When people have looked at differences like this across languages, one first response has been "Wow, languages really require different things from their speakers, therefore people who speak different languages must think differently." On the other side, people have argued, "Not so fast. Just because languages differ in what their speakers are required to say, doesn't mean that people have to think differently. The differences could be just on the surface, just in how people talk, not in how they think."

Here's an argument that lends that point of view some weight. Whenever we talk, whenever we say anything, we're only reporting a very small proportion of what we know. For example, if you were to say, "It's raining today," you can say that without having to say "It's raining today, but only outside and not inside," even though you very well know that it's only raining outside and not inside. The person hearing you also knows that you know that. You don't have to report everything that you know, and the sentence that you do say contains only a small proportion of the information that you actually know. Some people have argued just because Turkish, Korean, Japanese, and Russian speakers say different stuff and include different information in their sentences, doesn't mean they actually know different stuff. They could know the same things, remember the same things, see the world the same way, and just include different things in their sentences. That is, people could all think the same ways, but talk differently.

This question, often called the Whorfian question after Benjamin Lee Whorf (who was an American linguistic anthropologist) is something people have been speculating about and expressing very strong views on for thousands of years. Charlemagne, Holy Roman Emperor, said, "to have a second language is to have a second soul." One of his successors, Charles V, said "A man who knows four languages is worth four men." That's a very strong statement about the worth of a language. People have been making these kinds of arguments and making claims about the importance of language for centuries, some coming down on the side of language shaping thought, but others, quite on the opposite side. For example, Shakespeare had Juliette say, "A rose by any other name would smell as sweet," suggesting that what you call something doesn't matter.

What has lacked up until now, up until quite recently, is actual empirical evidence to answer the question: do speakers of different languages actually think differently? Is it because of the differences in language that these differences arise? Let me give you three of my favorite examples on how speakers of different languages think differently in important ways. I'm going to give you an example from space; how people navigate in space. That ties into how we think about time as well. Second, I'm going to give you an example on color; how we are able to discriminate colors. Lastly, I'm going to give you an example on grammatical gender; how we're able to discriminate objects. And I might throw in an extra example on causality.

Let’s start with one of my favorite examples, this comes from the work of Steve Levinson and John Haviland, who first started describing languages that have the following amazing property: there are some languages that don't use words like "left" and "right." Instead, everything in the language is laid out in absolute space. That means you have to say things like, "There is an ant on your northwest leg," Or "can you move the cup to the south southeast a little bit?" Now to speak a language like this, you have to stay oriented. You have to always know which way you're facing. And it's not just that you have to stay oriented in the moment, all your memories of your past have to be oriented as well, so that you can say things like "Oh, I must have left my glasses to the southwest of the telephone." That is a memory that you have to be able to generate. You have to have represented your experience in absolute space with cardinal directions.

What Steve Levinson and John Haviland found is that folks who speak languages like this indeed stay oriented remarkably well. There are languages like this around the world; they're in Australia, they're in China, they're in South America. Folks who speak these languages, even young kids, are able to orient really well.

I had the opportunity to work with a group like this in Australia in collaboration with Alice Gaby. This was an Aboriginal group, the Kuuk Thaayorre. One of my first experiences there was standing next to a five year old girl. I asked her the same question that I've asked many eminent scientists and professors, rooms full of scholars in America. I ask everyone, "Close your eyes, and now point southeast." When I ask people to do this, usually they laugh because they think, "well, that's a silly question. How am I supposed to know that?" Often a lot of people refuse to point. They don't know which way it is. When people do point, it takes a while, and they point in every possible direction. I usually don't know which way southeast is myself, but that doesn't preclude me from knowing that not all of the possible given answers are correct, because people point in every possible direction.

But here I am standing next to a five year old girl in Pormpuraaw, in this Aboriginal community, and I ask for her to point southeast, and she's able to do it without hesitation, and she's able to do it correctly. That's the case for people who live in this community generally. That's just a normal thing to be able to do. I had to take out a compass to make sure that she was correct, because I couldn't remember.

