Xenolinguistics: Towards a Science of Extraterrestrial Language 1032399600, 9781032399607

Citation preview

XENOLINGUISTICS

Xenolinguistics brings together biologists, anthropologists, linguists, and other experts specializing in language and communication to explore what non-human, nonEarthbound language might look like. The 18 chapters examine what is known about human language and animal communication systems to provide reasonable hypotheses about what we may find if we encounter non-Earth intelligence. Showcasing an interdisciplinary dialogue between a set of highly established scholars, this volume: • Clarifies what is and is not known about human language and animal communication systems • Presents speculative arguments as a philosophical exercise to help define the boundaries of what our current science can tell us about non-speculative areas of investigation • Provides readers with a clearer sense of the how our knowledge about language is better informed through a cross-disciplinary investigation • Offers a better understanding of future avenues of research on language This rich interdisciplinary collection will be of interest to researchers and students studying non-human communication, astrobiology, and language invention. Douglas A. Vakoch is Director of METI (Messaging Extraterrestrial Intelligence), a nonprofit research and educational organization dedicated to transmitting intentional signals to nearby stars. He is the editor of more than a dozen books, including Ecofeminist Science Fiction: International Perspectives on Gender, Ecology, and Literature (2021) and The Routledge Handbook of Ecofeminism and Literature (2022). Jeffrey Punske is Associate Professor and the director of undergraduate studies in linguistics at Southern Illinois University. He is the editor of Language Invention and Linguistics Pedagogy (2020).

XENOLINGUISTICS Towards a Science of Extraterrestrial Language

Edited by Douglas A. Vakoch and Jeffrey Punske

Designed cover image: © Getty Images | WhataWin First published 2024 by Routledge 4 Park Square, Milton Park, Abingdon, Oxon OX14 4RN and by Routledge 605 Third Avenue, New York, NY 10158 Routledge is an imprint of the Taylor & Francis Group, an informa business © 2024 selection and editorial matter, Douglas A. Vakoch and Jeffrey Punske; individual chapters, the contributors The right of Douglas A. Vakoch and Jeffrey Punske to be identified as the authors of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library ISBN: 978-1-032-39960-7 (hbk) ISBN: 978-1-032-39959-1 (pbk) ISBN: 978-1-003-35217-4 (ebk) DOI: 10.4324/9781003352174 Typeset in Times New Roman by Apex CoVantage, LLC

CONTENTS

List of Contributors 1

Goals of the Volume Jeffrey Punske

2

Many Ways to Say Things: What the Diversity of Animal Communication on Earth Can Tell Us About the Likely Nature of Alien Language Arik Kershenbaum

3

Recognizing Intentional Signals and Their Meanings in Non-Human Communication Catherine Hobaiter, Adriano R. Lameira, and Derek Ball

4

Getting Out of Our Skin: What Decoding Interspecies Communication and Nonhuman Intelligence Can Tell Us About Deciphering Alien Languages Denise L. Herzing

5

Communicative Resources Beyond the Verbal Tier: A View on Xenolinguistics From Interactional Linguistics Heike Ortner

viii 1

3

15

33

39

vi Contents

6

How Studies of Communication Among Nonhumans and Between Humans and Nonhumans Can Inform SETI Irene M. Pepperberg

51

7

Patterns of Communication of Human Complex Societies as a Blueprint for Alien Communication Anamaria Berea

62

8

Interstellar Competence: Applications of Linguistics and Communicative and Cultural Competencies to Extraterrestrial Communication Sumayya K.R. Granger, Judd Ethan Ruggill, and Ken S. McAllister

74

9

Why Do We Assume That We Can Decode Alien Languages? Con Slobodchikoff

10 Xenolinguistic Fieldwork Claire Bowern 11 Investigating the Foundations of Meaning in a Xenolanguage Andrew McKenzie 12 A Linguistic Perspective on the Drake Equation: Knowns and Unknowns for Human Languages and Extraterrestrial Communication Daniel Ross

87 96 110

123

13 Cognition, Sensory Input, and Linguistics: A Possible Language for Blind Aliens Sheri Wells-Jensen

138

14 The Design Features of Extraterrestrial Language: A Domain-General Approach Darcy Sperlich

152

15 Universal Grammar Ian G. Roberts, Jeffrey Watumull, and Noam Chomsky

165

Contents

vii

16 Where Does Universal Grammar Fit in the Universe? Human Cognition and the Strong Minimalist Thesis Bridget D. Samuels and Jeffrey Punske

182

17 Learning and Adaptation of Communication Systems in Biological Life Forms Jessie S. Nixon and Fabian Tomaschek

194

18 Writing Systems and METI: Off-the-Shelf Encoding of Human Physiology, Language, Cognition, and Culture Daniel Harbour

214

Index

230

CONTRIBUTORS

Derek Ball is a senior lecturer in philosophy at the University of St Andrews. He holds a PhD in Philosophy from the University of Texas at Austin. His research focuses on understanding the mind and its place in the natural world, with special focus on the ways in which our thought and language are shaped by the environment. During his time in St Andrews, he has served as Coordinator of the St Andrews/ Stirling joint Ph.D. program in philosophy, and as Director of the Arche Research Centre. He has held visiting fellowships at the Australian National University and the University of Oslo, as well as research grants from the Arts and Humanities Research Council. He is the co-editor of The Science of Meaning (Oxford University Press), and his work has appeared in Mind, Philosophy and Phenomenological Research, Philosophers’ Imprint, Inquiry, and elsewhere. Anamaria Berea is an associate professor in computer science at George Mason University. She holds Ph.D. degrees in economics (2010) and computational social science (2012), and her current research is focused on the emergence of communication in biological and social networks by applying theories and methods from economics, complex systems, and data science to reinterpret historical, anthropological, biological, and artistic data regarding the fundamental aspects of communication on our planet, from signaling in simple biological organisms to complex human and computer languages. Claire Bowern is a professor at Yale University. She is one of the leading scholars on language documentation, historical linguistics, and indigenous languages of Australia. She has authored numerous articles and chapters related to these topics. She is the author or co-author of two textbooks: An Introduction to Historical Linguistics and Linguistic Fieldwork: A Practical Guide. She is also the author of A

Contributors

ix

Grammar of Bardi and Sivisa Titan: Grammar, Texts, Vocabulary. She earned her Ph.D. in linguistics from Harvard in 2004. Noam Chomsky, often considered the founder of modern linguistics, is one of the most cited scholars in modern history. Among his groundbreaking books are Syntactic Structures, Language and Mind, Aspects of the Theory of Syntax, and The Minimalist Program, each of which has made distinct contributions to the development of the field. He has received numerous awards, including the Kyoto Prize in Basic Sciences, the Helmholtz Medal, and the Ben Franklin Medal in Computer and Cognitive Science. Chomsky is also one of the most influential public intellectuals in the world. He has written more than 100 books. Chomsky joined the University of Arizona in fall 2017, coming from the Massachusetts Institute of Technology (MIT), where he worked since 1955 as professor of linguistics, then as professor of linguistics, emeritus. Sumayya K.R. Granger is an assistant professor in the Department of Public and Applied Humanities at the University of Arizona. She is also the Associate Director of Program Administration at the Center for English as a Second Language. She earned her M.A. and Ph.D. in linguistics, writing her dissertation on the syntactic theory of modals, and she has written on other topics, such as plural formation in Nahuatl and Aktionsart in Navajo. She has taught classes in pedagogical grammar and intercultural competence, and she has presented several times on different aspects of language teaching/learning and intercultural competence. Daniel Harbour is Professor of the Cognitive Science of Language at Queen Mary University of London, where he currently holds a Leverhulme Research Fellowship exploring the impact of grammatical change on writing system evolution. He received his Ph.D. in linguistics from the MIT in 2003 for work on the complex number system of a Native American language. Since then, he has conducted indepth fieldwork on languages on four continents. He uses broad typological variation, coupled with language-specific investigation, to build mathematically precise models of the conceptual building blocks of human language, particularly in relation to systems of counting and deixis. His publications include An Intelligent Person’s Guide to Atheism, Morphosemantic Number, and Impossible Persons. Denise L. Herzing, Founder and Research Director of the Wild Dolphin Project, has completed more than 33 years of her long-term study of the Atlantic spotted dolphins inhabiting Bahamian waters. She is an affiliate assistant professor in biological sciences at Florida Atlantic University. Dr. Herzing is a 2008 Guggenheim Fellow, a fellow with the Explorers Club, a scientific adviser for the Lifeboat Foundation and the American Cetacean Society, and on the board of Schoolyard Films. In addition to many scientific articles, she is the co-editor of Dolphin Communication and Cognition, and author of Dolphin Diaries: My 25 Years with Spotted Dolphins in the Bahamas and The Wild Dolphin Project (2002).

x

Contributors

Catherine Hobaiter is a field primatologist who has spent the past 19 years studying wild primates across Africa. She received her Ph.D. from the University of St Andrews and is now a Reader in the School of Psychology and Neuroscience. She spends around half the year in the field and leads a team of researchers exploring great ape behavior. The main focus of her research is communication and cognition in wild apes. Through long-term field studies, she explores what the behavior of modern apes living in their natural environment tells us about their minds and also about the evolutionary origins of our own behavior. Her work revealed that apes use large repertoires of gestures with specific meanings in a language-like way. Arik Kershenbaum is a fellow of Girton College, University of Cambridge, and a college lecturer. He researches animal communication across diverse species, with a particular interest in the communication between cooperative predators. Using field experiments and mathematical simulations, he tries to understand what this communication tells us about the fundamental principles driving the evolution of language. Dr. Kershenbaum’s research focuses mostly on wolves, which he studies in Montana, Spain, and Italy, and on free-ranging human-habituated dolphins in the Red Sea. He held the Herchel Smith Research Fellowship in Zoology at the University of Cambridge, and prior to this, he was a postdoctoral research fellow at the National Institute for Mathematical and Biological Synthesis (NIMBioS) in Knoxville, Tennessee. Dr Kershenbaum received his Ph.D. from the Haifa University in Israel, where he studied the communication and behavior of the rock hyrax. Adriano R. Lameira started primate vocal research in the early 2000s in Borneo with wild orangutans, and quickly expanded his research in the years that followed to Sumatran orangutans. Today, he holds the largest collection of orangutan calls ever assembled, spanning multiple wild and captive populations and several tens of thousands of observation hours. Since the beginning of his work, orangutans have exhibited a level of vocal diversity, flexibility. And learnability that has surpassed traditional expectations. He is an assistant professor at the University of Warwick. Ken S. McAllister is Associate Dean for Research and Program Innovation for the College of Humanities at the University of Arizona, where he is also Professor of Public and Applied Humanities. He holds affiliate appointments in the departments of English and Teaching, Learning, and Sociocultural Studies, as well as in the School of Information. He co-founded and co-directs (with Judd Ethan Ruggill) the Learning Games Initiative and its attendant research archive. His research focuses on technologically enhanced modes of persuasion, particularly in transdisciplinary contexts.

Contributors

xi

Andrew McKenzie is an associate professor at the University of Kansas. He is the co-author of Plains Life in Kiowa: Voices from a Tribe in Transition along with numerous articles and chapters on natural language semantics and the structure and documentation of Kiowa. His work has been supported by two National Science Foundation (NSF) grants. He has also co-authored work regarding language change during long-distance space travel for the European Space Agency’s journal, Acta Futura. Prior to his service at the University of Kansas, he was employed at University of Texas at Arlington. He earned his Ph.D. in linguistics in 2012 from the University of Massachusetts Amherst. Jessie S. Nixon completed her Ph.D. in psycholinguistics at the University of Leiden in 2014. Dr. Nixon uses neuroscientific and behavioral methods, combined with advanced statistical modeling, to investigate speech processing and acquisition in a variety of languages. Dr. Nixon’s research has demonstrated that listeners’ level of perceptual uncertainty during speech perception is highly sensitive to the shape of statistical cue distributions in both native and non-native speech and in both segmental and prosodic cues. Dr. Nixon is currently based in the Quantitative Linguistics Group, University of Tübingen, where she is working on a project developing a computational model of speech perception and production based on learning theory and discriminative linguistics. Heike Ortner is an associate professor at the Department of German Studies, University of Innsbruck, Austria. She received her master degree at the Karl-Franzens-University Graz in German Philology and Applied Linguistics. In 2012, she completed her Ph.D. studies at the University of Innsbruck with a “Promotio sub auspiciis praesidentis rei publicae,” the highest honor for academic achievements in Austria. From 2011–2014, she held a postdoctorate position in the division “German Linguistics” at the Department of German Studies. She became assistant professor at the same department in 2014. In 2016, she was a short-term visiting professor at the Wirth Institute at the University of Alberta, Canada. Her research interests include language and emotion, interactional and multimodal linguistics, linguistic aspects of public health communication, computer-mediated communication, and discourse analysis. Irene M. Pepperberg is a research associate at Harvard, and has been a Northwestern University visiting assistant professor, a University of Arizona tenured associate professor, an MIT Media Lab visiting associate professor, a Brandeis University adjunct associate professor and an MIT senior lecturer. She has received John Simon Guggenheim, Whitehall, Harry Frank Guggenheim, Selby, and Radcliffe fellowships, was an alternate for the Cattell Award for Psychology, won the 2005 Frank Beach Award, and was nominated for Weizmann, L’oréal, and Grawemeyer awards and the Animal Behavior Society’s 2001 Quest and 2015 Exemplar awards. She received St. Johns University’s 2013 Clavius Award. She authored the

xii

Contributors

book, The Alex Studies, on grey parrot cognition and interspecies communication. Her memoir, Alex & Me, a New York Times bestseller, won a Christopher Award. She has published more than 100 scholarly articles in peer-reviewed journals and as book chapters. Jeffrey Punske received his Ph.D. in linguistics from the University of Arizona in 2012. He is an associate professor in the School of Languages and Linguistics at Southern Illinois University, With previous employment at Kutztown University in Pennsylvania and the University of Oklahoma. His primary research is one the interfaces of morphosyntax with other grammatical, and cognitive components. He is the co-editor of the book Language Invention in Linguistics Pedagogy. He has also co-authored work regarding language change during long-distance space travel for the European Space Agency’s journal, Acta Futura. Ian G. Roberts is a professor of linguistics at the University of Cambridge. His research is in theoretical linguistics, more specifically in comparative syntax. His work is set against the background assumptions advocated by Noam Chomsky: that there exists a specific human cognitive capacity for language that is present at birth and requires simple environmental stimulation for linguistic competence in the mother tongue to develop during the early years of life. He currently holds a European Research Council Advanced Grant funding a project whose goal is to investigate a specific hypothesis as to the way in which the grammatical options made available by universal grammar are organized. Refining and testing this hypothesis involves looking at languages from all over the world and assessing the extent to which certain patterns recur. Dr. Roberts received his B.A. from Bangor and his Ph.D. from University of Southern California. Daniel Ross is a lecturer of linguistics at the University of California, Riverside. He has taught courses on syntax, semantics, morphology, and historical linguistics. His dissertation (University of Illinois Urbana-Champaign, 2021) focuses on pseudocoordination, serial verb constructions and other multi-verb predicates as instances of form-structure mismatches in syntax; from a comparative perspective, these constructions are strikingly similar in function and syntactic properties despite variation in form, and from a theoretical perspective, the data from English and other languages proves difficult to explain in conventional syntactic theory. The research in the dissertation is supported by a 325-language comparison of an array of morphosyntactic features. Judd Ethan Ruggill is Head of the Department of Public and Applied Humanities at the University of Arizona, and an affiliate faculty member in the Africana Studies Program; the Department of English; the School of Information; the School of Theatre, Film, and Television; and the Graduate Interdisciplinary Program in Social, Cultural, and Critical Theory. In addition, he co-directs the Learning Games

Contributors

xiii

Initiative, a transdisciplinary, inter-institutional research group he co-founded in 1999 to study, teach with, and build computer games. He primarily researches play and the technologies, industries, and sociocultural phenomena that enable it. He has published and presented on topics ranging from game design for second language acquisition and teaching to the wicked problem of collaboration. Bridget D. Samuels received her Ph.D. in linguistics from Harvard University and is a member of the Center for Craniofacial Molecular Biology at the University of Southern California. Previously, she has held teaching positions at the University of Maryland and Pomona College. Her research focuses on the evolution of language and cognition in the history of humankind, phonological theory, and the interface between phonology and morphosyntax. She is the author of Phonological Architecture: A Biolinguistic Perspective (2011, Oxford University Press), as well as numerous other publications in evolutionary, biological, and theoretical linguistics. Con Slobodchikoff is a professor emeritus of biology at Northern Arizona University, Flagstaff. He received B.S. and a Ph.D. degrees from the University of California, Berkeley, and has been a Fulbright Fellow and a visiting professor of Zoology at Kenyatta University in Kenya. Slobodchikoff has authored two books on animal behavior (Harvard University Press and St. Martin’s Press); has edited three books on subjects of species concepts, ecology, and social behavior; and has published about 100 papers and book chapters in the scientific literature. His work with prairie dogs has drawn considerable interest, and he was the subject of a BBC one-hour documentary (Talk of the Town) and an Animal Planet documentary (Prairie Dogs). He has appeared in a variety of media programs such as NBC Dateline, ABC World News, the BBC, the CBC, Australian Broadcasting, German-Belgian television, CNN, and Turner Broadcasting. Darcy Sperlich is an associate professor of Linguistics at Xi’an Jiaotong – Liverpool University. He works both in theoretical and experimental linguistics with a cross-linguistic perspective, focusing on syntax and pragmatics, and their interfaces. He has published a number of articles, and published a book entitled Reflexive Pronouns: A Theoretical and Experimental Synthesis. Fabian Tomaschek obtained his Ph.D. in 2013 with a study on the neural mechanisms of vowel perception. In 2012, he became a postdoctoral research fellow at the Department of General Linguistics, University of Tübingen. Ever since, he has focused in his work on how humans learn to encode and decode messages using the acoustic signal. He investigates the articulation of vowels and uses advanced statistical modeling to predict their acoustic and articulatory shape. His work has been published among others in the Journal of Phonetics, PloS ONE, Brain and Language, and Linguistics Vanguard. On his website, he also published an introductory textbook on programming data analyses in linguistic corpora with R.

xiv

Contributors

Jeffrey Watumull is the artificial intelligence (AI) lead at Oceanit. He received B.S. degrees, summa cum laude, in mathematics and evolutionary biology from Harvard University in 2009, an M.Phil. in linguistics from the University of Cambridge in 2010, and a Ph.D. in artificial intelligence from MIT in 2015. His research has centered on the mathematical optimality of human cognition, specifically in the domain of language. He has published numerous papers in peer-reviewed journals and his work has been covered frequently in the media. At Oceanit, Dr. Watumull has embarked on research to develop strong AI: computational systems equipped with human-level intelligence in domains both specific (e.g., language) and general (e.g., problem solving). He is currently principal investigator on a Defense Advanced Research Projects Agency (DARPA) seedling, an Office of Naval Research Broad Agency Announcement (ONR BAA), and a number of Small Business Innovation Research grants (SBIRs). Sheri Wells-Jensen is a professor in the Department of English and coordinator of the minor in linguistics at Bowling Green State University in Ohio. Along with various aspects of issues pertaining to SETI (the search for extraterrestrial intelligence), her research interests include phonetics, braille, language preservation, TESOL (teaching English to speakers of other languages), language creation, and disability studies. She serves on the board of directors of Messaging Extraterrestrial Intelligence (METI). Dr. Wells-Jensen also coordinates BGSU’s graduate certificate in TESOL and teaches courses in general linguistics, applied phonology, and applied syntax.

1 GOALS OF THE VOLUME Jeffrey Punske

To date, there is no serious evidence of contact between humanity and non-terrestrial intelligence. In fact, as of writing, there is no confirmed evidence available to us of any form of life that did not originate on Earth. Given the vastness of the universe and the processes involved in both the development of life and the development of intelligence, it is generally assumed that this is due to limitations of our ability to find the evidence – not necessarily the lack entirely. It is unlikely that contact with any non-terrestrial intelligence will occur within the lifetimes of any of the contributors to this volume. This raises the question of the purpose of a volume like this. This volume seeks to explore the potential nature of a non-terrestrial intelligence with linguistic capabilities. We cannot observe this nature, which moves us into a different realm of inquiry than the standard scientific approach. Thus, rather than being a speculative volume about the potential nature of non-Earth intelligence/linguistic systems, this volume largely explores what we know about communication systems, languages, and other cognitive systems on Earth through the lens of what we might observe beyond it. In doing so, we hope to help define the limits of what we presently know about such systems and provide the foundation for future explorations into broader questions about the nature of language and communication. This volume draws together a range of experts from various fields and divergent perspectives. Our goal is not to provide any definitive answers. Rather, we seek to show the division points across these fields and perspectives. The volume may be roughly divided into three major themes: animal communication systems, human language and linguistics, and communication systems more generally. As each of these topics closely intersect, there is not a neat division from one topic to the next – rather, each contribution connects with others to try to define the question more broadly. DOI: 10.4324/9781003352174-1

2

Jeffrey Punske

It is our hope that in trying to define what we know and where the limitations of our knowledge lie, we can work towards an integrated science of language and communication that takes seriously the fundamental and universal natures of both. By utilizing our fascination of with stars and the unknown, we can perhaps enhance our own knowledge of ourselves – and we may continue to explore the nature of the unknown with imagination but also with a core knowledge of what might be out there.

2 MANY WAYS TO SAY THINGS What the Diversity of Animal Communication on Earth Can Tell Us About the Likely Nature of Alien Language Arik Kershenbaum

One of the most perplexing questions in modern evolutionary biology is this: Why has only one species on Earth evolved a true, rich language? In almost four billion years of evolution, humans stand out among all the species that have ever evolved as the only one able to communicate a limitless number of concepts between conspecifics. From an anthropocentric perspective, it seems obvious that language brings tremendous adaptive advantages, particularly in terms of the cooperation it enables, and humanity’s huge technological and intellectual advances testify to the power of such cooperation. But the question may not be a straightforward as it seems, because we need to examine the claim that humans are the only species with language. Indeed, our definition of language may be biased in such a way that only humans fit the definition – maintaining the qualitative difference between humans and other animals that has been a particular quest of philosophers and natural scientists for generations. Animals on Earth have exploited a staggering range of strategies to increase their fitness by communicating both with con- and hetero-specifics, and examples of these evolutionary strategies can aid us in understanding what extraterrestrial communication may be like (Herzing, this volume [Chapter 4]; Pepperberg, this volume [Chapter 6]; Slobodchikoff, this volume [Chapter 9]). Almost every physical modality has been explored, from the familiar acoustic and visual channels to the seemingly unlikely electromagnetic – and within each modality, different encoding strategies have been used. Indeed, it is hard to find a potential form of physical communication that has not been explored somewhere in evolutionary history on Earth. It therefore seems quite likely that life on other planets will make use of some of the underlying communicative techniques exploited on this planet, albeit with particular adaptations suited to the specific conditions of each habitat. For this reason, it is instructive to examine DOI: 10.4324/9781003352174-2

4

Arik Kershenbaum

the breadth of communicative strategies employed by animals on Earth, and to ask what is it about each strategy that would, or would not, favour it as a carrier for linguistic behaviour. Without doubt human language arose through natural selection. Whether language evolved from a precursor in the form of a simple system of communication in proto-humans, or arose independently following rapid social and cognitive development of our ancestors (Jackendoff 2011), evolutionary forces certainly shaped the uptake and development of the linguistic trait, as with any phenotypic trait. Therefore, certain selective pressures favoured the evolution of language in one lineage of hominins but not others, or in any other taxa across the animal kingdom. Nonetheless, most – if not all – animals communicate, and sometimes with phenomenally complex and precise signals that appear able to carry linguistic content – although they are not used in that way. What is it that caused one evolutionary trajectory to accelerate in the direction of language while other potential trajectories were discarded? Sensory systems are energetically expensive (Niven and Laughlin 2008), and every sensory system we know of makes use of a considerable part of an animal’s energy budget. For example, in the blowfly Calliphora vicina, energy consumption by the retina alone is as much as 8% of the resting metabolic rate (Howard et al. 1987). On Earth, such sensory systems, and the processing mechanisms attached to them, are energetically expensive largely due to the pumping of ions in and out of cells in neural tissue – a parochial mechanism that may not be used by creatures that have evolved on other worlds. However, whatever its implementation, communication is fundamentally energetic because it acts to reduce information entropy in a system. Therefore, as long as energy is a limiting resource in a particular ecosystem, we expect that communication systems will evolve on other planets only to address important selective pressures that justify investing a considerable proportion of an animal’s energy budget into producing, detecting, and processing communication signals. In a hypothetical habitat where energy is not, in fact, a limiting resource, the consequential reduction in competition and selective pressure may mean that communication – and many other things – would be unlikely to evolve at all! Because of the energetic cost of sensory systems, most such systems should be broadly tuned (Stevens 2013: 62), i.e., should respond in a general fashion to a wide range of signals rather than be dedicated to detecting signals of a particular type. However, communication requires a degree of sensor specificity because discrimination between signals with different meanings is a fundamental requirement of communication. Therefore, sensory systems should evolve in the face of strong selective pressure, and communicative systems should evolve only when selective pressures are even stronger, such that precious sensory resources can be dedicated to transmitting and receiving specific signals. This will also allow broadly tuned sensory systems (such as hearing) to respond to narrow signals with specific meaning, such as alarm calls.

Many Ways to Say Things

5

Communicative Modalities Sound and Vision

We are most familiar with the particular communication modality that has come to dominate animal communication on this planet and which forms the basis of spoken human language. Acoustic communication is not the most ancient form of communication but it is undoubtedly the most studied. Despite the dangerous tendency to attach more significance than it deserves to our own preference for a particular modality, it is nonetheless the case that acoustic communication appears to be so widespread because it provides a number of objective and significant advantages. Acoustic communication is fast: sound travels at 340 m/s in air, and 1500 m/s in water, far faster than the movement of any animal on this planet. Speed also means that acoustic signals can be temporally discriminated, which greatly increases the amount of information they can carry compared to slowly varying modalities such as olfactory cues. Critically, acoustic signals also diffract around objects commonly found in both the terrestrial and marine environments, which means that the signal can be detected even if the receiver is not in the line of sight of the sender – a key disadvantage of the visual modality. Visual signals can contain large amounts of information, particularly by manipulating the geometrical configuration of objects. This is most obviously demonstrated by the shapes of the letters on this page, which provide almost unlimited flexibility for information encoding. Visual signalling is widespread across the animal kingdom, quite likely because the production of complex geometrical configurations does not require specialist production mechanisms beyond those normally possessed by animals. The bristling of hair on an angry cat, or the thumbs-down signal of a Roman emperor, make use of anatomical features of the animals that are already evolved for manipulating the environment. Therefore, it seems likely that in any non-opaque alien environment, some form of visual signalling will exist. However, although visual signalling almost certainly preceded acoustic communication in ancient hominins, it is possible that the very fact that visual signalling relies on geometric configuration may limit its utility in many environments. Although sound and light both dissipate similarly with distance, decoding visual information also depends on angular resolution, so visual signals are more difficult to distinguish at long ranges. Finally, acoustic signals can be generated with varying frequency, which adds another layer of complexity to the channel and allows even more information to be encoded in signals. Although light also can be combined in different frequencies, the extremely short wavelength means that animals on this planet do not directly measure light frequency as we do sound frequency; rather, colours are distinguished by arbitrary filters. Human language and bird song make use of both amplitude and frequency modulation as well as formant manipulation (Fitch 2000), so that subtle differences in the acoustic structure of a signal can be detected, perceived, and interpreted by the receiver as having distinct

6

Arik Kershenbaum

meanings. However, organisms of our physical scale are unlikely to be able to perform a similar analysis of visual signals; an observation made by J.B.S. Haldane, who pointed out that, “It is a curious fact that we men can place musical notes in their natural order by intuition, while it required the genius of Newton to do the same for colour” (Haldane 1927: 282). The disadvantages of both acoustic communication (dependence on the density of a planet’s atmosphere, difficulty in localising the sound source, etc.) and visual communication (it can be obscured by solid objects or scattered by atmospheric particles) has led many animals to rely on multimodal communication, whereby different modalities are used either to provide a more robust communication channel, or to add extra and richer information. Indeed, both Herzing (this volume [Chapter 4]), and Ortner (this volume [Chapter 5]) argue that for animals on Earth and for aliens on another planet, respectively, multimodal channels are fundamental to the development of complex communication and the development of language. Acoustic communication is very powerful, and it may be that animals that evolved on other planets and live in a dense gaseous or rarefied liquid environment like Earth’s may have evolved a language based on an acoustic channel. Nixon and Tomaschek (this volume [Chapter 17]) explore the possible effects of different planetary environments on the suitability or otherwise of acoustic and visual communication – both the opacity of the atmosphere and the fluid density can have dramatic effects on whether particular signals propagate well or not. However, what other possibilities in other physical modalities exist for conveying sufficient information of sufficient complexity to constitute a language? Earth’s animals have exploited a number of other communications modalities, and the extent to which these do – or do not – support complex communication helps us consider the possible nature of language on other planets. Here I consider three candidates for linguistic communication that appear less likely on Earth but whose characteristics may provide diverse opportunities on other planets: olfactory or chemical signals, electrical signals, and magnetism. Chemical Sensing and Olfaction

Although chemical sensing is likely the most primitive communications channel, olfaction (which, for simplicity, I loosely define as chemical sensing at a distance) seems a poor candidate for linguistic communication. In typical environments on Earth (aqueous, terrestrial), chemicals diffuse slowly. As a result, although pheromones can be used to convey urgent signals, the complexity of such signals must be very low – barely more than indicating presence or absence. As Ross (this volume [Chapter 12]) points out, the linguistic potential in chemical signalling is very limited, as it fails to provide what Ross considers a crucial design feature, “rapid fading” (Hockett 1960). Although an alarm signal with low temporal specificity is easily conveyed (“a predator is near”), it is much harder to imagine olfactory cues giving information on fast-changing situations (“he’s coming up on your left!”).

