Skip to main content

Testing the Limits – A Special Research Area Investigates How Flexible Language Really Is

Thanks to its variability, language helps to build bridges. Photo: Fotolia/
Photo : Fotolia/
Thanks to its variability, language helps to build bridges. Photo: Fotolia/

It’s Friday morning at a weekly market in Berlin. Those who visit the Maybachufer market in the district of Neukölln are treated to a little journey around the world – or at least that’s what it sounds like: Market stalls are a-buzz with languages. And yet people somehow understand each other – most of the time, that is. Among the crowd is a team of Potsdam linguists led by German philologists Heike Wiese and Ulrike Freywald. They are investigating why communication succeeds despite the apparently Babylonian diversity of languages as well as the structures of the utterances used at the weekly market, although there seem to be none at first glance. The research project is part of the new Collaborative Research Cluster (CRC; SFB: Sonderforschungsbereich) 1287 at the University of Potsdam that investigates the “Limits of Variability in Language” from various perspectives. In an interview, spokesperson Prof. Isabell Wartenburger and her deputy Prof. Malte Zimmermann offer insight into the SFB’s background and objectives.

The SFB researches the “Limits of Variability in Language”. What does that mean?

Wartenburger: What people say in a particular situation depends on many factors. The same is true for how we understand and interpret linguistic utterances. For this to work, language needs variation and this variability exists on all levels: meaning, syntax (sentence formation), and even phonetics. Each sentence is a bit different, depending on who I’m speaking to and about what. I’m not even able to pronounce a sound exactly the same twice. But there are also certain limits. I can’t do anything I want. After all, “A” always needs to sound approximately like “A”, otherwise it won’t be recognized, and I won’t be understood. In the SFB, we explore such limits of variability at various levels to see what they have in common – and how they differ. This will enable us to learn more about how language actually works and how it is that we are able to use language so flexibly.

Zimmermann: You have to understand that for a long time theoretical linguistics operated on the false premise that there is something like “the ‘one’ language” – German, English, French, for instance – and that linguistics is about identifying its rules, its abstract representations. As a result, you get a rather rigid, black-and-white system, in which a sentence is either grammatically correct or isn’t. At the same time, theoretical linguists have always known that this is not quite the case, that language use is more flexible. So, this is the SFB’s focus. The question is to what extent language can be varied before it becomes incomprehensible. A popular example is the discussion around German youth dialects. Researchers like Heike Wiese and Ulrike Freywald have been able to demonstrate that these dialects are certaintly no pidgin. It differs from the German taught at school, but it does follow rules. Also, even if someone says, “I’m going Kudamm”, any speaker of standard German would understand.

Why is language variability so important?

Zimmermann: For it to serve its purpose, language needs to be variable. On the one hand, we use language like a tool – with very diverse communication objectives. These can only be achieved if our tool is versatile. On the other hand, though, flexibility contributes to effectiveness. All speakers have their idiosyncracies: voice modulation, pitch, clarity, etc. Such differences need to be tolerable, otherwise we’d always be asking: “What do you mean?” If language were a rigid system, it wouldn’t function very well. We have to be able to tolerate deviations from the norm and still understand each other.

What are the limits of variability?

Zimmermann: Every language has a standard version, like standard German. In many situations – as has been pointed out – deviations from this standard can be tolerated, even if they are systematic. At the same time, there are rules we would consider unbreakable. For instance, when children acquire language and start learning to conjugate verbs, they might add regular endings to irregular verbs. They have realized that there’s a rule about the inflection of verbs and apply it to all verbs, so they would say, for instance, “I eated” or “I goed”. With enough feedback, they quickly learn that this is incorrect. But the other way around – incorrectly conjugating a regular verb – doesn’t happen.

Wartenburger: Exactly. It’s similar when we look at the syntax. The German language has many options for word order – but some are ruled out. It’s exciting to see whether these limits also exist when the possibilities of variation are regularly exhausted, such as at the weekly market, where people use different languages or a mixture of languages. This is what Heike Wiese and Ulrike Freywald’s project in the SFB focuses on. She and her team are investigating: What is said? What is possible? The assumption is that even at the weekly market there are certain rules that aren’t broken – automatically, that is, since speakers don’t consciously think about it. This would be a limit to variability, a fixed parameter of the language that can’t be easily ignored – whether at the weekly market or in the lecture hall.

Why are these limits needed?

Zimmermann: Without limits to variability, language systems might never have evolved. Language came into existence to better coordinate larger social interrelations. For this to work, the speakers of a language needed a common framework: a conventionalized system of rules, an unconscious agreement on the use of language that every speaker has internalized, which is not completely arbitrary and is relatively stable. Yet this framework is not as rigid as linguists had long thought. After all, language exists in the interaction between speakers. This is also where it changes, but changes don’t happen overnight. There isn’t someone who gets up in the morning and says: “Hey, from now on, I’m gonna systematically change the position of the subject in the sentence! It does start somehow, and if enough speakers use it, it’ll gain acceptance. But there are also other reasons altogether for changes: In Old Icelandic, for instance, there used to be two variable positions for the verb and the object in a sentence: “object before verb” or “verb before object”. (By the way, a similar variation in word order also exists in the German language between main and subordinate clauses.) As it happened, many older speakers of Icelandic fell victim to a smallpox epidemic in the early 17th century. At the same time, Danish – with a more rigid “verb before object” word order – became more prevalent in administrative language, so “verb before object” became the norm in the language system in the following generations. So, extralinguistic factors resulted in one variant remaining and becoming the standard.

