Language Evolution and Computation Bibliography

Understanding how words change their meanings over time is key to models of language and cultural evolution, but historical data on meaning is scarce, making theories hard to develop and test. Word embeddings show promise as a diachronic tool, but have not been carefully evaluated. We develop a robust methodology for quantifying semantic change by evaluating word embeddings (PPMI, SVD, word2vec) against known historical changes. We then use this methodology to reveal statistical laws of semantic evolution. Using six historical corpora spanning four languages and two centuries, we propose two quantitative laws of semantic change: (i) the law of conformity—the rate of semantic change scales with an inverse power-law of word frequency; (ii) the law of innovation—independent of frequency, words that are more polysemous have higher rates of semantic change.

Humans spatialize time. This occurs within individual minds and also in larger, shared cultural systems like language. Understanding the origins of space-time mappings requires analyses at multiple levels, from initial individual biases to cultural evolution. Here we present a laboratory experiment that simulates the cultural emergence of space-time mappings. Dyads had to communicate about temporal concepts using only a novel, spatial signaling device. Over the course of their interactions, participants rapidly established semiotic systems that mapped systematically between time and space. These semiotic systems exhibited a number of similarities, but also striking idiosyncrasies. By foregrounding the interaction of mechanisms that operate on disparate timescales, laboratory experiments can shed light on the commonalities and variety found in space-time mappings in languages around the world.

Focal colors, or best examples of color terms, have traditionally been viewed as either the underlying source of cross-language color-naming universals or derived from category boundaries that vary widely across languages. Existing data partially support and partially challenge each of these views. Here, we advance a position that synthesizes aspects of these two traditionally opposed positions and accounts for existing data. We do so by linking this debate to more general principles. We show that best examples of named color categories across 112 languages are well-predicted from category extensions by a statistical model of how representative a sample is of a distribution, independently shown to account for patterns of human inference. This model accounts for both universal tendencies and variation in focal colors across languages. We conclude that categorization in the contested semantic domain of color may be governed by principles that apply more broadly in cognition and that these principles clarify the interplay of universal and language-specific forces in color naming.

Language's expressive power is largely attributable to its compositionality: meaningful words are combined into larger/higher-order structures with derived meaning. Despite its importance, little is known regarding the evolutionary origins and emergence of this syntactic ability. Although previous research has shown a rudimentary capability to combine meaningful calls in primates, because of a scarcity of comparative data, it is unclear to what extent analog forms might also exist outside of primates. Here, we address this ambiguity and provide evidence for rudimentary compositionality in the discrete vocal system of a social passerine, the pied babbler (Turdoides bicolor). Natural observations and predator presentations revealed that babblers produce acoustically distinct alert calls in response to close, low-urgency threats and recruitment calls when recruiting group members during locomotion. On encountering terrestrial predators, both vocalizations are combined into a "mobbing sequence," potentially to recruit group members in a dangerous situation. To investigate whether babblers process the sequence in a compositional way, we conducted systematic experiments, playing back the individual calls in isolation as well as naturally occurring and artificial sequences. Babblers reacted most strongly to mobbing sequence playbacks, showing a greater attentiveness and a quicker approach to the loudspeaker, compared with individual calls or control sequences. We conclude that the sequence constitutes a compositional structure, communicating information on both the context and the requested action. Our work supports previous research suggesting combinatoriality as a viable mechanism to increase communicative output and indicates that the ability to combine and process meaningful vocal structures, a basic syntax, may be more widespread than previously thought.