What I came to this community to find out was whether the way people think about space also affects the way that they think about time. I had done a lot of work trying to understand how we build our representations of time. One consistent claim has been that we build representations of abstract things, like time, out of more concrete things. The way we build complex knowledge is to take simpler building blocks and then use the power of language to combine these building blocks into more and more complex ideas. Time seemed to be an example of that; people take spatial representations, make an analogy or metaphor from space to time, and that gives us more complex ideas of time. But of course, that whole story I just told you about how ideas of space underlie our ideas of time suggests that if people think about space differently, they should also think about time differently. So if you find two cultures that think about space in very different ways, they should also think about time in very different ways.

That's what Alice Gaby and I came to Pormpuraaw to find out. I gave people a really simple task. I would give them a set of cards, and the cards might show a temporal progression, like my grandfather at different ages from when he was a boy to when he's an old man. I would shuffle them, give them to the person, and say "Lay these out on the ground so that they're in the correct order." If you ask English speakers to do this, they will lay the cards out from left to right. And it doesn't matter which way the English speaker is facing. So if you're facing north or south or east or west, the cards will always go left to right. Time seems to go from left to right with respect to our bodies. If you ask Hebrew speakers to do this, or Arabic speakers, they're much more likely to lay the cards out from right to left. That suggests that something about the writing direction in a language matters in how we imagine time. But nonetheless, time is laid out with respect to the body.

But these folks, the Kuuk Thaayorre, don't use words like "left" and "right." So what would they do? How will they lay out time? Well, it turns out they do it from East to West. If a person is sitting facing south, they will lay out the cards from left to right. But if they're facing north, they will lay the cards out from right to left. If they're facing east, the cards will come towards them. That's a pattern that we had just never seen with any American. That's a radically different way of organizing time. It's a way of organizing time that's in a different coordinate frame, in an independent coordinate frame from what you see with English speakers or Russian speakers, and so on. That shows a really big difference in cognitive ability, where a five year old in one culture can do something that an eminent professor in another culture cannot do. So that's one of my favorite examples, the way that people organize space and the way they organize time.

There are lots of other wonderful examples also from the domain of time, which seems to be especially flexible. For example, there are cultures that put the past in front and the future behind them. There are folks who organize time from left to right or right to left. Time could also take on a vertical dimension in some cultures. In Mandarin, the past is above and the future is below. You get this incredible variation in how people represent time, with complete reversals, taking on new dimensions, or going in a completely different coordinate frame.

Let me give you another set of examples. This one starts with an observation by Roman Jakobson - a Russian linguist. He was interested in grammatical gender and the ability to translate across languages. In languages that have grammatical gender, all the nouns are assigned to a grammatical category. In the simpler examples it would be masculine and feminine. Sometimes there is a third gender, masculine, feminine and neuter. In more complex cases there can be as many as 16 genders with a special grammatical category for hunting tools or for canines, depending on the language. George Lakoff made famous a grammatical gender category in an Aboriginal language that included women, fire and dangerous things. Those were the things that were all treated grammatically equivalently in this language.

What Jakobson was interested in was, does it matter if a word is assigned a masculine or feminine gender in the language? What if Monday is masculine and Wednesday is feminine? Does that matter for how you think about Monday and Wednesday? Well, one thing he did is he asked students at Moscow University to act out, to personify, days of the week. Act like Monday, or act like Wednesday. What he noticed in these Russian speakers is that when they were acting like Monday, they would act like a man, and when they were acting like Wednesday, they would act like a woman. That is, the way they personified these abstract things depended on the grammatical gender in Russian.

He also made another observation. This was an informal observation about art. He looked at the way artists portray abstract things like justice or death or time or love or charity in paintings and in sculptures, and he asked, how do they decide whether to make death a man or a woman? How do they decide to make truth or time a man or a woman? He noticed, for example, that Russians portrayed death as an old woman, whereas Germans are much more likely to portray death as a man. He again hypothesized that this has to do with grammatical gender.

Recently, a student and I did an analysis of ArtStor, a database with millions of images of art in it, and we went through and analyzed all of the images that were personifications and allegories. For each one, we said what is the gender being depicted in this painting or sculpture? And what is the grammatical gender of that word in the artist's native language? What we found was that Jakobson was right. There's an amazing correlation. Seventy eight percent of the time the grammatical gender of the word predicted the gender that the artist chose for the depiction.

And this surprised us because, of course, artists are supposed to be such idiosyncratic spirits, right? They're supposed to be not showing what the general culture might believe. And of course, one's ideas of love or death are likely to be very idiosyncratic and based on personal experiences. But nonetheless, this little quirk of grammar, whether something got assigned a masculine or a feminine gender in language had really good predictive power for how an artist was going to portray an abstract entity.