Many Ways to Say Things

7

Olfactory signals are also difficult to localise if they diffuse throughout a medium (although they can be deposited at specific points), particularly in any environment with turbulent fluid flow that tends to homogenise the signal within its transport medium (air or water). Despite this, several animal species have made widespread and rather complex use of chemical cues that never evolved to the complexity of a language on Earth but may give us clues about the kind of world where olfactory language may be possible. Stereo-olfaction, for instance, allows animals such as dogs (Craven et al. 2010), rats (Rajan et al. 2006), and sharks (Gardiner and Atema 2010) to localise chemical cues in space by measuring concentration differences between multiple detectors (in the case of Bilateria, usually two nostrils or antennae). Even in a complex and turbulent marine environment, animals such as lobsters can perceive remarkably complex “olfactory landscapes”, including deriving information on temporal and spatial variation in chemical cues (Atema 1995). As our understanding of animal olfactory perception increases, it becomes clear that the integration of multiple detectors can generate complex signals based on suites of chemical cues (Su et al. 2009). Nonetheless, the communicative potential of olfaction appears slight. What kind of alien environments might favour the evolution of more complex and possibly linguistic smells? Clearly, one requirement is a medium through which chemicals can be transported – probably by diffusion, but mass transfer currents could provide an alternative. Laminar flow of the medium would be helpful, as turbulence will tend to reduce information content and mask important cues such as location and time of signalling. Second, the slow nature of diffusion (compared to sound and light) may mean that olfactory language is realistic only in small niches, whereby communication over long distances is not required. Unfortunately, a world in which animals compose sonnets through the carefully controlled release of strong-smelling gases still seems very unlikely – and alien. Perhaps most significantly, the slow speed of diffusion – and the unidirectional nature of the passive transport of olfactory cues by mass flow – limit the potential for bidirectional chemical “conversations”. Hobaiter et al. (this volume [Chapter 3]) provide a convincing argument that cooperative, two-way signalling is essential for the evolution of complex communication that could eventually become something we would label “language”. Higher levels of communicative complexity are often linked to time-critical decisions that animals need to make, and it is difficult to see how the Gricean paradigm that Hobaiter et al. discuss would evolve from olfactory signals that are in essence slow and unidirectional. Electrical Signals

All life on Earth makes use of electrically charged ions to store and mobilise energy. When ions move, they create electric fields, so electrical signals are inherent to life on Earth. It is not clear that mobilisation of energy in the form of the movement of ions is necessarily a universal feature of life – and on other planets,

8

Arik Kershenbaum

storing and using energy could be accomplished in other force fields – but all life on Earth makes use of electrical potential to store and release energy, and animals in particular are a source of copious electrical fields, particularly from muscles and nerve cells. Given that all animals produce electrical signals, it is almost surprising that the use of passive electroreception, i.e., the detection of the fields created by other animals, is not more widespread both among predators detecting prey and prey avoiding predators. In fact, many animals do have weak electrosensing, although our understanding of the distribution of this ability in the animal kingdom is still expanding. The ability both to sense and generate electric fields as an active signal is, however, most developed among two groups of teleost fish, the Mormyriformes, a group of African electric fish, and the Gymnotiformes, electric fish from Central and South America (Hopkins 2009). Unlike olfaction, the electric sense has the complexity and fidelity needed to convey large amounts of information reliably and quickly, and could serve as a possible model for linguistic communication. Nonetheless, no electric fish possess language. What features of electrosensing make it amenable to complex communication and limit is usefulness? Although electric fields are pervasive among animal tissues, the nature of these indirectly generated fields is not particularly conducive to being co-opted for communicative uses. Detecting your prey (or your predator) by passive sensing of the electric field of their muscles requires detection of a low-frequency or direct current (DC) electromagnetic field, and the information content of such a field is necessarily low. Electric fish also have electric organs that actively generate an electrostatic field around them, and receptors on the fish’s body detect variations in that field caused by the presence of nearby animate and inanimate objects (Meyer 1982). Under the selective pressure of predatory animals searching for low-frequency electrical fields of prey animals, electric fish evolved active search using electrical fields at a higher frequency, which could not be as easily detected by predators (Stoddard 1999). In contrast to slowly varying or DC electric fields, high-frequency communication by electric fish holds the potential to encode large amounts of information and transmit it quickly and reliably. Both Mormyriformes and Gymnotiformes have diverse and complex electrical communication signals that have been shown to encode information such as species identity (Sullivan et al. 2000), individual identity (Scheffel and Kramer 2006), sex (Lorenzo et al. 2006), and social status (Hagedorn and Zelick 1989). There are, however, drawbacks to the use of an electrical channel for complex communication. Signal range in most media is limited, especially because the signal strength of a dipole falls as the cube of the distance, rather than the square as with most dissipative signals. Electrocommunication is therefore a short-range modality on Earth, with typical distances generally being less than 1 metre (Squire and Moller 1982). However, just as an oscillating electrical field such as those producing radio signals can propagate long distances through an insulating medium (e.g., air) from a dipole antenna, it is possible to theorise that certain environmental conditions could favour the evolution of high-frequency electrical communication

Many Ways to Say Things

9

that is not as range-restrictive as the aquatic solutions found on Earth. Other considerations, however, cause us to be sceptical of such a possibility. Decoding such signals would require a mechanism for frequency analysis, analogous to the basilar membrane in the ear, in which multiple detectors are tuned to different acoustic frequencies. Such an array of electrical receptors, each tuned to a different frequency, is not impossible, and electric fish do have a limited range of sensory cells tuned to different frequencies (Kawasaki 2005). However, the number and specificity of frequency detectors needed greatly exceeds those observed, and the complexity of such a frequency analyser would imply very strong selective pressure. Even the evolution of frequency-dependent acoustic communication on Earth appears to have been in response to strong selective pressures, because even the most diverse animal class, insects, have very little ability to discriminate acoustic frequency (Stebbins 1983; Michelsen 1973). Magnetic Communication

Unlike communication using electric fields, no animals on Earth – as far as we know – communicate using magnetism. This may seem surprising, given the closely related physical properties of magnetic and electric fields, and the fact that many animals – including possibly humans – can detect weak variations in magnetic field strength. However, it seems that no species has adopted this ability for communications purposes, probably because magnetic signals are difficult to generate and manipulate. Magnetic sensing in organisms on Earth is achieved by measuring the force exerted by external magnetic fields on small pieces of magnetised materials located in special detector cells in an animal’s body, or by measuring changes in the magnetic properties of photoreceptor chemicals in the retina, as they absorb photons (Gould 2008; Wiltschko and Wiltschko 2010). These subtle detector mechanisms also hint at another possible reason for not communicating via magnetic signals: the strength of any such signal is likely to be very small compared to Earth’s magnetic field, and thus difficult to detect and, more importantly, to quantify. But even if small variations in magnetic field strength are unlikely to be detectable, it is still possible that communication could occur via reversals in magnetic polarity – repeated changes in the north-south alignment of a magnetic field generating organ in a hypothetical magnetic signalling animal. Thus, we can speculate that in the presence of a weak planetary magnetic field, animals could communicate using a digital encoding of information as a series of north-south inversions (analogous to the 1–0 binary representation), perhaps by moving appendages containing permanent magnets of different orientations. However, despite the fact that we may wonder, “why didn’t evolution invent . . . ?”, such a mechanism remains unlikely. On a planet with a weak magnetic field, the prerequisite magnetosensing ability is unlikely to arise in the first place, because it may not provide a primary adaptive advantage. Similarly, digital communication is extremely rare in the natural world, and it too is likely to arise only as an adaptation

10

Arik Kershenbaum

on top of pre-existing analogue signalling, which is simpler to generate and interpret. Therefore, given that there is no plausible mechanism for analogue magnetic signalling, our hypothetical magnetic communicator seems increasingly unlikely in any planetary environment. Information Encoding

It seems clear that any effective communication requires that both parties agree on the nature of the communicative units that make up an interaction. Indeed, Ortner (this volume [Chapter 5]) argues that a two-way protocol is fundamental to the development of complex communication on any planet. Humans, with a welldeveloped language, are prone to look for similar information-encoding strategies in every species we examine. We all understand the semantic relationship between abstract symbols – words, or more strictly, morphemes – and their intended meaning. Indeed, this semantic relationship is often considered a fundamental feature of language, known as duality of patterning (Hockett 1960), and is often used as the baseline for analysing the nature of language not just as we understand it, but also how it must be (Fitch 2005; Hauser et al. 2002; Samuels and Punske, this volume [Chapter 16]). However, I believe that evidence from examining the communication of other species requires us to re-examine this most basic assumption, especially when considering the possibility of alien languages. In fact, many species do not appear to communicate using discrete encoding of information into distinct and distinguishable packages. Many species, including jumping spiders (Elias et al. 2005) and kangaroo rats (Randall and Lewis 1997), communicate using a series of seismic pulses, whereby the inter-pulse interval, as well as the frequency characteristics of the pulses themselves, contain information on individual identity and social status (Randall 1989). A unidimensional trait such as inter-pulse interval seems to be insufficient for linguistic complexity, but other species use more complex graded signals. Bottlenose dolphins (Tursiops truncatus), for example, use highly frequency modulated narrowband acoustic signals (whistles) as their main communication channel, and experimental evidence shows that the precise modulation pattern contains information such as individual identity (Sayigh et al. 2017; Sayigh et al. 1999). However, even whistles that ostensibly carry identical information show considerable variance in their modulation pattern, and the total repertoire of an animal could be anywhere between very small (if similar whistles are considered identical) to almost infinite (if minor differences between whistles are considered significant). Clearly, the perception of the receiving animal (and the cognitive implications of received signals are the true arbiters of information content) remain at least for now hidden to us. Nonetheless, it appears increasingly likely that these whistles do not fit the morpheme paradigm of traditional linguistics. Similarly, wolves (Canis lupus) communicate using a long-range signal (howling) that shares many acoustic properties with dolphin whistles. Although wolf howling is believed to contain many levels of information

Many Ways to Say Things

11

(Kershenbaum et al. 2016), no two wolf howls are the same, and it is not clear whether subtle variations in howl modulation indicate subtle variations in meaning (Déaux and Clarke 2013). Notwithstanding our lack of knowledge of the actual information content of dolphin and wolf communication, the use of non-discrete, continuously variable signals in animals on Earth raises important questions when we consider alien language. Is it possible that an entire language could be constructed on the basis of a signal that does not consist of morphemes, each attached to a particular meaning? The idea that dolphins may be the most sophisticated animal communicators after humans (Gregg 2013) leads us to consider this possibility very seriously. At first, it seems that the primary constraint acting on a non-discrete communication channel is that of the resolution between semantically distinct concepts. Two similar whistles may have different meanings, but as their acoustic patterns may be arbitrarily similar, a receiving animal may not acquire the signal-to-noise ratio needed to distinguish between different meanings. However, we must seriously consider the possibility that the cognition of the animals may operate in a non-discrete way, and rather than considering the use of continuous signals to convey discrete concepts, continuous signals may convey continuous concepts by which subtle variation in whistle modulation implies only subtle (but relevant) variation in semantic meaning. Experimental studies have shown that dolphins clearly possess the ability to make discrete semantic judgements (Herman et al. 1994; Harley 2008), and a more detailed understanding of the abilities of these and other animals to perceive nondiscrete information is essential to understanding the evolution of communication systems quite different from our own. Conclusion

Of the vast range of communicative strategies that we observe on Earth, only one very specific form of information encoding led to the evolution of language; namely discrete sematic symbols (words) transmitted through an acoustic medium. Animals communicate using many different modalities (visual, olfactory, vibratory), and many different encodings (discrete, continuous, pulsatile). While all of these other approaches are effective at communication, in that they evolved to provide an adaptive advantage, why did none of them result in language? Even the most complex communication seen in the animal world – the acoustic communication of dolphins and other cetaceans – does not appear to possess the properties of a true language. I propose that one possible explanation for this seemingly unlikely situation is the shared nature of their communication channels. Considering that passive sensory systems first evolved to gather information on the environment, and given the costly nature of the sensing and processing apparatus, sensory systems will have become more complex together with the need for more complex information. However, once a system became sufficiently complex to allow sophisticated communication, that system was already

12

Arik Kershenbaum

tied to a vital adaptive need of the animal to survive in its niche. For example, the complex biosonar system of bats requires such sophisticated processing that the animals have difficulty processing more than one stream of acoustic information at a time (Barber et al. 2003). Similarly, in electric fish, separate neural and sensory apparatus have evolved to process information originating from different sources (Pothmann et al. 2012). Thus, in bats, dolphins (which also use biosonar), and other animals, communicative content may be limited by the bandwidth of a sensory and cognitive channel that is essential for gathering information about their environments. Humans, of course, sense their environment primarily through the visual channel, and so the development of complex acoustic communication does not impede critical information about their surroundings. Naturally, other constraints on the evolution of language exist, but this constraint of a dedicated sensing channel remains a useful test for the degree of communicative complexity that we may expect in a particular modality. Given that one species that did evolve language, to what extent did the specific characteristics of the communication channel of our ancestors constrain or enable the rapid onset of such exceptional complexity? For the consideration of alien languages, the first question we must answer is whether the particular conditions that led to a tipping point that made the evolution of language inevitable were rare or unique, and whether those conditions might be expected to occur at some point on any inhabited planet. In particular, in our discussion, to what extent was the nature of the communication channel a factor in the selective pressure for language? From our review, it seems almost inevitable that a linguistic communication system on our planet would exist in the acoustic modality. Further, other modalities, perhaps with the exception of electrical fields, seem unlikely candidates for language on any planet. Nonetheless, perhaps the most important message from reviewing communication strategies on Earth is that virtually every communication modality that we can imagine evolving has in fact evolved, and as information on environmental niches on exoplanets becomes available, we can use – cautiously – analogy with adaptive strategies on Earth to speculate about what communicative solutions have been found elsewhere. References Atema, J. 1995. “Chemical signals in the marine environment: Dispersal, detection, and temporal signal analysis.” Proceedings of the National Academy of Sciences of the United States of America 92 (1): 62–66. Barber, J.R., K.A. Razak, and Z.M. Fuzessery. 2003. “Can two streams of auditory information be processed simultaneously? Evidence from the gleaning bat antrozous pallidus.” Journal of Comparative Physiology A 189 (11): 843–855. Craven, B.A., E.G. Paterson, and G.S. Settles. 2010. “The fluid dynamics of canine olfaction: Unique nasal airflow patterns as an explanation of macrosmia.” Journal of the Royal Society, Interface 7 (47): 933–943. Déaux, Éloïse C., and Jennifer A. Clarke. 2013. “Dingo (Canis Lupus Dingo) acoustic repertoire: Form and contexts.” Behaviour 150 (1): 75–101.

Many Ways to Say Things

13

Elias, Damian O., Eileen A. Hebets, Ronald R. Hoy, and Andrew C. Mason. 2005. “Seismic signals are crucial for male mating success in a visual specialist jumping spider (Araneae: Salticidae).” Animal Behaviour 69 (4): 931–938. Fitch, W. 2000. “The evolution of speech: A comparative review.” Trends in Cognitive Sciences 4 (7): 258–267. Fitch, W. Tecumseh. 2005. “The evolution of language: A comparative review.” Biology and Philosophy 20 (2–3): 193–203. Gardiner, Jayne M., and Jelle Atema. 2010. “The function of bilateral odor arrival time differences in olfactory orientation of sharks.” Current Biology 20 (13): 1187–1191. Gould, James L. 2008. “Animal navigation: The evolution of magnetic orientation.” Current Biology 18 (11): R482–R484. Gregg, Justin. 2013. Are Dolphins Really Smart?: The Mammal Behind the Myth. Oxford: Oxford University Press. Hagedorn, Mary, and Randy Zelick. 1989. “Relative dominance among males is expressed in the electric organ discharge characteristics of a weakly electric fish.” Animal Behaviour 38 (3): 520–525. Haldane, J.B.S. 1927. Possible Worlds. London: Chato and Windus. Harley, H.E. 2008. “Whistle discrimination and categorization by the atlantic bottlenose dolphin (Tursiops Truncatus): A review of the signature whistle framework and a perceptual test.” Behavioural Processes 77 (2): 243–268. Hauser, M.D., N. Chomsky, and W. Fitch. 2002. “The faculty of language: What is it, who has it, and how did it evolve?” Science 298 (5598): 1569–1579. Herman, Louis M., Adam A. Pack, and Amy M. Wood. 1994. “Bottlenose dolphins can generalize rules and develop abstract concepts.” Marine Mammal Science 10 (1): 70–80. Hockett, Charles F. 1960. “The origin of speech.” Scientific American 203 (3): 88–97. Hopkins, C.D. 2009. “Electrical perception and communication.” Encyclopedia of Neuroscience 3: 813–831. Howard, J., B. Blakeslee, and S.B. Laughlin. 1987. “The intracellular pupil mechanism and photoreceptor signal: Noise ratios in the fly Lucilia Cuprina.” Proceedings of the Royal Society of London. Series B, Biological Sciences 231 (1265): 415–435. Jackendoff, Ray. 2011. “What is the human language faculty?: Two views.” Language 87 (3): 586–624. Kawasaki, Masashi. 2005. “Physiology of tuberous electrosensory systems.” In Electroreception, edited by T. Bullock, C. Hopkins, A. Popper, and R. Fay, 154–194. Dordrecht: Springer. Kershenbaum, Arik, Holly Root-Gutteridge, Bilal Habib, Jan Koler-Matznick, Brian Mitchell, Vicente Palacios, and Sara Waller. 2016. “Disentangling canid howls across multiple species and subspecies: Structure in a complex communication channel.” Behavioural Processes 124: 149–157. Lorenzo, Daniel, Ana Silva, and Omar Macadar. 2006. “Electrocommunication in gymnotiformes: Jamming avoidance and social signals during courtship.” Communication in Fishes 2: 753–779. Meyer, J. Harlan. 1982. “Behavioral responses of weakly electric fish to complex impedances.” Journal of Comparative Physiology 145 (4): 459–470. Michelsen, Axel. 1973. “The mechanics of the locust ear an invertebrate frequency analyzer.” In Mechanisms in Hearing, edited by Aage R. Møller, 911–934. Dordrecht: Springer. Niven, J.E., and S.B. Laughlin. 2008. “Energy limitation as a selective pressure on the evolution of sensory systems.” The Journal of Experimental Biology 211 (Pt 11): 1792–1804.

14

Arik Kershenbaum

Pothmann, Leonie, Lon A. Wilkens, and Michael H. Hofmann. 2012. “Two modes of information processing in the electrosensory system of the paddlefish (Polyodon Spathula).” Journal of Comparative Physiology A 198 (1): 1–10. Rajan, Raghav, James P. Clement, and Upinder S. Bhalla. 2006. “Rats smell in stereo.” Science 311 (5761): 666–670. Randall, J.A. 1989. “Individual footdrumming signatures in banner-tailed kangaroo rats Dipodomys Spectabilis.” Animal Behaviour 38 (4): 620–630. Randall, J.A., and E.R. Lewis. 1997. “Seismic communication between the burrows of kangaroo rats, Dipodomys Spectabilis.” Journal of Comparative Physiology A 181 (5): 525–531. Sayigh, L.S., P.L. Tyack, R.S. Wells, A.R. Solow, M.D. Scott, and A.B. Irvine. 1999. “Individual recognition in wild bottlenose dolphins: A field test using playback experiments.” Animal Behaviour 57 (1): 41–50. Sayigh, Laela S., Randall S. Wells, and Vincent M. Janik. 2017. “What’s in a voice? Dolphins do not use voice cues for individual recognition.” Animal Cognition 20 (6): 1067–1079. Scheffel, Andreas, and Bernd Kramer. 2006. “Intra-and interspecific electrocommunication among sympatric mormyrids in the Upper Zambezi River.” In Communication in Fishes, edited by F. Ladich, S.A. Collins, P. Moller, and B.G. Kapoor. Enfield, New Hampshire: Science Publishers. Squire, Ann, and Peter Moller. 1982. “Effects of water conductivity on electrocommunication in the weak-electric fish brienomyrus niger (Mormyriformes).” Animal Behaviour 30 (2): 375–382. Stebbins, William C. 1983. The Acoustic Sense of Animals. Cambridge, MA: Harvard University Press. Stevens, Martin. 2013. Sensory Ecology, Behaviour, and Evolution. Oxford: Oxford University Press. Stoddard, Philip K. 1999. “Predation enhances complexity in the evolution of electric fish signals.” Nature 400 (6741): 254. Su, Chih-Ying, Karen Menuz, and John R. Carlson. 2009. “Olfactory perception: Receptors, cells, and circuits.” Cell 139 (1): 45–59. Sullivan, John P., Sébastien Lavoue, and Carl D. Hopkins. 2000. “Molecular systematics of the African electric fishes (Mormyroidea: Teleostei) and a model for the evolution of their electric organs.” Journal of Experimental Biology 203 (4): 665–683. Wiltschko, Roswitha, and Wolfgang Wiltschko. 2010. “Avian magnetic compass: Its functional properties and physical basis.” Current Zoology 56 (3).

3 RECOGNIZING INTENTIONAL SIGNALS AND THEIR MEANINGS IN NONHUMAN COMMUNICATION Catherine Hobaiter, Adriano R. Lameira, and Derek Ball

Introduction

How can we tell if someone is trying to talk to us – especially when that someone is not a member of our species (or even from our solar system)? And once we have detected a signal, how can we tell what it means? Every organism leaves evidence of its activity, and in some cases these cues can be exploited to gain rich information. But in the study of communication, signals sent with the intent to communicate are of special interest because they indicate the existence of human-like forms of intelligence. The study of pre-verbal children and non-human animal communication has provided toolkits for the diagnosis of intentional communication. At present, these typically employ a Gricean approach, which explores evidence that the signaler directs their communication to a specific recipient, in order to achieve a particular goal. We will discuss the potential to adapt the Gricean strategy to the interstellar context, where our knowledge of the environment, interests, ideas, and purposes of our possible interlocutors is extremely limited, and explore the idea that signal simplicity may play a crucial role. Grice on Communication and Meaning

Imagine driving along a country road and noticing the driver of a passing car flashing her headlights at you. You would probably assume that she was trying to tell you something. But what? Figuring this out requires a bit of thought. The other driver intends to help you – to give you information that you can use. Is there something wrong with your car? Suppose that a quick check makes this unlikely – your lights are on, and there is no sign of smoke. Most likely, you might conclude, the passing driver knows something about the road ahead; the driver wants to inform DOI: 10.4324/9781003352174-3

16

Catherine Hobaiter, Adriano R. Lameira, and Derek Ball

you that there is reason to be cautious, which means you should reduce your speed and be careful. Similar reasoning is often involved in understanding linguistic communication. To adapt a famous example (Grice 1991: 33), fully understanding a letter of recommendation for an academic position that reads only “The candidate has excellent handwriting and is always very punctual” requires us to reason along the following lines: the author of the recommendation intends to give the reader information relevant to the issue of whether the candidate is suitable for the position, and is in possession of such information; information about the candidate’s handwriting and punctuality is not relevant; therefore, there must be further information that the author is unwilling (perhaps for reasons of politeness) to share; and this can only be information to the effect that the candidate is grossly unsuitable. The philosopher Paul Grice developed a view of meaning that provides several insights into how communication works in this kind of case. For our purposes, two of Grice’s themes will be particularly important: (i) the idea that communication is a cooperative endeavor, and (ii) the idea that meaning (in one interesting sense) is a matter of acting with particular intentions. To get an idea of the motivation for these themes, let’s unpack the very beginning of our reasoning about the passing driver: “The other driver intends to help you – to give you information that you can use.” Note first that we assume that the other driver is trying to help: we have certain desires (to drive safely, avoid accidents, etc.), she knows this, and she intends to help us satisfy those desires. This assumption is crucial to the reasoning that follows: without it, even if we could somehow come to know that the passing driver is trying to communicate, we would have no way of making even a reasonable guess about what she is trying to say. Grice formulates the idea that communication is cooperative as an injunction to communicators: The Cooperative Principle: “Make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged.” (1991: 26) Grice identifies several “maxims,” observing which (he claims) “will, in general, yield results in accordance with the Cooperative Principle” (1991: 26); these include principles such as “Try to make your contribution one that is true” and “Be perspicuous.” (We will return to the status of these maxims in the context of interstellar communication.) Note also what we took the passing driver’s cooperation to consist of: she intends to give us certain information, to make us form a certain belief about the conditions of the road ahead. Grice took this to be crucial to a certain kind of meaning that he called “non-natural meaning” (to contrast with cases of “natural

Intentional Signals, Their Meanings in Non-Human Communication

17

meaning” or cues – roughly, reliable correlation – ascribed in claims like “smoke means fire”) (1991: 213–214). But not every action performed with the intention that others form a belief is meaningful in Grice’s sense. I may plant misleading evidence at a crime scene, intending that investigators will form the belief that my rival is the culprit; but Grice would deny that the misleading evidence (or my planting it) is meaningful in the relevant sense. What is crucial to communication that the non-naturally meaningful is that the communicator intends that her audience form a certain belief on the basis of recognizing that very intention. Non-natural meaning is thus very much connected to interaction, an engagement between entities. For example, when we saw the passing driver flash her lights, we recognized that she was trying to tell us something – in other words, that she intended us to take up some bit of information or form some belief – and the driver intended us to recognize this. In Grice’s view, this is what distinguishes her light flashing from other signals that might (inadvertently) alert us to the danger ahead (such as the screech of her brakes in the distance, or her concerned expression). As the passing driver example shows, communication involving non-natural meaning in the Gricean sense can take place without language. But typical uses of language are paradigmatic cases of intentional communication, and Grice maintains that the notion of non-natural meaning is crucial to understanding the meaningful use of language. Grice (1991: 89–90) distinguishes between occasion meaning – roughly, what a particular communicator meant by a use of a particular word or phrase in a particular set of circumstances – and timeless meaning – roughly, the meaning that every use of that word or phrase has in common (sometimes called the literal or semantic meaning). For example, the occasion meaning of the recommendation letter is (roughly) that the candidate is unsuitable; but that is no part of the timeless meaning of the sentence “The candidate has excellent handwriting and is always very punctual,” because many possible uses of that sentence would not mean that the candidate is unsuitable. In Grice’s view, occasion meaning is non-natural meaning; one occasion means “the candidate is unsuitable” just in case one acts with the intention that one’s audience forms the belief that the candidate is unsuitable on the basis of recognizing that very intention. And Grice argues further that timeless meaning should be accounted for in terms of occasion meaning: abstracting away from the details, his idea is that a sentence has a particular timeless meaning for a group just in case the members of that group have what he calls a “procedure in their repertoire” – roughly, a disposition – to use that expression with that occasion meaning, and it is common knowledge that this is so. Thus, for example, the sentence “The candidate has excellent handwriting” timelessly means (for English speakers) that the candidate has excellent handwriting because we are disposed in appropriate circumstances to use the sentence to occasion-mean that the candidate has excellent handwriting, and we all know this fact about each other.