How universal is human conceptual structure? The way concepts are organized in the human brain may reflect distinct features of cultural, historical, and environmental background in addition to properties universal to human cognition. Semantics, or meaning expressed through language, provides indirect access to the underlying conceptual structure, but meaning is notoriously difficult to measure, let alone parameterize. Here, we provide an empirical measure of semantic proximity between concepts using cross-linguistic dictionaries to translate words to and from languages carefully selected to be representative of worldwide diversity. These translations reveal cases where a particular language uses a single "polysemous" word to express multiple concepts that another language represents using distinct words. We use the frequency of such polysemies linking two concepts as a measure of their semantic proximity and represent the pattern of these linkages by a weighted network. This network is highly structured: Certain concepts are far more prone to polysemy than others, and naturally interpretable clusters of closely related concepts emerge. Statistical analysis of the polysemies observed in a subset of the basic vocabulary shows that these structural properties are consistent across different language groups, and largely independent of geography, environment, and the presence or absence of a literary tradition. The methods developed here can be applied to any semantic domain to reveal the extent to which its conceptual structure is, similarly, a universal attribute of human cognition and language use.

The ability to abstract a regularity that underlies strings of sounds is a core mechanism of the language faculty but might not be specific to language learning or even to humans. It is unclear whether and to what extent nonhuman animals possess the ability to abstract regularities defining the relation among arbitrary auditory items in a string and to generalize this abstraction to strings of acoustically novel items. In this study we tested these abilities in a songbird (zebra finch) and a parrot species (budgerigar). Subjects were trained in a go/no-go design to discriminate between two sets of sound strings arranged in an XYX or an XXY structure. After this discrimination was acquired, each subject was tested with test strings that were structurally identical to the training strings but consisted of either new combinations of known elements or of novel elements belonging to other element categories. Both species learned to discriminate between the two stimulus sets. However, their responses to the test strings were strikingly different. Zebra finches categorized test stimuli with previously heard elements by the ordinal position that these elements occupied in the training strings, independent of string structure. In contrast, the budgerigars categorized both novel combinations of familiar elements as well as strings consisting of novel element types by their underlying structure. They thus abstracted the relation among items in the XYX and XXY structures, an ability similar to that shown by human infants and indicating a level of abstraction comparable to analogical reasoning.

The naming of colors has long been a topic of interest in the study of human culture and cognition. Color term research has asked diverse questions about thought and communication, but no previous research has used an evolutionary framework. We show that there is broad support for the most influential theory of color term development (that most strongly represented by Berlin and Kay [Berlin B, Kay P (1969) (Univ of California Press, Berkeley, CA)]); however, we find extensive evidence for the loss (as well as gain) of color terms. We find alternative trajectories of color term evolution beyond those considered in the standard theories. These results not only refine our knowledge of how humans lexicalize the color space and how the systems change over time; they illustrate the promise of phylogenetic methods within the domain of cognitive science, and they show how language change interacts with human perception.

Identifying universal principles underpinning diverse natural systems is a key goal of the life sciences. A powerful approach in addressing this goal has been to test whether patterns consistent with linguistic laws are found in nonhuman animals. Menzerath's law is a linguistic law that states that, the larger the construct, the smaller the size of its constituents. Here, to our knowledge, we present the first evidence that Menzerath's law holds in the vocal communication of a nonhuman species. We show that, in vocal sequences of wild male geladas (Theropithecus gelada), construct size (sequence size in number of calls) is negatively correlated with constituent size (duration of calls). Call duration does not vary significantly with position in the sequence, but call sequence composition does change with sequence size and most call types are abbreviated in larger sequences. We also find that intercall intervals follow the same relationship with sequence size as do calls. Finally, we provide formal mathematical support for the idea that Menzerath's law reflects compression-the principle of minimizing the expected length of a code. Our findings suggest that a common principle underpins human and gelada vocal communication, highlighting the value of exploring the applicability of linguistic laws in vocal systems outside the realm of language.

We share our thoughts with other minds, but we do not understand how. Having a common language certainly helps, but infants' and tourists' communicative success clearly illustrates that sharing thoughts does not require signals with a pre-assigned meaning. In fact, human communicators jointly build a fleeting conceptual space in which signals are a means to seek and provide evidence for mutual understanding. Recent work has started to capture the neural mechanisms supporting those fleeting conceptual alignments. The evidence suggests that communicators and addressees achieve mutual understanding by using the same computational procedures, implemented in the same neuronal substrate, and operating over temporal scales independent from the signals' occurrences.

