Usually, pointing is thought to help the process of bootstrapping a symbolic system. You can point to stuff to help people agree on what certain symbols refer to. This process has been formalised in the ‘naming game’ (see Matt Spike’s talk):
I request an object by naming it (with an arbitrary symbol)
You guess what I mean and give me an object
I point to the object that I meant you to give me (feedback)
We remember the name that referred to this object
This game is the basis for many models of the emergence of shared symbolic systems, including iterated learning experiments (e.g Feher et al., and Macuch Silva & Roberts). Here’s some robots playing the naming game in Luc Steels’ lab:
However, the setup of these experiments assumes one crucial thing: that the individuals can’t use pointing to make the request in the first place. Most experiments are set up so that participants must communicate symbolically before they can use pointing. If you allowed pointing to be used in a naming game, then it would probably go something like this:
I point at the object I want.
I request an object by naming it (with an arbitrary symbol)
You guess what I mean and give me an object
I point to the object that I meant you to give me (feedback)
We remember the name that referred to this object
That is, if we’re good enough at pointing then we don’t need a symbolic language for this task.
Of course, there must have been some task in our evolutionary history that provided a pressure for us to develop language. We set out to explore what kind of task this might have been by running an experiment in Minecraft.
The origins of language, and how they change over time, are tricky topics. We can’t travel back in time to observe how it happened, and we’re only just beginning to understand the range of variation in existing languages. Traditionally, the study of language evolution was more of a philosophical enterprise, with many educated guesses and a lot of debate about theoretical distinctions. But these days it’s clear that a much wider approach is needed. Thinking about how so many diverse ways of communicating could have emerged in a single species (and that species alone) involves thinking about topics as diverse as genetics, animal communication, cultural evolution, emerging sign languages, and the history of human migration and contact (even Chomsky recently wrote of the importance of acquisition, pragmatics, computer science and neuroscience in understanding the language faculty!).
But if this isn’t deep enough into the frontiers of language evolution for you, there is also a debate on humidity and tone. Caleb Everett, Damian Blasí and myself discuss the potential effects of our ecology on language evolution. This includes obvious differences such as some languages having more specific words for relevant climatic factors (not just words for snow, but watch this space for news on that front), to the way the climate affects population movement. We focussed on one controversial idea: dry air affects phonation accuracy, so some sounds should be harder to produce accurately in dry climates. Over a long period of time, this might lead to languages changing to avoid these sounds.
This year’s Nijmegen lectures were given by David Poeppel on his work linking language processing to low-level neural mechanisms. He called for more “muscular” linguists to step up and propose a “parts list” of linguistic primitives that neuro researchers could try and detect in the brain. In this post, I cover the generativist answer to this, as proposed by Norbert Hornstien, who appeared as a panelist at the Nijmegen lectures, and why it bothered me (TLDR: I think there’s more to language science than syntax, and other areas can also draw up a “parts list”).
The move to self-publishing is a bit of an experiment, but hopefully it’ll mean that the papers are more accessible to a wider audience. To aid this, the papers are published under Creative Commons licenses. Some papers also include supplementary materials.
The full list of papers will be updated as revisions come in, but here are some interesting papers available so far:
The MPI for Psycholinguistics in Nijmegen, Netherlands is inviting scholars to visit in 2016 and 2017 to participate in Language Evolution research.
Under the running guest programme of the Institute, special opportunities are available in 2016-2017 for visitors – both senior scholars and post-docs – participating in the research theme “Evolution of Language”, starting at any time in 2016. The theme will be hosted by the Language & Cognition department (run by Stephen Levinson) in collaboration with the other departments at the Institute.
The MPI for Psycholinguistics and its partners in Nijmegen offer an exciting environment in which to explore the evolution of language, both in its biological and cultural dimensions, and the interactions between them. The MPI has departments in language and genetics, the neurobiology of language, the psychology of language, language and cognition, and (in development) the acquisition of language. The MPI is a partner in the Donders Institute for Brain, Cognition & Behaviour, Centre for Language Studies, and the Language in Interaction consortium, which bring together many different perspectives on language evolution.
