 Is linguistics a science? And if it isn't, what else could it be? In this video I want to present three arguments that show parallels between linguistics and natural sciences such as biology or physics. I'll also highlight three aspects in which linguistics is very different from those disciplines. My name is Martin Hilpert. I'm a professor of linguistics at the University of Neuchâtel in Switzerland. At university, linguistics is usually found in the humanities or the social sciences. So when I go to my office, I meet colleagues from philosophy, from sociology, history or literature. And that of course makes perfect sense. All of these disciplines work together to provide a better understanding of human life. Linguistics, as the scientific study of language, is one part of that. Linguists try to understand the human capacity for acquiring and using language. When someone asks, is linguistics a science? That question invites us to think about how the study of language is different from the study of matter and energy in the natural sciences. And that of course is an interesting question to think about. However, it's also the case that this question is rarely asked in a neutral way. Humanities and social sciences have a reputation of being soft and fuzzy and sometimes even politically charged. So I'll come back to this implicit criticism at the end of this video. For now, let's just focus on three similarities between linguistics and the natural sciences and then three differences. So let's go. The most important shared characteristic, I should say, is the scientific method, which means that researchers start with a theory that then gives rise to a hypothesis, which is tested in an experiment on the basis of empirical observation. Experiments can show us, for instance, that a hypothesis that we had is false, which then allows us to correct our misconceptions about the world, come up with another theory and repeat the cycle as needed. And through that process, slowly but surely, we arrive at a more accurate understanding of what the world is actually like. Let me illustrate that with an example from linguistics. Linguists have all kinds of theories and one theory that is widely accepted is the idea that speakers' knowledge of words is organized in a mental lexicon in which words are connected through associative links. A hypothesis that we can derive from that is that hearing a word such as doctor would send activation to semantically associated words such as hospital. And this hypothesis can be tested with an instrument that linguists call a lexical decision task. Participants have to say as fast as they can whether or not a string of letters is an actually existing word of English. And in the case of hospital, well, yes, hospital is a word. By contrast, despotyl is not a word. What can be shown with a lexical decision task is that positive responses to hospital are indeed faster when participants have seen the word doctor before. So doctor has sent activation to the word hospital. And this result is consistent with the idea that we have a mental lexicon that contains associative links. We need further experiments to establish how exactly these links work. And for these new experiments, new hypotheses have to be generated. Some of them will turn out to be correct and others will turn out to be false. So our understanding of the mental lexicon advances one experiment at a time. Right. So that's similarity number one, the scientific method. Similarity number two is the quest for laws of nature. Natural sciences study processes in the world and they try to model them with the tools of mathematics. So if you think of motion or gravity or the conservation of energy in a closed system, all of this you try to understand. And understanding a natural process means that you can express it in a mathematical formula, in a precise mathematical formula. That formula further needs to be consistent with other laws of nature that have been already observed. So if there is a conflict, it means that either the formula that you have proposed is not correct or the entire system needs to be adjusted so that everything is consistent again. Right. Are there any linguistic laws of nature? Well, there actually are linguistic insights that can be expressed and formalized mathematically. One example is what's known as Ziff's law. Ziff's law states that in any body of text, in any book, in any newspaper, in any newspaper article, the frequency of a word is inversely proportional to its frequency rank. That means in a nutshell that the most frequent word in a corpus is 10 times more frequent than the 10th most frequent word. And it is a hundred times more frequent than the 100th most frequent word and so on and so forth. That's Ziff's law. Another example would be Heap's law, which describes the mathematical relation between the length of a text and the number of different words that you find in that text. Linguists call this the type frequency of a text. So the longer a text, the more different words you find, but crucially the relation is not a straight line. As the text gets longer and longer, you find fewer and fewer new types because you've already found so many words. Now, linguists have found generalizations that can be expressed mathematically, so this we can count as the second similarity between linguistics and the natural sciences. Let me come to similarity number three, big data. Researchers in the natural sciences are often confronted with amounts of data that represents millions and millions of data points, millions of measurements. In all fairness, language scientists are far away from the kind of data that is generated, for example, by a large Hadron collider. However, linguistics has experienced a significant change in its empirical foundations over the last 50 years or so. Many linguists are working with corpora, which are digitized text collections that represent authentic language use. So over the last decades, corpora have become larger and larger from the one million word brown corpus that you can hardly see on this slide to the 100 million word British national corpus, to the 10 billion word news on the web corpus, the growth has been exponential. That means that linguists have to change their methods of