Why are there different languages? A common explanation is that different languages arise from the gradual accumulation of random changes. Here, we argue that, beyond these random factors, linguistic differences, from sounds to grammars, may also reflect adaptations to different environments in which the languages are learned and used. The aspects of the environment that could shape language include the social, the physical, and the technological.

In biological evolution traits may rise and fall in frequency due to genetic drift, where variant frequencies change by chance, or by selection where advantageous variants will rise in frequency. The neutral model of evolution, first developed by Kimura in the 1960s, has become the standard against which selection is detected. While the balance between these two important forces - drift and selection - has been well established in biology there are other domains where the contribution of these processes is still coming together. Although the idea of natural selection has been applied to the cultural domain since the time of Darwin, it has proven more challenging to positively identify cultural traits under selection both because of a lack of established tests for selection and a lack of large cultural data sets. However, in recent years with the accumulation of large cultural data sets many cultural features from pre-history pottery to modern baby names have been shown to evolve according to the neutral theory. But there is accumulating empirical evidence from cultural processes suggesting that the neutral theory alone cannot account for all features of the data. As such, there has been a renewed interest in determining whether there is selection amidst drift. Here we analyze a subset English word frequencies, and determine whether frequency change reveals processes of selection. Inspired by the Moran and Wright-Fisher models in population genetics, we developed a neutral model of word frequency variation to assess when linguistic data appears to depart from neutral evolution. As such, our model represents a possible "test for selection" in the linguistic domain. We explore how the distribution of word use has changed for sets of words in English for more than 100 years (1901-2008) as expressed in vocabulary usage in published books, made available by Google Ngram. When comparing empirical word frequency changes to our neutral model we find pervasive and systematic departures from neutrality.

Interest in the origins and evolution of language has been around for as long as language has been around. However, only recently has the empirical study of language come of age. We argue that the field has sufficiently advanced that it now needs its own journal—the Journal of Language Evolution.

We make the case that, contra standard assumption in linguistic theory, the sound systems of human languages are adapted to their environment. While not conclusive, this plausible case rests on several points discussed in this work: First, human behavior is generally adaptive and the assumption that this characteristic does not extend to linguistic structure is empirically unsubstantiated. Second, animal communication systems are well known to be adaptive within species across a variety of phyla and taxa. Third, research in laryngology demonstrates clearly that ambient desiccation impacts the performance of the human vocal cords. The latter point motivates a clear, testable hypothesis with respect to the synchronic global distribution of language types. Fourth, this hypothesis is supported in our own previous work, and here we discuss new approaches being developed to further explore the hypothesis. We conclude by suggesting that the time has come to more substantively examine the possibility that linguistic sound systems are adapted to their physical ecology.

In recent years, laboratory studies of cultural evolution have become increasingly prevalent as a means of identifying and understanding the effects of cultural transmission on the form and functionality of transmitted material. The datasets generated by these studies may provide insights into the conditions encouraging, or inhibiting, high rates of innovation, as well as the effect that this has on measures of adaptive cultural change. Here we review recent experimental studies of cultural evolution with a view to elucidating the role of innovation in generating observed trends. We first consider how tasks are presented to participants, and how the corresponding conceptualization of task success is likely to influence the degree of intent underlying any deviations from perfect reproduction. We then consider the measures of interest used by the researchers to track the changes that occur as a result of transmission, and how these are likely to be affected by differing rates of retention. We conclude that considering studies of cultural evolution from the perspective of innovation provides us with valuable insights that help to clarify important differences in research designs, which have implications for the likely effects of variation in retention rates on measures of cultural adaptation.