Senior Visitorships are available for any period from one month up to one year until the end of 2017, and Postdocs are available for up to one year starting at any time in 2016.
Gregory Mills and I are running a workshop on language evolution and interaction. Details are on the workshop website, including a call for papers.
Language Adapts to Interaction
Language has been shown to be adapted to constraints from many domains such as production, transmission, memory, processing and acquisition. These adaptations and constraints have formed the basis for theories of language evolution, but arguably the primary ecology of language is interaction – face-to-face conversation (Levinson, 2006). Taking turns at talk, repairing problems in communication and organising conversation into contingent sequences seem completely natural to us, but are in fact highly organised, tightly integrated systems (Sacks et al., 1974) which are not shared by any other species. Therefore, the infrastructure for interaction may provide an insight into the origins of our unique communicative abilities (Mills, 2014a; Micklos, 2014).
Indeed, recent studies on interaction have shown that an approach that emphasises interaction can sharpen our understanding of linguistic universals (Kendrick et al., 2014; Dingemanse et al., 2015), ontogeny and acquisition (Hilbrink, et al. 2015; Vogt, 2014), on-line processing constraints (Bögels et al., 2015) and animal signalling (Levinson & Holler, 2014).
The emerging picture is that the infrastructure for interaction is an evolutionary old requirement for the emergence of a complex linguistic system, and for a cooperative, cumulative culture more generally. These issues are also being integrated into computational models of the cultural evolution of linguistic systems (Vogt and Haasdijk, 2010; Roberts, & Levinson, 2015) and are being explored through studies of experimental semiotics (Mills, 2014b; Christensen et al., 2016).
This workshop is interested in addressing the following questions:
How did the infrastructure for interaction emerge and how did it shape the emergence of language?
What evidence is there that language structures show adaptation to interaction?
How do interactional constraints interact with other domains such as processing?
What are the limits of interactional abilities in non-human animals?
Bögels, S., Magyari, L., & Levinson, S. C. (2015). Neural signatures of response planning occur midway through an incoming question in conversation. Scientific Reports, 5: 12881.
Christensen, P., Fusaroli, R., and Tylén, K. (2016). Environmental constraints shaping constituent order in emerging communication systems: Structural iconicity, interactive alignment and conventionalization. Cognition, 146:67–80.
Dingemanse, M., Roberts, S. G., Baranova, J., Blythe, J., Drew, P., Floyd, S., Gisladottir, R. S., Kendrick, K. H., Levinson, S. C., Manrique, E., Rossi, G., & Enfield, N. J. (2015). Universal Principles in the Repair of Communication Problems. PLoS One, 10(9).
Hilbrink, E., Gattis, M., & Levinson, S. C. (2015). Early developmental changes in the timing of turn-taking: A longitudinal study of mother-infant interaction. Frontiers in Psychology, 6: 1492.
Kendrick, K. H., Brown, P., Dingemanse, M., Floyd, S., Gipper, S., Hayano, K., Hoey, E., Hoymann, G., Manrique, E., Rossi, G., & Levinson, S. C. (2014). Sequence organization: A universal infrastructure for action. 4th International Conference on Conversation Analysis. UCLA, CA.
Levinson, S. (2006). On the human interaction engine. In Enfield, N. and Levinson, S., editors, Roots of Human Sociality: Culture, Cognition and Human Interaction, pages 39–69. Oxford: Berg.
Levinson, S. C., & Holler, J. (2014). The origin of human multi-modal communication. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 369(1651): 2013030.
Micklos, A. (2014) The Nature of Language in Interaction. Proceedings of the 10th Evolution of Language conference. World Scientific: Vienna, Austria.
Mills, G. (2014a) Establishing a communication system: Miscommunication drives Abstraction. Proceedings of the 10th Evolution of Language conference. World Scientific: Vienna, Austria.