Other people have done work looking at how even young kids are affected by these little quirks of grammar. For example, young kids are told, "We're making an animated film and you have to help us figure out what voices to give different characters. So here's a fork, and here is an alarm clock, and here's an accordion. Tell us what voices you would like these characters to have?" Studies by Maria Sera showed that even young kids start assigning voices to objects that are consistent with grammatical gender in their language. Even relatively early, grammatical gender seems to be shaping the way people think about objects.

There's another aspect of gender, of course, and that's how we talk about people, not objects. Some languages mark gender very, very strongly. English is somewhere in the middle. Some languages like Hebrew mark gender all over the place both for people and for objects, and some languages don't mark gender at all. Finnish would be an example of a language with nearly no gender information in it. So you might ask, well is it really even true for how people think about people's gender? Is it possible that if you don't mark gender in a language, that somehow changes how people think about biological gender?

Now I want to put a limit on this. There are some things that we can discover outside of the bounds of language. It's not the case that Finnish speakers are only able to reproduce by virtue of randomly bumping into each other, and then once in awhile a happy miracle occurs. Clearly, Finnish speakers have figured out that there are men and women. But here's an interesting question, how long does it take to figure that out? How long does it take to figure out whether for example you, yourself, are a boy or a girl? There's a wonderful study by Alexander Guiora who looked at kids learning Hebrew, kids learning Finnish, and kids learning English as their first language. He asked all of them, "Are you a boy or a girl?" And he had all kinds of other clever ways of figuring out how aware they were of what gender they were. What he found was that kids in these three groups actually figure it out at different rates. The Hebrew learning kids got it first, and then the English kids, and then the Finnish kids last. Of course eventually everyone figured it out, but there was this difference in the developmental timeline.

In a recent set of studies, another group compared English speakers and Mandarin speakers on their ability to remember and respond to questions about gender in stories that they heard. In this study, you get a story about a bunch of people, and then you’re asked some quick questions. Some of those questions had to do with the gender of the people involved. Mandarin doesn't mark gender nearly as much as English does. What these studies found is that whereas Mandarin speakers were perfectly fast at answering all kinds of other questions, when it came to questions about gender, they were slower than English speakers, suggesting that gender information wasn't as salient, wasn't as important in their minds.

While it isn't the case that language is the only source of our information, clearly we can figure things out outside of language, when something is marked in your language that seems to give you an extra boost. It allows you to figure something out faster. It allows that information to be more salient.

Lastly, how do we think about causality and responsibility? All events that happen around us are quite complicated and require us to construe them. Often when you see something, like an apple falls off a table, it seems, "well, that's a really simple physical event. I should just have a simple way of thinking about it." But actually, we bring a huge amount of knowledge to be able to understand the event, and languages give us lots of tools for interpreting what went on.

When Dick Cheney went hunting with Harry Whittington and had an accident, and accidentally shot Whittington in the face, that was an event that took a split second. It was a really simple physical event, but there are many, many different ways that we could describe it. When the European Herald had to write about it, they wrote "Cheney Bangs Lawyer." Whittington was a lawyer, and so that gives the sense of "Oh, Cheney went out hunting for lawyers, and he got one." Of course, more prosaically we could just say, "Cheney shot Whittington." Or, take Cheney out of it a little bit, so we could say, "Whittington got shot by Cheney." We could take Cheney out of it altogether, and just say "Whittington got shot." We could say something similar to what Texas newspapers said at the time, which was, "Whittington got peppered pretty good."

Listen to what Cheney actually said. He was giving an interview in which he took full responsibility for the event, and he said, "Ultimately I'm the guy who pulled the trigger that fired the round that hit Harry." Think about how many events there are in that statement. "I'm the guy that pulled the trigger that fired the round that hit Harry." This is a split second event but he's just broken it up into all these different steps. That makes him so far removed from the eventual outcome. Bush actually did one better. Bush said, "He heard a bird flush, and he turned and pulled the trigger, and saw his friend get wounded." Now "saw his friend," that's one sentence in which Cheney transforms from agent to mere witness by the end of the sentence. It is a masterful exculpation.