Evolution is marked by well-defined events involving profound innovations that are known as 'major evolutionary transitions'. They involve the integration of autonomous elements into a new, higher-level organization whereby the former isolated units interact in novel ways, losing their original autonomy. All major transitions, which include the origin of life, cells, multicellular systems, societies or language (among other examples), took place millions of years ago. Are these transitions unique, rare events? Have they instead universal traits that make them almost inevitable when the right pieces are in place? Are there general laws of evolutionary innovation? In order to approach this problem under a novel perspective, we argue that a parallel class of evolutionary transitions can be explored involving the use of artificial evolutionary experiments where alternative paths to innovation can be explored. These 'synthetic' transitions include, for example, the artificial evolution of multicellular systems or the emergence of language in evolved communicating robots. These alternative scenarios could help us to understand the underlying laws that predate the rise of major innovations and the possibility for general laws of evolved complexity. Several key examples and theoretical approaches are summarized and future challenges are outlined.This article is part of the themed issue 'The major synthetic evolutionary transitions'.

Human languages are extraordinarily complex adaptive systems. They feature intricate hierarchical sound structures, are able to express elaborate meanings and use sophisticated syntactic and semantic structures to relate sound to meaning. What are the cognitive mechanisms that speakers and listeners need to create and sustain such a remarkable system? What is the collective evolutionary dynamics that allows a language to self-organize, become more complex and adapt to changing challenges in expressive power? This paper focuses on grammar. It presents a basic cycle observed in the historical language record, whereby meanings move from lexical to syntactic and then to a morphological mode of expression before returning to a lexical mode, and discusses how we can discover and validate mechanisms that can cause these shifts using agent-based models.This article is part of the themed issue 'The major synthetic evolutionary transitions'.

The evolution of life in our biosphere has been marked by several major innovations. Such major complexity shifts include the origin of cells, genetic codes or multicellularity to the emergence of non-genetic information, language or even consciousness. Understanding the nature and conditions for their rise and success is a major challenge for evolutionary biology. Along with data analysis, phylogenetic studies and dedicated experimental work, theoretical and computational studies are an essential part of this exploration. With the rise of synthetic biology, evolutionary robotics, artificial life and advanced simulations, novel perspectives to these problems have led to a rather interesting scenario, where not only the major transitions can be studied or even reproduced, but even new ones might be potentially identified. In both cases, transitions can be understood in terms of phase transitions, as defined in physics. Such mapping (if correct) would help in defining a general framework to establish a theory of major transitions, both natural and artificial. Here, we review some advances made at the crossroads between statistical physics, artificial life, synthetic biology and evolutionary robotics.This article is part of the themed issue 'The major synthetic evolutionary transitions'.

We make the case for developing a Computational Comparative Neuroprimatology to inform the analysis of the function and evolution of the human brain. First, we update the mirror system hypothesis on the evolution of the language-ready brain by (i) modeling action and action recognition and opportunistic scheduling of macaque brains to hypothesize the nature of the last common ancestor of macaque and human (LCA-m); and then we (ii) introduce dynamic brain modeling to show how apes could acquire gesture through ontogenetic ritualization, hypothesizing the nature of evolution from LCA-m to the last common ancestor of chimpanzee and human (LCA-c). We then (iii) hypothesize the role of imitation, pantomime, protosign and protospeech in biological and cultural evolution from LCA-c to Homo sapiens with a language-ready brain. Second, we suggest how cultural evolution in Homo sapiens led from protolanguages to full languages with grammar and compositional semantics. Third, we assess the similarities and differences between the dorsal and ventral streams in audition and vision as the basis for presenting and comparing two models of language processing in the human brain: A model of (i) the auditory dorsal and ventral streams in sentence comprehension; and (ii) the visual dorsal and ventral streams in defining "what language is about" in both production and perception of utterances related to visual scenes provide the basis for (iii) a first step towards a synthesis and a look at challenges for further research.

