicon, what drives the iconic depiction of semantic features, and whether iconic signs are perceived and processed differently than non-iconic signs. We have used Event-Related Potentials (ERPs) to examine whether iconic signs exhibit a distinct “neural signature” for deaf fluent signers or for hearing new learners and whether effects of iconicity are task-dependent. Thus far, this work indicates a) iconic signs have a unique lexical distribution that differs from that of iconic spoken words, b) salient/distinctive semantic features are more likely to be iconically depicted in signs, c) there appears to be no neural signature that tracks with the strength of iconicity during sign recognition for fluent signers, d) the effects of sign iconicity are task-dependent, facilitating picture-naming but not word-to-sign translation, and e) neural effects of iconicity diminish with learning. Overall, the results reveal linguistic consequences for grounding language in the body that shape the structure of the lexicon, but neural effects of iconicity may only occur under certain circumstances.

There is a variety of different evolutionary scenarios that hypothesize how human language might have evolved. While some suggest that language evolved from scratch in humans only, others propose that precursors to human language were already present in our last common ancestor shared with other primates. Consequently, comparative researchers suggest that at least some of building blocks necessary for language to evolve are shared with other primates. However, which aspects of primate communication are studied to shed light on language evolution heavily depends on which communicative modality – gesture, facial expressions, or vocalizations – is studied. I will focus on the gestural communication of nonhuman primates and the ongoing debate about how language might have evolved by evaluating findings from research on the different building blocks of language, and discuss if and how these data support a gestural origin of human language.