Mills, G. J. (2014b). Dialogue in joint activity: complementarity, convergence and conventionalization. New Ideas in Psychology, 32:158–173.
Roberts, S. G. & Levinson (2015) On-line pressures for turn-taking constrain the cultural evolution of word order. Workshop on Cognitive Linguistics and the Evolution of Language, International Cognitive Linguistics Conference, Newcastle University, UK.
Sacks, H., Schegloff, E. A., and Jefferson, G. (1974). A simplest systematics for the organization of turn-taking for conversation. Language, 40(4):696–735.
Vogt, P. (2014) Language acquisition through interactions: A cross-cultural corpus of multimodal interactions with 1-to 2-years old infants. 8th International Conference on Construction Grammar. Osnabrück University, Germany.
Vogt, P. and Haasdijk, E. (2010). Modeling social learning of language and skills. Artificial Life, 16(4):289–309.
I was interviewed for about an hour, but the quotes from me in the piece are actually a bit out of context. For instance, I claim that this is the first study of its kind, when there have been several studies which looks at climate and language (including one I co-authored with Caleb Everett and Damian Blasi). But, to be fair, Maddieson and Coupe’s study is probably the one with the greatest range of ecological variables and most sophisticated linguistic measure (although I’m not sure yet how they control for historical relatedness).
You can read about the study in the article above, but I wanted to address another thing I’m quoted as saying. The interviewer asked if it was possible to see these effects in a single language or a single speaker, and I said that it was very unlikely, but that I’d tried to do this with transcripts of Larry King. I went on to say that I absolutely wouldn’t publish this because it was a crazy idea that nobody would believe.
But now the cat’s out of the bag, so here’s what I found.
Does Larry King use more vowels when it’s warmer?
If we see language change as natural selection operating on individual utterances in conversation (a la Croft and others), then we should be able to see selective pressures at work in the utterances of an individual speaker. This should also apply to the influence of climate. Given enough data, you should be able to see an individual adapting over the seasons. In light of Maddieson and Coupe’s hypothesis, speakers should use proportionately more vowels compared to consonants when it’s warmer.
CNN provides transcripts for every show broadcast between 2000 and 2012. Larry King has done an interview practically every day on the show ‘Larry King Live’ (which has been used before in linguistic studies), so I extracted Larry King’s speech transcripts (excluding the guest’s speech, mentions of the location of the studio and the guest’s names). For each transcript I counted the frequency of each letter, then calculated the ratio of vowels (aeiouy) to consonants. Then, for each air date, I found the actual temperature and humidity data for that date and the location of the show (CNN studios in LA). The show is occasionally recorded in Washington DC or New York. I tried to detect these automatically and matched them with the climate measures for the CNN studios in those cities.
There are about 3,500 transcripts over 11 years, about 90% of which were aired on consecutive days. (I know nothing about this show, and am a bit surprised by its frequency! I’ll have to check whether the transcripts include repeats).
Here is a depiction of the results for temperature and humidity:
The Black lines show King’s vowel ratio (higher = more vowels) and bars are 95% confidence intervals around the mean for each week of the year. The maximum and minimum temperature are shown in red and blue. Below is a similar graph for specific humidity.
Surprisingly, there is some variation in proportion of vowels, and it looks like there’s a trend in the right direction.
To analyse the data, I used a linear mixed effects model, predicting vowel ratio by (log) text length and maximum temperature, with random effects for year and each week (580 separate weeks, to try to control for topical issues).
Maximum temperature significantly improves the fit of the model over a null model with text length (Chi Squre = 23.7, df = 1, p < 0.00001).
Estimate Std. Error t value
(Intercept) 4.100e-01 8.114e-04 505.3
maxTempC.loc 9.628e-05 1.973e-05 4.9
text.total.log.center 5.340e-03 8.400e-04 6.4
King uses proportionately more vowels when it’s warmer. The effect is very small: On average, there is a difference of about 15 vowels used in an hour of conversation between summer and winter. A model with location-specific maximum temperature improves the model fit over one with just LA-specific maximum temperature (Chi Sqaured = 11.59, df=1 , p=0.0007).
