At the Edge of Grammar
Researchers at the University of Aarhus are looking at how people process different types of sentence structures and probing the limits of grammar. The focus of this research is the possibility of extracting elements across long distances from embedded clauses to the left edge of main clauses in Danish and English, as Professor Anne Mette Nyvad explains.
Observing the grammatical rules of a language is essential to clear and effective communication, but sometimes it is unclear whether a given structural output is in fact part of a language or the result of a performance error. As Principal Investigator of a research project based at Aarhus University in Denmark, Professor Anne Mette Nyvad is part of a team investigating the limits of human grammar.
The project deals specifically with the borderlands between what is possible and impossible in the grammatical systems of language, focusing on the borders between clauses, more specifically their left edge. Theoretical and experimental syntax are combined in pursuit of the overall goal of investigating how a fundamental property of human language interacts with unique characteristics of Danish and English.
The focus is on one of the defining characteristics of human language, namely the possibility of extracting elements across long distances from embedded clauses to the left edge of main clauses, e.g Who did you say that John met at the conference? Here, the fronted element who is spelt out at one location (at the left edge of the main clause) and interpreted in another (as the object of met in the embedded clause). The theoretical and experimental investigations of the project all more or less explicitly revolve around issues related to the link between these two realizations of the fronted constituent.
The project team are now comparing the Danish and English languages, specifically with respect to restrictions on the possibility of moving an element out of an embedded clause to the left edge of the matrix clause. Embedded clauses that ban movement from within them are also referred to as syntactic islands, because the elements inside them are effectively ‘marooned’. “For instance, if was to make the embedded clause in the example above into an embedded question, movement from within it would be unacceptable: Who did you say where you met?
However, counterexamples to the proposed universal constraints on long-distance dependencies abound in Danish and the
other Mainland Scandinavian languages and therefore these languages have been assumed to be more ‘tolerant’ of these long-distance dependencies than for instance English. In this project, we are investigating whether Danish is in fact more ‘lenient’ when it comes to movement out of embedded clauses than English,” Professor Nyvad outlines.
Embedded clause
This movement of an element out of an embedded clause to the front of a sentence, or its left edge, can theoretically speaking occur in three different ways. One is by effectively reorganising a sentence to turn it into a question. “For example, you might ask the question: Who did John say that I saw? In that case, you’re moving the element out
which 35 native speakers of Danish underwent brain scans, from which Professor Nyvad hopes to gain fresh neurolinguistic insights. “We looked at the brains of people as they processed these long-distance extractions, that is, movement from within the embedded clause. We wanted to see which parts of the brain are activated more when you process these very complex sentence structures,” she outlines. The project’s agenda also encompasses psycholinguistic research, which involves looking at behavioural data. “For instance, we can look at acceptability judgments, how people rate this type of sentence. We can also look at their response times or their error rates,” says Professor Nyvad. “We’ve done around seven psycholinguistics experiments, the fMRI study, and also quite a few corpus searches.”
“We looked at the neural activation patterns associated with the processing of these longdistance dependencies in order to determine which parts of the brain are activated more when it tries to understand these very complex sentence structures.“
of the embedded clause by way of so-called wh-movement,” says Professor Nyvad.
An element can also be moved out of an embedded clause by relativisation, which can produce sentences that may sound a little unnatural to an English speaker. “You might say something like; This is the exercise that I would be surprised if she completed, that’s an example of movement by relativisation,” continues Professor Nyvad.
“In Danish we can also move an element out through what’s called topicalisation, where we can have the topic in the first position, like; Him will get really mad if you talk to, which is not possible in English. We have a wider range of possibilities in Danish when it comes to moving an element out of the embedded clause.”
The project team is now looking at how people process these different types of sentence structures, work which involves several strands of research. In one part of the project researchers conducted an fMRI experiment, in
Researchers are looking at text corpora in the project and assessing what people actually produce, in both English and Danish, which provides a snapshot of how people communicate on a daily basis. This angle of research can be viewed as more ‘clean’ in a sense, as people produce these text corpora naturally. “We know that they are actually part of the language,” says Professor Nyvad. This research is split roughly equally between English and Danish. Bilingual people may also transfer some grammatical structures from their own native tongue to their second language, another topic of interest to Professor Nyvad and her colleagues. “At the outset, we thought we would find that Danish is more tolerant, that it allows more extractions from embedded clauses than English,” she outlines. “We wanted to then see whether native speakers of Danish and English transfer their constraints on long-distance dependencies to a second language? The evidence suggests that this is indeed the case.”
Mainland Scandinavian languages
This tolerance of extraction from embedded clauses has been prevalent in Danish for at least a hundred years, and while Norwegian and Swedish are similarly liberal, it hasn’t really been observed in English until relatively recently. It was previously thought there was a clear divide between English and the Mainland Scandinavian languages in this respect, but Professor Nyvad says evidence suggests this is not the case. “Our most recent experiment shows that English is more similar to the Mainland Scandinavian languages than has previously been assumed. So maybe there is more common ground than the theoretical literature will have you believe,” she explains. Beyond the scope of the current project, Professor Nyvad hopes to look at languages from outside the Indo-European family, which may have different features.
“The Indo-European languages – including English, Danish, Swedish and Norwegian – all have very similar traits. But the non-Indo-
European languages may be completely different when it comes to the possibility of island extractions,” she says.
This is a topic Professor Nyvad hopes to explore further in future by examining or comparing ten different languages spoken in Europe, looking for similarities and differences with respect to restrictions on extractions from island structures. This would be based on a similar experimental framework to the current project, yet it would be impractical to conduct fMRI studies, so Professor Nyvad plans to use electroencephalography (EEG) technology to measure activity in the brain and probe the limits of syntactical structures. “An EEG cap can tell you, when you’re listening to particular syntactic structures, whether you are responding to something that is for instance a syntactic or semantic anomaly,” she outlines.
“You can get an idea as to whether these complex syntactic structures are perceived by the brain – without us really knowing it –as something that is restricted in the syntax, semantics, or pragmatics of a language.”
At thE EdGE Of L ANGuAGE
At the Edge of Language - An Investigation into the Limits of human Grammar Project Objectives
This project deals with the borderlands between what is possible and impossible in the grammatical systems of language. More specifically, it is an interdisciplinary exploration of dependencies across clausal boundaries, focusing on how this fundamental property of human language interacts with special characteristics of Danish and English.
Project funding
This project is funded by The Danish Council for Independent Research. Grant ID: DFF9062-00047B.
Project Participants
Senior researchers:
• Dr Ken Ramshøj Christensen (Associate Professor, Dept. of English, Aarhus University)
• Dr Douglas Saddy (Professor, CINN, University of Reading)
Junior researchers:
Postdocs Dr Christiane Müller and Dr Katrine Rosendal Ehlers • PhD students Julie Maria Rohde and Maria Mørch Dahl
Contact details
Principal Investigator, Anne Mette Nyvad
Associate Professor
Department of English/Scandinavian Studies School of Communication and Culture Aarhus University
JensChr. Skous Vej 4 8000 Aarhus C
Denmark
t: +45 87163016
E: amn@cc.au.dk
W: https://projects.au.dk/at-the-edge-oflanguage W: au.dk/en/amn@cc.au.dk
published 2 books and 31 peer-reviewed papers on a range of linguistic topics within the fields of comparative linguistics, neurolinguistics, language processing and second language acquisition.