18

Catherine Hobaiter, Adriano R. Lameira, and Derek Ball

Intentional Communication in Non-Human Animals: The Gricean Toolkit

Communication with other members of our own species can be very challenging. Even when we share the same language, we must infer the occasion meaning a signaler intends to communicate. Even small differences in our life experience and day-to-day events (an argument with a friend, our team winning the game) affect the layers of meaning in any phrase. Our ideas shape our language, which may in turn shape our ideas (Lupyan and Dale 2010). Communication across more substantial cultural divides presents an even greater challenge. What hope do we have of decoding speaker meaning (in the Gricean sense) in another species? One approach has been to try to teach other species to use human systems (Gardner and Gardner 1969; Pepperberg, this volume [Chapter 6]) but doing so has required not only that these individuals not only share a human environment, but that they are also raised in one from infancy. Not only is this situation improbable in the context of communication with extraterrestrial species (Kryptonians aside), it does not provide us with a means to communicate with others of their species. An enculturated chimpanzee raised as a ‘human child’ may be almost as poorly equipped to understand the ideas and purposes of a wild chimpanzee as we are. An alternative is to decode the information in a species’own communication, and while doing so try to eliminate (or at least recognize the limitations of) our humancentric perceptual biases (see also Slobodchikoff, this volume [Chapter 9]; Herzing, this volume [Chapter 4]; Pepperberg, this volume [Chapter 6] Grund et al. 2023). The first problem any signaler or recipient faces is how to distinguish a signal from the noise. How do we ensure that our signal is recognized as such? Many other authors in this volume focus on this part of the puzzle. There is increasingly widespread evidence for ‘universal’ patterning of information (see Berea, this volume [Chapter 7]); for example, Zipf’s Law of brevity and Menzerath’s law – universal to human languages – are also present in other systems of information from gibbon calls to genomic structure (Zipf 1949; Gustison et al. 2016; Ferrer-i-Cancho and Lusseau 2009; Huang et al. 2020; Heesen et al. 2019; Safryghin et al. 2022). Humans, like many species, acquire the fundamentals of communication easily, intuitively, and without a rich input – suggesting that there are biologically inherited components to our language acquisition (Grice 1991; Senghas et al. 2004; Hauser 2002; Nixon and Tomaschek, this volume [Chapter 17]; Roberts et al., this volume [Chapter 15]; Samuels and Punske, this volume [Chapter 16]) or language production (Harbour, this volume [Chapter 18]). From continuous streams of sound or movement, human and non-human animals parse out discreet signals (e.g., Barutchu et al. 2008; Kuhl 1979) in both our own and other species’ communication (e.g., Kuhl and Miller 1975; Kamiloğlu et al. 2020). Our species is able to recombine these signals into hierarchically structured, recursive sequences (syllables, words, sentences) that, from a finite set of – for example – sounds, allow us to communicate any idea we can imagine. While humans have a bias toward auditory and visual communication

Intentional Signals, Their Meanings in Non-Human Communication

19

(Ortner, this volume [Chapter 5]), across species on Earth, information from cues and signals is decoded from a far more diverse range of modalities, from touch to magnetism (see Kershenbaum, this volume [Chapter 2]). However, once we have passed the step of discriminating a signal from background noise, as generated by someone, successful communication of ideas requires we also recognize whether it was intended as a message. Many things signal information: the color of a tree’s leaves tells us something about season; the color, smell, and feel of a berry tell us whether we should eat it. This information is broadcast to the world irrespective of whether or not an audience is there or can see it: a tree does not change the color of its leaves depending on who is looking at it, and a ripe berry tastes just as good at night when its color is hidden. Most animal signals are similar; the bright pink genital swelling of a female chimpanzee is on display throughout her fertile period, irrespective of which males are around (or if there are any there at all). It is easy to make a distinction between these fixed signals and human language – but what about those that seem superficially more similar? There may be rich nuanced information in a bird’s song, a wolf’s howl, or a monkey’s alarm call (Templeton et al. 2005; Kershenbaum et al. 2016; Pepperberg, this volume [Chapter 6]; Ouattara et al. 2009). But when a vervet monkey gives an eagle alarm call, does he intend for the other members of his group to understand that there is an eagle there and that they need to flee? These calls reliably elicit this behavior from other vervet monkeys (Cheney and Seyfarth 1992), and careful experiments have shown that similar alarm calls across primate species generate a mental representation of the threat in the recipients (Zuberbühler et al. 1999). They are called functionally referential – they function as if they refer to an eagle. There is a reason for the caveat: as yet, we have no evidence to suggest that the signaler intends to refer the recipients’ attention to the eagle. Absence of evidence for intention cannot confirm that these signals are produced without intention, and they are certainly produced with greater flexibility than we find in fall leaf colors or female sexual swelling. Signalers may call more loudly or for longer if an audience is present (Schel, Machanda et al. 2013; Cheney and Seyfarth 1992; Wich and de Vries 2006), or if specific types of individuals are present (Schel, Machanda et al. 2013; Crockford et al. 2012; Slocombe and Zuberbühler 2007). However, these effects can typically be explained without invoking the communicative intentions that appear to be fundamental to language use – for example, variation in whether and how a call is produced may be modulated by similar variation in physiological arousal, generated by the presence of (relevant) social peers. Humans show the same behavioral flexibility in our broadcast signals. In a context whereby you would smile when alone, you smile more – or laugh longer – if others are present (Gervais and Wilson 2005; Devereux and Ginsburg 2001; Wild et al. 2003; Provine 1992). In our species, language did not replace this system of broadcast signals – we continue to broadcast information, whether an involuntary blush, laugh, or yelp (see also: Ortner, this volume [Chapter 5], for how non-linguistic signals can be co-opted when language’s original modality becomes

20

Catherine Hobaiter, Adriano R. Lameira, and Derek Ball

unavailable). Instead, language provides us with a different system of signaling, one in which we can choose whether and how to share information depending on our audience, and on what we know about them, including what they already know. Grice’s view of meaning requires that, in meaningful communication, the signal be produced with the intent that the audience form a belief on the basis of recognizing that intention. But even among humans, not all interesting communicative behaviors need be accompanied by intentions with such complex structure. Following Dennett (1983), we may distinguish different levels of complexity in the intentions associated with a signal. In the simplest case, a signal may be produced with no intentions at all. Signals like this could be produced even by what Dennett calls a zero-order intentional system: a creature with no understanding of other minds, no attitudes at all about what their audience believes or does. Fully Gricean communication would require what he describes as a second-order (or higher) intentional system, which is capable of thinking about the mental states of other creatures: the signaler intends that the recipient (at least) recognize their intention. But it is possible that signals be produced with some intention, but not with a full-blown Gricean intention. For example, a signaler may intend that her audience change her behavior without having any intentions (or any attitudes at all) about the audience’s mental state – about what the audience recognizes or believes. To date, the exploration of intentional communication in non-human species has largely focused on whether they meet the criteria for this broader, first-order use of intentional communication; and we will go on to speak of the meanings of intentional communication of this kind, while recognizing that these may fall short of Gricean non-natural meaning. On first observing her, we have no way of interrogating a primate signaler (or an extraterrestrial life form, or a pre-verbal human infant for that matter) to find out what she means when she signals; but we can employ behavioral cues that together indicate intentional use. Typically, these include: checking whether or not the audience can see visual signal and, if not, adjusting signal selection to other modalities or changing signaler position to bring the signal into view; waiting’; waiting for a response after an initial signal; and, if this signal fails, persisting or perhaps elaborating with further signals. Consider a human example: if I want you to pass me the coffee, I should not point to it if you are facing in the wrong direction; I would either use an audible or tactile signal (speaking or tapping you on the shoulder), or move into a position where you can see me (Tanner and Byrne 1996; Pika et al. 2005; Tomasello et al. 1985). Similarly, and despite the fact that my desire for coffee is, frequently, a continuous stimulus, I would be unlikely to repeat an endless string of “give-me-coffee-give-me-coffee-give-me-coffee.” I would ask once and wait to see what you did next. If you did not pass it to me, I might then ask again and then, if I have failed a couple of times, elaborate by adjusting the signals I use (for example: adding a point to the coffee pot). I would also stop communicating once you pass me the coffee (my intended goal), despite the fact that my desire for coffee would not be satisfied until after I start drinking it. The same behavior accompanies ape gestural communication. They adjust their selection of

Intentional Signals, Their Meanings in Non-Human Communication

21

signal types to the audiences’ visual attention and wait for a response after signaling (Tanner and Byrne 1996; Pika et al. 2003; Tomasello et al. 1985; Genty et al. 2009; Hobaiter and Byrne 2011a; McCarthy et al. 2013); and when signals fail, they persist (Hobaiter and Byrne 2011b; Liebal et al. 2004; McCarthy et al. 2013; Leavens et al. 2005). They also elaborate (Hobaiter and Byrne 2011b; Leavens et al. 2005), distinguishing between likely reasons for failure – whether the recipient partially or completely misunderstood the initial signal (Cartmill and Byrne 2007). While there is no single panacea with which we can diagnose a signaler’s intent, taken together, these steps provide a Gricean toolkit for the recognition of intentional communication. The intention to communicate to a particular audience is what takes us beyond broadcasting information and to a point where we can ask: what does the signaler mean to say? Early work exploring intentional communication looked at how language develops in young children. Bates and colleagues distinguished illocutory acts, in which an infant employed a conventionalized signal toward a recognizable goal, from perlocutory acts, in which a signal changed a recipient’s behavior, but without any evidence that this effect was intended by the signaler (Bates et al. 1975). At around the same time, the first field studies of wild chimpanzees were providing descriptions of their vocal and gestural behavior. In early investigations, vocalizations appeared relatively fixed in terms of their content, and strongly linked to signaler affect (Goodall 1986). In contrast, gestures were combined and used across behavioral contexts in a way that suggested “openness . . . one of the most characteristic design features of human language” (Plooij 1978: 127). Today, there is a more nuanced understanding of the different types of signals produced by non-human apes (hereafter apes), with increasing flexibility demonstrated in their

BOX 3.1 DEFINITION OF TERMS USED IN COMMUNICATION Cue

Passive trait that provides information to a recipient, but on which there was no selection for it to serve this function. For example: the color of fall leaves acts as a cue that provides information on the season.

Signal

Behavior or characteristic that has evolved to convey information to a potential recipient. For example: the color of a berry, or the pink of a female primate’s genital swelling, have both evolved to provide information on ripeness, or female ovulation.

Intentional signal

Signals produced by a signaler toward a specific recipient with the intent to change their behavior or state of knowledge. For example: the signaler intends for that the recipient to understand she should go away.

22

Catherine Hobaiter, Adriano R. Lameira, and Derek Ball

vocal repertoire (Slocombe and Zuberbühler 2007; Schel, Townsend, et al. 2013; Crockford et al. 2012; Wich et al. 2012). Nevertheless, despite extensive research effort across species, it remained challenging to demonstrate intentional signal use in non-human communication (Rendall et al. 2009; Seyfarth and Cheney 2003). Supporting evidence found was typically limited to single signals used in evolutionarily important contexts; for example, a chimpanzee snake-alarm call (Schel, Townsend, et al. 2013; Crockford et al. 2012). One consistent exception is the large repertoires of great ape gestures – in which we find abundant evidence for intentional use across eighty or more gesture types used in everyday communication (Tomasello et al. 1985; Pika et al. 2003, 2005; Cartmill and Byrne 2007, 2010; Pollick and de Waal 2007; Genty et al. 2009; Hobaiter and Byrne 2011a; Graham et al. 2017; Bard et al. 2014; Bard 1992; Leavens et al. 1996). From this body of research, we have been able to explore not only the context in which apes gesture, but the intended meanings for which they use individual gestures in their repertoires (Hobaiter and Byrne 2014; Graham et al. 2018 Hobaiter et al. 2022). Today, the distinction between gestural and vocal signals is blurring; for example, orangutan signalers use their hands to modify the sound of their calls (Hardus et al. 2009), and all apes regularly combine different signal types in a single message (Wilke et al. 2017; Hobaiter et al. 2017; Genty et al. 2014). Given that the capacity for intentional communication has been established in their gestures and at least one vocalization, it is likely that future research will extend the range of signals and channels (and species) in which we are able to investigate ape signalers’ intended meaning. Fundamentally, language-like communication can occur independently of modality. While vocal communication is concentrated in the acoustic channel (although, see McGurk and MacDonald 1976), gestural communication employs visual, acoustic, and tactile information, and many other channels (for example olfactory, electrical, and magnetic: see Kershenbaum, this volume [Chapter 2]; Ortner, this volume [Chapter 5]) can be employed. A recent study found intriguing evidence of an intentional hunting gesture in grouper fish (Vail et al. 2013). This finding could be interpreted in two ways. On the one hand, it might be that intentional communication is much more widespread in non-human communication than is currently recognized – a possibility that increases the likelihood that intentionality is also present in living systems beyond Earth. If this is true, then with time and tinkering, the current Gricean toolkit may enable the detection of these species through intergalactic signals. On the other hand, this evidence could also indicate that while representing a particularly demanding test of an animal’s cognition, the Gricean toolkit remains hackable in specific communicative circumstances by species that meet our current criteria – but nevertheless lack the human-like cognition our own attempts at interstellar communication are aiming to reach. Implications for Interstellar Communication

Importantly, the detection of intentionality is independent of signal complexity. While some complex features may be at the core of human language, such as

Intentional Signals, Their Meanings in Non-Human Communication

23

hierarchical structural architecture (Hauser 2002; Senghas et al. 2004; Roberts et al., this volume [Chapter 15]), these tell us little about whether the signaler is capable of intentional language-like communication. Take bird song, for example, or the intricate sand patterns generated by courting pufferfish; these complex, structured signals provide rich information for a passing recipient about mate fitness, motivation, and vigor, but there is no evidence, to date, that these are produced with an intent to communicate to a specific recipient. Similarly, a computer algorithm may generate complex signal patterns in, again to date, the pure absence of intention.

BOX 3.2 THE PROBLEM WITH GRICE Gricean communication appears to require the ability to make complex inferences about others’ mental states for even simple requests (e.g., “I want you to recognize that I intend that you pass me the coffee”; Scott-Phillips 2015). If true, we are faced with the conundrum that a 2- or 3-year old child, capable of using language, may not be – explicitly – able to attribute these mental states (Wimmer and Perner 1983; Liddle and Nettle 2006). How do children learn to use language before they have the cognitive capacity to understand it? One possibility is that the standard interpretation of Gricean intent has overstated the cognitive skills needed to acquire intentional language (Gómez 1994; Moore 2016, 2017; Townsend et al. 2017). Alternatively, infants and great apes may employ non-Gricean forms of intentional communication for which we do not have an adequate toolkit (Leavens et al. 2017; Bar-On 2013). Alternative defnitions for intentional signal use tend to suffer from the same issue as the original toolkits for detecting intentional use developed from the Gricean defnition: a bias toward recording visual attention. In doing so, they limit our ability to explore intentional use across a wide range of signal types. If the information in the signal is primarily audible, as is the case in ape vocalizations and long-distance gestures (such as buttress-drumming, for example), how do we measure this? The signaler has no need to check the recipient’s auditory state of attention; as long as they are aware of the recipient’s location, they are aware of whether the signal can be received. As a result, we likely underestimate the frequency with which non-humans employ intentional communication. The absence of any evidence to date in areas such as cetacean or elephant communication, or in a wider range of vertebrate communication, may have more to do with the restrictive nature of the Gricean toolkit than with the true absence of intentional communication in these species.

24

Catherine Hobaiter, Adriano R. Lameira, and Derek Ball

Would a Gricean framework allow us to decode intentional communication from another world? As Grice illustrates with his discrimination of occasion meaning from timeless meaning, decoding intent can depend to a very large extent on context. Our (still limited) knowledge of ape behavior, body plans and ecology provides us with a foundation from which to explore signaler intent and meaning on Earth. However, across interstellar space, two interlocutors would share virtually no common ground. Nevertheless, there is hope – the recognition of intentional communication depends on the behavioral interaction of the interlocutors rather than decoding the signal itself. We should therefore distinguish two steps in the process of understanding intentional communication – detecting the presence of an intentional signal, and coming to understand the meaning of that signal – and the Gricean account can give us some purchase on each step. Step 1: Detecting Intention

When we consider the question of detecting intention in interstellar signals, there are two importantly different cases to consider. First, it may be that extraterrestrials are sending a signal to us, or just broadcasting a signal in an attempt to make their presence known. We will call this case extraterrestrial–human communication. Alternatively, we may inadvertently pick up on a signal that is intended for other extraterrestrials. We will call this case extraterrestrial–extraterrestrial communication. Most empirical work on non-human animals in the Gricean paradigm is focused on communication between animals of the same species: for example, chimpanzees gesturing to other chimpanzees. It might therefore be thought that the most promising possible application of Gricean tools is to the case of extraterrestrial–extraterrestrial communication. However, the extent to which the Gricean toolkit can be applied depends on exactly what we have detected. We are able to detect intention in ape communication because we can see the two-way interaction between communicator and audience. Information from both sides of the interaction is required in order to pick up on the characteristic features of intentional communication: making sure the audience can detect the signal, waiting for a response, and persisting or elaborating. With a single signal, the opportunities for investigating its intentional nature would be limited; it is the interaction between two or more parties that allows us to employ our Gricean toolkit to investigate whether the signaler intended to communicate. Ideally, what is required to build up a richer evidence base of the kind that might help us decide whether a signal is intentional is an exchange of signals. If we could somehow pick up on an entire exchange of signals between extraterrestrials, we would have a greater chance of gathering evidence (such as repetition and elaboration) that would bear on whether the signals were intentional. But the case of extraterrestrial–human communication would provide even better opportunities, since in that case we could generate our own exchange. Signal repetition and elaboration are strategies that can be easily explored in the exchange of signals in which you

Intentional Signals, Their Meanings in Non-Human Communication

25

are not yet able to decode meaning, but can nevertheless recognize intent (Moore 2014). In investigating the intentional nature of the communication rather than attempting to understand what it means, signal simplicity may be key. Through the exchange of simple signals, we are able to explore the structure of the communication without the risk of error incorporated by signal variation or flexibility. A common solution found in nature to establish communicative contact is to bounce back an exact copy of the original signal several times (before starting to adjust it). Here, signal simplicity prevents information leakage before the nature of the social relationship is established. The likelihood of even a simple signal being repeated exactly back and forth by chance is low. If the form of one signal is further randomly altered with each iteration being sent individually before an exact response is received, then an agent’s intentional action would be a (or perhaps the only) plausible explanation of any such repetition. Here it is not the signal content that is language-like; it is the way in which the signal is used. The repeated exchange of a single simple signal allows you to indicate cooperation as a partner, positive engagement in the communication, and your intent to respond. (Thus, the Gricean approach justifies conclusions reached on other grounds by Granger et al., this volume [Chapter 8], and Ortner, this volume [Chapter 5], among others.) With this step, we may be able to establish our intent to communicate; however, we cannot establish meaning by repeating back to someone only what they have already said. Step 2: Decoding Meaning

In the study of non-human ape communication, researchers initially identify the meaning of specific signals by exploring patterns in repeated signal use across individuals and contexts. As in the study of many animal signals, including primate vocalizations, initial exploration of ape gestures used behavioral context as a proxy for meaning: feeding gestures or traveling gestures (Goodall 1986; Crockford and Boesch 2003; Ouattara et al. 2009; Cheney and Seyfarth 1992; Genty et al. 2009; Hobaiter and Byrne 2011a; Pika et al. 2005; Pollick and de Waal 2007; Tomasello et al. 1985). Using established behavioral ethograms, these meanings can then be reliably established by the direct observation of behavior that we easily recognize: you can see that your signaler is feeding, or resting. However, not only is this approach not possible in interstellar communication, it may also not be well suited to exploring meaning on Earth. Take the example when a chimpanzee intends to communicate the meaning: !Stop. We observe her signaling this with a gesture produced when she is feeding (and someone tries to take her food), when she is resting (and a boisterous juvenile tries to play with her), and when she is traveling (and her young infant tries to run off). By recording the context of use, her gesture’s meaning appears to be ambiguous or flexible – while in fact it is both clear and consistent. Instead, to establish her meaning in the Gricean sense of what she intends, we must employ both signaler behavior and recipient response, specifically: the behavioral change in the recipient that stops the signaler from signaling (Cartmill and Byrne

26

Catherine Hobaiter, Adriano R. Lameira, and Derek Ball

2010; Hobaiter and Byrne 2014; Graham et al. 2018). A recipient’s response may be a refusal, or indicate misunderstanding, in which case we would expect a signaler to persist toward her goal. Take our earlier example: I ask you to pass me the coffee; if you refuse, or pass me the tea, I will ask you again. I may even – after several attempts – elaborate. Occasionally, I may give up. But the one thing that will consistently and across different attempts and partners stop me from asking for the coffee, is when I am passed the coffee. In this way, we have been able to establish that great apes employ their gestures toward at least nineteen meanings (Hobaiter and Byrne 2014; Graham et al. 2018; Byrne et al. 2017). To date, this means of exploring signaler meaning is itself limited to particular types of meaning – imperative demands that require a consistent behavioral response from the recipient, but other meanings may be expressed (Hobaiter et al. 2022). In the case of extraterrestrial–extraterrestrial communication, our very limited knowledge makes it hard to see how we could apply Gricean resources to attempt to interpret or decode a signal. Even in the case of extraterrestrial–human communication, in which a two-way exchange of signals has been established, very significant challenges remain. First, the standard applications of the Gricean toolkit do not straightforwardly extend to declarative statements that do not require a consistent behavioral response. (Because a declarative message such as “what a lovely planet you have” does not call for any particular response, it is unlikely to be followed by repetition or elaboration, or indeed any observable change in behavior at all) Second, decoding meaning in this way requires hundreds – if not thousands – of exemplars (Hobaiter and Byrne 2014). Obtaining such a rich base of signals to work with will be difficult in the interstellar case, wherein signal return time would likely be measured in years – if not decades, centuries, or even longer. These considerations suggest that, while we may be able to establish intentional communication, taking the step to understanding occasion meaning may be more challenging. Without both signaler and recipient having access to information on the physical and social context in which a communication system evolved, we appear to be stuck. Occasion meaning only emerges in context. We can draw, for example, a comparison to the decoding of ancient languages. In order to decode such a language, we first needed to establish the repertoire of possible signals and symbols used. In the case of human language, we could assume some basic shared syntactic presuppositions, which emerge across human languages even without explicit instruction (e.g., Senghas et al. 2004; Roberts et al., this volume [Chapter 15]). However, the transition to decoding the meaning of a specific script was often based on some understanding of communicative context and behavior. We were aware of not only the tools available to communicate with, but also of what was likely to be communicated in that context. A Rosetta stone for other animal species (see Slobodchikoff, this volume [Chapter 9]) can be imagined in part because we recognize how particular behavioral responses or ideas – fear, arousal, ‘predator,’ etc. – might be expressed in species in which we share some context in common; for example, physiology or socio-ecological environment (Graham et al. 2022). Cracking scripts

Intentional Signals, Their Meanings in Non-Human Communication

27

of extinct languages or unknown or secret codes typically starts with breaking a single item within the code in relation its possible meaning given away by context. Once the first item is decoded, it opens the door to further decoding of other items that appear in combination with the first. Simplicity in both signal and meaning of the initial information received may be key to taking these first steps. Even if one accepts the contentions of Samuels and Punske, this volume [Chapter 16], and Roberts et al., this volume [Chapter 15], that many features of human language may be shared with extraterrestrials, our attempts at parsing alien messages are more likely to meet with success if we can begin with simple messages and work our way up. Even with the advantage of knowing the context of a signal, there remain challenges. Systems of communication are adapted to the needs of the species employing them (Bradbury and Vehrencamp 1998). For example, a change in the environment (such as increased traffic noise) leads to a change in the signaling (e.g., in bird song; Slabbekoorn and Peet 2003; Nemeth and Brumm 2010; Gross et al. 2010). Communication is also bound to the socio-ecological niche of a particular cultural group, or species, whether humans, primates, mammals, or Earthlings. Thus, interpreting meaning requires that both parties in the exchange can make a series of presuppositions about the context in which the signal is given. Without these, the signal is at best ambiguous, at worst meaningless. We can see the importance of this in cross-cultural exchanges within our species, when even the basic positive or negative valence of an item varies: is a thumbs-up gesture supportive or obscene? Grice tells us that communication is cooperative, but we can cooperate only when we have a sense of what our audience wants; and we can exploit the fact that someone is cooperating in order to interpret her, only if we know what she thinks we want. If we humans typically obey Grice’s maxims to speak truthfully and be perspicuous, that is because doing so typically serves what we want and what we take others to want. The presupposition that other species share these desires may be problematic; extraterrestrials may greatly prefer praise to truth, or politeness (as they see it) to perspicuity. (Thus, the Gricean account casts some doubt on the distinction between understanding the message and understanding from a cultural context suggested by Ross, this volume [Chapter 12]; understanding the message may require much information about cultural context.) In short, understanding what someone means does not require sharing their same communicative context; but we do need sufficient information about our interlocutors to either select or interpret even simple signals appropriately. Gaining such information is extremely problematic when our interlocutors are light years away (see Slobodchikoff, this volume [Chapter 9], for discussion of related problems). But perhaps the task is not hopeless. We should not underestimate the common ground with which we begin: an exchange of signals in which there is evidence of the intent to communicate already establishes some presuppositions about the recipient: for example, that they are capable of communication in which we can potentially recognize their intentions. And perhaps we can gain

28

Catherine Hobaiter, Adriano R. Lameira, and Derek Ball

leverage on the problem by exploring the full breadth of human and non-human animal signal systems on Earth. In doing so, we may be able to establish patterns of consistency (natural probabilistic universals) across species that provide a starting place for exploring the context of communication with an interstellar species; for example, an increase in relative speed or pitch of signaling typically indicates an increase in arousal, with the reverse being also true. A mirroring or synchronizing of behavior typically signals the intention to strengthen and extend a positive social interaction in humans and non-humans. In contrast, an increase in the distance between signalers indicates neutral or negative affect, while elaboration may signal both positive interest and social challenge. If we have strong reasons to expect that an alien language is culturally inherited and transmitted socially across individuals, then some of the same features that are known to universally enable the learning of languages could also be expected to have emerged across different planets. Such a systematic comparison across species signaling may also provide us with possible methods to expand or evolve our Gricean toolkit. Similarly, new techniques from the study of related areas of cognition, such as theory of mind, may allow us to recognize a mind capable of intentional communication as indicated in other areas of behavior outside of signal exchange. Conclusion

Grice’s idea that communication is a cooperative activity involving intentional action has been productively utilized in research on animal communication, and the resulting Gricean toolkit has promise for recognizing attempted interstellar communication, as well. But coming to understand what someone is telling us requires understanding a great deal, not only about their communicative faculties, but also about their aims and desires, and potentially also how they think about our aims and desires; and it hardly needs to be said that understanding the aims and desires of extraterrestrials prior to establishing an effective means of communication is an extremely difficult (though, as we have suggested, perhaps not impossible) task. The Gricean toolkit does not provide a simple recipe for understanding xenolanguage – but perhaps it can point us towards useful directions for further inquiry. References Bard, Kim A. 1992. “Intentional behavior and intentional communication in young freeranging orangutans.” Child Development 63 (5): 1186–1197. Bard, Kim A., Sophie Dunbar, Vanessa Maguire-Herring, Yvette Veira, Kathryn G. Hayes, and Kelly McDonald. 2014. “Gestures and social-emotional communicative development in chimpanzee infants: Gestural development in chimpanzees.” American Journal of Primatology 76 (1): 14–29. Bar‐On, Dorit. 2013. “Origins of meaning: Must we ‘go gricean’?” Mind & Language 28 (3): 342–375.

