Connecting cognitive biases and typological universals in syntax

Lead Research Organisation: University of Edinburgh
Department Name: Sch of Philosophy Psychology & Language

Abstract

Languages can be very different from each other. For example, just focussing on the order of words, languages like English put adjectives before nouns ('red house') while languages like Thai put them afterwards ('house red'). Similarly, languages like Vietnamese put Numerals before nouns ('three houses'), while others, like the Kitharaka (spoken in Kenya), put numerals after ('houses three'). If word ordering was simply due to happenstance, we would expect to see all different orders appearing in equal proportion across languages, but we don't find that. In fact, some orders are very common, some are very rare, and some don't seem to appear at all. For example, many languages are ordered like English ('three red houses'), and many are also ordered like Thai, which is exactly the reverse ('houses red three'). But the Kitharaka order ('houses three red') is much rarer, and its mirror image ('red three houses') never seems to occur. Why is this?

One of the major controversies in the language sciences is whether we need to appeal to the basic set-up of the human mind to explain the ways languages can vary, or whether these properties are instead a result of cultural differences in communication and social interaction. A great deal of recent work coming from the perspective of psychology assumes the latter: that the properties of language can be boiled down to communication, interaction and the vagaries of history, while most work in linguistics assumes the former: there must be biases in the human mind that allow us to learn languages of particular types more easily than others. This project seeks to resolve that issue.

In order to do this, we test how well people learn languages of various types, to see whether their behaviour follows the general tendencies we see across real languages. Importantly, we use artificially constructed languages, rather than natural languages, in order to make sure that they only differ in the crucial respects. For example, we present English speakers with artificial languages that use word orders from Thai and Kitharaka. If Thai orders are more common across languages than Kitharaka ones because the former are easier to learn, then we should see this reflected in the behaviour of learners in our experiments. We can also see whether such patterns are always harder to learn, or if speaking a language which uses them-like Kitharaka-makes them easier to pick up in a new language. To do this, our experiments compare English, Thai, Vietnamese and Kitharaka speakers. If our learners all show the same kinds of patterns in how they learn our artificial languages that we find across real languages, that will suggest that the way languages vary is not random, nor is it entirely a product of historical facts. Rather it would suggest that there are universal cognitive biases at play.

We plan to look at not just the basic question of what orders appear, but also two other well-known cases where languages don't seem to vary randomly. The first relates to how words like adjectives and numbers are placed relative to the nouns they modify: most languages place them both before or after (like English and Thai), rather than putting them on opposite sides (e.g., 'two houses red', like Vietnamese). We will test whether this type of pattern is always easier to learn in a new language. Second, we will look at whether people prefer to learn languages with suffixes (e.g., 'cat-s') rather than prefixes (e.g., 'un-happy'). Both types are present in English, but most languages have (more) suffixes. Our project we will shed light on whether there are universal cognitive biases in language learning, if such biases are at play for the particular phenomena we look at, and how people's native languages affect these biases.

Planned Impact

This project investigates how cognitive biases shape language learning, and thus ultimately the distribution of linguistic patterns across the languages of the world. The approach we use is distinctly interdisciplinary: we target hypotheses put forward by linguistics using methods developed by psychologists and psycholinguistics in order to answer questions that have broad implications for the cognitive science of language. Beyond its relevance to this academic community, the project speaks to questions that are of interest to the general public. For example, what are the deep common features of human psychology that shape the unique trait of language that we share? How do these features interact with our experience, culture, and history to produce the rich variety we see across the world's languages?