Of course, the temperature is not independent from day to day, so I also tried a lagged regression, predicting vowel ratio by text length (total) and maximum temperature of the recording location (maxTempC.loc). Lagging back in time by days.
The temperature of actual day is still significant, taking into account previous days. Note that the coefficient is negative after 3 lagged days. (please forgive the rough analysis- it’s all I have left after my computer broke recently).
I’m not really sure what to make of this. Given the data above, there is an argument that King is adapting the way he speaks to the climate. However, a LOT more needs to be done in order to show this. There are several confounding factors, such as the show being recorded in an air conditioned studio, the topics or guests might be different, there might be seasonal topics or key-words which affect the results (though one might argue that the lexicon for things related to cold climates has adapted). The graphs show big jumps around week 32, which I can’t explain yet. Then there is the question of the mechanism – how exactly is King adapting? By choice of alternative words, or constructions? And, of course, the transcripts are orthorgraphic. And, of course, the idea is crazy.
Still, I think it’s amazing that we’re now in a position where we can even start asking these kind of questions with data.
Readers of this blog are likely to be familiar with controversy surrounding Keith Chen’s (2013) findings suggesting that the way a language deals with the future tense affects speakers’ economic decisions. A new study by a group of Austrian and German economists has recently added experimental evidence to the growing body of research on the issue.
The study focuses on German and Italian. As with English, Italian requires a speaker to mark the future tense, e.g. “it will rain tomorrow” not simply “it rains tomorrow.” German, on the other hand, requires no such marking—speakers are free to say the equivalent of “it rains tomorrow.”
Chen hypothesized that speaking about future event using the present tense, as in languages like German, could make the future seem more immediate. He suggested this might cause speakers to value future events more highly than speakers of languages like Italian. Indeed, using large-scale statistical methods, Chen found that speakers of languages like German (with no separate future tense) tend to smoke less, be less obese, engage in safer sex practices, retire with more money, and save more per year.
However, most criticism has centred around the potential spuriousness of the correlation between future tense marking and future orientated behaviour. Commentators have been quick to point out that since cultural traits tend to be inherited in packages, it’s likely that future orientation and future tense reference are causally unrelated. In fact, Roberts, Winters and Chen (2015) found that the correlation dropped below significance when controlling for cultural relatedness (for some if not all their statistical tests). As such, the jury may yet be hung until experimental evidence confirms (or doesn’t) Chen’s findings.
The new study attempts to do exactly this. Though not yet peer reviewed, it was recently opened to comments and criticism as a discussion paper, as apparently is common in the discipline of economics. You can read it here.
The paper takes advantage of bilingualism in the city of Meran in the autonomous province of South Tyrol in Northern Italy, where roughly 50% of the population speaks German and the other 50% speaks Italian. Hoping to mitigate the cultural confounds that plagued Chen’s (2013) findings, the authors argue that since the German- and Italian-speaking citizens of Meran share a home city, they also share a similar enough cultural milieu that experimenting on them can isolate the effects of their language on how they value future events.
The experiment worked like this: children were presented with three choices. In each of the choices they could either accept two tokens at the time of the test or opt instead to receive a greater number of tokens four weeks in the future (the ‘patient’ choice). The future rewards were three, four, or five tokens. So in other words, children had to choose three times, between: 1) two tokens now and three tokens in four weeks, 2) two tokens now and four tokens in four weeks, and 3) two tokens now and five tokens in four weeks.
One of these choices was randomly selected for their actual pay-out, and they could then trade the tokens for a selection of small toys and items the experimenters had on offer. As control variables, they also measured IQ relative to cohort, the income and education of the parents, the children’s risk-taking propensity, average housing prices in the area of the school, and whether the children had friends who spoke another language.