Intentional Signals, Their Meanings in Non-Human Communication

29

Barutchu, Ayla, Sheila G. Crewther, Patricia Kiely, Melanie J. Murphy, and David P. Crewther. 2008. “When /b/ill with /g/ill becomes /d/ill: Evidence for a lexical effect in audiovisual speech perception.” European Journal of Cognitive Psychology 20 (1): 1–11. Bates, Elizabeth, Luigia Camaioni, and Virginia Volterra. 1975. “The acquisition of performatives prior to speech.” Merrill-Palmer Quarterly of Behavior and Development 21 (3). Bradbury, Jack W., and Sandra Lee Vehrencamp. 1998. Principles of Animal Communication. Sunderland, MA: Sinauer Associates. Byrne, Richard W., Erica Cartmill, Emilie Genty, Kirsty E. Graham, Cathrine Hobaiter, and Joanne Tanner. 2017. “Great ape gestures: Intentional communication with a rich set of innate signals.” Animal Cognition 20 (4): 755–769. Cartmill, Erica A., and Richard W. Byrne. 2007. “Orangutans modify their gestural signaling according to their audience’s comprehension.” Current Biology 17 (15): 1345–1348. Cartmill, Erica A., and Richard W. Byrne. 2010. “Semantics of primate gestures: Intentional meanings of orangutan gestures.” Animal Cognition 13 (6): 793–804. Cheney, Dorothy L, and Robert M Seyfarth. 1992. How Monkeys See the World: Inside the Mind of Another Species. Chicago, IL: University of Chicago Press. Crockford, Catherine, and Christophe Boesch. 2003. “Context-specific calls in wild chimpanzees, pan troglodytes verus: Analysis of barks.” Animal Behaviour 66 (1): 115–125. Crockford, Catherine, Roman M. Wittig, Roger Mundry, and Klaus Zuberbühler. 2012. “Wild chimpanzees inform ignorant group members of danger.” Current Biology 22 (2): 142–146. Dennett, Daniel C. 1983. “Intentional systems in the cognitive ethology: The ‘Panglossian paradigm’ defended.” The Behavioral and Brain Sciences 6: 343–390. Devereux, Paul G., and Gerald P. Ginsburg. 2001. “Sociality effects on the production of laughter.” The Journal of General Psychology 128 (2): 227–240. Ferrer-i-Cancho, Ramon, and David Lusseau. 2009. “Efficient coding in dolphin surface behavioral patterns.” Complexity 14: 23–25. Gardner, R. Allen, and Beatrix T. Gardner. 1969. “Teaching sign language to a chimpanzee.” Science 165: 664–672. Genty, Emilie, Thomas Breuer, Catherine Hobaiter, and Richard W. Byrne. 2009. “Gestural communication of the Gorilla (Gorilla Gorilla): Repertoire, intentionality and possible origins.” Animal Cognition 12 (3): 527–546. Genty, Emilie, Zanna Clay, Catherine Hobaiter, and Klaus Zuberbühler. 2014. “Multi-modal use of a socially directed call in bonobos.” PLoS One 9 (1): e84738. Gervais, Matthew, and David Sloan Wilson. 2005. “The evolution and functions of laughter and humor: A synthetic approach.” The Quarterly Review of Biology 80 (4): 395–430. Gómez, Juan Carlos. 1994. “Mutual awareness in primate communication: A gricean approach.” In Self-Awareness in Animals and Humans: Developmental Perspectives, 61–80. New York, NY, US: Cambridge University Press. Goodall, Jane. 1986. The Chimpanzees of Gombe: Patterns of Behavior. Cambridge, IL: Cambridge University Press. Graham, Kirsty E., Takeshi Furuichi, and Richard W. Byrne. 2017. “The gestural repertoire of the wild bonobo (Pan Paniscus): A mutually understood communication system.” Animal Cognition 20 (2): 171–177. Graham, Kirsty E., Catherine Hobaiter, James Ounsley, Takeshi Furuichi, and Richard W. Byrne. 2018. “Bonobo and chimpanzee gestures overlap extensively in meaning.” PLoS Biology 16 (2): e2004825.

30

Catherine Hobaiter, Adriano R. Lameira, and Derek Ball

Graham, Kirsty E., Gal Badihi, Alexandra Safryghin, Charlotte Grund, and Catherine Hobaiter. 2022. “A socio-ecological perspective on the gestural communication of great ape species, individuals, and social units.” Ethology, Ecology, & Evoltuion 34 (3): 235–259. Grice, H. Paul. 1991. Studies in the Way of Words. Cambridge, MA: Harvard University Press. Gross, Karin, Gilberto Pasinelli, and Hansjoerg P. Kunc. 2010. “Behavioral plasticity allows short‐term adjustment to a novel environment.” The American Naturalist 176 (4): 456–464. Grund, Charlotte, Gal Badihi, Kirsty E. Graham, Alexandra Safryghin, and Catherine Hobaiter. 2023. “GesturalOrigins: A bottom-up framework for establishing systematic gesture data across ape species.” Behavior Research Methods doi:10.3758/ s13428-023-02082-9. Gustison, Morgan L., Stuart Semple, Ramon Ferrer-i-Cancho, and Thore J. Bergman. 2016. Gelada vocal sequences follow Menzerath’s linguistic law.” Proceedings of the National Academy of Sciences 113: e2750–2758. Hardus, Madeleine E., Adriano R. Lameira, Carel P. Van Schaik, and Serge A. Wich. 2009. “Tool use in wild orang-utans modifies sound production: A functionally deceptive innovation?” Proceedings of the Royal Society B: Biological Sciences 276 (1673): 3689–3694. Hauser, Marc D. 2002. “The faculty of language: What is it, who has it, and how did it evolve?” Science 298 (5598): 1569–1579. Heesen, Raphaela, Catherine Hobaiter, Ramon Ferrer-i-Cancho, and Stuart Semple. 2019. “Linguistic laws in chimpanzee gestural communication.” Proceedings of the Royal Society B 286: e20182900. Hobaiter, Catherine, and Richard W. Byrne. 2011a. “The gestural repertoire of the wild chimpanzee.” Animal Cognition 14 (5): 745–767. Hobaiter, Catherine, and Richard W. Byrne. 2011b. “Serial gesturing by wild chimpanzees: Its nature and function for communication.” Animal Cognition 14 (6): 827–838. Hobaiter, Catherine, and Richard W. Byrne. 2014. “The meanings of chimpanzee gestures.” Current Biology 24 (14): 1596–1600. Hobaiter, Catherine, Richard W. Byrne, and Klaus Zuberbühler. 2017. “Wild chimpanzees’ use of single and combined vocal and gestural signals.” Behavioral Ecology and Sociobiology 71 (6): 96. Hobaiter, Catherine, Kirsty E. Graham, and Richard W. Byrne. 2022. “Are ape gestures like words? Outstanding issues in detecting similarities and differences between human language and ape gesture.” Philosophical Transactions of the Royal Society B 377: 20210301. Huang, Mingpan, Haigang Ma, Changyong Ma, Paul A. Garber, and Pengfei Fan. 2020. “Male gibbon loud morning calls conform to Zipf’s law of brevity and Menzerath’s law: Insights into the origin of human language.” Animal Behaviour 160: 145–155. Kamiloğlu, Roza G., Katie E. Slocombe, Daniel, B.M. Haun, and Disa A. Sauter. 2020. “Human listeners’ perception of behavioural context and core affect dimensions in chimpanzee vocalizations.” Proceedings of the Royal Society: Biological Sciences. doi:10.1098/ rspb.2020.1148. Kershenbaum, Arik, Holly Root-Gutteridge, Bilal Habib, Janice Koler-Matznick, Brian Mitchell, Vicente Palacios, and Sara Waller. 2016. “Disentangling canid howls across multiple speices and subspecies: Structure in a complex communication channel.” Behavioural Processes 124: 149–157. Kuhl, Patricia K. 1979. “The perception of speech in early infancy.” Speech and Language 1: 1–47. Kuhl, Patricia K., and James D. Miller. 1975. “Speech perception by the chinchilla: Voicedvoiceless distinction in alveolar plosive consonants.” Science 190 (4209): 69–72.

Intentional Signals, Their Meanings in Non-Human Communication

31

Leavens, David A., Kim A. Bard, and William D. Hopkins. 2017. “The mismeasure of ape social cognition.” Animal Cognition. doi:10.1007/s10071-017-1119-1. Leavens, David A., William D. Hopkins, and Kim A. Bard. 1996. “Indexical and referential pointing in chimpanzees (Pan troglodytes).” Journal of Comparative Psychology 110 (4): 346–353. Leavens, David A., Jamie L. Russell, and William D. Hopkins. 2005. “Intentionality as measured in the persistence and elaboration of communication by chimpanzees (Pan troglodytes).” Child Development 76 (1): 291–306. Liddle, Bethany, and Daniel Nettle. 2006. “Higher-order theory of mind and social competence in school-age children.” Journal of Cultural and Evolutionary Psychology 4 (3–4): 231–244. Liebal, Katja, Josep Call, and Michael Tomasello. 2004. “Use of gesture sequences in chimpanzees.” American Journal of Primatology 64 (4): 377–396. Lupyan, Gary, and Rick Dale. 2010. “Language structure is partly determined by social structure.” pLoS One 5 (1): e38559. McCarthy, Maureen S., Mary Lee Abshire Jensvold, and Deborah H. Fouts. 2013. “Use of gesture sequences in captive chimpanzee (pan troglodytes) play.” Animal Cognition 16 (3): 471–481. McGurk, Harry, and John MacDonald. 1976. “Hearing lips and seeing voices.” Science 5588: 746–748. Moore, Richard. 2014. “Ape gestures: Interpreting chimpanzee and bonobo minds.” Current Biology 24 (14): R645–R647. Moore, Richard. 2016. “Meaning and ostension in great ape gestural communication.” Animal Cognition 19 (1): 223–231. Moore, Richard. 2017. “Social cognition, stag hunts, and the evolution of language.” Biology & Philosophy 32 (6): 797–818. Nemeth, Erwin, and Henrik Brumm. 2010. “Birds and anthropogenic noise: Are urban songs adaptive?” The American Naturalist 176 (4): 465–475. Ouattara, Karim, Alban Lemasson, and Klaus Zuberbühler. 2009. “Campbell’s monkeys use affixation to alter call meaning.” pLoS One 4 (11): e7808. Pika, Simone, Katja Liebal, and Michael Tomasello. 2003. “Gestural communication in young gorillas (Gorilla gorilla): Gestural repertoire, learning, and use.” American Journal of Primatology 60 (3): 95–111. Pika, Simone, Katja Liebal, and Michael Tomasello. 2005. “Gestural communication in subadult bonobos (Pan paniscus): Repertoire and use.” American Journal of Primatology 65 (1): 39–61. Plooij, Frans X. 1978. “Some basic traits of language in wild chimpanzees?” In Action, Gestures, and Symbol: The Emergence of Language, edited by Andrew Lock, 111–132. Pollick, Amy S., and Frans B.M. de Waal. 2007. “Ape gestures and language evolution.” Proceedings of the National Academy of Sciences 104 (19): 8184–8189. Provine, Robert R. 1992. “Contagious laughter: Laughter is a sufficient stimulus for laughs and smiles.” Bulletin of the Psychonomic Society 30 (1): 1–4. Rendall, Drew, Michael J. Owren, and Michael J. Ryan. 2009. “What do animal signals mean?” Animal Behaviour 78 (2): 233–240. Safryghin, Alexandra, Catharine Cross, Brittany Fallon, Raphaela Heesen, Ramon Ferrer-i-Cancho, and Catherine Hobaiter. 2022. “Variable expression of linguistic laws in ape gesture: A case study from chimpanzee sexual solicitation.” Royal Society Open Science 9: Online. Schel, Anne Marijke, Zarin Machanda, Simon W. Townsend, Klaus Zuberbühler, and Katie E. Slocombe. 2013. “Chimpanzee food calls are directed at specific individuals.” Animal Behaviour 86 (5): 955–965.

32

Catherine Hobaiter, Adriano R. Lameira, and Derek Ball

Schel, Anne Marijke, Simon W. Townsend, Zarin Machanda, Klaus Zuberbühler, and Katie E. Slocombe. 2013. “Chimpanzee alarm call production meets key criteria for intentionality.” PLoS One 8 (10): e76674. Scott-Phillips, Thomas C. 2015. “Meaning in animal and human communication.” Animal Cognition 18 (3): 801–805. Senghas, Ann, Sotaro Kita, and Asli Özyürek. 2004. “Children creating core properties of language: Evidence from an emerging sign language in nicaragua.” Science 305 (5691): 1779–1782. Seyfarth, Robert M., and Dorothy L. Cheney. 2003. “Signalers and receivers in animal communication.” Annual Review of Psychology 54 (1): 145–173. Slabbekoorn, Hans, and Margriet Peet. 2003. “Ecology: Birds sing at a higher pitch in urban noise.” Nature 424 (6946): 267. Slocombe, Katie E., and Klaus Zuberbühler. 2007. “Chimpanzees modify recruitment screams as a function of audience composition.” Proceedings of the National Academy of Sciences 104 (43): 17228–17233. Tanner, Joanne E., and Richard W. Byrne. 1996. “Representation of action through iconic gesture in a captive lowland gorilla.” Current Anthropology 37 (1): 162–173. Templeton, Christopher, N., Erick Greene, and Kate Davis. 2005. “Allometry of alarm calls: Black-capped chickadees encode information about predator size.” Science 308: 1934–1937. Tomasello, Michael, Barbara L. George, Ann Cale Kruger, and Andrea Evans Farrar. 1985. “The development of gestural communication in young chimpanzees.” Journal of Human Evolution 14: 175–186. Townsend, Simon W., Sonja E. Koski, Richard W. Byrne, Katie E. Slocombe, Balthasar Bickel, Markus Boeckle, Ines Braga Goncalves, et al. 2017. “Exorcising Grice’s ghost: An empirical approach to studying intentional communication in animals: Intentional communication in animals.” Biological Reviews 92 (3): 1427–1433. Vail, Alexander L., Andrea Manica, and Redouan Bshary. 2013. “Referential gestures in fish collaborative hunting.” Nature Communications 4 (1). Wich, Serge A., Michael Krützen, Adriano R. Lameira, Alexander Nater, Natasha Arora, Meredith L. Bastian, Ellen Meulman, et al. 2012. “Call cultures in orang-utans?” pLoS One 7 (5): e36180. Wich, Serge A., and Han de Vries. 2006. “Male monkeys remember which group members have given alarm calls.” Proceedings of the Royal Society of London B: Biological Sciences 273 (1587): 735–740. Wild, Barbara, Michael Erb, Michael Eyb, Mathias Bartels, and Wolfgang Grodd. 2003. “Why are smiles contagious? An FMRI study of the interaction between perception of facial affect and facial movements.” Psychiatry Research: Neuroimaging 123 (1): 17–36. Wilke, Claudia, Eithne Kavanagh, Ed Donnellan, Bridget M. Waller, Zarin P. Machanda, and Katie E. Slocombe. 2017. “Production of and responses to unimodal and multimodal signals in wild chimpanzees, Pan Troglodytes Schweinfurthii.” Animal Behaviour 123: 305–316. Wimmer, Heinz, and Josef Perner. 1983. “Beliefs about beliefs: Representation and constraining function of wrong beliefs in young children’s understanding of deception.” Cognition 13 (1): 103–128. Zipf, George Kingsley. 1949. Human Behaviour and the Principle of Least-Effort. Cambridge, MA: Addison Wesley. Zuberbühler, Klaus, Dorothy L. Cheney, and Robert M. Seyfarth. 1999. “Conceptual semantics in a nonhuman primate.” Journal of Comparative Psychology 113 (1): 33–42.

4 GETTING OUT OF OUR SKIN What Decoding Interspecies Communication and Nonhuman Intelligence Can Tell Us About Deciphering Alien Languages Denise L. Herzing

Introduction

Although researchers have studied intraspecific communication signals in many nonhuman taxa and across different sensory systems, interspecies communication remains largely uncharted territory. Organisms across taxa often live in close proximity to each other and may learn to hear and interpret signals from their allospecific neighbors. In some cases, decoding signals can aid in survival. Many species use the calls and signals of nearby species to their survival advantage (Munn 1986). Some species take advantage of their neighbors’ monitoring abilities to learn the meaning of appropriate alarm calls (birds, primates). In other cases, it can help one species interact socially with another species. Dolphins in the wild have a history of interaction, both competitive and cooperative, with other cetaceans. Atlantic spotted dolphins in the Bahamas interact with bottlenose dolphins (Tursiops truncatus) on a regular and intimate basis (Herzing and Johnson 1997). Such interactions have also occurred throughout the nonhuman world, and also between humans and nonhumans. Domestic animals, in some cases our closest working partners, have tuned into human signals for decades. For example, McConnell (1990) describes specific cross-cultural signals work between domestic dogs and their owners. Some universal rules across species have been described for mammals and birds (Morton 1977) but have yet to be explored adequately. Even wild animals, such as dolphins, have a history interaction with humans in the wild (Pryor and Lindberg 1990). One place to begin exploring the mechanisms for interpreting alien signals may reside in this area of research. Here, many species have either incorporated allospecific signals into their own interpretative systems or have developed new communication repertoires as a way to interpret and transmit information. DOI: 10.4324/9781003352174-4

34

Denise L. Herzing

Passive decoding or active interactions can both be present during interspecific communication. How far taxa can be nonrelated and physically different can vary, and their success in cross-species communication may depend on similarities between both their signals and their sensory systems. Complex communication can include one or more sensory modalities, including visuals, tactile, chemical, and others. Humans have focused on acoustic communication primarily because of our understanding of the complexity of our own acoustic linguistic language. However, it may be worth some investigation to look for other encoded information in these nonhuman interspecific communication systems. Deciphering Nonhuman Communication Signals

How do nonrelated species understand each other? Some examples from the wild suggest that although listening and deciphering may occur with signals such as acoustic alarm calls, studies may also suggest that it is more efficient to create a new communication system with another species than to learn the intricacies of the others. For example, two examples in the dolphin world illustrate the strategy of creating shared/mutual calls during interaction. Resident killer whale pods in the Pacific Northwest have unique pod dialects. When pods interact, they use a small repertoire of shared calls (Ford 1991). There have been recent reports of complex dynamics of vocalization used between two species of sympatric dolphins in Costa Rica, suggesting that they alter aspects of their calls when together and revert back to their own species’ calls when apart (May-Collado 2010). Creating a shared system of information has also proven successful in experimental settings for bonobo chimpanzees (Savage-Rumbaugh et al. 1986), African grey parrots (Pepperberg 1986) and, to a limited degree, dolphins (Delfour and Marten 2005; Herzing et al. 2012; Xitco et al. 2001). Past studies have focused on acoustic communication, but even in the analysis of complex human language, context and information are distributed and interactive (Johnson 2001, Forster 2002), and the interplay of multi-modal signals and social dynamics is the essence of complex information. As we can see, acoustic channels of communication have been the focus of most linguistic studies. This is primarily due to our bias, as humans, to both the acoustic and visual channels of communication. Arik Kershenbaum (in this volume [Chapter 2]) takes a look at less likely sensory modalities that we have thought might be available for linguistic communication. Although chemical, electric, and magnetic senses are not usually discussion in the depth that acoustic channels are, these alternative sensory systems remain a viable option for exploring alien communication. The ability to identify appropriate sensory channels for communication, along with frameworks that include species-specific and within-species signals and rules, will be critical to assess alien communication signals, whether on Earth or elsewhere.

Getting Out of Our Skin

35

Categorizing Signals

Classifying communication signals to determine the natural boundaries of signal units has been for a variety of taxa (e.g., Ehret 1992; Slobodchikoff et al. 1991; Marler 1982; May et al. 1989; Seyfarth et al. 1980). Whether nonhuman animals use a referential or graded systems of communication is still unknown. Although it is likely that most nonhuman animals use a graded system, referential signals are known only to a few taxa; dolphin signature whistles (e.g., Caldwell et al. 1990; Smolker et al. 1993), vervet monkey alarm calls (Seyfarth et al. 1980, Seyfarth and Cheney 1993), ground squirrels (Robinson 1981), and prairie dogs (Slobodchikoff et al. 1991). Cutting-edge computer techniques (e.g., Kohlsdorf et al. 2014; Kershenbaum et al. 2013) have recently been applied to nonhuman acoustic signals. Keeping in mind that referential and graded communication are not necessarily mutually exclusive, such techniques may be valuable for categorization. Other possible outcomes of machine learning/artificial intelligence (AI) might include the discovery of smallest or fundamental units of an animal’s repertoire, unhindered by human biases. The Importance of Metadata for Interpretation

What are metadata, and how are they important? In most animal studies, metadata may include things like age, sex, relationships, life history, and other parameters. They are often used as dependent variables during analysis. However, such metadata plays a large role in the interpretation and function of signal use in many studies. In my own work with dolphins, such metadata is overlaid during different developmental periods, in relation to quantifiable personality traits, etc. Another important component of interpretation and meta-analysis lies in the prosodic features of communication. These features of communication (frequency, duration, amplitude, relative signals, rhythm, synchrony, etc.) should not be overlooked for their importance (Herzing 2015). If there are any universals across species, it may be found in these types of modulation, and ascribing meaning to communication signals is critical to any meaningful study of nonhuman signals. Observational and Experimental Verifcation of Meaning

For a complete understanding of communication signals and their meaning, we may have to turn to other techniques based in the cognitive realm. Distributed cognition, which suggests that cognition occurs not just within an individual mind, but also between individuals (Johnson 2001; Forster 2002), is one example. In this framework, measurable behaviors are considered the “media” that are exchanged between individuals. Because such interactions can be recorded (e.g., behavior),

36

Denise L. Herzing

they become measurable phenomena, unlike mental states and concepts like “intention” that are difficult to assess. Such techniques applied to both nonhuman species and our interaction with them may further illuminate the communication process, critical for any alien signal interpretation. Researchers have also used experimental tests to study both behavioral and cognitive flexibility. Experiments in laboratories with a variety of taxa have shown that many animals – including dolphins – can understand word order (syntax), word meaning (semantics), and abstract thought, and show self-awareness (Herman et al. 1990; Pack and Herman 1995; Marino et al. 1994; Delfour and Marten 2001), common and pygmy chimpanzees (Savage-Rumbaugh et al. 1986) bottlenose dolphins (Herman et al. 1984), and African grey parrots (Pepperberg 1986). Some of these experimental techniques have also involved the creation of a technological bridge that allows and enhances communication between species (Herzing 2016). Conclusions

Interspecific communication analysis is a viable avenue to explore and exercise our abilities to categorize and interpret potential future alien communications. Scientists have new machine learning techniques that can expedite data mining. However, metadata still remain an important element in the interpretation of communication signals. Understanding a species sensory system and it species-specific cultures and social systems will enhance these interpretations. However, such information is likely to be unavailable – or unrecognizable – in the reception of alien communications, and experimental and cognitive-based tests will not be an option for signals received remotely. Imagining different sensory, perceptual, and social systems when interpreting alien communication signals will be critical for our understanding. Utilizing Earth’s great repertoire of other species communication signals in an exercise to extract and decode information would be a worthy endeavor in preparation for non-terrestrial signal interpretation in the future. References Caldwell, Melba C., David K. Caldwell, and Peter L. Tyack. 1990. “Review of the signature whistle hypothesis for the Atlantic bottlenose dolphin.” In The Bottlenose Dolphin, edited by Stephen Leatherwood and Randall R. Reeves, 199–233. San Diego: Academic Press. Delfour, Fabienne, and Ken Marten. 2001. “Mirror image processing in three marine mammal species: killer whales (Orca orcinus), false killer whales (Pseudorca crassidens) and California sea lions (Zalophus californianus).” Behavioral Processes 53: 181–190. Delfour, Fabienne, and Ken Marten 2005. “Inter-modal learning task in bottlenosed dolphins (Tursiops truncatus): A preliminary study showed that social factors might influence learning strategies.” Acta Ethologica 8: 57–64. Ehret, Günter. 1992. “Categorical perception of mouse-pup ultrasounds in the temporal domain.” Animal Behaviour 43: 409–416.

Getting Out of Our Skin

37

Ford, John K.B. 1991. “Vocal traditions among resident killer whales (Orcinus orca) in coastal waters of British Columbia.” Canadian Journal of Zoology 69: 1454–1483. Forster, Debbie. 2002. “Consort turnovers as distributed cognition in olive baboons: A systems approach to mind.” In The Cognitive Animal: Empirical and Theoretical Perspectives on Animal Cognition, edited by Marc Bekoff, Collin Allen, and Gordon Burghardt, 163–172. Cambridge: MIT Press. Herman, Louis M., Palmer Morrel-Samuels, and Adam A. Pack. 1990. “Bottlenosed dolphin and human recognition of veridical and degraded video displays of an artificial gestural language.” Journal of Experimental Psychology 119: 215–230. Herman, Louis M., Douglas G. Richards, and James P. Wolz 1984. “Comprehension of sentences by bottlenosed dolphins.” Cognition 16: 129–219. Herzing, Denise L. 2015. “Synchronous and rhythmic vocalizations and correlated underwater behavior of free-ranging Atlantic spotted dolphins (Stenella frontalis) and Bottlenose dolphins (Tursiops truncatus) in the Bahamas.” Animal Behavior and Cognition 2 (1): 14–30. Herzing, Denise L. 2016. “Interfaces & keyboards for human/dolphin communication: What have we learned?” Animal Behavior and Cognition 3 (4): 243–254. truncatu Herzing, Denise L., Fabienne Delfour, and Adam A. Pack. 2012. “Responses of human-habituated wild Atlantic spotted dolphins to play behaviors using a two-way human/dolphin interface.” International Journal of Comparative Psychology 25: 137–165. Herzing, Denise L., and Christine M. Johnson. 1997. “Interspecific interactions between Atlantic spotted dolphins (Stenella frontalis) and bottlenose dolphins (Tursiops truncatus) in the Bahamas, 1985–1995.” Aquatic Mammals 23: 85–99. Johnson, Christine M. 2001. “Distributed primate cognition: A review.” Animal Cognition 4: 167–183. Kershenbaum, Ari, Layla S. Sayigh, and Vincent M. Janik. 2013. “The encoding of individual identity in dolphin signature whistles: How much information is needed?” PLoS One. doi:10.1371/journal.pone.0077671. Kohlsdorf, Daniel, Celeste Mason, Denise Herzing, and Thad Starner. 2014. “Probabilistic extraction and discovery of fundamental units in dolphin whistles.” Acoustics, Speech and Signal Processing (ICASSP), 2014 IEEE International Conference, 8242–8246. IEEE, Piscataway, New Jersey. Marino, Lori, Diana Reiss, and Gordon G. Gallup, Jr. 1994. “Mirror self-recognition in bottlenose dolphins: Implications for comparative investigations of highly dissimilar species.” In Self-Awareness in Animals and Humans: Developmental Perspectives, edited by S. Taylor Parker, Robert W. Mitchell, and Maria L. Boccia, 380–391. New York: Cambridge University Press. Marler, Peter R. 1982. “Avian and primate communication: the problem of natural categories.” Neuroscience and Biobehavior 6: 87–92. May, Brad, David B. Moody, and William C. Stebbins. 1989. “Categorical perception of conspecific communication sounds by Japanese macaques, Macaca fuscata.” Journal of the Acoustical Society of America 85: 837–847. May-Collado, Laura. 2010. “Changes in whistle structure of two dolphin species during interspecific associations.” Ethology 116: 1065–1074. McConnell, Patricia B. 1990. “Acoustic structure and receiver response in domestic dogs, Canis familiaris.” Animal Behaviour 39: 887–904. Morton, Eugene S. 1977. “On the occurrence and significance of motivation: Structural rules in some bird and mammal sounds.” American Naturalist 111: 855–869.