Wartenburger: This is exactly the type of historical changes the SFB is researching. Ulrike Demske and Claudia Felser are exploring why certain structures of Early Modern High German died out while others are still in use. The most fascinating part is that they are testing out the extinct forms on contemporary speakers to find out whether our brains “accept” them as legitimate or “sort them out” as unnecessarily complicated. 

Was language more variable in the past than it is now?

Wartenburger: Yes and no. Without a doubt, Early Modern High German had linguistic options that disappeared over the centuries. People became increasingly mobile. There was more language contact, which contributed to more variability. Of the many options, some turned out to be more efficient than others and survived. But if speakers of different languages are able to communicate at the weekly market, this again testifies to the flexibility of language. Or take, for example, the language used in completely different channels, like on the internet – in blogs, on Twitter, Facebook, and others, as Tatjana Scheffler and Manfred Stede are researching in their project. Online, linguistic forms and structures that would've been considered impossible and completely ungrammatical 10 years ago are being tested out – and seem to become accepted. It’s a sort of feedback loop.

Zimmermann: In this cycle of change, there will always be phases when two competing variants exist until one of them gains the upper hand. Negation in French is a very good example: According to the standard grammar, it takes two parts, “ne” and “pas” – on either side of the finite verb – to create a negative sentence, but in colloquial French nowadays only the more economical simple “pas” is used. It is predicted that the “ne pas” variant will become extinct and that only “pas” will remain. This actually also happened to the German language. Until Middle High German, a small negative adjective was used in negative clauses in addition to the adverbial “nicht” – analogous to the French “ne”. As a matter of fact, new requirements in language contact constantly lead to new variants. For reasons of efficiency and conventionalization, some of these are later dropped. All of this happens unconsciously.

For you as researchers, what is more fascinating, the limits or the variability?

Zimmermann: (Laughing) The limits, absolutely.

Wartenburger: I agree. All the more so because variability has already been heavily researched. What makes our SFB special are the many projects that enable us to look at our research topic from very different angles. Some of us go to the weekly market and study the language used there with sociolinguistic methods; others study what the brain does by analyzing eye movements during reading. Ultimately – and this is very exciting – we will pool our results to find the links, correlations, and common features. 

Does that mean that limits are more difficult to establish?

Zimmermann: Yes, because they’re not as rigid as theoretical linguistics long thought them to be. Variations are actually easy to establish empirically – and to describe. You take a recording device and find people in village A speaking differently than those in village B. To a certain extent, this is trivial and was even sneered at by theoretical linguists for quite some time. However, it gets really interesting when you combine the two – the empirical study of variability and the theoretically reconstructable structures that limit them.

Wartenburger: Each project in the SFB is trying to establish these limits empirically or experimentally, to push variability as far as possible to see: Okay, it works up to this point but no further.

The research projects of the SFB are divided into three clusters. What are they?

Wartenburger: Cluster A looks at variability in language and its limits in situations of language contact and interaction. What variability do we see in such situations? How has it developed historically? What do contemporary speakers make of structures that are no longer standard? And how much variation is there in how people use language? That is, how does linguistic usage vary in situations like sitting here and giving an interview, writing an email, using Twitter, or doing some shopping at the weekly market …

Zimmermann: This includes – in the broadest sense – the social function of language. Cluster B covers the biological-cognitive component: how people process language in various situations or based on their cognitive abilities: babies and young children, people with aphasia, people who grow up bilingual, etc. Cluster C focuses on grammar theory. It deals with the intermediary system between the individual speaker and the interpersonal construct of a grammar. Grammar models are not yet well suited to capture linguistic variability. We hope to better understand and describe the mental system of language by combining empirical data and theoretical models. I think this is also one of the things that convinced the reviewers at the German Research Foundation (DFG).

Wartenburger: Combining the various levels is one of our major objectives. Ultimately, any of the projects could just as well have been assigned to another cluster. For instance, Gisbert Fanselow and Reinhold Kliegl are trying to teach people different syntactic structures that are not part of standard German. Of course, they’re interested in limits: What structures can I learn, and at what point does the brain say: “No, this is complete nonsense! I refuse to reproduce or accept this.” With exactly the same approach or result of this study, you could also go to the weekly market and try to find out: Are any of these unacceptable structures being used here? And how do speakers deal with them when they hear them? What do the brain and the eyes do? The good thing is that the projects can all be combined and benefit from each other. For this reason, it’s enormously helpful that almost the entire SFB is at Potsdam. We meet often and talk to each other. This is a big advantage over SFBs in which research projects are distributed all over Germany.

What are you researching in your own projects?