In presenting their results they begin with a sample of just those students both of whose parents speak either German or Italian (n=860). They then progressively add students from different linguistic groups to their sample. They start with (n=203) students from bilingual German-Italian households, then add (n=261) students from household where both parents speak a language which has a future tense (like Italian) but which is not Italian. They then treat these delineations as categorical variables and use both regressions and non-parametric tests to see whether a child’s language category affects their tendency to wait for future rewards.
Their results suggest that it does. In the first sample, the German-speaking children were more likely to wait for the future reward than the Italian-speaking children, across all age groups and when including the host of control variables. Moreover, when they added the German-Italian bilingual group to the sample, their tendency to wait for future rewards was intermediate between the all-Italian and all-German children. Finally, they find that the (n=261) student who speak a language which marks the future tense, but which is not Italian, are significantly less likely to wait for future rewards compared to German speaking children, but are not significantly different from the Italian cohort.
They also separately analyse (n=91) students one of whose parents speak German or Italian and the other of whose parents (except for 5 cases) speaks a language which, like Italian, mandates marking of the future tense. They find that children with one German-speaking parent and one future-tense-marking parent are more likely to wait for future rewards than children with one Italian-speaking parent and another future-tense-marking parent.
Should these results be trusted? There are some strange statistics in the paper; throughout, they use multiple Mann-Whitney U Tests when comparing more than two groups, but never report any kind of post-hoc correction for multiple comparisons. With some p-values reported at the < .05 level, this could be problematic for some of their findings. Whilst their regressions would not be affected, I have to assume they present non-parametric tests because at least some of their data violate normality assumptions.
However, their results are corroborated by a second experiment which uses a slightly different discounting task and which only has two groups (all German and all Italian) and is therefore not affected by any apparent failure to correct for multiple comparisons.
They also include several other robustness tests. They make sure risk attitudes (which they find predict propensity to wait for future pay outs) are not correlated with language group (they aren’t). Although, whether this reveals anything more than their multiple regressions is unclear, since they already revealed a result whilst controlling for any correlation between the two variables. Additionally, they survey (n=177) Meraners to check whether there are any differences between the Italian and German populations in terms of their attitudes towards the importance of “thrift” and “patience.” Finding no such differences, they argue that this supports the conclusions that there are no cultural differences confounding their findings.
However, can we really trust self-reported estimations of the importance of thrift and patience? Most people know that they should say these are important, so is it really possible to rule out implicit demand characteristics that might undermine actual differences between the two language groups?
In fact, even according to the authors themselves, the German- and Italian-speaking populations in Meran remain linguistically and culturally “fairly segregated, with different media (like newspapers or TV channels) and leisure activities (like different football clubs)” (Sutter et al, 2015 p. 5). Schools are evidently also segregated by language, teaching either in Italian or German, and no schools to-date have an equal number of Italian and German students.
Moreover, the experiment does not actually manipulate children’s responses; it merely applies the same test to children from different language groups. What is strange is that there may actually be an unreported manipulation in the experiment. Since the authors only state that the test prompts were given in the child’s “mother tongue”, I would assume that some of the (n=203) bilingual children were tested in Italian and some in German. This means that the researchers hopefully have (or might be able to get) data on whether the language of the test prompt affected the children’s propensity to wait for future rewards.
Until they include these results in the analysis, their findings seem to be subject to the same criticisms levelled against Chen (2013). In other words, they may be picking up on a spurious correlation resulting from the fact that thriftiness and tense structure, whilst causally unrelated, could have been co-inherited from antecedent cultural groups.
If the Italian and German populations in South Tyrol are as alienated and segregated as they seem to be, it’s entirely possible that the German-speakers inherited increased thriftiness as well as a language without a future tense, whilst the Italian-speakers inherited decreased thriftiness as well as a language with a future tense. They need not be causally related, and the speakers’ geographical proximity may not have been enough to override these packages of traits. Until we see a language-based manipulation of future-orientated behaviour, the issue of whether tense affects how we value future events will remain unresolved.
Chen, M. K. (2013). “The Effect of Language on Economic Behavior: Evidence from Savings Rates, Health Behaviors, and Retirement Assets.” American Economic Review103(2): 690-731.