38

Denise L. Herzing

Munn, Charles A. 1986. “The deceptive use of alarm calls by sentinel species in mixedspecies flocks of neotropical birds.” In Deception: Perspectives on Human and Nonhuman Deceit, edited by Robert W. Mitchell and Nicholas S. Thompson, 169–176. Albany: New York State University Press. Pack, Adam A., and Louis M. Herman. 1995. “Sensory integration in the bottlenose dolphin: Immediate recognition of complex shapes across the sense of echolocation and vision.” Journal of the Acoustical Society of America 98: 722–733. Pepperberg, Irene M. 1986. “Acquisition of anomalous communicatory systems: Implications for studies on interspecies communication.” In Dolphin Cognition and Behavior: A Comparative Approach, edited by Ron. J. Schusterman, Jeanette A. Thomas, and Forrest G. Wood, 289–302. Hillsdale: Lawrence Erlbaum Associates. Pryor, Karen, and Jon Lindberg. 1990. “A dolphin-human fishing cooperative in Brazil.” Marine Mammal Science 6: 77–82. Robinson, Scott R. 1981. “Alarm communication in Belding’s ground squirrels.” Zier Tierpsychologie 56: 150–168. Savage-Rumbaugh, Sue, Kelly McDonald, Rose A. Sevcik, William D. Hopkins, and Elizabeth Rupert. 1986. “Spontaneous symbol acquisition and communicative use by pygmy chimpanzees [Pan paniscus].” Journal of Experimental Psychology 112: 211–235. Seyfarth, Robert M., and Dorothy L. Cheney. 1993. “Meaning, reference, and intentionality in the natural vocalizations of monkeys.” In Language and Communication: Comparative Perspectives, edited by Herbert R. Roitblat, Louis M. Herman, and Paul Nachtigall, 195–219. Hillside: Erlbaum. Seyfarth, Robert M., Dorothy L. Cheney, and Peter Marler. 1980. “Vervet monkey alarm calls: Semantic communication in a free-ranging primate.” Animal Behaviour 28: 1070–1094. Slobodchikoff, Con N., Judith Kiriazis, C. Fischer, and E. Creef. 1991. “Semantic information distinguishing individual predators in the alarm calls of Gunnison’s prairie dog.” Animal Behaviour 42: 713–719. Smolker, Rachel A., Janet Mann, and Barbara B. Smuts. 1993. “Use of signature whistles during separations and reunions by wild bottlenose dolphin mothers and infants.” Behavioural and Ecological Sociobiology 33: 393–402. Xitco, Jr. Mark J., John D. Gory, and Stan A. Kuczaj II. 2001. “Spontaneous pointing by bottlenose dolphins (Tursiops truncatus).” Animal Cognition 4: 115–123.

5 COMMUNICATIVE RESOURCES BEYOND THE VERBAL TIER A View on Xenolinguistics From Interactional Linguistics Heike Ortner

Introduction

When we try to anticipate contact with extraterrestrial intelligence (ETI), we speculate about communication systems other than human language. We develop assumptions concerning the grammatical structures of extraterrestrial languages (cf. Roberts et al., this volume [Chapter 15]), the types of signals that might be used for transmission (cf. Herzing, this volume [Chapter 4], Granger et al., this volume [Chapter 8]), and the impact of what is almost certain to be different cognitive abilities (cf. Samuels and Punske, this volume [Chapter 16]). In this chapter, I will reflect on the corporality of our terrestrial existence and its impact on our everyday interactions, drawing from the theoretical background of interactional linguistics. This conception was prompted by two experiences I had while I was preparing a project about multimodal instructions in physical therapy at a clinic for neurorehabilitation. There were two patients who I remember vividly. The first was an elderly woman with global aphasia, which means that she could neither fully understand the therapist’s utterances nor produce intelligible speech. However, she was fairly well able to communicate with the therapist by attentively watching him, mainly interpreting his facial expressions, gestures, and the pitch of his voice (Goodwin 2000, 2003 for detailed analyses of interactions with aphasic patients). With sufficient communicative success, she replied by gaze, body posture, and gestural hand and head movements. She relied on these clues so much that the therapist deliberately toned down his para-verbal and non-verbal behavior,1 forcing her to train her remaining productive and receptive linguistic competence. The second patient was a young man with locked-in syndrome, which left him unable to move or talk, while his mental abilities were probably undamaged. In the DOI: 10.4324/9781003352174-5

40

Heike Ortner

session that I observed, the speech therapist concentrated on teaching the patient how to swallow food on cue. For this patient, communicating via eye movement was not the solution. His interactions and the shared meaning-making with his fellow human beings were diminished not only by his lack of a voice but also by his almost fully paralyzed body. The therapist was nonetheless able to read the current mood and compliance of the patient and got him to work with her on the task at hand. These two extraordinary interactions lead me to the topic of this chapter: the multimodality of human interaction and its implications for xenolinguistics. Some current trends in humanities and social studies indicate a heightened interest in the human body and its functions in meaning-making – so much so that there is talk of an “embodied turn” (Nevile 2015). Interactional linguistics is a relatively new research program dedicated to the practices of shared meaning-making with all available communicative resources in face-to-face interaction: verbal utterances, voice, gaze, body postures and body orientation, positioning in space, facial expression, object manipulation, gesture and other body movements, or generally any kind of action (Couper-Kuhlen and Selting 2018: 9, 12). If we ever get to establish immediate contact or at least interpret video signals of visible bodily behavior, this aspect of the interaction is going to be of utmost importance, yet also of utmost extraneousness. This chapter gives an overview of some key terms and topics in interactional linguistics and their significance for speculation on extraterrestrial languages, xenolinguistics, and messages to extraterrestrial intelligence (METI). Some helpful insights from studies on human–animal interaction will be included, as well. Multimodality in Interaction

Interactional linguistics is rooted in functional theories (Bühler 1978 [1934]; Halliday 1973), constructionist and ethnomethodologist traditions (Garfinkel 1967), sociology (Goffman 1983), linguistic anthropology (Goodwin 1981), and conversation analysis (Sacks et al. 1974; Heath 1986) (see Couper-Kuhlen and Selting 2018; Streeck et al. 2011 for a historical overview). Conversation analysis (CA) was originally mainly concerned with verbal processes such as turn-taking, repairs, and the specifics of linguistic constructions in speech (Sacks et al. 1974; Goodwin 1981). While facial expressions and gestures were recognized as integral to conversation early on, there was still a strong focus on the “verbal tier”2 of conversation. Catalyzed by the advancements in recording and transcribing visual data, interactional linguistics became a popular research program. By these technological means, it is possible to analyze all facets of interaction in more detail: sequential and collaborative processes in several modes at once and their coordination in social interaction, and technology-mediated interaction or resources in the environment, such as the use of tools (Deppermann and Schmitt 2007; Haddington et al. 2013).

Communicative Resources Beyond the Verbal Tier

41

Social interaction is not chaotic: By the strictly empirical and qualitative sequential micro-level analysis of natural data, it is possible to find systematic patterns in the use of “verbal, vocal and visible resources in the service of performing actions and activities” (Couper-Kuhlen and Selting 2018: 9). One of the key terms is multimodality; it expands “the social interpretation of language and its meanings to the whole range of representational and communicational modes or semiotic resources for making meaning with employed in culture – such as image, writing, gesture, gaze, speech, posture” (Jewitt 2014: 1). I would like to stress the aspect of culture in this definition to make clear that multimodal signs may have iconic and indexical properties, yet their forms (e.g., appropriate amplitude) and pragmatic functions in interaction are learned and culturally diverse. For CETI (communication with extraterrestrial intelligence) and METI, this means that we should have little hope of easily understanding bodily expressions like gesture or facial expressions of ETI, even if their outer morphology were similar to ours. After all, the sensorimotor system of our body is an integral part of the Chomskyan “faculty of language” (cf. Samuels and Punske, this volume [Chapter 16]). Growing research in interactional linguistics aims at formulating a grammar of interactional displays and their coordination. Beyond linear sequentiality and systematicity, another guiding principle of interactional research is the notion of joint meaning-making: Participants in interaction constantly negotiate and co-construct meaning in cooperation; for example, in processes of multimodal identity construction (Norris 2011), in narratives (Quasthoff and Becker 2005), and in cross-cultural settings. Summing up the interactional linguistics program, Couper-Kuhlen and Selting (2018: 551) name several design features of language that are very different from a structuralist approach: [Language is] a dynamic process rather than a static entity; it is physically embodied and publicly displayed; it delivers actions; its resources are organized as paradigmatic alternatives for specific positions in the syntagma of turns and sequences; it is temporally emergent, allowing for projection, extension, and repair; it is shared and continuously provides for co-participation.

Preconditions of Interaction With ETI

First, I would like to distinguish three types of interaction that have very different interactional consequences: 1) direct face-to-face interaction, or being at the same place at the same time; 2) technology-mediated communication, e.g., through a screen, but still in a way “face-to-face” by partly visible displays of communicative resources; and 3) using only abstract symbols without any direct perception of “the other.” Henceforth, I will deal mainly with the first notion (Granger et al., this volume [Chapter 8], thoroughly discuss the case of “hyper-asynchronicity” of contact).

42

Heike Ortner

What are the preconditions of interacting with ETI in a meaningful, cooperative way? The human visual and auditory systems and our cognitive abilities such as attention, memory, and judgment are highly developed (Evans 2001). To us, perceiving an object as moving is the basis of perceiving it as living and animated. The perception of a “dynamic congruency of affectivity and movement” – moving toward something or moving away from it as an index of emotion – is the basis of any relationship we have with other life forms (Sheets-Johnstone 2009: 376, emphasis in original). However, a working visual sensory apparatus is not necessary for bodily interaction (cf. Herzing, this volume [Chapter 4]). Children who were born blind, for example, produce gestures at the same rate as their sighted counterparts, but use more body-focused types and rather refer to objects and persons that are near them. All in all, they use more auditory gestures and cues and other modalities (Avital and Streeck 2011: 170, 180; cf. also Wells-Jensen, this volume [Chapter 13], on the abilities of blind persons and their implications for ETI). There are many resources that could be central to the interaction of ETI: touch, colors, scents, and so on. To give an example of interactional argumentation, let us look at an element that is thought to be universal in all human languages, according to Dingemanse et al. (2013): the interjection “huh?” The universality of this element, according to them, is not due to an innate mechanism or a genetic given but to a process of “convergent cultural evolution”: The “huh?” simply has the perfect articulatory syllabic structure in any language to account for a swift initiation of a repair. Of course, it is highly unlikely that ETI will repeat their message because we say “huh?” to them, just as it is unlikely that our gestures and facial expressions will be meaningful to them. When thinking about future applications of CETI and METI, we ought to develop strategies to complete communicative tasks such as initiating repair creatively, starting from what we know so far. This is because there are further important preconditions for mutual understanding (additionally to those discussed in depth by Ross, this volume [Chapter 12]). There has to be some common ground or a starting point for intersubjectivity (see what follows). Even more basically, we need to recognize relevant displays in the different modes of interaction (Deppermann and Schmitt 2007: 35) – in other words, the ability and opportunity to isolate functional units. There has to be some kind of systematicity of cues. Fuchs and De Jaegher (2009: 471, emphasis in original following Lyons-Ruth et al. 1998) speak of “implicit relational knowing” – in CETI and METI, the implicit aspects are not clear-cut from the start. Perceivable changes in the different modes of interaction (e.g., gaze, gesture, facial expression) are the basic unit of interactional analysis. The proposed co-construction of meaning implies that a communicative signal is “what the participant treats as a signal” (Norris 2011: 49) – but this depends a lot on our communicative experiences and is therefore deeply rooted in cultural and social structures (Ross, this volume [Chapter 12], also highlights the difficulty of cultural understanding, especially with little context given).

Communicative Resources Beyond the Verbal Tier

43

For thinking about CETI and METI, we need to ask the following questions. What constitutes a change in a communicative resource – to us and to ETI? How will we recognize the units of interactional displays, and how will we give cues on how to disassemble our signals within one single mode or in different modes? Human–robot interaction faces a similar dilemma that is solved by orientation toward human habits and practices. For example, a robot in a museum can be programmed to pause and restart at a certain rate to guide the gaze of visitors, acting on the assumption that gazing at something equals attention (Kuzuoka et al. 2008). In general, the idea of “Gestalt” is very important to interaction (Mondada 2014: 37), meaning that our perception of a functional display depends on whether we see a figure that becomes eye-catching before the ground. Incentives from human–animal interaction (HAI) can help us better understand the challenges of CETI. I will not engage in a discussion of the linguistic competence of animals (see Pearson 2013 for a comprehensive overview). It is problematic to talk about animal signals as conveying meaning, transmitting information, or being representational. Horisk and Cocroft (2013), for example, argue against any representational notion to get away from any speculation on the existence of a theory of mind in animals. They propose to discuss signals of animals from the viewpoint of their influence on other animals and humans and the responses. Actually, this is a very interactional linguistics’ type of reasoning: The aboutness of a signal is not representational, but co-constructed in interaction. Whatever we interpret as a communicative sign of an animal conveying meaning is shared meaning in interaction, whether or not the interpretation is “accurate.” In this sense, intentionality is less important than interpretation (but cf. Hobaiter et al., this volume [Chapter 3], for more profound thoughts on the interplay between intentionality, cooperativity, meaning, behavior, and context from a Gricean perspective). There is a lot of research on multimodal displays of animals. While smell and taste are central to animal communication (chemical signaling), many animals can also fine-tune their multimodal displays – for example, birds coordinating their dancing choreography and their singing voices (Miles and Fuxjager 2018), gibbons adjusting their facial expressions depending on social context (Scheider et al. 2016), and cephalopods attuning their body postures and patterns of colors and spots to express specific meanings, not to mention the astounding abilities of dolphins (Ballesteros 2010: 114; cf. Pepperberg, this volume [Chapter 6] and Kershenbaum, this volume [Chapter 2], for more detailed discussions on HAI, the interpretation of different types of signals and the likeliness of ETI using specific modes other than the acoustic one). There are even attempts to apply multimodal CA to the study of animal communication, e.g., to determine vocal and gestural turn-taking of birds and great apes (Fröhlich 2017). To me, studies of HAI show how we interpret animal signals according to our experiences with humans. In general, the behavior of animals is interpreted as emotional, based on the notion that the affective neuronal and hormonal mechanisms – as well as the social functions of affective displays – can be compared (Watanabe and Kuczaj 2013; Kotrschal 2013; cf. also Pepperberg,

44

Heike Ortner

this volume [Chapter 6], on the functions of social interaction). On the other side, our closest allies in the animal kingdom, dogs, can use gaze as a communicative tool and comprehend many human bodily expressions, having an overall “sensitivity to human stimuli” (Wynne et al. 2011: 106). We can only speculate on our sensitivity for biologically more different life forms. Intersubjectivity Is Key

Intersubjectivity is a key term in interactional linguistics. I will turn only to one of its many meanings, described by Deppermann (2014: 65) as “intersubjectivity as practical accomplishment.” It has a certain sequential organization: first, a participant in the interaction conducts an “action-be-understood”; second, the recipient displays his or her understanding of the action; and third, the producer acknowledges this understanding (Deppermann 2015: 66) – in CETI, of course, we might need to find a different “flow” of intersubjective understanding. Another interesting branch of research deals with processes of accommodation and mimicry in interaction. The communication accommodation theory (CAT) states that “Accommodation – as a process – refers to how interactants adjust their communication so as to either diminish or enhance social and communicative differences between them” (Giles 2009: 278); one function is “reducing uncertainties about the other” (Giles 2008: 163). This can be done by verbal means such as word choices, but also by interactive displays – for example smiling, nodding, and gesturing. According to Giles 2008), the main strategies are the following. • Convergence: upward or downward adjustment to the other’s language or dialect, rhythm, posture, and their interpretative abilities. • Maintenance or non-accommodativeness: not shifting one’s language use. • Divergence: accentuating one’s social identity by emphasizing differences. Related concepts are recipient design, i.e., orientation toward co-participants (Sacks et al. 1974), audience design; i.e., linguistic choices to seek approval of the audience (Bell 2009); and behavioral mimicry, i.e., “the automatic imitation of gestures, postures, mannerisms, and other motor movements” (Chartrand and Lakin 2013: 285). Verbal aspects such as syntax may be subject to mimicry, as well. For any of these concepts, the connection to interactional linguistics lies in the notion that meaning “unfolds only in the context of an embodied contextual configuration,” building “on prior interactional histories and on both shared and private experiences” (Deppermann 2015: 96), so we are led by assumptions and expectations regarding other participants at first, but we can quickly establish common ground (Clark 1996) molded by our interactive negotiations. When trying to apply these concepts to communication with ETI, behavioral mimicry seems be a good starting point for building rapport. Only after the establishment of a common “language” in any meaning of the world may we think

Communicative Resources Beyond the Verbal Tier

45

about more complex strategies such as convergence, maintenance, and divergence to communicate social meanings toward ETI. The Intricate Temporality and Spatiality of the Human Condition

Multimodal resources are grounded in time and space (Mondada 2014: 37). Regarding time, there are both successive and simultaneous temporal relationships of multimodal displays in human–human interaction (Deppermann and Schmitt 2007); for example, we can combine a bodily demonstration with a verbal explanation (Mondada 2014; Haddington et al. 2014). Regarding space, there are specific practices on how to refer to physical as well as social positions of interactants (Stukenbrock 2014; Haddington et al. 2013: 20). A very basic human action is pointing to objects in the immediate spatio-temporal surrounding (Bühler 1978 [1934] calls this “demonstratio ad oculos”) but also to objects that are not present (Bühler’s “deixis am phantasma”) (Stukenbrock 2014: 89). I would not call pointing a universal gesture, even though it is one of the earliest communicative means infants make use of. Dogs that are socialized in a human environment and have continuous experience with it can follow pointing gestures (Wynne et al. 2011). We may assume that pointing at something can easily guide attention and interrelate objects or persons with terms and names, facilitating language learning or more general understanding. The cognitive foundations and perceptions of time and space may be very different in ETI, however. While we perceive interactions as a sequence of utterances, the different layers of multimodal interaction are produced and processed simultaneously. How may we get a grasp on ETI’s “flow”? Coordination and Projection

Multimodal resources are not separate layers but interwoven, flexible, and dynamic. Their coordination is so complex that we only start to understand the processes of human mutual calibration in interaction: on the one hand, intrapersonal coordination, the coordination of modes by one person; on the other hand, interpersonal coordination, i.e., the coordination with multiple participants or the multiplicity of modes of expression while interacting (Deppermann and Schmitt 2007: 35). Still, we are very capable of coordinating our own bodily expressions and verbal utterances, as well as ourselves with other participants in the interactional ensemble. This is in part due to our ability to predict (in linguistic terms project) “what is likely to come next” (Auer 2015: 28) – for example, there are syntactic restrictions in verbal utterances that make it easy for us to complete a sentence before it is uttered. Therefore, we can also co-construct utterances by taking over a construction. It is more difficult to imagine how this process could work in CETI. Completions might not only turn out completely wrong because predictions of things to come are more difficult, but what is absolutely normal in human interaction could be interpreted as unfriendly interruption. Communication with ETI would probably

46

Heike Ortner

stay very explicit and careful for a long time. To cite an example from human–animal interactions, there seems to be a gradual adjustment between cattle and their herders after long-lasting contact (Ellingsen et al. 2014). With this in mind, coordination between humans and ETI may come about slowly, but steadily. The Body as a Resource of Meaning-Making

In interactional linguistics, “embodiment” refers to the bodily resources of communication, the embodied experiences of making sense in real time (Nevile 2015; Streeck et al. 2011). Sheets-Johnstone (2009) calls for a closer look at kinesthesia, i.e., a sensory-kinetic understanding of the body, rather than a senorimotor definition of the body as the “place” of cognition. The notion of intercorporeality was introduced by Merleau-Ponty (2005 [originally published in 1945]: 413–422) referring to the coordination of human bodies in interaction as a “together-ness,” some sort of merge of object and subject. Fuchs and De Jaegher (2009: 470) use the terms “mutual incorporation” and “coupling:” Social understanding is achieved dynamically by the coupling of “two embodied subjects” and their interactive coordination through voice, facial expressions, touch, gesture, and other features. The other person’s body that we experience through perception becomes incorporated into one’s own, so to say as an extension of ourselves. This incorporation facilitates “participatory sense-making” (Fuchs and De Jaegher 2009: 477), which means providing the participants with knowledge, experience, and domains not available to the individual. This also implies mutual affection; similarly, Trevarthen (1987: 194) defines meaning as “meaning in relationships” and “interpersonal symbiosis.” From research on infant–caregiver interactions, Fuchs and De Jaegher (2009: 482) draw the conclusion that mutual incorporation also allows us to get a hold of another person’s experiences, even if they are not capable of verbal expression. Whether this not only applies to infants, but also to ETI, is very questionable. A pessimistic consequence of mutual incorporation might be that we are not able to understand someone else’s body if this body and its perceptions are fundamentally different from ours. Meyer and von Wedelstaedt (2017: 13) emphasize the importance of “tacit bodily coordination.” Unfortunately, anything that is deemed “tacit” in human interaction is tricky for communication with ETI. All of our long-standing, acquired presumptions on interaction may mislead us when we have to interpret a different mindset and its bodily expressions – or, to apply a more accurately interactional attitude, when we strive toward understanding the relations between thinking and body. There is a high risk of “sending the wrong message” or subtext without bodies and voices (Harbour, this volume [Chapter 18]). However, as Berea (this volume [Chapter 7]) points out, communication pathologies have the potential to be transformative in the most helpful way when trying to find common ground with ETI. I also second the remarks by Granger et al. (this volume [Chapter 8]) in favor of a systematic, creative, and empathic approach devoid of judgmental good/ bad evaluations of ETI and premature assumptions.

Communicative Resources Beyond the Verbal Tier

47

So far, my remarks mainly concerned a dyadic participation framework – yet the term “multiactivity” gained a lot of attention in the last couple of years, i.e., paying reference to more complex participation frameworks, the coordination of multiple bodies and activities, and temporal and sequential aspects of concurrent activities (Haddington et al. 2014; Deppermann 2014; Mondada 2014). I can point only to the increasing complexity of an interactional ensemble consisting of several human beings and representatives of ETI that would probably be the more probable case anyway. So, the participation framework does not have only two fundamentally dissimilar parties (human–ETI) but calls for efficient communication between interlocutors of the “same” communicative universe, as well – a feat not as trivial as it may sound, given the vast cultural differences on Earth and the pitfalls of coconstructing meaning in everyday interaction. Conclusion

“Combine and repeat” is a central principle of language mentioned in several contributions to this volume (cf. Roberts et al. [Chapter 15], Samuels and Punske [Chapter 16], and Sperlich [Chapter 14]). It can also be a helpful strategy in finding patterns in interactional displays of any mode, communicative resource, signal type, and code. Summing up, I would like to repeat the following main challenges we would face in IETI (interactions with extraterrestrial intelligence). • Recognizing relevant displays and even systems of displays as codes and cues for understanding – given the fact that corporeal displays are not iconic, it would be extremely hard to translate them without some sort of Rosetta stone (cf. Slobodchikoff, this volume [Chapter 9], on thoughts on the irreplaceability of a tool like this). • Getting a grasp on the temporal “flow” of ETI, taking into account the option that spatial and temporal concepts and therefore coordination and projective forces might be completely different. • Making ourselves comprehensible, not taking our corporal design for granted. • Trying to think with a different body as probably the biggest challenge. Given these fascinating questions, it is a pity that direct “face-to-face” interaction with extraterrestrial intelligence is even less likely than other forms of contact. I still think that it is stimulating to make this endeavor. Interactional linguistics refrains from the thought of “body language” as being more honest, an exhibition of the subconscious, or mainly constituted by iconic signs. The most helpful implications of applying the methodology and terminology of interactional linguistics on CETI is a warning: we will need to be careful to draw deterministic conclusions on the basis of ETI’s cognitive make-up because of their bodily experiences. After all, meaning is meaning in relationships, meaning in interaction – we as humans, embedded in our own corporeality (!), are part of the equation.

48

Heike Ortner

Notes 1 In interactional linguistics, the terms “para-verbal” and “non-verbal” are avoided because they imply the separation of codes that interactionist studies wish to overcome (Streeck et al. 2011: 9). In the same vein, there is no mention of “body language”: the body is fundamentally involved in the human language facility. 2 “Tier” refers to linguistic transcripts that dissect interactions into different tiers or lines, e.g., speech (word-by-word transcription of utterances) and gesture.

References Auer, Peter. 2015. “The temporality of language in interaction: Projection and latency.” In Temporality in Interaction, edited by Arnulf Deppermann and Susanne Günthner, 27–56. Amsterdam, Philadelphia: Benjamins. Avital, Sharon, and Jürgen Streeck. 2011. “Terra incognita: Social interaction among blind children.” In Embodied Interaction: Language and Body in the Material World, edited by Jürgen Streeck, Charles Goodwin, and Curtis LeBaron, 169–181. Cambridge: Cambridge University Press. Ballesteros, Fernando J. 2010. E.T. Talk: How Will We Communicate with Intelligent Life on Other Worlds? Astronomer’s Universe. New York: Springer. Bell, Allan. 2009. “Language style as audience design.” In The New Sociolinguistics Reader, edited by Nikolas Coupland and Adam Jaworski, 265–275. Basingstoke, New York: Palgrave Macmillan. Bühler, Karl. 1978 [1934]. Sprachtheorie: Die Darstellungsfunktion der Sprache [The Theory of Lanuage: The Representational Function of Language], edited by Ungekürzte Ausgabe. Ullstein-Buch. Frankfurt: Ullstein. Chartrand, Tanya L., and Jessica L. Lakin. 2013. “The antecedents and consequences of human behavioral mimicry.” Annual Review of Psychology 64: 285–308. Clark, Herbert H. 1996. Using Language. Cambridge: Cambridge University Press. Couper-Kuhlen, Elizabeth, and Margret Selting. 2018. Interactional Linguistics: Studying Language in Social Interaction. Cambridge: Cambridge University Press. Deppermann, Arnulf. 2014. “Multimodal participation in simultaneous joint projects: Interpersonal and intrapersonal coordination in paramedic emergency drills.” In Multiactivity in Social Interaction: Beyond Multitasking, edited by Pentti Haddington, Tiina Keisanen, Lorenza Mondada, and Maurice Nevile, 247–281. Amsterdam, Philadelphia: Benjamins. Deppermann, Arnulf. 2015. “When recipient design fails: Egocentric turn-design of instructions in driving school lessons leading to breakdowns of intersubjectivity.” Gesprächsforschung. Online-Zeitschrift zur verbalen Interaktion 16: 63–101. Deppermann, Arnulf, and Reinhold Schmitt. 2007. “Koordination. Zur Begründung eines neuen Forschungsgegenstandes.” In Koordination. Analysen zur multimodalen Interaktion, edited by Reinhold Schmitt, 15–54. Tübingen: Narr. Dingemanse, Mark, Francisco Torreira, and Nick J. Enfield. 2013. “Is huh? a universal word? Conversational infrastructure and the convergent evolution of linguistic items.” PLoS One 8 (11): Online. Ellingsen, Kristian, Grahame J. Coleman, Vonne Lund, and Cecilie M. Mejdell. 2014. “Using qualitative behaviour assessment to explore the link between stockperson behaviour and dairy calf behaviour.” Applied Animal Behaviour Science 153: 10–17. Evans, Dylan. 2001. Emotion: The Science of Sentiment. Oxford: Oxford University Press. Fröhlich, Marlen. 2017. “Takng turns across channels: Conversation-analytic tools in animal communication.” Neuroscience and Biobehavioral Reviews 80: 201–209.