The claim that Eskimo languages have words for different types of snow is well-known among the public, but has been greatly exaggerated through popularization and is therefore viewed with skepticism by many scholars of language. Despite the prominence of this claim, to our knowledge the line of reasoning behind it has not been tested broadly across languages. Here, we note that this reasoning is a special case of the more general view that language is shaped by the need for efficient communication, and we empirically test a variant of it against multiple sources of data, including library reference works, Twitter, and large digital collections of linguistic and meteorological data. Consistent with the hypothesis of efficient communication, we find that languages that use the same linguistic form for snow and ice tend to be spoken in warmer climates, and that this association appears to be mediated by lower communicative need to talk about snow and ice. Our results confirm that variation in semantic categories across languages may be traceable in part to local communicative needs. They suggest moreover that despite its awkward history, the topic of "words for snow" may play a useful role as an accessible instance of the principle that language supports efficient communication.

Despite being a paradigm of quantitative linguistics, Zipf's law for words suffers from three main problems: its formulation is ambiguous, its validity has not been tested rigorously from a statistical point of view, and it has not been confronted to a representatively large number of texts. So, we can summarize the current support of Zipf's law in texts as anecdotic. We try to solve these issues by studying three different versions of Zipf's law and fitting them to all available English texts in the Project Gutenberg database (consisting of more than 30 000 texts). To do so we use state-of-the art tools in fitting and goodness-of-fit tests, carefully tailored to the peculiarities of text statistics. Remarkably, one of the three versions of Zipf's law, consisting of a pure power-law form in the complementary cumulative distribution function of word frequencies, is able to fit more than 40% of the texts in the database (at the 0.05 significance level), for the whole domain of frequencies (from 1 to the maximum value), and with only one free parameter (the exponent).

We consider the problem of the evolution of a code within a structured population of agents. The agents try to maximize their information about their environment by acquiring information from the outputs of other agents in the population. A naive use of information-theoretic methods would assume that every agent knows how to interpret the information offered by other agents. However, this assumes that it knows which other agents it observes, and thus which code they use. In our model, however, we wish to preclude that: It is not clear which other agents an agent is observing, and the resulting usable information is therefore influenced by the universality of the code used and by which agents an agent is listening to. We further investigate whether an agent that does not directly perceive the environment can distinguish states by observing other agents' outputs. For this purpose, we consider a population of different types of agents talking about different concepts, and try to extract new ones by considering their outputs only.

Vocalizations, and less often gestures, have been the object of linguistic research for decades. However, the development of a general theory of communication with human language as a particular case requires a clear understanding of the organization of communication through other means. Infochemicals are chemical compounds that carry information and are employed by small organisms that cannot emit acoustic signals of an optimal frequency to achieve successful communication. Here, we investigate the distribution of infochemicals across species when they are ranked by their degree or the number of species with which they are associated (because they produce them or are sensitive to them). We evaluate the quality of the fit of different functions to the dependency between degree and rank by means of a penalty for the number of parameters of the function. Surprisingly, a double Zipf (a Zipf distribution with two regimes, each with a different exponent) is the model yielding the best fit although it is the function with the largest number of parameters. This suggests that the 2 HERNÁNDEZ-FERNÁNDEZ & FERRER-I-CANCHO worldwide repertoire of infochemicals contains a core which is shared by many species and is reminiscent of the core vocabularies found for human language in dictionaries or large corpora.

For robots to effectively bootstrap the acquisition of language, they must handle referential uncertainty-the problem of deciding what meaning to ascribe to a given word. Typically when socially grounding terms for space and time, the underlying sensor or representation was specified within the grammar of a conversation, which constrained language learning to words for innate features. In this paper, we demonstrate that cross-situational learning resolves the issues of referential uncertainty for bootstrapping a language for episodic space and time; therefore removing the need to specify the underlying sensors or representations a priori. The requirements for robots to be able to link words to their designated meanings are presented and analyzed within the Lingodroids-language learning robots-framework. We present a study that compares predetermined associations given a priori against unconstrained learning using cross-situational learning. This study investigates the long-term coherence, immediate usability and learning time for each condition. Results demonstrate that for unconstrained learning, the long-term coherence is unaffected, though at the cost of increased learning time and hence decreased immediate usability.