In addition to these big picture questions, our research also has important implications for language learning-in particular how cognitive biases and our prior experience together impact how we learn a new language. For example, our results will show how prior language experience-coming from both first and second language knowledge-affects adults' ability to acquire linguistic patterns hypothesized to present a learning challenge. These findings will have clear implications for language education practitioners who work with learners from a range of linguistic backgrounds (e.g. English for Speakers of Other Languages (ESOL) support workers, or primary school staff teaching modern languages to children with English as an Additional Language). The research also speaks directly to an important strand of current research which seeks to uncover the effects, both positive and negative, of multilingualism. Bilingualism Matters is a Centre at the University of Edinburgh, based in the same school as the PI. The Centre is dedicated to research and knowledge exchange around language education and bilingualism and has a proven track record in shaping public policy in this field (e.g., advising Scottish Government on their 1+2 policy for language learning in primary schools). The proposed grant will benefit from the Centre's existing networks (including language practitioners, key organisations such as Education Scotland and Scotland's National Centre for Languages, and members of local and national government). These non-academic stakeholders will be invited to contribute to the project from the outset (e.g., by discussing how the research and tools developed in the project might feed into Continued Professional Development for practitioners), as well as being key audiences for the dissemination of research results.
 
Description The major aim of this grant is to uncover universal cognitive and perceptual biases that shape why languages look the way they do. By bias we mean implicit preferences that all humans share, driven by features of our cognitive or perceptual system. This project is focused on how such biases might explain common grammatical patterns we find in the languages of the world. The main method we use to investigate this involves miniature artificial "languages", which we ask people to listen to and then answer questions about, or use themselves. So far, the project has produced two key findings.

First, we have shown that when people learn nouns like 'apple' modified by adjective words like 'red' and number words like 'two', in an artificial language, they are better at learning how to order these words when the language has both modifiers on the same side of the noun. For example, a language which uses 'red apples' and 'two apples' or 'apples red' and 'apples two' is easier to learn than a language that uses 'apples red' and 'two apples', or 'red apples' and 'apples two'. This preference pattern, or bias, is the same regardless of the native language of the person learning--strikingly, even French and Hebrew speakers, whose language uses the equivalent of 'apples red' and 'two apples', still have an easier time learning a language where both modifying words are on the same side of the noun. This bias is even stronger in 6-7 year old children than it is in adult learners, and crucially, it aligns with a trend we see among the languages of the world. To summarize, the trend we see in languages for consistent placement of modifiers relative to the nouns they modify can be explained by a universal cognitive bias active during learning. This key finding largely puts to rest a longstanding debate about the role of human cognition in shaping this aspect of language.

Our second key finding similarly takes as its starting point an observation about the languages of the world, in this case about the order of words and grammatical morphemes--small parts of words like the English plural -s in 'apples', or un- in 'unhappy'. In particular, it has long be noticed that grammatical morphemes tend to follow words rather than precede them. In English for example, there are many more so-called suffixes like -s, that attach to the ends of words, than prefixes like un-, that attach to the beginnings of words. Most languages are like English in this way, and this has been claimed to reflect a universal perceptual bias that all humans have. Evidence for this bias comes from studies where people hear a novel (artificial) 'word' like 'tebo', and are asked which of two subsequent words is more similar to it 'teboga' or 'gatebo'. The first involves adding a small piece to the end of the word (akin to a suffix), while the second involved adding it to the beginning (akin to a prefix). In previous studies, English speakers were found to judge 'teboga' as more similar to 'tebo', suggesting that possibility that people group words together based on how similar they are their beginnings. This offers an explanation for why most languages have words that are constructed in this way. However, this result would simply reflect how English speakers are used to perceiving words, since most English words relate to each other in just this way (e.g., 'apple' and 'apples'). We showed that speakers of Kîîtharaka, a Bantu language spoken in Kenya, in fact have the opposite preference from English speakers: they judge words like 'gatebo' to be more similar to 'tebo'. This likely reflects the fact that Kîîtharaka is one of the rare languages that has more prefixes than suffixed. In other words, we showed that the behavior of English speakers observed in previous studies does not reflect a universal perceptual bias, but instead people's behavior in this task reflects their experience with words in their native language. This key finding calls into question a widely accepted claim about human perception of words, and points to the importance of research with diverse populations, beyond English and other European languages.
Exploitation Route These key findings have already been taken forward in a number of collaborations between the PI, other teams members and collaborating researchers. For example, the preference for consistent order of nouns and their modifiers is one example of a broader trend across languages that extends beyond nouns. We are working on testing whether this broader trend also reflects a cognitive bias, again using diverse populations of speakers. In light of our finding that words are not universally grouped based on similarity at the beginnings, we are also already collaborating with other researchers to better understand why it is that most languages use suffixes rather than prefixes. These findings also have implications for 'deep learning' models of language used in applications like Siri. Recent work in deep learning is focused on whether human biases and the biases of these neural network models are aligned. To the degree that they are, this suggests these models are successful in learning 'like humans'; to the degree that they are not, researchers have considered why and how to change this. Finally, these findings may be of interest to language teachers, since they show which features of language learners may be more likely to struggle with.
Sectors Digital/Communication/Information Technologies (including Software),Education,Other