Communicative Resources Beyond the Verbal Tier

49

Fuchs, Thomas, and Hanne De Jaegher. 2009. “Enactive intersubjectivity: Participatory sensemaking and mutual incorporation.” Phenomenology and the Cognitive Sciences 8: 465–486. Garfinkel, Harold. 1967. Studies in Ethnomethodology. Englewood Cliffs: Prentice Hall. Giles, Howard. 2008. “Communication accommodation theory.” In Engaging Theories in Interpersonal Communication. Multiple Perspectives, edited by Leslie A. Baxter and Dawn O. Braithwaite, 161–173. Los Angeles: Sage. Giles, Howard. 2009. “The process of communication accommodation.” In The New Sociolinguistics Reader, edited by Nikolas Coupland and Adam Jaworski, 276–286. Basingstoke, New York: Palgrave Macmillan. Goffman, Erving. 1983. “The interaction order.” American Sociological Review 48: 1–17. Goodwin, Charles. 1981. Conversational Organization, Language, thought, and Culture: Advances in the Study of Cognition. New York: Academic Press. Goodwin, Charles. 2000. “Pointing and the collaborative construction of meaning in aphasia.” Proceedings of the seventh annual Symposium About Language and Society, Austin. Goodwin, Charles, ed. 2003. Conversation and Brain Damage. Oxford: Oxford University Press. Haddington, Pentti, Tiina Keisanen, Lorenza Mondada, and Maurice Nevile. 2014. “Towards multiactivity as a social and interactional phenomenon.” In Multiactivity in Social Interaction: Beyond Multitasking, edited by Pentti Haddington, Tiina Keisanen, Lorenza Mondada, and Maurice Nevile, 3–32. Amsterdam, Philadelphia: Benjamins. Haddington, Pentti, Lorenza Mondada, and Maurice Nevile. 2013. “Being mobile: Interaction on the move.” In Interaction and Mobility: Language and the Body in Motion, edited by Pentti Haddington, Lorenza Mondada, and Maurice Nevile, 3–61. Berlin, Boston: de Gruyter. Halliday, Michael A.K. 1973. Explorations in the Functions of Language, Explorations in Language Study. London: Arnold. Heath, Christian. 1986. Body Movement and Speech in Medical Interaction. Cambridge: Cambridge University Press (Studies in Emotion and Social Interaction). Horisk, Claira, and Reginald B. Cocroft. 2013. “Animal signals: Always influence, sometimes information.” In Animal Communication Theory: Information and Influence, edited by Ulrich E. Stegmann, 259–280. Cambridge: Cambridge University Press. Jewitt, Carey. 2014. “An introduction to multimodality.” In The Routledge Handbook of Multimodal Analysis, edited by Carey Jewitt, 15–30. London, New York: Routledge. Kotrschal, Kurt. 2013. “Emotions are at the core of individual social performance.” In Emotions of Animals and Humans: Comparative Perspectives, edited by Shigeru Watanabe and Stan A. Kuczaj, 3–21. Tokyo: Springer. Kuzuoka, Hideaki, Karola Pitsch, Yuya Suzuki, Ikkaku Kawaguchi, Keiichi Yamazaki, Akiko Yamazaki, Yoshinori Kuno, Paul Luff, and Christian Heath. 2008. “Effect of restart and pauses on achieving a state of mutual orientation between a human and a robot.” Proceedings of the 2008 ACM Conference on Computer Supported Cooperative Work, San Diego. Lyons-Ruth, Karlen, Nadia Bruschweiler-Stern, Alexandra Harrison, Alexander C. Morgan, Jeremy P. Nahum, Louis Sander, Daniel N. Stern, and Edward Z. Tronick. 1998. “Implicit relational knowing: Its role in development and psychoanalytic treatment.” Infant Mental Health Journal 19 (3): 282–289. Merleau-Ponty, Maurice. 2005 [1945]. Phenomenology of Perception. Translated by Colin Smith. London, New York: Routledge. Meyer, Christian, and Ulrich von Wedelstaedt. 2017. “Intercorporeality, interkinesthesia, and enaction: New perspectives on moving bodies in interaction.” In Moving Bodies

50

Heike Ortner

in Interaction – Interacting Bodies in Motion: Intercorporeality, Interkinesthesia, and Enaction in Sports, edited by Christian Meyer and Ulrich von Wedelstaedt, 1–23. Amsterdam, Philadelphia: Benjamins. Miles, Meredith C., and Matthew Fuxjager. 2018. “Animal choreography of song and dance: A case study in the Montezuma oropendola, Psarocolius montezuma.” Animal Behaviour 140: 99–107. Mondada, Lorenza. 2014. “The temporal orders of multiactivity: Operating and demonstrating in the surgical theatre.” In Multiactivity in Social Interaction: Beyond Multitasking, edited by Pentti Haddington, Tiina Keisanen, Lorenza Mondada, and Maurice Nevile, 33–75. Amsterdam, Philadelphia: Benjamins. Nevile, Maurice. 2015. “The embodied turn in research on language and social interaction.” Research on Language and Social Interaction 48 (2): 121–151. Norris, Sigrid. 2011. Identity in (Inter)action: Introducing Multimodal (Inter)action analysis, Trends in Applied Linguistics. Berlin, New York: de Gruyter Mouton. Pearson, Susan. 2013. “Speaking bodies, speaking minds: Animals, language history.” History and Theory Theme Issue 52: 91–108. Quasthoff, Uta M., and Tabea Becker, eds. 2005. Narrative Interaction, Studies in Narrative. Amsterdam, Philadelphia: Benjamins. Sacks, Harvey, Emanuel A. Schegloff, and Gail Jefferson. 1974. “A simplest systematics for the organization of turn-taking for conversation.” Language 50: 696–735. Scheider, Linda, Bridget M. Waller, Leonardo Oña, Anne M. Burrows, and Katja Liebal. 2016. “Social use of facial expressions in hylobatids.” PLoS One 11 (3): 1–11. Sheets-Johnstone, Maxine. 2009. “Animation: The fundamental, essential, and properly descriptive concept.” Continental Philosophy Review 42 (3): 375–400. Streeck, Jürgen, Charles Goodwin, and Curtis LeBaron. 2011. “Embodied interaction in the material world: An introduction.” In Embodied Interaction: Language and Body in the Material World, edited by Jürgen Streeck, Charles Goodwin, and Curtis LeBaron, 1–26. Cambridge: Cambridge University Press. Stukenbrock, Anja. 2014. “Pointing to an ‘empty’ space. Deixis am Phantasma in face-toface interaction.” Journal of Pragmatics 74: 70–93. Trevarthen, Colwyn. 1987. “Sharing making sense: Intersubjectivity and the making of an infant’s meaning.” In Language Topics: Essays in Honour of Michael Halliday, edited by Ross Steele and Terry Threadgold, 177–199. Amsterdam, Philadelphia: Benjamins. Watanabe, Shigeru, and Stan A. Kuczaj, eds. 2013. Emotions of Animals and Humans: Comparative Perspectives: The Science of the Mind. Tokyo: Springer. Wynne, Clive D.L., Nicole R. Dorey, and Monique A.R. Udell. 2011. “The other side of the bond: Domestic dogs’ human-like behaviors.” In How Animals Affect Us: Examining the Influence of Human-Animal Interaction on Child Development and Human Health, edited by Peggy McCardle, Sandra McCune, James A. Griffin, and Valerie Maholmes, 101–115. Washington: American Psychological Association.

6 HOW STUDIES OF COMMUNICATION AMONG NONHUMANS AND BETWEEN HUMANS AND NONHUMANS CAN INFORM SETI Irene M. Pepperberg

Introduction

Humans have long wished to “talk” with alien beings – first with creatures with whom they shared their world and later with those in worlds beyond. Legends and stories, from those involving King Solomon (who purportedly had a ring that enabled him to communicate at will with all the birds and beasts in his realm; Lorenz 1952), to many children’s books (e.g., Lofting 1948; King-Smith 1984), to those of Native Americans (who supposedly could change into various animals and thus share their lives; Rasmussen 1972), to various Hollywood blockbusters (e.g., Arrival, Contact, Close Encounters of the Third Kind), demonstrate humans’ deep interests in these possibilities. For most of human history, understanding nonhuman codes and communicating with animals were indeed reserved for individuals like the fictional Dr. Doolittle – though people such as trackers could predict some nonhuman behavior based on, for example, spoor and probabilistic reasoning. True two-way communication, however, was impossible. Nevertheless, starting in the twentieth century, humans began to crack these codes (described in what follows), and some even trained nonhumans to use limited forms of human systems (e.g., Pepperberg 1999). Although these human endeavors have not been completely successful, some progress has been made, and I review what is known – or more accurately, what we are beginning to understand – about some Earthly nonhuman communication systems, and I propose how this knowledge might be used to understand possible extraterrestrial communicative systems and how we might establish two-way communication systems with other beings. Decoding the World Around Us

Around the middle of the twentieth century, researchers began using rigorous scientific protocols to decipher the meaning of nonhuman communication systems. DOI: 10.4324/9781003352174-6

52

Irene M. Pepperberg

Decades of work led to some knowledge of information conveyed in systems such as birdsong (Kroodsma and Miller 1996; Catchpole and Slater 2008), vervet and prairie dog alarm calls (respectively, Cheney and Seyfarth 1990; Slobodchikoff, this volume [Chapter 9]), chimpanzee vocalizations (Goodall 1986; Laporte and Zuberbühler 2010, Hobaiter et al., this volume [Chapter 3]), and dolphin whistles (McCowan and Reiss 1997; Ferrer-i-Cancho and McCowan 2009; Herzing, this volume [Chapter 4]). Initially, experiments simply involved engaging these nonhumans in playback experiments in which researchers would record various vocalizations, then play them back to individuals of the same species as those they recorded and document what ensued. Based on behavioral reactions of numerous subjects, scientists could then classify a recording as, for example, aggressive or affiliative – for example, as being used, respectively, to defend territory or attract a mate. Clearly these findings were but a first step, somewhat akin to a visitor to Earth playing two recordings outside a theater, one stating “Free beer!” and the other “FIRE!”, and seeing how humans reacted. Current research has, however, provided a better understanding of nonhuman systems, in part because of a number of technological breakthroughs and the availability of extensive computational power. Of particular interest were findings that, of all creatures studied, birds were likely to have some of the most complicated systems because, unlike nonhuman primates, they were vocal learners (Marler 1967). That is, for avian species known as ocsines, communication systems were not innately specified, but are acquired, much as for humans, via learning from older, competent models (e.g., Okanoya 2017; Todt et al. 1979). Depending upon the species, the numbers of songs in a repertoire can vary from one to thousands, and the ways in which such songs convey meaning usually differ across species. Even songbirds that do not learn their vocalizations (the suboscines) vary how they used their songs in order to convey numerous different meanings. Given that our expectation is that the extraterrestrial signals we receive will likely be converted to auditory frequencies for human benefit (e.g., as were the Laser Interferometer Gravitational-Wave Observatory [LIGO] signals of two black holes colliding), and that such transposition will – if performed properly – maintain the basic characteristics of the original signal, birdsong might indeed be an exceptionally good model for study. Notably, after more than sixty years of research, we are still learning about birdsong. Having found so many differences among species in a single taxa on Earth, we must be open to an immense number of possibilities when examining signals from beyond Earth. For starters, let us examine possibilities presented by the avian mode of communication. The first concerns basic attributes of the signal itself. For example, some birdsong (e.g., hummingbird vocalizations: Olson et al. 2018; Pytte et al. 2004) occurs at the very limits of human auditory thresholds – and, if we go beyond birds, to bats, cetaceans, and elephants (see, respectively, Prat et al. 2016; Finneran et al. 2008; McComb and Reby 2009) – we thus must examine frequencies well above and below those that humans hear. Avian signals may also be visual (e.g., the flash

How Studies of Communication Can Inform SETI

53

of a redwing blackbird’s epaulets, Peek 1972), so we must be open to any sort of incoming energetically interesting signals. We also might expect that the signal (in whatever modality it is delivered) will be repetitive – the equivalent of “Hello . . . are you there?” sent over and over (see, however, Harbour, this volume [Chapter 18], for many other intriguing possibilities). We would have little problem deciphering a signal so semantically and syntactically similar to human languages; however, aliens’ ideas about everything will likely be quite different from those of humans, and thus their communication systems for conveying those ideas will also likely differ from those of humans. We must, for example, be open to the possibility that the number of repetitions of a signal (or possible combinations of different types of signals) rather than the individual units are meaningful: Templeton et al. (2005) demonstrated that the number of “dee” syllables in black-capped chickadee alarm calls indicate the extent and severity of the threat; Suzuki et al. (2016) show that for the Japanese great tit, “dee” calls when produced alone are used to attract a caller. Researchers like Smith (1988) found that combinations of flights and song repetitions indicate different meaning such as various levels of aggression for flycatchers. Thus, unlike human words that themselves have meaning, might specific patterns of similar signals have different meanings in extraterrestrial broadcasts? Interestingly, Balaban (1988) found that although swamp sparrow song consists primarily of repetitions of three-note syllables wherever the singer lives, the order of the syllables specifies the geographic location of the singer – that is, constitutes a dialect, not the song of a different species. Clearly such possibilities must be taken into account in our interpretation of incoming information. Humans might also miss meaning contained in the fine structure of signals. Possibly what initially appears to be a pure tone or a simple sine wave might have additional levels of complexity. Several studies showed that birdsongs that appeared to consist of just a few notes, when played back at much reduced speeds, contained a huge wealth of additional information. For example, Dooling and Prior (2017) demonstrated that zebra finches are extremely sensitive to temporal fine structure, although these authors do not examine what types of information may be encoded in this fine-structure variation. Vélez et al. (2015) revealed seasonal plasticity in the fine structure of the songs of three different songbirds; would such variation exist unless it designates something important? What might we miss if we failed to actively search for such information in incoming signals? Although sometimes different songs, somewhat like different human sentences, are used to convey different meanings (e.g., warblers’ use of one song type for territorial defense and another for mate attraction, Demko et al. 2013), it is possible that a particular vocal pattern is not important, but rather its contextual use. Song sparrows, for example, have repertoires of about five to seven songs, used predominantly for territorial defense. Different songs themselves do not have separate meanings. The meaning of a bird’s choice of song, however, depends on the one that its neighbor has sung, and the physical origin of the heard song – that is, the

54

Irene M. Pepperberg

spot from which the neighbor sings in their respective territories. Stoddard (1996) demonstrated that the same song sung on the edge of a territory will evoke a completely different response – and neighbor’s song choice – from one sung during an intrusion just over the territorial border. Beecher et al. (2000) refined this further, showing that a sparrow will exactly match the song of a neighbor that is singing at or past the territorial boundary, using the song to present a strong vocal challenge; in contrast, the sparrow will choose to sing a different song – one merely from the repertoire of a neighbor that is singing deep within its own territory, thereby acknowledging the presence of the singer as a familiar, non-threatening entity. For some species such as nightingales, it is not song choice but the temporal pattern of overlapping that acts as a threat (Hultsch and Todt 1982). Marsh wrens seem to play a kind of poker, in which bird A not only matches the song of bird B whom he is trying to dominate, but then acts to “raise him one” by singing the next song in bird B’s repertoire, which is not the next song in his own (Kroodsma 1979). For other species, repertoire size might indicate fitness, so that the point is to produce an extremely large number of different signals (reviewed in Todt and Naguib 2000). Would we be smart enough to recognize such patterns in extraterrestrial signals in our attempts to respond appropriately? The preceding discussions do not examine issues of any possible direct parallels between what could be syntactic arrangements in avian song elements and human language, which is currently a matter of debate (e.g., see Buckingham and Christman 2010; Lipkind et al. 2013). My point is not to emphasize the similarities between birdsong and human language (of which there are many) but some of the numerous differences, so that we are aware of the many various possible forms that communicative structures can take. Such knowledge would be crucial if we are to make sense of alien signaling systems. Learning to Communicate

Although the likelihood of an actual physical encounter between humans and alien species is far less than that of our having to decode their broadcast signals, the possibility is non-zero, and if such is the case, we would need to attempt some form of direct two-way communication. Such beings, having figured out how to get to us, would likely be far more advanced than humans, and thus would probably have already developed protocols to try to learn our communication code, or teach us theirs. Having our own procedures in place to instruct such visitors would be prudent, but lacking any information as to the most appropriate modality (auditory, visual, tactile, olfactory, or some form we cannot yet even imagine) puts us at a disadvantage. A likely basis for such procedures might, however, be methods developed by researchers who have already taught nonhumans some elements of human communication systems – those of us who have worked with nonhuman primates, cetaceans, and parrots. Given that communication systems are by definition social in nature, the expectation is that some form of social learning system would be

How Studies of Communication Can Inform SETI

55

most appropriate. Researchers actually tasked to establish human-ET communication will likely develop a system based on some combination of forms proposed by me and other authors in this volume. To make such material available, I describe in detail procedures used with my grey parrots. The techniques are based on social modeling theory, which has its origins in the work of researchers such as Piaget (1952), Vygotsky (1962), Todt (1975), and Bandura (1977). Space does not permit a full discussion of their findings, but the principles of the theory involve providing input that is optimal – one that is tailored to the learners’ level of competence and adjusted as competence increases (for those interested, see summary in Pepperberg 1999). Social modeling theory provides a framework for devising training procedures, but this framework must be put into practice. To whit: what factors characterize optimal input? Based on ideas of Piaget and Vygotsky, and competence issues, optimal input should (1) correlate well with specific aspects of an individual’s environment (i.e., be “referential”); (2) have functional meaning relevant to the individual’s environment (also known as “contextual applicability”); and (3) be socially interactive. I briefly described these factors, quoted from Pepperberg (1999). Reference – Reference is, in general, what signals “are about” (Smith 1991). Reference concerns the direct relationship between a signal and an object or action. Reference is not always easily determined. For example, “ape” can refer to a subset of nonhuman primates, but may also mean an action, as to “ape,” or imitate, a behavior. Similarly, a bird’s alarm call may refer to either the predator nearby, the action the bird is about to take, or both. Thus, not all information contained in a signal involves a single referent. The more explicit the signal’s referent, however, the more easily the signal can be learned. Functionality – Functionality (contextual applicability) involves pragmatics of signal use: when a signal is to be used and the effects of using information in the signal. Explicit demonstration of functionality shows when using a signal is advantageous and the specific advantage gained by its use. The way a signal is used and its effect on recipients may depend upon environmental context – the phrase “My, don’t we look nice today?” has one meaning and effect for a little girl in a party dress and a different meaning and effect for a hungover friend. Functionality also helps define reference; that is, context defines “ape” as noun versus verb. The more explicit a signal’s functionality, the more readily the signal can be learned. Social interaction – Social interaction has three major functions that can be clarified by examples. First, social interaction can highlight which environmental components should be noted; a subject can be directed to an object’s color to learn color labels (“Look at the blocks. The color of this one is blue; the color of that one is green”). Second, social interaction can emphasize common attributes – and thus possible underlying rules – of diverse actions (i.e., “Give me the ball” versus “Take the block”). Third, social interaction allows input to be

56

Irene M. Pepperberg

continuously adjusted to match the receiver’s level (“Yes, you found a block among these toys! Now can you find the green block?”). Interaction can provide a contextual explanation for an action and demonstrate its consequences (“I don’t know what toy you want . . . do you want the ball or the block? Tell me what you want, and you can have it.”). Interactive input thus facilitates learning. In sum, reference and functionality refer to real world use of input, social interaction highlights various components of input, and all are necessary for meaningful learning. I thus reasoned that to teach a bird to communicate with humans, my training procedure needed to take these factors into account. The critical point, however, was my hypothesis that a parrot’s acquisition of a human-based code was a form of exceptional learning: I believed that despite these birds’ abilities to reproduce all sorts of sounds, some strong inhibition existed toward learning to use allospecific sounds in a functional manner; I further believed that, to overcome this inhibition, training would have to be carefully adjusted to the parrot’s abilities and include intense interactions and extremely clear demonstrations of reference and functionality. I decided that the best approach would be to modify a procedure used by Todt (1975), called the Model/Rival or M/R technique. He had demonstrated the effectiveness of social interaction for training parrots to produce human speech in simple dialogues without requiring them to understand the meaning of these dialogues; what if I adapted his method to incorporate referentiality and functionality in order to establish communication? My training system, because of its similarity to Todt’s, is also called the M/R technique. In my procedure, however, an interaction is not only modeled; it also involves three-way interactions among two human speakers and the avian student. I provide details of the procedure, although the material is available elsewhere (e.g., Pepperberg 1981). During M/R training, humans demonstrate to the bird the types of interactive responses that are to be learned. In a typical interaction, the bird is on a perch, cage, or the back of a chair, and observes two humans handling an object in which it has already demonstrated interest (e.g., as a preening implement). While the bird watches, one human “trains” the second human. The trainer presents an object, asks questions about the object (e.g., “What’s here?”, “What color?”, “What shape?”), and gives praise and the object itself as a reward for a correct answer – thereby showing the direct connection between the label and the item to which it refers. Unlike Todt’s procedure, our technique demonstrates referential and contextual use of labels for observable objects, qualifiers, quantifiers, and, on occasion, actions. As in Todt’s procedure, the second human is a model for the bird’s responses and a rival for the trainer’s attention. The model/rival occasionally errs (i.e., produces garbled utterances, partial identifications, etc., that are similar to mistakes being made by the bird at the time). Disapproval for an incorrect response is demonstrated by scolding and temporarily removing the object from sight. Because the human model/rival is, however, encouraged to try again or talk more clearly (e.g.,

How Studies of Communication Can Inform SETI

57

“You’re close; say better”), the procedure also allows the bird to observe “corrective feedback,” which also assists acquisition (see Goldstein 1984; Vanayan et al. 1985; Moerk 1994). As part of the demonstration of functionality and relevance, the M/R protocol also repeats the interaction while reversing roles of trainer and model/rival, and occasionally includes the bird in interactions. Thus, unlike Todt’s subjects (see also Goldstein 1984), our birds do not simply hear stepwise vocal duets, but rather observe a communicative process that involves reciprocity. We show that interaction is indeed a “two-way street” in that one person is not always the questioner and the other always the respondent, and how the process can be used to effect environment change. I surmised, based on Bandura’s (1977) studies of interactive modeling, that Todt’s failure to demonstrate role reversal between trainer and model/rival explained why his birds could not transfer their responses to anyone other than the particular human who posed the questions, and why his birds never learned both parts of the interaction. In contrast, my birds respond to, interact with, and learn from all their trainers. Three actions by trainers insure that our birds indeed attend to sessions (Pepperberg 1992); these actions are consistent with the principles of social modeling theory. First, trainers adjust the level of modeling to match a bird’s current capacities. If, for example, a label being trained (“wool”) resembles one already in the repertoire (“wood”), trainers praise but do not reward the bird’s likely initial use of the existent label and clearly demonstrate how the two labels differ with respect to both sound and referent. Trainers then adjust their rewards as a bird practices its utterances to challenge it to achieve correct pronunciation. Second, a bird must be interested in obtaining the items used in training. Trainers working on a numerical task who choose, for example, corks rather than keys are more likely engage one bird’s attention, whereas another bird might prefer keys. Third, trainers must act as though they themselves find the task interesting. A bird is less likely to ignore the session and begin to preen if the emotional content of the trainers’ interactions suggests there is real relevance to the task, and trainers who actively engage the bird in the task are more likely to be successful. The M/R technique is the primary method for introducing new labels and concepts and for shaping correct pronunciation, but another procedure helps clarify pronunciation (Pepperberg 1981). Because this technique does encourage some imitation, it is used only after a bird begins to attempt a new label in the presence of a new object (i.e., after a bird makes some connection between sound and object). We present the new object along with a string of “sentence frames” – phrases like “Here’s paper!”, “Such a big piece of paper!” The target label, “paper,” is consistently stressed and the one most frequently heard, but not as a single, repetitive utterance. The target label is also consistently at the end of the phrase; conceivably parrots, like humans (e.g., Lenneberg 1971), most easily remember ends of word strings. This combination of nonidentical but consistent vocal repetition and physical presentation of object resembles parental behavior for introducing labels for new

58

Irene M. Pepperberg

items to very young children (Berko-Gleason 1977; de Villiers and de Villiers 1978), and appears to have two effects: a bird (1) hears the label employed as it is to be used in normal, productive speech; and (2) learns to reproduce the emphasized, targeted label without associating word-for-word imitation of its trainers with reward. On occasion, birds experiment with labels in their repertoire and produce novel vocalizations. To encourage such recombinations of – or variants upon – parts of labels and enlarge the referential repertoire, we reward these utterances (when possible) with appropriate objects and use a variant of the M/R technique to associate the novel vocalizations and objects. The technique, called “referential mapping,” is described in detail in Pepperberg (1990). Thus, by integrating Bandura’s social modeling theory with the ideas and results of scientists as diverse as Piaget, Vygotsky, and Todt, I devised training procedures for teaching a parrot to communicate with humans. Whether such procedures could be adapted for use with creatures far more “alien” than a grey parrot remains to be seen. Clearly, extraterrestrials, unlike parrots, likely will not share similarities with us in auditory and visual systems, may not respond to the same forms of social interaction (e.g., like nonhuman primates, may see direct gaze as a threat rather than a socially acceptable form of interaction), and thus “optimal input” for an Earth-dwelling creature maybe completely suboptimal for ET. However, I propose these procedures as a possible basis, not as a rigid protocol, for the establishment of interspecies communication. Conclusions

From the material discussed here and in many other chapters, it is clear that organisms do not need all the elements of human language to establish some level of inter- or intraspecific communication, and that communication systems may involve aspects completely foreign to human language. Granted, human language enables levels of communication that have so far not been found to exist in nonhumans, but one can argue that humans have not been clever enough to unearth all possible complexities inherent in nonhuman systems. We nevertheless can understand the gist of nonhuman systems; we have taught them elements of ours. Such knowledge could be used to establish rudimentary interplanetary dialogues. Another possibility is that extraterrestrial communication systems have layers of complexity far beyond those of human systems. That is, might our system seem lacking in many essential components when compared to theirs? Might we find ourselves struggling to learn a system far more advanced than ours? Will we have the humility to accept instruction if necessary? Only by being open to all possibilities and eventualities might we have some chance of success. References Balaban, Evan. 1988. “Bird song syntax: Learned intraspecific variation is meaningful.” Proceedings of the National Academy of Sciences, USA 85: 3657–3660.

How Studies of Communication Can Inform SETI

59

Bandura, Albert. 1977. Social Modeling Theory. Chicago: Aldine-Atherton. Beecher, Michael D., Elizabeth Campbell, John M. Burt, Christopher E. Hill, and J. Cully Nordby. 2000. “Song-type matching between neighboring sparrows.” Animal Behaviour 59: 21–27. Berko-Gleason, Jean. 1977. “Talking to children: Some notes on feedback.” In Talking to Children; edited by Catherine E. Snow and Charles A. Ferguson, 199-205. Cambridge: Cambridge University Press. Buckingham, Hugh W., and Sarah S. Christman. 2010. “Charles Darwin and the evolution of human grammatical systems.” Journal of the History of Neuroscience 19: 121–139. Catchpole, Clive K., and Peter J.B. Slater. 2008. Bird Song: Biological Themes and Variation. Cambridge: Cambridge University Press. Cheney, Dorothy L., and Robert M. Seyfarth. 1990. How Monkeys See the World: Inside the Mind of Another Species. Chicago: University of Chicago Press. Demko, Alana D., Leonard R. Retsma, and Cynthia A. Staicer. 2013. “Two song categories in the Canada warbler (Cardellina canadensis).” Auk 130: 609–616. de Villiers, Jill G., and Peter A. de Villiers. 1978. Language Acquisition. Cambridge, MA: Harvard University Press. Dooling, Robert J., and Nora H. Prior. 2017. “Do we hear what birds hear in birdsong?” Animal Behaviour 124: 283–289. Ferrer-i-Cancho, Ramon, and Brenda McCowan. 2009. “A law of word meaning in dolphin whistle types” Entropy 11: 688–701. Finneran, James J., Dorian S. Houser, Dave Blasko, Christie Hicks, Jim Hudson, and Mike Osborn. 2008. “Estimating bottlenose dolphin (Tursiops truncates) auditory thresholds from single and multiple evoked potentials.” Journal of the Acoustical Society of America 123: 542–551. Goldstein, Howard. 1984. “The effects of modeling and corrected practice on generative language and learning of preschool children.” Journal of Speech and Hearing Disorders 49: 389-398. Goodall, Jane. 1986. The Chimpanzees of Gombe: Patterns of Behavior. Cambridge, MA: Harvard. Hultsch, Henrike, and Dietmar Todt. 1982. “Temporal performance roles during vocal interactions in nightingales (Luscinia megarhynchos).” Behavioural Ecology & Sociobiology 11: 253–260. King-Smith, Dick. 1984. Harry=s Mad. London: Victor Gollantz Ltd. Kroodsma, Donald E. 1979. “Vocal dueling among male marsh wrens: Evidence for ritualized expression of dominance/subordinance.” Auk 96: 506–515. Kroodsma, Donald E., and Edward H. Miller, eds. 1996. Ecology and Evolution of Acoustic Communication in Birds. Ithaca, NY: Cornell University Press. Laporte, Marion N.C., and Klaus Zuberbühler. 2010. “Vocal greeting behavior in female chimpanzees.” Animal Behaviour 80: 467–473. Lenneberg, Eric H. 1971. “Of language, knowledge, apes, and brains.” Journal of Psycholinguistic Research 1: 1-29. Lipkind, Dina, Gary F. Marcus, Douglas K. Bemis, et al. 2013. “Stepwise acquisition of vocal combinatory capacity in songbirds and human infants.” Nature 498: 104–109. Lofting, Hugh. 1948. The Voyages of Dr. Doolittle. Philadelphia: Lippincott. Lorenz, Konrad. 1952. King Solomon’s Ring. Translated by M.K. Wilson. New York: Harper & Row. Marler, Peter. 1967. “Comparative study of song development in sparrows.” Proceedings of the XIVth International Ornithological Congress, 231–244. Blackwell, Oxford.