These examples give you a sense of how many different ways we can frame and construe events in language. What's important is that different languages encourage different ways of framing and construing these events. In English, in fact, the kind of language that Cheney used and Bush used, we find it suspicious, this kind of linguistic wiggling. It sounds like you're trying to get out of something. It sounds evasive. It's the kind of thing that kids say and politicians say when they're trying to get out of something. In English we prefer direct causative statements. Like "He broke the vase."

But in other languages, when something is an accident, when something wasn't intentional, you wouldn't use a phrase like "He broke the vase" or "He lost the book." You would say something more like, "the vase broke" or "the book lost itself to him." Something more indirect. Something where the person involved isn't an agent. English is quite strange in that it doesn't distinguish very strongly between intentional events and accidental events. We're supposed to talk about both of them in the same way, to take responsibility for even accidental events. In some languages you can't say things like "I broke my arm" unless you're crazy and you went out trying to actually break your arm, and so you broke it.

One thing we got interested in, my student Caitlin Fausey and I, is how do the structures in language affect whether you think someone is responsible? Does it matter for whether you even remember who did it? In these studies, we showed people a bunch of videos of intentional and accidental actions, and then later tested their memory for who did it. We compared speakers of English, Spanish and Japanese. English doesn't do a very good job of distinguishing between intentional and accidental actions, in all cases we tend to describe the agent. Whereas in Spanish and Japanese, there's a stronger distinction made between accidents and intentional events. For accidents, Spanish and Japanese are less likely to talk about the agent. What we found was a difference in memory that perfectly mirrored this pattern in language. When it came to intentional events, everyone remembered really well who did it and everyone talked about who did it. But when it came to accidents, a very different pattern emerged. English speakers still remembered really well who did it, whereas speakers of Spanish and Japanese remembered less well. They didn't mention the agent as often, that wouldn't be something they would normally talk about. They also didn't think it was as important to remember who did it. They allocated their attention elsewhere to other aspects of the scene.

One thing you could ask in all of these examples that I've given you is, how do you know that it's really language that's important and not something else? How do you know that speakers of Spanish, Japanese, Mandarin, Russian, really think differently because of the structures of language and not because of other cultural differences, differences in the environment, because maybe it's something in their past, or maybe it's something in their olive oil? How do you know it isn't something else?

Another way of phrasing that question is, how do you know that it's that language shapes thought and not the other way around? Maybe these folks think differently for whatever reason, and then as a result of that, they speak differently? How do you determine the arrow of causality? Well, the answer to that is to do studies in the lab that are really targeted. We change how people talk and see if that changes how they think. That's what makes it a real experiment. In study after study that's exactly what we've done. We bring people into a lab, and we teach them a new way of talking, and we see if that changes how they think. We find that if you teach people a new way of talking about time, that changes how they think about time. If you get English speakers to talk agentively or non-agentively about events, that will change what they remember about those events. In other labs people have trained people to talk differently about colors, and that changes people's ability to remember and discriminate colors across color boundaries.

In all these studies, what emerges is that language really can have this casual power. It can change the way you think. Now that doesn't mean that the other direction of causation is unimportant. Because sometimes people think, "oh, just because language shapes thought doesn't mean that the other way around isn't also true." That's absolutely right. Language shapes thought and also the way that we think importantly shapes the way we talk, and aspects of culture importantly shape aspects of language. It's a bi-directional cycle. I think that the fact that these two things can influence each other, and can exist in a mutual cycle of influence allows humans to create complex knowledge so quickly and to be so flexible and agile in how we think about the world.

Think about it this way. We have 7,000 languages. Each of these languages encompasses a world-view, encompasses the ideas and predispositions and cognitive tools developed by thousands of years of people in that culture. Each one of those languages offers a whole encapsulated universe. So we have 7,000 parallel universes, some of them are quite similar to one another, and others are a lot more different. The fact that there's this great diversity is a real testament to the flexibility and the ingenuity of the human mind. The fact that we're able to take so many different perspectives and create such an incredibly diverse set of ways of looking at the world, that is something first to be celebrated, but also something to learn from: flexibility and diversity are at the very heart of what makes us human and what makes us so smart. I think the more we understand how people are able to take all these different perspectives, and able to change the way they think, the better we'll understand the nature of being human.