Grounded language acquisition is an important issue, particularly to facilitate human-robot interactions in an intelligent and effective way. The evolutionary and developmental language acquisition are two innovative and important methodologies for the grounding of language in cognitive agents or robots, the aim of which is to address current limitations in robot design. This paper concentrates on these two main modelling methods with the grounding principle for the acquisition of linguistic ability in cognitive agents or robots. This review not only presents a survey of the methodologies and relevant computational cognitive agents or robotic models, but also highlights the advantages and progress of these approaches for the language grounding issue.

The interest in language evolution by various disciplines, such as linguistics, computer science, biology, etc., makes language evolution models an active research topic and many models have been defined in the last decade. In this work, an overview of computational methods and grammars in language evolution models is given. It aims to introduce readers to the main concepts and the current approaches in language evolution research. Some of the language evolution models, developed during the decade 2003–2012, have been described and classified considering both the grammatical representation (context-free, attribute, Christiansen, fluid construction, or universal grammar) and the computational methods (agent-based, evolutionary computation-based or game theoretic). Finally, an analysis of the surveyed models has been carried out to evaluate their possible extension towards multimodal language evolution.

Infants' own activities create and actively select their learning experiences. Here we review recent models of embodied information seeking and curiosity-driven learning and show that these mechanisms have deep implications for development and evolution. We discuss how these mechanisms yield self-organized epigenesis with emergent ordered behavioral and cognitive developmental stages. We describe a robotic experiment that explored the hypothesis that progress in learning, in and for itself, generates intrinsic rewards: The robot learners probabilistically selected experiences according to their potential for reducing uncertainty. In these experiments, curiosity-driven learning led the robot learner to successively discover object affordances and vocal interaction with its peers. We explain how a learning curriculum adapted to the current constraints of the learning system automatically formed, constraining learning and shaping the developmental trajectory. The observed trajectories in the robot experiment share many properties with those in infant development, including a mixture of regularities and diversities in the developmental patterns. Finally, we argue that such emergent developmental structures can guide and constrain evolution, in particular with regard to the origins of language.

Two computer simulations are investigated that model interaction of cultural evolution of language and biological evolution of adaptations to language. Both are agent-based models in which a population of agents imitates each other using realistic vowels. The agents evolve under selective pressure for good imitation. In one model, the evolution of the vocal tract is modeled; in the other, a cognitive mechanism for perceiving speech accurately is modeled. In both cases, biological adaptations to using and learning speech evolve, even though the system of speech sounds itself changes at a more rapid time scale than biological evolution. However, the fact that the available acoustic space is used maximally (a self-organized result of cultural evolution) is constant, and therefore biological evolution does have a stable target. This work shows that when cultural and biological traits are continuous, their co-evolution may lead to cognitive adaptations that are strong enough to detect empirically.

If human language must be squeezed through a narrow cognitive bottleneck, what are the implications for language processing, acquisition, change, and structure? In our target article, we suggested that the implications are far-reaching and form the basis of an integrated account of many apparently unconnected aspects of language and language processing, as well as suggesting revision of many existing theoretical accounts. With some exceptions, commentators were generally supportive both of the existence of the bottleneck and its potential implications. Many commentators suggested additional theoretical and linguistic nuances and extensions, links with prior work, and relevant computational and neuroscientific considerations; some argued for related but distinct viewpoints; a few, though, felt traditional perspectives were being abandoned too readily. Our response attempts to build on the many suggestions raised by the commentators and to engage constructively with challenges to our approach.

Memory is fleeting. New material rapidly obliterates previous material. How, then, can the brain deal successfully with the continual deluge of linguistic input? We argue that, to deal with this "Now-or-Never" bottleneck, the brain must compress and recode linguistic input as rapidly as possible. This observation has strong implications for the nature of language processing: (1) the language system must "eagerly" recode and compress linguistic input; (2) as the bottleneck recurs at each new representational level, the language system must build a multilevel linguistic representation; and (3) the language system must deploy all available information predictively to ensure that local linguistic ambiguities are dealt with "Right-First-Time"; once the original input is lost, there is no way for the language system to recover. This is "Chunk-and-Pass" processing. Similarly, language learning must also occur in the here and now, which implies that language acquisition is learning to process, rather than inducing, a grammar. Moreover, this perspective provides a cognitive foundation for grammaticalization and other aspects of language change. Chunk-and-Pass processing also helps explain a variety of core properties of language, including its multilevel representational structure and duality of patterning. This approach promises to create a direct relationship between psycholinguistics and linguistic theory. More generally, we outline a framework within which to integrate often disconnected inquiries into language processing, language acquisition, and language change and evolution.