60

Irene M. Pepperberg

McComb, Karen, and David Reby. 2009. “Communication in terrestrial animals.” In Encyclopedia of Neuroscience, Vol. 2, edited by Larry R. Squire, 1167–1171. Oxford: Academic Press. McCowan, Brenda, and Diana L. Reiss. 1997. “Vocal learning in captive bottlenose dolphins: A comparison with human and nonhuman animals.” In Social Influences on Vocal Development, edited by Charles T. Snowdon and Martine Hausberger, 178–207. Cambridge: Cambridge University Press. Moerk, Ernst L. 1994. “Corrections in first language acquisition: Theoretical controversies and factual evidence.” International Journal of Psycholinguistics 10: 33–58. Okanoya, Kazuo. 2017. “The biopsychology and development of birdsong.” In APA Handbook of Comparative Psychology, edited by Josep Call, Gordon M. Burghardt, Irene M. Pepperberg, Charles T. Snowdon, and Thomas R. Zentall, 539–555. Olson, Christopher R., Marcella Fernández-Vargas, Christine V. Portfors, and Claudio V. Mello. 2018. “Black Jacobin hummingbirds vocalize above the known hearing range of birds.” Current Biology 28: R187–R207. Peek, Frank W. 1972. “An experimental study of the territorial function of vocal and visual displays in the male Red-winged Blackbird (Agelaius phoeniceus).” Animal Behaviour 20: 112-118. Pepperberg, Irene M. 1981. “Functional vocalizations by an African Grey parrot (Psittacus erithacus).” Zeitschrift für Tierpsychologie 55: 139-160. Pepperberg, Irene M. 1990. “Referential mapping: A technique for attaching functional significance to the innovative utterances of an African Grey parrot.” Applied Psycholinguistics 11: 23-44. Pepperberg, Irene M. 1992. “Social interaction as a condition for learning in avian species.” In The Inevitable Bond, edited by Hank Davis and Dianne Balfour, 178–204. Cambridge: Cambridge University Press. Pepperberg, Irene M. 1999. The Alex Studies. Cambridge, MA: Harvard University Press. Piaget, Jean. 1952. The Origins of Intelligence in Children. Translated by Margaret Cook. New York: International Universities Press. (Original written in French in 1936). Prat, Yosef, Mor Taub, and Yossi Yovel. 2016. “Everyday bat vocalizations contain information about emitter, addressee, context, and behavior.” Scientific Reports 6: 39419. Pytte, Carolyn L., Millicent S. Ficken, and Andrew Moiseff. 2004. “Ultrasonic singing by the blue-throated hummingbird: A comparison between production and perception.” Journal of Comparative Physiology A 190: 665–673. Rasmussen, Knud. 1972. The Netsilik Eskimos. In Shaking the Pumpkin, edited by J. Rothenberg, 45. Garden City, NY: Doubleday. Smith, W. John. 1988. “Patterned daytime singing of the eastern wood-pewee (Contopus virens).” Animal Behaviour 36: 1111-1123. Smith, W. John. 1991. “Animal communication and the study of cognition.” In Cognitive Ethology: The Minds of Other Animals, edited by Carolyn A. Ristau, 209-230. Hillsdale, NJ: Erlbaum. Stoddard, Philip. 1996. “Vocal recognition of neighbors by territorial passerines.” In Ecology and Evolution of Acoustic Communication in Birds, edited by Donald E. Kroodsma and Edward. L. Miller, 356–374. Ithaca: Cornell University Press. Suzuki, Toshitaka N., David Wheatcroft, and Michael Griesser. 2016. “Experimental evidence for compositional syntax in bird calls.” Nature Communications 7: 10986. Templeton, Christopher N., Erick Greene, and Kate Davis. 2005. “Allometry of alarm calls: Blackcapped chickadees encode information about predator size.” Science 308: 1934–1937.

How Studies of Communication Can Inform SETI

61

Todt, Dietmar. 1975. “Social learning of vocal patterns and modes of their applications in Grey parrots.” Zeitschrift für Tierpsychologie 39: 178-188. Todt, Deitmar, Henrike Hultsch, and Dietmar Heike. 1979. “Conditions affecting song acquisition in nightingales.” Zeitschrift für Tierpsychologie 51: 23-35. Todt, Dietmar, and Mark Naguib. 2000. “Vocal interactions in birds: The use of song as a model in communication.” Advances in the Study of Behavior 29: 247–296. Vanayan, Marina, Heather A. Robertson, and Gerald B. Biederman. 1985. “Observational learning in pigeons: The effects of model proficiency on observer performance.” Journal of General Psychology 112: 349-357. Vélez, Alejandro, Megan D. Gall, and Jeffrey R. Lucas. 2015. “Seasonal plasticity in auditory processing of the envelope and temporal fine structure of sounds in three songbirds.” Animal Behaviour 103: 53–63. Vygotsky, Lev S. 1962. Thought and Language. Cambridge, MA: MIT Press.

7 PATTERNS OF COMMUNICATION OF HUMAN COMPLEX SOCIETIES AS A BLUEPRINT FOR ALIEN COMMUNICATION Anamaria Berea

Introduction

In this chapter, I explore patterns of ubiquitous communication as a sign of successful communication evolution that started local and became global. These patterns are the framework within which an alien language can develop and would bring us to a better understanding of what alien languages might look like. The common ways in which we communicate today, whether language-based or symbol-based or sounds-based – or any other forms – are essentially patterns of communication that have emerged due to the coevolution of our languages, cultures, and technologically inclined civilization. A great example is the growth of the English language (and not only English; Latin is another historical example) with words such as “to google” or “meme” in recent years, while in the digital realm the growth of the menu of emoticon symbols has been quite astonishing during the past couple of decades, concurrent with a specific technological evolution that is very recent in our history. But at a larger, global point of view, what I am arguing for is that the globalization of certain systems of communication which transcend boundaries and cultures, such as the emergent ones I exemplified or the designed ones, such as those used in transportation (e.g., aviation or maritime symbols, traffic lights), are – in effect – great representations of how language coevolves with our complex society. Therefore, this is an invitation to think about xenolanguages as emergent and coevolving phenomena that can be thought of only within the context of the social and technological alien civilization, as well. While thinking about xenolinguistics in this framework might complicate things further, it also helps us consider which coevolutionary patterns have endured the passing of time, civilization rises and falls, and crossed cultural borders, and DOI: 10.4324/9781003352174-7

Patterns of Communication of Human Complex Societies

63

therefore transcended the context of time and space, becoming more and more generalizable and potentially universal. Besides this coevolutionary aspect between languages and the complexity of societies, I am also exploring a methodological reversal: Can the laws that describe social and collective behavior function as proxies for universal language patterns, as well? Can we extrapolate from empirical laws and statistical regularities of the social and collective phenomena features specific to the evolution of language, as well? These are undoubtedly interesting questions, and answering them will help us get an even better understanding with respect to what we can expect from an alien language. Universals of Human Languages are Coevolving With Technology and Society

My argument is that intelligent communication is a coevolutionary process in which innovation in artifactual communication (i.e., writing, the telegraph, the Internet, etc.) and types of collective behavior that represents a civilization (i.e., infrastructure systems) intertwine and produce large patterns of communication, such as transportation-system signals, brand logos (symbols of global organizations that transcend specific human languages), universal computer languages, universal data-storage systems, data encoding and analyses, and many more. At the same time, our actual languages incorporate these changes and become more symbolic and more global (Finn 2017). Words related to technology are more likely to be universal across languages (Chun et al. 2016). We can prepare to understand the language of another intelligence by using the best technological tools that our science and socio-technological advancements provide. Technology is very likely to be a “universal” in the cosmos (Cohen and Stewart 2002), meaning that there is a high probability that if another intelligence willing to make contact exists, they will have technology and that their language is technologically based. Some of these points are also addressed in more depth by Ross in this volume (Chapter 12). The current age of data and information in the history of our civilization is the result of two major forces: the global spread of artifactual means of communication (i.e., the commonality of human-computer interactions, the globalization of the English language) and the global spread of physical systems of recording and storing communication (if we think about the long evolution from Mesopotamian cuneiform tablets to current data-server farms and supercomputers). But while information as means of communication is global, its meaning is still local. In case of alien detection, we would similarly expect their language to have coevolved with the social and technological evolution of that particular civilization and, while the meaning of its information can still be highly contextual or parochial, the means of communication might have universal features that we can connect with and understand. A great such example is the fact that a German person who speaks

64

Anamaria Berea

only German can use the same computer as an English person who can speak only English. On Earth, technology has evolved as an extension of our intelligence (Hubbard 2015), while remaining constrained by our intelligence. There are socio-economic evolutionary factors that triggered or constrained technological innovation (Arthur 2014), but at the same time technological innovation depends on natural laws as well as on our discovery and understanding of these laws through science (McClellan and Horn 2015). Therefore, it is likely that the technological advancement of another civilization will depend both on their understanding of the universe (their science), as well as on independent constraints that pertain to the evolution of their own social and collective behavior (for example, a single, non-social entity would pose an entirely different set of questions – could such an entity even need language?). This is why I believe that current technologies – particularly in computing and information theory (artificial intelligence, cellular automata, quantum computing, etc.) – are on track to give us a deeper understanding about universals in technologically based communication, or at least an understanding of the universal laws of information theory (i.e., Shannon boundary, the physical limits of information, etc.). At the same time, a deeper understanding of what is universal in the coevolution of our collective behavior with the current technological and scientific discoveries (i.e., information-based technologies seem to be common from the Antikythera mechanism to modern computing) versus what is contextual (i.e., technologies that did not survive after a civilization crashed or disappeared, such as mechanical technologies of ancient agriculture) will help us narrow down the framework for xenolinguistics. In other words, an understanding of the evolution of technology in our history, with an emphasis on what transcended cultures and histories versus what did not, would help us frame the technology advancements in terms of universals versus contextual/parochials. We know that language has largely evolved organically and endogenously, and with few exceptions (scholarly rules of linguistics imposed by the academies in a few European countries), language is essentially an informational representation of the society and culture where it was used (Goodenough 1981). Before digital means of communication were invented, the economies of the world were localized, embedded in the communities and the cultures of their specific societies (salt in North Africa, silk in China, wine in France, tourism in Italy, etc.) and with large variations in geography (Castells 1997). The embeddedness of language in culture is also addressed in depth by Granger et al. in this volume (Chapter 8). On another hand, we have designed informational systems that are universal to some degree (e.g., mathematics, computer languages), but devoid of any meaning or content, and therefore used only in small scientific and high-tech communities (Berea 2018). One organic language that has become a standout among others due to trade and the proto-globalization phase in our civilization is English, while in the digital age, various computer languages are still competing for global hegemony

Patterns of Communication of Human Complex Societies

65

and this effect may largely be due to the fact that computers cannot choose the language they operate with (yet) or that computational challenges are evolving much more rapidly than human skills for designing such languages (Berea 2018). Another way to think about the language/civilization coevolution is to look at the effect of different informational structures (networks vs. symbols vs. images vs. text vs. probabilistic vs. Boolean vs. sound vs. any other form of information representation) into the evolution and emergence of communication patterns, particularly linguistic ones, and to create a taxonomy of languages that includes both the organically emerged and the designed communication (as well as any mix in between) as a better representation of our civilization as a whole. In other words, we can look deeper into how information representations are being used today and classify them so that they are representative of all the complexities of a language of an advanced civilization. These classes of information representation are easier to generalize into patterns of language and communication in the context of a globalized civilization (Raber and Budd 2003). Globalized languages and large systems of communication are products of selection and adaptation mechanisms that also emerged from the economic systems of the growth of a civilization (McCann 2008). This seems intuitive. At the large scale of our modern civilization on Earth, the planetary systems of interconnectivity for transportation and communication show some of the artifactual, global patterns of language of the largest magnitude we know to date. But data is global only as a communication means: its meaning is local, as previously mentioned. Examples of systems that use truly global means and meanings of communication are transportation infrastructures, such as aviation, maritime, and financial systems, or Internet protocols. Therefore, these transportation systems and their embedded communication are products of selection and adaptation mechanisms emerged from the economic systems of our civilization. At the large scale of the civilization, these planetary systems of interconnectivity show artifactual, global patterns of signals/communication of the largest magnitude known to date. Ever since cuneiform writing began in Mesopotamia to record bookkeeping for crops and animals, thus with an economic functionality, the evolution of our civilization’s economy and the emergence of artifactual communication have been intertwined and influencing each other. This spurs another question we should address: To what degree can the currently designed communication such as computer languages evolve organically (i.e., bots inventing their own language) or become organic, endogenously evolving – and to what degree does any type of organic communication become standardized (English words imports into other languages, Internet memes, etc.)? Similarly to the classifications and taxonomies described here, a guiding map of past and current communication patterns would help us understand the evolution of communication in the near future. Will we invent a new type of human–computer language that would change the current software language paradigm, which has standardized for now the way humans interact digitally?

66

Anamaria Berea

Language (in the broadest sense) has deep effects on the selection and adaptation of a species, whether this species is ants, dolphins, or humans, and the more social a species is, the more important language is for its survival (Pinker 2003). Even in non-human species on Earth, communication and “language” between species has evolved deeply intertwined with the evolution of collective behavior, networks, and communities of these species (McGregor 2005). But in the case of a civilization in which the biological evolution standards of selection and adaptation are more elusive and not yet fully understood (we don’t really know why we create art or why we are creating technology that makes us more comfortable and less fit for survival – whether everything we do as a species has a biological evolutionary component), while the social evolution standards are more acute (we compete not only for physical resources but also for networks of influence and power, whether these networks are finance, firms, institutions, or friends), the language patterns that will give more fitness to networks of influence would be the ones that take advantage of both the organic and the designed patterns of communication. More than languages, which deeply depend on localized communities, art and symbols have evolved organically as efficient ways of global communication before we had emoticons or social media (emotions and visual human behavior are perceived as such in any place on the globe). Therefore, the methodological question is whether we can integrate organic, global ways to store information (qualitative visual data) with designed global ways to communicate (digital, mathematical data) in a way that can become a universal standard for storing, understanding, and communicating data and information. The big question I would like to see answered is this: How will language evolve in the near future? The applied question I would like to see answered is what will be the next unifying standard in digital communication? – is it a blend of visual symbols and data science techniques? Answering these questions is valuable for two reasons: to improve our basic scientific understanding (since we have not really bridged the evolution of signaling in biological systems with the evolution of grammar in human languages and with the evolution of computer language structures), and to provide economic insights, because in the long run, it may advance the next innovations in global technologies or communication while giving us an understanding of the clashes between the physical economies and the informational economies. Such lags or rifts in coevolution (Cabrol 2016) of large systems are important to understand, as they are also the fertile ground for the next innovations in either languages or technologies, in a somewhat similar fashion to biological mutations: a continuous feedback between the physical and the informational aspects of an organism or a system of organisms. Therefore, as there is also a lag between the human use of computer languages and the human use of human languages, and a lag in the physical and informational economy that has the potential to create both societal crises and opportunities for large-scale innovations, societies that will continue to have this physical/ informational duality will also have communication systems that blend organically with the organisms, the environment, and the informational/technological world.

Patterns of Communication of Human Complex Societies

67

Identifable Patterns of Communication for Xenolinguistics

Life’s laws follow the laws of physics, but they have a higher degree of granularity and variation (more “local” laws that fall into the “general” laws – i.e., cellular laws do not violate atomic laws), and civilization’s laws follow life’s laws, with an even higher degree of granularity and variation (even more “local” laws – i.e., macroeconomic convergence does not violate evolution). Therefore, communication – a phenomenon strictly dependent on living organisms – is life-dependent and intelligent communication is civilization-dependent, which means that our best way to find general patterns of communication in the universe is to look for general patterns of communication in life and civilization on Earth. Additionally, we need to understand the limits of variations in the laws of communication within life from those within civilizations. Essentially, this is a data/information science problem that involves cleaning, cataloging, mapping, and analyzing the evidence of all patterns (not observations, which most probably would be impractical) of communication that we know of today, but with an emphasis on “local” versus “global” transitions in communication patterns. Communication is as ubiquitous as information, yet while there is an integrated discipline of information science with theories and applications, there is no integrated discipline of communication science that studies not only “local” patterns and observations of human communication or the broader phenomenon of communication in the living world. Therefore, this chapter is also a rationale for curating and mapping our current but rich collection of communication evidence, scattered throughout multiple disciplines and sciences into a taxonomy as mentioned previously, but with a focus on local/global patterns, on means/meaning of communication, and on natural/artifactual communication. The vast majority of scientific disciplines study communication, from biology (species-specific communication) to linguistics (human languages), to communication theory (context-specific human communication – i.e., parent–children, political and social communication) to business and economics (by organization specific communication). This domain-specific research on communication has given us very rich details about specific, local communication patterns and the role of communication in other phenomena (i.e., predation, mating, immunotherapy, organizational functions, social movements, etc.) (Bradbury and Vehrencamp 1998). This means that we now have many pieces of a big puzzle to help us create a comprehensive map of communication on Earth, based on taxonomy, ontology, phylogeny, and evolution (Berea 2019). Having a broad idea about how communication has emerged and evolved in the living systems on our planet, from cells to firms and societies, would enable us to highlight: 1) the common patterns across species, ecosystems, and civilization trends, and 2) the differences/variations and the boundaries/limits where we know that communication does not and cannot occur. Given the large datasets and big data techniques available now from various research projects and federal agencies, we can harvest and create a database of such

68

Anamaria Berea

evidence once the protocols for storing and realigning the data and meta-data are set. Such a large classification endeavor will help create the protocols for aligning and re-encoding the current evidence in a unified framework for what is possible and what is not possible for a xenolanguage. Following are some examples for focusing classification and categorization efforts. 1 Locality, globalism, and transition phases in communication patterns. A taxonomy would place local patterns of communication of the living systems, from cell-to-cell interaction, to organism-specific interaction, to human languages and human–computer interactions, paying attention to transition phases or boundaries between these phenomenon’s “localities,” within same paradigm or framework of thinking. This paradigm is outlined by viewing communication through the following lenses: natural/artifactual, meaning/means, and local/global. 2 Natural versus artifactual communication. While all living systems on Earth communicate through natural, genetically evolved means (chemical communication in non-vertebrate species, ultrasound in bats, olfactory, auditory and visual in most vertebrate species), only humans have developed artifactual means of communication – writing, drawing, and computational technologies. Kershenbaum’s contribution to this volume (Chapter 2) is dedicated to the understanding of natural/animal forms of communication that can inform us about xenolinguistics. Moreover, humans have developed complex languages that combine natural (sounds) and artifactual (grammar) means of communication. The hallmark of an advanced civilization is its ability to combine the natural and the artifactual means and meaning of communication in a way that reflects selection and adaptation processes both in natural communication and in artifactual one. Pepperberg deepens some of these ideas in this volume (Chapter 6), on human–non-human communication. 3 The “wave-particle” theory of communication: communication as both means and meaning of information exchange. Shannon explored only the “wave” theory of communication – communication as a means of exchanging information, devoid of meaning (Shannon 1949). On another hand, semiotics and cryptography explore the “particle” theory of communication – the meaning of communication devoid of the means of communicating. Therefore, mapping communication from cells to societies in distinct terms of means and meaning of communication would help understand where the two aspects of the phenomenon of communication converge and diverge. Based on the large, core ideas outlined in the preceding list, chronologically and logically, I envision this endeavor of establishing a framework of research for xenolinguistics in the following three parts. 1 Establish the current facts about large-pattern evolution. For example, visual language started only with species that developed eyes or visual perception,

Patterns of Communication of Human Complex Societies

69

artistic language emerged only with species that understood symbolism and interpretation, language was developed due to grammar, and digital communication was developed only after both math and grammar were used, etc. 2 Create a model for categorizing and classifying language structures. Which language patterns are networks of information? How much of language is physical or informational representation? Are they data-rich structures? Are languages structures or processes? Are they symbols or rules? After classifications based on these types of meta-understanding of language in the context of communication and information, we can determine which would be the best model for bridging different types of language representations (i.e., visual and artistic with textual with 0/1 or mathematical representations). 3 Explore the future scenarios. Given such a meta-classifying model as a “good enough” (never perfect) representation of the communication evolution for humans and for computers, we can design scenarios that explore the future of a linguistic standard. Alternatively, given that we have enough data about these language patterns (computational semiotics), we can simply look at the commonalities versus differences in these patterns and assume that what they have in common will persist (i.e., perhaps metaphors and leitmotifs) while the differences will mutate and thus not able to become universal. An inventory of communication standards in art (as they already exist, we would only need to quantify them), in infrastructure systems, and in computer languages might be sufficient to create a dataset of factors and attributes that we can analyze and process. Such a xenolinguistic framework has the potential to succeed if in the process we end up learning something that we did not know before. Alternatively, the project will succeed if we can establish a believable scenario about what the next language our technology and/or civilization will speak. And alternatively again, the project will still succeed if we end up establishing a clear link between the informational economy and the physical economy that can be tracked and understood in terms of crises or growth. The Role of Communication Pathologies for Xenolinguistics

In any scientific and scholarly endeavor, there is another side of the coin when we approach the answer to a question or hypothesis. In this case, the other side of the coin question is: Are there communication pathologies, and what can we learn from them? In general, in science, we aim to explain and understand the what, when, how, who, when, and why of a phenomenon. We are less likely inclined to explain the what does not, when it does not, how it does not, who does not, when not, and why not of the phenomenon – in other words, to put the non-occurrence of the phenomenon in the same epistemological terms of study as the occurrence of it. The payoffs are higher and the search space is usually smaller for explaining a phenomenon than

70

Anamaria Berea

for explaining the not of same phenomenon. In the case of language, though, the search space might be similar in size to the search space for non-language explaining when there is no interspecies communication, why we cannot find meaning in certain messages, why we do not have a unified language, why computers cannot understand metaphors, etc. The rationale for such an approach in a framework of patterns as thus described is that miscommunication is known to trigger cellular pathologies in molecular biology, but also useful mutations, to cause wars in the human civilization, but also to have languages converge and evolve. By the nature of information (information negation creates new information; therefore, it is still information), communication pathologies lead not to the death of communication but to the evolution and transformation of communication. Linguistic pathologies lead to a transformation of language and not the death of language. The framework previously described will essentially help identify the boundaries in interspecies communication, the life cycles of communication, and the transition phases – and from these unrelated local communication failures, we can draw global patterns of communication pathologies as useful as the global patterns of language and communication described previously. Methodologies for Pattern Identifcation

Another way to draw a useful, guiding framework for xenolinguistics is to explore a methodological reversal, as I mentioned in the beginning: Can the laws that describe social and collective behavior function as proxies for universal language patterns, as well? Can we extrapolate from empirical laws and statistical regularities of the social and collective phenomena into the evolution of language? For example, we know that Zipf’s Law applies both to languages and to complex systems, such as cities or companies (Zipf 1949); we know that marginal (cost/benefit) analysis in economics also applies to metabolism and chemical communication in other species (Searcy and Nowicki 2005). What other laws from collective and social behavior of animals, humans, and computers can we infer with respect to their effect on language? For example, the “edge of chaos” model in complex systems behavior shows us where noise becomes signal or signal becomes noise, and that the “edge of chaos” is where both the means and meaning of communication happen (Kauffman and Johnsen 1991). The “edge of chaos” model is both global and local for different complex systems phenomena; therefore, an “edge of chaos” model of communication can potentially give us insights into the transition from local to global communication patterns. Another useful methodological approach would be to create a bridge between semiotics and data science: an epistemological transition from locality to globalism. In data science and machine learning, natural language processing techniques are notoriously bad at identifying meaning in clusters of words and semantic distances (Berea 2018). But they have the power to search for signals through large sets of noise, as long as we identify these signals a priori – as long as we know what

Patterns of Communication of Human Complex Societies

71

to search for. At the same time, computational semiotics techniques are better at identifying signals in small data sets and reconfiguring algorithms based on semantic identifiers. A computational tool that would combine natural language processing with computational semiotics would be able to search through large datasets of noise and adapt the signal search based on dynamic semantic identifiers, thus enhancing the probability of identifying signals globally, not only locally. A third approach involves patterns again. In this case, the patterns of common, widespread communication function as signals of successful communication that survived evolution, started locally and became global, as mentioned in the beginning of this chapter. Classification and cataloging the current evidence of communication from cells to societies into an integrated view can today easily be done using machine learning techniques, and the results would be referential for understanding what can be classified as “common communication” in the living universe, what possible scenarios of communication can be developed with small variations in the living forms, and – most importantly – to prune out the scenarios or possibilities of communication that can never lead to intelligent, decipherable communication. While using patterns, classifications, catalogs and taxonomies are extremely helpful for us to begin to understand the “space” and boundaries for how a xenolanguage could look like, they are also helpful for us to design and explore various alternative scenarios. Particularly in the case of xenolinguistics and alien communication, for which we do not yet have evidence and validation protocols, as we do in other sciences, scenarios and experiments are some of the best suited methodologies we can employ given we can use a robust database of patterns, identified as explained in this chapter. Following are a few examples of such scenarios based on patterning: 1 Scenarios of “relaxing”/varying the assumptions from the bottom up. Given that we have a good map or guide about how our civilization actually communicates globally, we can explore potential scenarios whereby we relax one or more assumptions from this larger picture, such as: What if cells on a different world do not communicate chemically? Would that imply different forms of organization between higher organized intelligent species? What if organisms do not develop sight or visual communication and cannot be recorded? How would a civilization without visual records communicate? Does intelligent communication always depend on social or collective forms of behavior, or can there be entities that do not connect with peers – that do not have a local, bounded communication, but have only a global framework of communication? 2 Scenarios of nullifying assumptions from the top down. On the null hypothesis side, does a civilization need an economic system in order to achieve global systems of commonly widespread, artifactual communication? Does a civilization need an economic system in order to emerge general – not contextually localized – communication? Humans have developed a diversity of forms of organizing their social or economic or political behavior that cannot be reproduced

72

Anamaria Berea

or recognized in any other species. But civilization as we know it comprises culture and social behavior that also gives us not only global communication, but also ultra-specialized (large variation) communication in technology, art, or economics that is not widespread and understood across the planet. We cannot find both global and high-variance local communication in other living systems, at the same time. Are these statistical patterns crucial for the development of a civilization and its communication? 3 Scenarios of variation in the natural/artifactual coevolution (Simon 1996). The interplay between the natural and the artifactual worlds through our economic systems have determined, for better or for worse, the current state of our civilization. Therefore, we can design scenarios whereby we can explore questions such as the following. What types of economic systems with coevolving artifactual versus natural forms of organized behavior are the ones that can lead to many types of sustainable civilizations? What types of civilizations that are based on the natural/artifactual or physical/informational duality can develop long-term, sustainable communication? What is the interplay between communication of a species as a collective (whole) and the many types of organizing behavior (natural/artifactual communication topologies)? These are only a few of the many scenarios and possibilities that can be explored by formulating and exploring hypotheses, once a baseline or referential guide and a unified dataset of evidence are in place. But in order to reduce the search space of exploration with many “ifs” into “probables” and “possibles,” a unified referential map would help guide these efforts. Conclusions

In this chapter, I have tried to explore a few ideas with respect to determining a framework for xenolinguistics or for identifying the possibilities and boundaries of a xenolanguage. I am exploring the ideas of language in the context of communication and information that can take many shapes and forms, but that produces potentially universal patterns only in the context of technological survival, globalization, and transition from local to global and from contextual to universal. We now have at hand methods we can use for classifying and categorizing or creating taxonomies of languages and communication with a focus on these large historicalscale processes of coevolution, adaptation, and selection that could help us better understand what can be universal in terms of language and what are the boundaries and contextualities that are specific to our planet and could not be found elsewhere. References Arthur, W.B. 2014. Complexity and the Economy. Oxford: Oxford University Press. Berea, A. 2018. Emergence of Communication in Socio-Biological Networks. Dordrecht: Springer International Publishing.