I grew up in the former Soviet Union. I mostly lived in Minsk, in Belarus, but I also spent my summers with my grandparents in Ukraine. My first set of linguistic experiences were speaking Russian and hearing a lot of Belarusian and Ukrainian. When my family came over to America, I was 12, and I was of course faced with the project of learning English. In learning to express my ideas and my thoughts in English, I first started noticing just how important connections between words and other elements of language were in creating meaning. One of the first things I got interested in was trying to get rid of those elements of language from thinking. I was very interested in social justice issues and I thought well, if we're really going to get to the bottom of what is truth and what is justice, then we have to get rid of this nuisance of language that seems to be only clouding the way that we think. I would notice how many arguments hinged on particular aspects of language.

I thought the more I studied language the more I'd discover the real truths that language is clouding. And actually exactly the opposite thing happened. I went to college at Northwestern to study cognitive science and started doing research there. I ended up going to graduate school in psychology at Stanford. The more I focused on language and started looking at cross-linguistic differences, the more it became apparent that systems of meaning exist within systems of language. Those interconnections between words are not simply the webbing on top of an otherwise pure logical knowledge system. Rather, in fact, meaning exists in the way that we use words; the patterns of word use create the system of meaning. There's no getting away from language in getting to complex meanings.

Even though I failed at the initial quest I formulated as a teenager to get to the bottom of truth and justice without language, I got very interested in how we create meaning in these complex social networks as we talk to one another. So after I finished my graduate work at Stanford, I went off to MIT, where I was on the faculty in brain and cognitive sciences. I got to do more work looking at cross-linguistic differences and make interesting connections in Indonesia and Australia doing fieldwork. I was then hired back at Stanford where I continued doing more work on complex knowledge and language, and on how language shapes thought.

When I started thinking about the question of language and thought, my first set of thoughts on this was quite naïve. For one thing, I didn't realize how controversial an issue it was. To me, it was just a curiosity and seemed like, "wow, there are so many people who speak different languages and learn different languages, why can't we just find out whether language shapes thought?" It seemed like a perfectly reasonable question. Also from a naive standpoint, it seemed like cognitive psychologists had discovered that all kinds of expertise mattered. If you became a chess expert, you would end up thinking differently, or if you became a doctor you would end up thinking differently, or if you became a master musician, you would end up thinking differently. Language seemed quite similar to that. You would become a master in a particular way of viewing the world, and that would give you the cognitive tools that came with that way of viewing the world.

I also thought, "well it will just be a matter of doing an experiment or two, and then we'll resolve that issue and move on to something else." Well, things didn't quite work out that way. It's an incredibly controversial issue. It's tied in with just about every controversy in cognitive science. This is both a good thing and a bad thing. The good thing is, it's a really interesting question. There's a lot on the line in how we answer the question about how languages and thinking interact with one another. But the bad thing is that everyone has a stake in it, so the ratio of people who do work on the question to people who express very strong views on it is not very favorable.

The kinds of big questions that this question is tied in with, other than itself, include: are there modules of mind that are separate from one another? Or is the mind more of an interactive system? Is it possible for something like grammar, which some people hypothesize as a separate organ, to influence something like the representation of meaning, or how we make perceptual discriminations? On some theories, that kind of interaction between different aspects of mind should be impossible. Another question that this obviously ties in with is the nature/nurture divide. Where do our ideas come from? To what extent are ideas innate, to what extent are they universal, and to what extent are they learned based on social convention? And that, again, is a huge split. The question of whether language shapes thought lives squarely within that divide.

Another big set of questions that language and thought interacts with are questions about what is involved in understanding language? When you're trying to process sentences and translate sounds into meaning, you’re getting a bunch of air vibrations on your eardrums, and you have to translate those into meaning, what are the set of steps and what kind of meanings do you arrive at? Is the surface structure of the language actually going to make a difference for the underlying meaning structure that you create when you understand the language? The idea that language shapes thought suggests that surface structure, quirks of language, actually can have consequences for how people construct meaning and what kinds of things people notice and encode and think about.

So this is a question that has been incredibly controversial, and every finding in this field has been vigorously debated. This is actually really fun because it means that every aspect of the question gets considered.