 
Description Collaboration with Chulalongkorn University 
Organisation Chulalongkorn University
Country Thailand 
Sector Academic/University 
PI Contribution Our team brought a series of research questions along with proposed experimental designs to test them.
Collaborator Contribution Our collaborators, specifically Theeraporn Ratitamkul, helped to adjust the designs for Thai participants, and coordinated data collection.
Impact This collaboration has resulting so far in the publication: Martin et al. (in press), Linguistics Vanguard.
Start Year 2018
 
Description Inbal Arnon, Hebrew University of Jerusalem 
Organisation Hebrew University of Jerusalem
Country Israel 
Sector Academic/University 
PI Contribution My research team helped design an implemented studies, data was then collected at the University of Geneva. We analyzed these results and wrote them up for publication. The collaboration is currently continuing.
Collaborator Contribution Collaborator collected data and helped with write up.
Impact Culbertson, J., Franck, J., Braquet, G., Barrera Navarro, M., and Arnon, I. (2020). A learning bias for word order harmony: evidence from speakers of non-harmonic languages. Cognition, 204.
Start Year 2018
 
Description Julie Franck, University of Geneva, Switzerland 
Organisation University of Geneva
Country Switzerland 
Sector Academic/University 
PI Contribution My research team helped design an implemented studies, data was then collected at the University of Geneva. We analyzed these results and wrote them up for publication. The collaboration is currently continuing.
Collaborator Contribution Our collaborator and her team in Geneva collected data for the original study and helped with write-up. An extension to this study in ongoing.
Impact Culbertson, J., Franck, J., Braquet, G., Barrera Navarro, M., and Arnon, I. (2020). A learning bias for word order harmony: evidence from speakers of non-harmonic languages. Cognition, 204.
Start Year 2018
 
Description Tharaka University, Kenya 
Organisation Tharaka University
Country Kenya 
Sector Academic/University 
PI Contribution Our team brought a series of research questions along with proposed experimental designs to test them.
Collaborator Contribution Our collaborators, specifically Peter Muriungi and Patrick Kanampiu, helped to adjust the designs for Kîîtharaka-speaking participants, and coordinated data collection.
Impact None yet.
Start Year 2019
 
Description Queen Mary LingLunch Series 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Other audiences
Results and Impact Talk entitled: Order and Structure in Noun Phrases: preliminary results from Kîîtharaka
Year(s) Of Engagement Activity 2021
 
Description APA Inside Grants: UK Economic and Social Research Council Research Grant 
Form Of Engagement Activity A magazine, newsletter or online publication
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Magazine article in the American Psychological Society Observer Magazine about the grant.
Year(s) Of Engagement Activity 2021
URL https://www.psychologicalscience.org/observer/uk-economic-and-social-research-council-research-grant
 
Description Bantu 8 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Talk entitled: Order and structure in the Kîîtharaka Noun Phrase
Year(s) Of Engagement Activity 2021
URL https://sites.google.com/view/bantu8/home?authuser=0
 
Description Birmingham Lectures on Language Structure and Language Use 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Public/other audiences
Results and Impact Talk entitled: Grammar in Language Use: A Minimalist Perspective
Year(s) Of Engagement Activity 2021
URL https://www.youtube.com/watch?v=9OcfdBeKL3I&ab_channel=LinguisticsatTheUniversityofBirmingham
 