This paper presents an evolutionary approach that, given a performance goal, produces a communication strategy that can improve a multi-agent system’s performance with respect to the desired goal. The evolved strategy determines what, when, and to whom agents communicate. The proposed approach further enables tuning the trade-off between the performance goal and communication cost, to produce a strategy that achieves a good balance between the two objectives, according the system’s designer needs. Experiments are designed to evaluate the approach using the Wumpus World application domain, with variations of three factors: fitness parameters (including objectives’ weights and action and communication costs), fitness goal, and simulation environment. Results show that the system’s performance can be highly tuned by controlling communication, and that the presented approach has significant utilization in improving the performance with respect to the goal.

Explaining the evolution of speech and language poses one of the biggest challenges in biology. We propose a dual network model that posits a volitional articulatory motor network (VAMN) originating in the prefrontal cortex (PFC; including Broca's area) that cognitively controls vocal output of a phylogenetically conserved primary vocal motor network (PVMN) situated in subcortical structures. By comparing the connections between these two systems in human and nonhuman primate brains, we identify crucial biological preadaptations in monkeys for the emergence of a language system in humans. This model of language evolution explains the exclusiveness of non-verbal communication sounds (e.g., cries) in infants with an immature PFC, as well as the observed emergence of non-linguistic vocalizations in adults after frontal lobe pathologies.

Human language has unusual structural properties that enable open-ended communication. In recent years, researchers have begun to appeal to cultural evolution to explain the emergence of these structural properties. A particularly fruitful approach to this kind of explanation has been the use of laboratory experiments. These typically involve participants learning and interacting using artificially constructed communication systems. By observing the evolution of these systems in the lab, researchers have been able to build a bridge between individual cognition and population-wide emergent structure. We review these advances, and show how cultural evolution has been used to explain the origins of structure in linguistic signals, and in the mapping between signals and meanings.

Social conventions govern countless behaviors all of us engage in every day, from how we greet each other to the languages we speak. But how can shared conventions emerge spontaneously in the absence of a central coordinating authority? The Naming Game model shows that networks of locally interacting individuals can spontaneously self-organize to produce global coordination. Here, we provide a gentle introduction to the main features of the model, from the dynamics observed in homogeneously mixing populations to the role played by more complex social networks, and to how slight modifications of the basic interaction rules give origin to a richer phenomenology in which more conventions can co-exist indefinitely.

In language, recombination of a discrete set of meaningless building blocks forms an unlimited set of possible utterances. How such combinatorial structure emerged in the evolution of human language is increasingly being studied. It has been shown that it can emerge when languages culturally evolve and adapt to human cognitive biases. How the emergence of combinatorial structure interacts with the existence of holistic iconic form‐meaning mappings in a language is still unknown. The experiment presented in this paper studies the role of iconicity and human cognitive learning biases in the emergence of combinatorial structure in artificial whistled languages. Participants learned and reproduced whistled words for novel objects with the use of a slide whistle. Their reproductions were used as input for the next participant, to create transmission chains and simulate cultural transmission. Two conditions were studied: one in which the persistence of iconic form‐meaning mappings was possible and one in which this was experimentally made impossible. In both conditions, cultural transmission caused the whistled languages to become more learnable and more structured, but this process was slightly delayed in the first condition. Our findings help to gain insight into when and how words may lose their iconic origins when they become part of an organized linguistic system.