analysis. Linguists have borrowed and adapted methods from many other fields, including the natural sciences. For example, biologists have used hierarchical clustering to study the DNA of different beetle species. Linguists have borrowed that in order to distinguish between different dialects on the basis of their morphosyntactic features. So there are conceptual similarities and methodological similarities between linguistics on the one hand and the natural sciences on the other. But of course, there are also many differences. The first difference that I want to talk about is that linguistics as such is not a unified field. What I mean by that is that there are remarkably few points on which all linguists and I really mean all of them would agree. This concerns quite fundamental issues. For example, the question are human beings born with a universal grammar that allows them to learn language? Chomsky and linguists say yes. Other linguists, myself included, would say no. And still others would say that this is not really relevant to their day-to-day research. Another question is whether the structures of language are shaped by functional pressures that result from human physiology, from cognition, or from social interaction, or whether language is something that should be viewed as independent of all of these factors. The bottom line is that linguistics is a very heterogeneous field and you can find researchers who are working in one area who don't understand, let alone agree with, what researchers in another area are doing. And that state of affairs is more typical of a humanities field than it is of the natural sciences. Let me come to difference number two. When I discussed the similarities between linguistics and the natural sciences, I said that language scientists have identified generalizations that can be expressed mathematically. And one example that I gave was Ziv's law. Now rather importantly, most linguistic theories work with constructs that do not have a mathematical expression and that would actually be difficult to formalize and quantify. Take for example the idea of a linguistic sign. The linguistic field that deals with signs and how they are interpreted is called semiotics. And semiotics has produced several models of how signs are supposed to work. The model that you see on this slide has been developed by Carl Bühler. It's a quite famous one. Now I'm oversimplifying here of course, but in the sciences you cannot really claim to have understood a phenomenon unless you provide a precise mathematical formula that is a mathematical model of what is going on. Linguists typically deal with models that are not formalized to this extent and that are not mathematical. That doesn't make these models any less important, but it shows that there's a difference between linguistics on the one hand and the natural sciences on the other. And with that I'd like to come to the third difference politics. I promised I would return to that. Now the laws of nature that are studied by natural scientists try to work out how the universe works across time and space independently of human observers. Linguistics on the other hand is very much concerned with human behavior. Important aspects of what we study are social and political issues and here of course I have to include phenomena such as sexism and racism which cause a tremendous amount of suffering around the world. These issues are reflected in language and that makes them the business of linguists. For example, language use has the power to perpetuate injustice. I learned in school that Christopher Columbus discovered America and the word discover downplays the violence and suffering that actually occurred. Language further regulates access to education, to material well-being and to trust. So we are judged by the language variety that we speak. In a courtroom how we speak affects the credibility of our testimonial and university students find the content of a recorded lecture less comprehensible if they are told that the instructor is a non-native speaker. Let me just bring up one more example. This week in Switzerland a supermarket was called out because they carried a candy product with a name that is racist. I'm sure that many of you know what I'm talking about. The supermarket chain took a decision to pull the product which is a very reasonable decision but you can imagine the backlash. The arguments were as follows. People said well I'm not racist so if I use that term it cannot be racist right? Second it's just candy you're focusing on the wrong issues we have much bigger problems to be worried about and third don't restrict my freedom to use language in whatever way I want. Now as a linguist I have a modest proposal. Why don't we just ask the people who are affected and listen to them what they have to say? It turns out they don't particularly like the term and as speakers we often fail to notice that an expression that we're using is hurtful to others because it's not hurtful to us. So when we find out that others are in fact bothered the appropriate reaction would be to say I'm sorry. This is an honest mistake thanks for telling me, thanks for setting me straight. Let's make some changes. It's only candy. And about that freedom argument. Changing the name of a candy does not restrict your freedom of speech. Everyone can say what they want. It's just that they're not entitled to everyone's approval when they do that. We're in the process of deciding together what we as a community would prefer and we prefer to use non-racist language. That shouldn't be that big of a deal but discussions of this kind show that there still is a long way to go. Language is not just about candy when it's racist and it affects our well-being as a community and for that reason linguists can and should weigh in on the issue. I'm coming to an end here. Is linguistics a science? Well, yes and no. It's a bit of everything and honestly that is why I'm drawn to it personally. If you spend any amount of time around researchers of different disciplines, you know that there is more than one way of trying to understand the world and I honestly believe that you can learn from everybody. That's it for this video. Thanks so much for watching and I'll see you next time. Bye.