Description Colloquia: Yale University, UC Irvine, TedLab MIT 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Other audiences
Results and Impact Talks to Linguistics/Cognitive Science departments entitled: Word and morpheme ordering preferences are affected by meaning, transparency, and perceptual experience
Year(s) Of Engagement Activity 2021,2022
 
Description Conference on Computational Natural Language Learning 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Keynote talk entitled: Linking language learning and typology
Year(s) Of Engagement Activity 2021
URL https://www.conll.org/2021
 
Description Dubrovnik Conference on Cognitive Science 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Keynote talk entitled: From semantic primitives to conceptual structure: Experimen- tal investigations into the role of meaning in grammar
Year(s) Of Engagement Activity 2021
URL https://ducog.cecog.eu/programme
 
Description Edge Asymmetries in Morphophonology workshop at DGfS 43 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Talk entitled: A domain-general bias cannot explain the suffixing preference: Experimental evidence from English and Ki^i^tharaka
Year(s) Of Engagement Activity 2022
 
Description European Summer School for Logic, Language and Information 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Plenary lecture entitled: Cognitive biases in learning and generalization shape language
Year(s) Of Engagement Activity 2021
URL https://esslli2021.unibz.it/
 
Description Futurum Careers 
Form Of Engagement Activity A magazine, newsletter or online publication
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Schools
Results and Impact Materials for Futurum Careers
Year(s) Of Engagement Activity 2021
URL https://futurumcareers.com/what-if-we-could-understand-how-and-why-languages-evolve
 
Description IGRA Colloquium Universit├Ąt Leipzig, Germany 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Other audiences
Results and Impact Talk entitled: Hierarchy and order in the Kîîtharaka nominal phrase.
Year(s) Of Engagement Activity 2022
 
Description Jabberwocky Words in Linguistics Online Workshop, February 11th-12th, 2022, Amherst, Massachussetts & Bucharest, Romania 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Talk entitled: Cross-linguistic artificial language learning: Designing for diverse populations
Year(s) Of Engagement Activity 2022
 
Description LAGB Annual General Meeting 2021 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Talk entitled: Hierarchy and Predication in the Kîîtharaka Nominal Phrase
Year(s) Of Engagement Activity 2021
 
Description MPI Future of Linguistics Webinar 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Talk entitled: The future of linguistics is multidisciplinary, multimodal, multicultural
Year(s) Of Engagement Activity 2021
URL https://www.mpi.nl/events/future-linguistics-44
 
Description Symposium on the Syntax of the Bantu Noun Phrase 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact This symposium was organised as a way to disseminate our project to, and take on board ideas from a wider audience who are studying Bantu languages. This is an audience we do not typically engage with. The result was a very fruitful set of talks and discussion, which impacted the course of our research.
Year(s) Of Engagement Activity 2018
URL http://bantunp.ppls.ed.ac.uk/
 
Description Tharaka University College 2nd International Research E-conference 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Talk entitled: Does a universal hierarchical structure underlie word order typology? Experimental evidence from Kîîtharaka
Year(s) Of Engagement Activity 2020
URL https://www.facebook.com/Tharaka-University-College-113400900029992/videos/2nd-international-researc...
 
Description Utrecht Linguistics Colloquium 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Other audiences
Results and Impact Talk entitled: Order and Structure in Noun Phrases: the mystery of Kîîtharaka
Year(s) Of Engagement Activity 2021
 
Description Why Are There Differing Preferences for Suffixes and Prefixes Across Languages? 
Form Of Engagement Activity A press release, press conference or response to a media enquiry/interview
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Public/other audiences
Results and Impact Why Are There Differing Preferences for Suffixes and Prefixes Across Languages?, Association for Psychological Science news release, August 2020.
Year(s) Of Engagement Activity 2020
URL https://www.psychologicalscience.org/news/releases/suffixes-and-prefixes.html