Roberts, S. G., Winters, J. & Chen, M. K. (2015). “Future Tense and Economic Decisions: Controlling for Cultural Evolution.” PLoS One10(7): e0132145.
Sutter, M., et al. (2015). “The Effect of Language on Economic Behavior: Experimental Evidence from Children’s Intertemporal Choices.” IZA Discussion Paper Series9383.
Cole Robertson lives and works in Oxford, where he is completing a PhD focusing on inter-linguistic tense differences and future-orientated behavior, but he’s generally interested in how language interfaces with other cognitive, perceptual, and behavioural systems, as well as language revitalization and language evolution. He’s an avid rock climber and mountaineer and once spent two days eating nothing but sausages cooked over a candle whilst waiting out a storm. He grew up in Vancouver, Canada.
Insofar as the field of language evolution has even had time to spill a lot of ink about anything – if you take Pinker & Bloom’s seminal 1990 paper as the starting point, the field is only just a carefree twenty-something – the focus of the field has primarily been about the finer points of language: syntax, the lexicon, the physiology of speech production, etc. But one’s mid-twenties are a time for exploration, so, with my colleagues at the Social Dynamics Unit, we’re looking towards something relatively unexplored in language evolution: stories.
It’s almost impossible to imagine telling a story without language – how would you even begin? (Even Emoji Dick had to be translated from the original, and might not even be successful, and emojis arguably are language…but I digress…) So the questions emerge: why and how do we tell stories? Do stories simply take advantage of our ability to speak about things that are not present and/or are not concrete (or even real), or are they a key part in how language evolved the ability to do this? How do stories evolve over time and respond to cultural pressures? What kinds of features of stories survive and replicate, and what features peter off and die? What selection pressures underlie this?
This is, of course a whole host of questions, none of which we can expect to find definitive answers to anytime soon – a feature shared by a lot of work in language evolution (and an exciting one, in my opinion). And of course, we’re not jumping into a void: already there is work that focuses on the phylogeny of stories, the potential evolutionary function of stories, and also a fair bit of work on evolution in literature more generally, some of it featured right here on Replicated Typo. But we’re taking a new experimental approach: we’re crowd sourcing collaborative stories. We hope this will contribute to answering the last two questions in particular: how do stories evolve over time and respond to different constraints, and what features survive and replicate?
We’re doing this using an experimental game called CreaStoria – and the more players we have, the better! So please play! The game is a hybrid of choose your own adventure, Twitter, and a creative writing workshop. It works like this: we start with a bunch of single-sentence “story prompts” created in collaboration with Piano Piano Book Bakery in Rome, and these stories become the “root” for collaborative story trees. At each turn, a player is presented with three potential stories and has to choose which one to continue with their own short story, creating an element of preferential selection. After you’ve played, or between bouts of play (I hear it’s great for procrastination, so feel free to come and go as you need it), you can look at the growing story tree and vote on stories you like (or don’t).
The inner workings are a bit more complicated than this, of course – I could tell you, but I’d have to “know your IP address and exclude any stories you write from our data” (if you know what I mean). We’ve had the opportunity to exhibit the game at a couple of live events in Italy, so the tree of Italian stories is pretty well populated – but we would really like more English data (and having a lot of both could lead to some interesting contrasts), so play now and tell your friends! If you’re curious as to how the data pans out, like the game on Facebook or follow @creastoria on Twitter to get updates.
Christine Cuskley is a linguist/psychologist/nerd type who currently researches the evolution of language and communication in the Social Dynamics Unit at the Institute for Scientific Interchange in Turin, Italy, and will take up a position as a British Academy Postdoctoral Fellow at the University of Edinburgh from January 2016. She mostly retweets at @nerdpro.
MPI researchers Mark Dingemanse, Francisco Torreira and Nick Enfield from the Max Planck Institute for Psycholinguistics have won a coveted Ig Nobel Prize for their work which shows that ‘Huh?’—a word people use when they missed what someone just said—may well be a universal word.