Wartenburger: The project that Sandra Hanne and I are pursuing is at the interface of syntax and prosody: Speech melody or prosody can play an important role in how we interpret a sentence. For example, in the German sentence, “Die Mutter küsst das Kind” [“The mother kisses the child”], there is subject-object ambiguity, so it could be understood as the mother kissing the child or the child kissing the mother, depending on my intonation. We are interested in how participants of varying ages register such prosodic cues – the intonations or stresses – in various settings, so we would test, for instance, an older person or a child in a noisy environment. Last but not least, we want to find out whether prosody could help people with speech impairments like aphasia to better or more quickly comprehend sentences – for instance, when prosodic cues are enhanced. This could ultimately have a benefit in speech therapy.

Zimmermann: Alexander Koller – who has since gone to Saarland University – and I are researching the limits of variability in semantic interpretation, that is, how we actually understand what’s been said. Basically, we want to find out if there are limits to the interpretation of statements imposed by syntactic structure. One example is the interpretation of sentences with two quantifying expressions such as: “Ein umgestürzter Baum blockiert jede Zufahrtsstraße” (literally: “A fallen tree blocks every access road”). English readers can understand this to mean that each road is blocked by a different tree, next to the implausible reading on which there is only one very big tree. In German, however, this is disputable. We want to empirically quantify the actual situations in German, English, and the West African language Akan. If there are differences, they have to be explained by the language system, since we can well imagine the situation described by this particular way of reading the sentence.

In addition to the 13 subprojects ...

Wartenburger: ... there is an internal PhD program chaired by Tatjana Scheffler to ensure that young researchers receive structured doctoral training so they can network better – including special events, retreats, and guest lectures for which they invite or suggest speakers. There is also the Q project for service and information infrastructure led by Shravan Vasishth and Ralf Engbert. It provides quality assurance and advises the participating projects on, for example, statistical analysis, data collection, data management, and ensuring sound scientific practice.

Zimmermann: The Q project in particular provides excellent methodological training for the doctoral students – probably the best in linguistics in Germany.

Looking back, how did the SFB come into being?

Wartenburger: It took many, many years ….

Zimmermann: When the SFB “Information Structure” expired, many here at the University had the idea of starting a new SFB – also in our field, in which a whole generation of students had grown up with this system. So we sat down and collected ideas. The topic had to be broad enough to “encompass” all prospective projects. On the other hand, it couldn’t be so vague that it no longer had any meaning. The first linguistics SFB in Constance in the 1970s was simply called “Linguistics”. Those days are over…

Wartenburger: Once we found the topic, we needed to draft an outline – a good 100 pages. It was written, revised, critiqued, rewritten, revised, critiqued, and rewritten again… An initial internal examination you might say. The paper was submitted to the DFG before Christmas 2015. After the assessment, we traveled to Bonn to present our outline and answer the examiners’ questions. About two months later, it was clear that we were going to be able to continue. That’s when the real work began. The final application has to be about 20 pages – for each project. So, it all started again: writing, revising …

Is it hard to push yourself to the limit for years without knowing whether it will work out in the end?

Zimmermann: The possibility of failure is certainly always present. And it’s very real, since there’s a lot of competition for research funding, and at the level of SFB’s, we’re competing with every other discipline. I’m very happy that we were successful in the end. 

Wartenburger: Yes, that would’ve been really frustrating. But I can say that everyone was highly motivated throughout the process. It’s been really fun. Even more so once the SFB started in July 2017. 

In an SFB like this, how do you make researchers and projects come together in the end?

Zimmermann: (Laughing) By force! Joking aside, it can’t hurt to constantly initiate discussion among the SFB members. You can also try out new formats, in which we all put our heads together and come up with suggestions of what else we could do together. These may well develop into small sub-projects or ideas for a second SFB phase.

And how can the results of 13 sub-projects be brought together?

Wartenburger: We don’t expect to end up with one model that explains everything; four years just isn’t long enough for that. We’re focusing on phenomena, structures, and topics for joint work and joint theoretical approaches, which then could lead to a potential second funding phase.

The researchers

Prof. Dr. Isabell Wartenburger studied psychology at the University of Bielefeld and received her doctorate from Charité - Universitätsmedizin Berlin in 2004. Since 2013, she has been Professor of Patholinguistics/Neurocognition of Language at the University of Potsdam.
E-Mail: isabell.wartenburgeruni-potsdamde

Prof. Dr. Malte Zimmermann studied German language and linguistics, English language and literature, and philosophy at the University of Cologne. In 2002, he received his doctorate in general linguistics from the Universiteit van Amsterdam. Since 2011, he has been Professor of Semantics and Grammar Theory at the University of Potsdam.
E-Mail: mazimmeruni-potsdamde

The project

SFB 1287 – Limits of Variability in Language: Cognitive, Grammatical, and Social Aspects
Funding: German Research Foundation (DFG) Head: Prof. Dr. Isabell Wartenburger (spokesperson), Prof. Dr. Malte Zimmermann (deputy spokesperson)
Duration: 2017–2021

Text: Matthias Zimmermann
Translation: Monika Wilke
Published online by: Marieke Bäumer
Contact to the online editorial office: onlineredaktionuni-potsdamde