Like all great questions in science, it's so interesting because it's not a yes or no question. Often it's phrased as, "Does language shape thought?" Of course there's no chance that we'll come up with an answer like "Yes" or "No" and that will be the end of it. The answer will be, "well here are the mechanisms through which language can shape thought, and here are cases where it does, and here are cases where it doesn't, and here's why." Ultimately what we're trying to figure out is how does the mind work? How do we come to be the incredibly complex and sophisticated creatures that we are?

Wittgenstein clearly made the case that the meanings of words reside in their use and reside in the systems of inter-relationships between word uses. In the “Philosophical Investigations” he disowned his first set of proposals about word meaning that he made in the "Tractatus." In the Investigations he argued that what words mean is the way you use them, and the uses that they have in the system of social interactions. That doesn't mean that you can't study word meaning, but what it does mean is that the system you have to study is bigger and you have to study the whole system. You can't study individual words. You can't study individual people set apart from their social context. In some ways that makes the problem harder, because the things you have to measure are more complicated. But on the other hand, it may actually make the problem solvable, whereas before it wasn't solvable.

I think often in science, what we do is we try to simplify the problem, and we separate and remove elements, because we think that will make it easier for us to solve the problem. In language research for a while one tack was, "let's focus just on syntax. We'll solve syntax. And let's not think about pragmatics. Let's not think about semantics. Let's not think about cross-linguistic differences. Because those will just cloud the picture. We'll take those things entirely out of our space of explanation. We'll just focus on the syntax, and that will be the way forward." It seemed like a logical step to simplify. But it turns out that you can't solve the problems of syntax without letting in information from semantics and from pragmatics and from culture and social context. So instead of simplifying the job, they actually made their job impossible in doing that.

I think the way forward is to consider all the complex contexts that language exists in. Cognitive scientists have to work kind of like spies. We're trying to understand something that's invisible and very complex, and what we have to do is figure out the right set of experiments to get at those hidden truths lurking in patterns under the surface.

As an undergraduate I worked with Dedre Gentner who is a world expert in analogy and relational reasoning. Dedre started me off on the path of research that I ended up on. She had at the time an almost secret interest in whether language shaped thought. She would talk about it in the lab, but then she would warn us, "you know, don't really talk about this outside of the lab, because people think this is kind of crazy stuff." She had a sense of how controversial the issue was. I did not heed her advice, and didn't realize just how controversial the issue was going to be. When I went off to graduate school, my advisor was Gordon Bower, a world expert in memory research. I had been doing some work on metaphor and metaphors between space and time, and I brought Gordon the first draft of the paper that I had written about the work I did. He read it and he said, "You know, this is all fine, but if you're right about this, then it suggests that if people have different metaphors for time, they should think about it differently. And we know that language doesn't shape thought. And so I think you're probably wrong about this. So you might want to go back and reconsider this."

I thought to myself then, well, I could be wrong about this, or, it could be the case that language actually shapes the way we think, and maybe I should just do an experiment on that, and find out. And just around that time Doug Hofstadter was visiting Stanford. My advisor has a very avuncular manner, and he grabbed me at a reception and walked me over to Doug Hofstadter and said "Doug, I want you to meet Lera. Lera, tell Doug about your work." So I stood there and told Doug about what I was working on. He told me about metaphors for time in Mandarin that were different from English (one of his students had worked on that), and he encouraged me to look into it. So I did. And from that first set of studies everything else unfolded. So in some ways it was a very fortunate set of accidents. I first got started in looking at time and metaphors for time in Dedre Gentner's lab, and then I continued doing that in graduate school, and this chance meeting with Doug Hofstadter gave me a specific topic to look into. George Lakoff's ideas were extremely influential in starting the first path of experiments on space and time and trying to understand how the metaphors we use for talking about time shape our understanding of time.

I haven't yet met someone who doesn't care about the question of whether language shapes thought. One reason is that all of us have experiences of trying to convince other people with our words, and either being successful or not being successful. Everyone intuitively thinks about the relationship between language and thought, and how they can use their own language more effectively, or how other people can use their language more effectively to convey thoughts. People also intuitively have a lot of theories about how differences between languages matter. A lot of people are bilingual, and as soon as you start learning another language you start noticing differences and start wondering about how people who speak that language must see the world. The most common reaction I get from the general public is not why would anyone care about this, but rather, why don't we know about this already hundreds of years ago? It seems like such an obvious question to investigate. People seem surprised that this isn't something that has been thoroughly investigated by scientists. That seems to be the main surprise.