A distinguishing property of human intelligence is the ability to flexibly use language in order to communicate complex ideas with other humans in a variety of contexts. Research in natural language dialogue should focus on designing communicative agents which can integrate themselves into these contexts and productively collaborate with humans. In this abstract, we propose a general situated language learning paradigm which is designed to bring about robust language agents able to cooperate productively with humans. This dialogue paradigm is built on a utilitarian definition of language understanding. Language is one of multiple tools which an agent may use to accomplish goals in its environment. We say an agent “understands” language only when it is able to use language productively to accomplish these goals. Under this definition, an agent’s communication success reduces to its success on tasks within its environment. This setup contrasts with many conventional natural language tasks, which maximize linguistic objectives derived from static datasets. Such applications often make the mistake of reifying language as an end in itself. The tasks prioritize an isolated measure of linguistic intelligence (often one of linguistic competence, in the sense of Chomsky (1965)), rather than measuring a model’s effectiveness in real-world scenarios. Our utilitarian definition is motivated by recent successes in reinforcement learning methods. In a reinforcement learning setting, agents maximize success metrics on real-world tasks, without requiring direct supervision of linguistic behavior.

The current mainstream approach to train natural language systems is to expose them to large amounts of text. This passive learning is problematic if we are interested in developing interactive machines, such as conversational agents. We propose a framework for language learning that relies on multi-agent communication. We study this learning in the context of referential games. In these games, a sender and a receiver see a pair of images. The sender is told one of them is the target and is allowed to send a message from a fixed, arbitary vocabulary to the receiver. The receiver must rely on this message to identify the target. Thus, the agents develop their own language interactively out of the need to communicate. We show that two networks with simple configurations are able to learn to coordinate in the referential game. We further explore how to make changes to the game environment to cause the “word meanings” induced in the game to better reflect intuitive semantic properties of the images. In addition, we present a simple strategy for grounding the agents’ code into natural language. Both of these are necessary steps towards developing machines that are able to communicate with humans productively.

An important reason to investigate dolphins is that they exhibit striking similarities with humans. Like us, they use tools: dolphins break off sponges and wear them over their rostrum while foraging on the seafloor (Smolker et al 1997). Dolphins are also capable of recognizing their body in front of a mirror (Reiss & Marino 2001). Closely related with their capacity to see through sound is their capacity to form abstract representations that are independent from modality (Herman et al 1998). Dolphins share with us other traits that are appealing from the perspective of language theory. First, they exhibit spontaneous vocal mimicry (Reiss & McCowan 1993) which suggests a predisposition to learn a vocal communication system. Second, they live, in general, in fission‐fusion societies and display complex social behaviours (Lusseau et al 2003, Connor & Krützen 2015) while converging research supports that the complexity of a society and the complexity of communication are correlated (Freeberg et al 2012). Third, they can learn a signal to innovate, namely to show a behavior not seen in the current interaction session (Foer 2015). This tells us something about the limits on memory and creativity in dolphins and is challenging from a theoretical perspective: many researchers believe that a crucial difference between humans and other species is our unbounded capacity to generate sequences, e.g., by embedding sentences into other sentences (e.g., Gregg 2013, Hauser et al 2002), or a capacity for large lexicons (Hurford 2004). In short, bottlenose dolphins share many traits we associate as pre‐requisite for our complex linguistic abilities. Although possessing such an infinite capacity makes a qualitative difference compared to a species with a finite capacity, the fact is that (a) a species being able to generate a huge number of sentences may not be distinguishable from a species that has infinite capacity (supposing that the latter is really true) and (b), humans have problems with parsing sentences with just a few levels of embedding (Christiansen & Chater 2015). The point is that the problem of infinite vs finite capacity does seem to be well poised and that dolphin capacity to innovate is being overlooked. We humans are fascinated by infinity (perhaps for purely aesthetical reasons) and may have rushed to steal the flag of infinity to keep it in some anthropocentric fortress where other species are not allowed to get in. In a recent book, the parallel in cognitive abilities between …