Patterns of Communication of Human Complex Societies

73

Berea, A. 2019. A Complex Systems Perspective of Communication from Cells to Societies. Norderstedt, Germany: BoD – Books on Demand. Bradbury, J.W., and S.L. Vehrencamp. 1998. Principles of Animal Communication. Sunderland, MA: Sinauer Associates. Cabrol, N.A. 2016. “Alien mindscapes—A perspective on the Search for Extraterrestrial Intelligence.” Astrobiology 16 (9): 661–676. Castells, M. 1997. “An introduction to the information age.” City 2 (7): 6–16. Cohen, J.S., and I. Stewart. 2002. Evolving the Alien: The Science of Extraterrestial Life. London: Ebury Press. Chun, D., R. Kern, and B. Smith. 2016. “Technology in language use, language teaching, and language learning.” The Modern Language Journal 100 (S1): 64–80. Finn, E. 2017. What Algorithms Want: Imagination in the Age of Computing. MIT Press. Goodenough, W.H. 1981. Culture, Language, and Society. San Francisco: Benjamin Cummings. Hubbard, B.M. 2015. Conscious Evolution: Awakening the Power of Our Social Potential. Novato, CA: New World Library. Kauffman, S.A., and S. Johnsen. 1991. “Coevolution to the edge of chaos: Coupled fitness landscapes, poised states, and coevolutionary avalanches.” Journal of Theoretical Biology 149 (4): 467–505. McCann, P. 2008. “Globalization and economic geography: The world is curved, not flat.” Cambridge Journal of Regions, Economy and Society 1 (3): 351–370. McClellan III, J.E., and H. Dorn. 2015. Science and Technology in World History: An Introduction. Baltimore: John Hopkins University Press. McGregor, P.K. ed. 2005. Animal Communication Networks. Cambridge: Cambridge University Press. Pinker, S. 2003. “Language as an adaptation to the cognitive niche.” Studies in the Evolution of Language 3: 16–37. Raber, D., and J.M. Budd. 2003. “Information as sign: Semiotics and information science.” Journal of Documentation 59 (5): 507–522. Searcy, W.A., and S. Nowicki. 2005. The Evolution of Animal Communication: Reliability and Deception in Signaling Systems. Princeton, NJ: Princeton University Press. Simon, Herbert A. 1996. The sciences of the artificial, 3rd ed. Cambridge, MA: MIT Press. Shannon, C.E. 1949. “Communication theory of secrecy systems.” Bell System Technical Journal 28 (4): 656–715. Zipf, G.K. 1949. Human Behaviour and the Principle of Least Effort. Boston: AddisonWesley Publications.

8 INTERSTELLAR COMPETENCE Applications of Linguistics and Communicative and Cultural Competencies to Extraterrestrial Communication Sumayya K.R. Granger, Judd Ethan Ruggill, and Ken S. McAllister

Introduction

Part of what makes xenolinguistics so challenging is that virtually nothing can be known – or even safely assumed – about the non-human elements of first contact. Indeed, a good portion of humankind’s eventual and thrilling initial exchanges with extraterrestrials are likely to be baffling. Foreknowledge of this dilemma is what has given rise to the planet’s current policies on extraterrestrial contact, most (if not all) of which are consistently confined to post-detection/disclosure protocols (e.g., the Rio and San Marino scales); such efforts’ operative premise is “Human beings cannot know what extraterrestrials are going to say and do; however, humans can determine in advance their own range of responses.”1 We agree that there will be many unknowns at first contact, but we contend that the context for that interaction is not necessarily one of them, even considering the non-Homo sapiens side of the encounter. As specialists in intercultural competence and transdisciplinary communication and translation, we have often witnessed the process whereby strangers learn how to communicate with each other: economists and poets, children and adults, animals and humans, management and workers, Nigerians and Germans, progressives and conservatives, academics and office workers, farmers and urbanites. Without question, the fact that human beings are at the center of all but one of these discursive conjunctions means that sense-making within them is a good deal easier than if the interlocutors were, for instance, humans and beings from exoplanet Kepler-62f. That said, there are a number of useful communication features that can usually be assumed to be true in any new encounter among intelligent species.2 This chapter begins by identifying a core set of these assumptions, then builds from them to outline a prolegomenon to alien communication. Along the way, we DOI: 10.4324/9781003352174-8

Interstellar Competence

75

attend to several essential components of the xenolinguistic complex: establishing and maintaining regular contact; the ways in which the constraints of remote contact sketch a roadmap toward (and ethical considerations about) the development of an adaptive process of learning, understanding, and communication; and designing a working plan for bridging pre-physical and physical contact communication scenarios. We begin with a discussion of what tool use says about culture and communication. Every Tool a Symbol

Regardless of the physical stature or interpretive frame of an alien interlocutor, we contend that at first contact, all parties will share at least five developmental features: tool use, symbol use, communication, culture-making, and curiosity.3 The fact that all parties will be tool users is self-evident; to send and receive interstellar communications strongly suggests a culture that is able to design and build complex devices. The very existence of such tools also suggests a culture capable of metonymy – that is, seeing the whole in the part (e.g., the symbol “interstellar transmitter” stands for “planet teeming with intelligent life”). Such complex symbol production and consumption indicates a communicative species – one that, given both its inward- and outward-facing communication practices, must also practice the making of culture. Finally, whatever motives might stand behind any interstellar messaging (e.g., wonder, expansion, catastrophe), such messages imply an intelligence that is both curious and expects surprise. Taken together, these features clarify the opacity of first contact just enough for us to begin piecing together a xenolinguistic approach that is both practical and adaptable – core elements in any successful intercultural exchange.4 For the purpose of context, consider humanity’s brief history of deliberately attempting to send and receive extraterrestrial messages. The primary goal of sending such signals into space, of course, has been to initiate an intercultural exchange. Practically, this means: (1) establishing contact, (2) knowing that contact has occurred, and (3) maintaining contact. It has been more than half a century since humans sent the “MIR, LENIN, SSSR” messages in Morse code to Venus, and there has not yet been a recognizable reply.5 It is also unclear if any of the numerous subsequent interstellar messages humans have sent were ever received, or if they were, what was made of them. Finally, there is the possibility that extraterrestrial beings have attempted to contact Earth (either in response to the aforementioned messages or in their own search for distant neighbours), but that their communications have so far proven undetectable.6 In other words, human attempts at interstellar cultural exchange do not appear to be advancing quickly, even with innovations such as SETI’s collaboration with IBM and its Spark application (a project specifically designed to discover communicative needles in the cosmic haystack).7 Such slow progress, however, is axiomatic for xenolinguistics. The nearest potentially inhabited planets are many light years away, creating a challenging

76

Sumayya K.R. Granger, Judd Ethan Ruggill, and Ken S. McAllister

hyper-asynchronous communication paradigm. Consider, for instance, the “Cosmic Call” message sent to 16 Cyg in 1999.8 Though it has been travelling for almost two decades, it is still more than half a century from its destination. Even if the message were received and replied to immediately, Earthlings would not be aware of achieving Phase 2 of intercultural exchange (“knowing contact has occurred”) until 2138 at the earliest, more than five generations removed from the original sender, Charlie Chafer. For this reason (among others), Seth Shostak’s argument for transmitting “short messages sequentially, but repeatedly” toward “a million star systems once a day” (Shostak 2011: 367) is compelling. Using 1-micron infrared signals set at 1 bit per pulse, such “pings” could daily “convey the equivalent of the Encyclopedia Britannica to each of these systems.” This would not be a lot of information relative to the human timeline, but perhaps it would be enough for an equally advanced species to decode and eventually master the transmission language. While such a process would still require significant patience and commitment to extraterrestrial messaging on the part of humanity, this kind of shotgun approach has the advantage of extending a high number of conversational invitations, as well as maximizing humanity’s chances to have multiple concurrent interstellar exchanges. We emphasize the time scale and the complexity of sending, receiving, and decoding messages because they speak directly to the five shared developmental features among all first-contact partners. Specifically, while there are many hypotheses about what makes the best first contact message – math, science, photographs, or descriptions of humanity’s “exquisite balance of joy and sorrow,” as Douglas A. Vakoch has proposed (Sutton 2015) – most likely the central transfixing element to interstellar receivers of initial outbound communications will be the tool of message transmission itself. As will be the case when humans receive their own first interstellar message, an anomaly in the fabric of deep space signals will be detected first. When a pattern emerges, symbols will be discovered if not decoded. When the symbols become recursive and systematic, communication may be presumed among the senders – and perhaps one day, between senders and receivers. Fantasies of culture will not be far behind, largely devoid of facts and context. Curiosity will drive the exchange forward, confusion will slow it down, and understanding will deepen both the exchange and the emerging relationship. Yet in the beginning, the question that receivers of an alien message will necessarily focus on is not, “Who are these beings?” or even “What are these symbols?” Rather, it will be “What is this signal anomaly, and how do we make sense of it?” Marshall McLuhan’s observation that “the medium is the message” (McLuhan 1964: 7) will never be truer than upon first contact. The Importance of Error Correction

The fact that extraterrestrial linguistic interaction will likely entail extended time frames in the beginning, at least – and possibly multiple, concurrent, and extended

Interstellar Competence

77

time frames if humans establish regular contact with multiple interstellar intelligences – means that confusion and error will have a spacious petri dish within which to grow before any clarification can be received. From an Earthly context, this portends nothing good.9 Given this likelihood, we suggest that xenolinguistics must not only continue coming to terms with the cunctatious nature of its practice (as this volume is doing), but also institutionalize a respect for this nature in its policies and professional practices. To not do so would be to give tacit approval for rushed xenolinguistic activities, the consequences of which could be devastating on a planetary scale. In this section, we consider what the institutionalization of a fastidious xenolinguistics might look like, as well as offer an explanation for why patience – already the watchword for intercultural exchange on Earth – must be doubly mandated for interstellar cultural exchange. To begin, because interstellar messages will not enjoy the benefit of being exchanged in the same physical space, they will face hyper-asynchronicity (at least from the human point of view), and will likely be truly attended to only after the receivers have reverse engineered the communication technology used to encode and send the message. That is why humans need to think strategically about what information can be meaningfully conveyed in any potential first-contact message. As we assess this situation, there are three primary factors to consider: (1) medium, (2) message robustness, and (3) degree of mutual intelligibility. The first of these factors, medium, is simply an admonition to recognize that before the content of a message can be decoded, it must first be discovered, stored, and studied. Light-based signals may seem fast and efficient to most humans, for example, but they will not be detectable by beings who lack the ability to apprehend and process them.10 Because the medium is the message carrier, interstellar communicators are well-advised to deploy media that offer a wide spectrum of communicative formats, from visible and invisible (to humans) light, to sound waves, to physical objects designed to be touched, looked at, and/or listened to. Second, tempting as it will be (and has been) to tell interstellar beings about the human animal and its cultures, discerning a recognizable pattern in such foreign data is likely to be exceptionally difficult. In order to increase the robustness of first-contact communication, therefore, a programme of discoverable mirroring is advisable. In such a programme, signals would be designed specifically to stand out from all other ambient space noise (discoverable), and would mirror information that is already familiar and meaningful to the intended recipient (e.g., orbital patterns or star charts of the recipients’ solar system). By giving first-contact partners something (presumably) easy to recognize, humans will simultaneously create a differential signal (as compared to the “noise” of deep space), convey Earthly knowledge of the partners’ general location, and provide a template for an equally legible reply. The exchange could then proceed along similar lines of comparison and identification, with each modest change signalling an intelligent and attentive interlocutor at the other of end of the pathway. Such an intercultural exchange strategy – part baby talk, part “cocktail party effect,” and part interpellation (or “hail,”

78

Sumayya K.R. Granger, Judd Ethan Ruggill, and Ken S. McAllister

as Louis Althusser [Althusser 2001 [1970]: 86] terms it) – could prove a remarkably efficient way to catalyze a productive (if protracted) conversation. As Daniel Ross (this volume [Chapter 12]) so elegantly concludes, human beings “should prepare to teach as much as to learn,” a point that Pepperberg (this volume [Chapter 6]) also makes and expands upon. Once a minimal form of contact and awareness has been established, including a working knowledge of what Denise L. Herzing (this volume [Chapter 4]) terms “metadata,” an important next objective would be to increase contact quality.11 By working to share – whether by signals or tangible goods – increasingly diverse and meaningful information, the interstellar relationship would grow, as would the channel of communication. That is, we predict that, as among humans, the communication channel itself would gradually transform to maximally accommodate the kinds of messages its users want it to carry (e.g., the telephone). Failure to establish such robust contact will, again as with humans, likely cause the stagnation of the communication channel, effectively arresting the relationship’s development. Detected sequences and patterns from space may well signal intelligence, but robust communication requires that curiosity never be entirely frustrated or sated. Such an unfortunate communicative plateau would be akin to castaways waving to each other from separate islands, aware of each other’s existence but never able to know more. Finally, at each one of these communication building stages, there must be a vigorous effort to discover and correct errors of translation, interpretation, and cordiality.12 Another common practice for Earth-based intercultural exchange that would seemingly transfer well to an interstellar variety is the practice of collaboratively developing a set of ground rules according to which all parties abide. These ground rules might include agreements about which words, topics, actions, and objects are off-limits, as well as practices that are encouraged, such as “When you are offended, make this clear to the offender and permit an opportunity for apology and clarification.” In our experience, it is rarely too late to apologize and make amends, and doing so can often have the effect of strengthening a relationship by building trust. Given the fermentation time between messages in a xenolinguistic context, “rapid” error correction will be imperative. As more significant conversation becomes possible, opportunities to connect and mutually develop will multiply. Concomitantly, establishing a mutually intelligible mode of communication will become increasingly important. In the next section, we explore how this process might evolve, particularly in a scenario wherein extraterrestrial interlocutors are similar enough to humans that consistent intercourse is possible, but from whom humans differ significantly in terms of physical form and communication apparatuses. Along with these guidelines for co-creating a common language given the constraint of divergent (physical) morphology, we also address the compounding complications that hyper-asynchronicity adds to the development of a hybrid language corpus.

Interstellar Competence

79

Mutual Intelligibility: Building a Common Language

In addition to considering interspecies similarity, vigilant mindfulness of – and respect for – difference is essential to xenolinguistics. Humans, for example, are routinely confounded by each others’ messages: “Did he mean what he said . . . or was he kidding?”; “Does she not understand I don’t speak g33k?”; “I can’t even pronounce that language, let alone understand it.” Interpretive gaps grow more profound beyond the species – think of the communication challenges between humans and terrestrial flora, for example.13 Researchers only recently discovered that beech trees are able to communicate to nearby relatives using chemical signals, alerting neighbours that they are becoming malnourished, for instance. Using fungi in their roots, surrounding beeches are then able to alter how soil nutrients are shared, optimizing the health of the entire stand of trees (Wohlleben 2016). Given the challenges of interspecies communication in and among Earth’s biomes, it is reasonable to surmise that interplanetary lifeform communication may well be even more formidable.14 For one thing, the recipients of humans’ first communiques may exist in an entirely different corporeal state (e.g., non-solid), may communicate through sensory organs humans do not possess, or have unfathomable body chemistries. How are humans to guess what kinds of messages will be meaningful? How does one build a communication bridge with no knowledge of what is on the other side of the chasm?15 Two fundamental building blocks of intercultural communication theory provide tentative answers: first, be systematic, creative, and (to the best of one’s ability) empathic about the kinds of tools that could be deployed to understand who the other communicant is, how they communicate, and what they communicate about. Second, humans need to make it as easy as possible for interlocutors to establish the same communicative building blocks. Put simply, humans must remember that intercultural exchange, on Earth or in the stars, is always about trying to solve two problems simultaneously: (1) What do they mean? and (2) How can we help them understand what we mean? We have already discussed, to a limited degree, how the first of these problems could be addressed; doubtless, other scholars in this volume will have more specific ideas about such work. Identifying instances of shared knowledge and experience, using high-performance computing to search for language patterns (first in ambient space, then in known but untranslated messages), and developing a regime of patience among all interlocutors are certainly among the most immediate applicable strategies. The second problem – helping alien contacts understand human messaging – is arguably what humans have tried to do in the variety of interstellar messages that have already been sent. By illustrating the human form and its functions, representing Earth’s galactic location, cataloguing geophysical details ranging from the subatomic to the planetary, and so on, humans have aimed to say “this is who we are.” Again, without a functional equivalent for the referents in these

80

Sumayya K.R. Granger, Judd Ethan Ruggill, and Ken S. McAllister

offerings, however, such messages are likely to remain opaque. Given that humans and extraterrestrials have stellar phenomena in common, at least – nebulae, suns, black holes, gravity, radiation – these will likely be reliable shared analogues for early exchanges. Even if terrestrials and extraterrestrials inhabit the universe in very different ways, mutual experience with the physical world should offer possibilities for quickly (relatively speaking) establishing rich communication on a variety of topical entry points. By first identifying as many shared data points as possible, we believe that, as Slobodchikoff (this volume [Chapter 9]) proposes, a Rosetta Stone of sorts can be established from which other details can be confidently linked by association. Top Down or Bottom Up?

While a number of interstellar messages sent to date have emphasized representations of the atomic and subatomic contexts, and while such contexts are presumably universal, they violate the second rule of intercultural exchange we proposed in this chapter; namely, do everything possible to help message receivers understand the nature of the message itself (i.e., its semantics). Such messages assume that alien interlocutors will, like humans, be unable to directly experience the activities of single atoms or even single molecules, yet given universal laws of gravitation and mass, that seems a safe assumption given that such beings will (like humans) be too large to be cognizant of such microdynamics. For this reason, we recommend that initial exchanges emphasize the largest, most unsubtle shared phenomenon, preferably using a combination of nonlinguistic (e.g., diagrams, models, sounds) and linguistic representations. Once a shared conceptual and terminological vocabulary has begun to emerge, both parties can begin to make associative steps toward greater levels of abstraction and more subtle sensory experiences. We also believe that once the paradigm of deductive reasoning has become practiced, avenues for inductive and abductive reasoning will subsequently emerge. Written language – including mathematical symbols, formulae, constants, and curiosities such as Fibonacci sequences (cf. Sutton 2015; Medeiros 2017) – can be gradually introduced, again, always remaining mindful of how best to make information (especially new information) comprehensible to others. Another communicative approach we recommend – one that admittedly requires the (even slower) exchange of objects, as well as messages – is the creation of associative information packets (AIPs), physical containers holding a number of labelled and connected physical specimens. Consider, for example, a carbon AIP containing: (1) a sample of graphite, (2) a sample of diamond, (3) diagrams of the molecular and crystalline structures of both materials, (4) a diagram of the atomic weight of each, (5) the weight of each of these items on Earth, and (6) the mass of each of these items. Such a packet, designed as it is around the premise of interconnectedness, creates a much smaller solution set to the question “what connects these things” than if they were to appear as discrete items and/or data points. Such

Interstellar Competence

81

AIPs can be developed for virtually any configuration of information, from genetics to political, social, and family structures to popular culture. As basic communication between humans and extraterrestrials unfolds through the identification of shared knowledge and experience, a sensible next step would be the construction of a Swadesh list or core vocabulary list (CVL), a register of words with “common meanings, such as ‘mother’, ‘lake’, ‘mountain’, ‘three’, ‘red’, ‘green’, ‘to vomit’, ‘to kill’, ‘dirty’, and ‘dull’” (Calude and Pagel 2011: 1102). Using this common vocabulary – and the shared concepts and phenomena they represent – humans and aliens would then be able to begin revisiting past conversations, creating more intuitive exchanges and assigning key semiotic elements within them to appropriate words drawn from the CVL. Proceeding in this way, the initial ad hoc Rosetta Stone would become a consistent and nuanced dictionary. To imagine what such a list might look like, it is first necessary to consider what forms of communication would be successful soon after first contact. If, for instance, the early signals were light-based (e.g., infrared bursts), then humans would want to consider using (or at least mimicking) these same signals for creating a preliminary CVL. If early messages were acoustic, the CVL would form around that mode. The aim of mediumic consistency here is both to expedite communication by keeping all exchanges within a narrowly focused communicative paradigm, as well as to (hopefully) provide an advantage to interlocutors who – given their initial ability to receive, identify, and respond to humans’ initial signals – have tacitly indicated a preference for the exchange’s underlying format. Moreover, because humans must, as much as possible, remove their linguistic biases from the exchanges, it would be of paramount importance to create a CVL that represented the look, feel, and/or sound of the communicants’ native language. Proceeding slowly and patiently, we believe that in the para-contact moment (e.g., the first two millennia), humans should deprioritize attempts to communicate using full language, opting instead for solidification of an accurate CVL, labelling system and set of rules by which one idea, object, or concept may be connected to another. A slow, methodical approach such as this diminishes the danger of what W.V.O. Quine characterizes as the “gavagai” problem (1969: 32). This problem arises when a speaker points to an object and utters an unfamiliar word – “rabbit” and “gavagai” in Quine’s example – and confused listeners must somehow determine what “gavagai” refers to: the rabbit itself? A running rabbit? A living rabbit? A brown rabbit? A young rabbit? The rabbit’s legs? The more methodical the process by which the CVL is assembled, the less likely it will be that subsequent exchanges will be undermined by communicative misfires. Such patient efficiency is particularly important when messages are being passed over interstellar distances. Despite what we see in so many science fiction movies and television shows, the experience of fluent interspecial communication seems highly unlikely until many millennia after first contact. We suspect, in fact, that communications will plateau for a long period, perhaps even indefinitely. During this phase, a best-case scenario might well be that there is a consistent “A1” fluency level, defined by The Common

82

Sumayya K.R. Granger, Judd Ethan Ruggill, and Ken S. McAllister

European Framework of Reference (CEFR) as a level of linguistic competence in which one: Can understand and use familiar everyday expressions and very basic phrases aimed at the satisfaction of needs of a concrete type. Can introduce him/herself and others and can ask and answer questions about personal details such as where he/she lives, people he/she knows and things he/she has. Can interact in a simple way provided the other person talks slowly and clearly and is prepared to help. (Council of Europe 2018: n.p.) Such communications with beings from another world – despite being “very basic” – promise to be extraordinarily thrilling nonetheless. They have the added advantage of minimizing the chances of misunderstandings (though these are still bound to occur), while at the same time building confidence in both the mechanisms of exchange and among interlocutors themselves. Over time, they also promise to lay a solid foundation for increasingly complex exchanges about ever more sophisticated topics. What Does It Mean to Be Me?

It is inevitable that as exchanges develop, all parties will begin to discern valuations embedded in topic choices, the CVL, and any other elements of the communication stream. Fundamentally, such assessments hinge on judgements about what is good and bad, and consequently entangle not miscommunications but actual disagreements. This is a highly abstract discursive realm, to put it mildly, and one that humans themselves often disagree about (cf. Sutton 2015). When combined with the problematics of nascent linguistic aptitude, xenophobia, and hyper-asynchronicity, a true conundrum is born: how can human beings effectively navigate contention with a vastly different species that is very far away? One solution is to use a variation of Optimality Theory, an approach to phonological theory whereby constraints within a language’s sounds compete for the “best” outcome. In the context of human language, Optimality Theory helps explain variations among languages (e.g., between Spanish and English) in terms of what linguistic constraints are in place when spoken language is expressed. For instance, Alan Prince and Paul Smolensky (1991) and John McCarthy and Alan Prince (1993) discuss two major types of linguistic constraint – that is, constraints present within a variety of a language that shape how it is spoken (how the surface forms produced vary from the underlying forms from which they are derived). The first of these constraints is faithfulness of the surface form to the underlying form (i.e., do a word’s underlying constituent parts logically add up to its accreted appearance?), while the second is markedness (i.e., the extent to which a word is “well-formed” by syllables of certain shapes and sounds). As languages – human

Interstellar Competence

83

TABLE 8.1 Illustration of outcomes of competing constraints

/in-/ + /mediate/

Make the result easy to pronounce

Keep the surface form similar to the underlying form

Option 1: [inmediate] Option 2: [immediate] (winner)

No Yes

Yes No

ones, at least – evolve, faithfulness and markedness often compete, and the winning option in any given language provides evidence for what qualities of linguistic expressions are favoured by speakers of the language. Consider, for instance, the word immediate, formed from the negating prefix / in-/, plus the root /mediate/; literally, the word means “not mediated,” that is, “right away.” Yet in English, the word is not “inmediate” but “immediate.” Most simply, Optimality Theory would represent this phenomenon thusly: Evident here is that English seems to prioritize word formations that are easier to pronounce (two adjacent labial sounds, rather than an alveolar immediately followed by a labial), over words that clearly reveal their underlying phonemes in their surface forms. Charting the same word in Spanish, however, reveals the opposite preference: inmediato. While Optimality Theory is concerned with phonological forms, it is also important to consider the possibility of socio-cultural applications of the theory. Geert Hofstede (2001), for example, proposes that human cultures can be understood as varying along different continua (e.g., individualism vs. collectivism; low-power distance vs. high-power distance). At the core, this means focusing on how humans in groups interact with one another. This kind of Optimality Theoretic approach might be helpful in conveying to interstellar interlocutors the variance in human cultures, but variance according to a finite set of parameters. Structuring dialogue on complex and abstract topics such as culture in this way could be key to avoiding conflict and extending discussion. The Danger of Oversharing

We have one last recommendation to offer xenolinguistics from the field of intercultural competence. An underappreciated aspect of negotiating intercultural exchange is knowing how much to say about what. For reasons that science fiction has elaborated extensively – from weaponized modifications of the human genome to exploitations of military systems to manipulations of the human psyche – the indiscriminate sharing of information about life on Earth is highly inadvisable.16 This is so, not for fear of planetary conquest but because even small sociocultural differences can have dramatic effects on how relationships do or do not develop. This is another reason to keep initial exchanges relatively neutral, focusing on factual observations about interstellar space until communicative pathways

84

Sumayya K.R. Granger, Judd Ethan Ruggill, and Ken S. McAllister

are sufficiently robust to handle the discovery and processing of more freighted cultural and personal expressions of meaning. At the same time, the whole proposition of entering into contact with other lifeforms presupposes a certain openness, at least on humanity’s part. Despite how vastly different humans may be from beings beyond the stars, failing to make unflagging efforts to understand and be understood – and doing so thoughtfully, patiently, and generously – will significantly diminish the probability of developing a persistent and rewarding communication channel. Final Thoughts

In this chapter, we have drawn on work from a number of interrelated fields to outline an approach to a (highly probable) hyper-asynchronous, physically removed first-contact scenario. On the chance that first contact is face-to-face, we still recommend the set of practices discussed here, though with appropriate modifications to accommodate the change from an asynchronous to a synchronous communication mode. In such a case, a patient, methodical approach to every exchange is essential. Particularly in the para-contact period – when trust is fragile, anxiety and eagerness are uneasy emotional partners, and the ground of understanding is, at best, precarious – all parties must remember how volatile new relationships can be. Whether built near or distantly, communication styles and strategies must be readily adjustable based on whatever feedback is discernible. If humans are hindering rather than helping their interlocutors understand and be understood, then a re-evaluation of all applicable communication strategies is required. And at every turn, interlocutors must be slow to assume and quick to forgive. A seemingly delightful similarity might quickly unravel, while an unfathomable difference might snap into clarity when its context is fully revealed. As a useful reminder, humans will do themselves a favour by remembering the vast differences even among human cultures, where miscommunications and dangerous assumptions often yield considerable strife. Extraterrestrials will be at least as different from humans as human cultures are from each other. And should interstellar conversationalists make the same mistake toward humans – unwarranted assumptions; misunderstanding signals; expressing voluble surprise or dismay at human values, beliefs, and practices; or otherwise getting the wrong end of the stick – humankind must its redouble efforts to work patiently to build bridges and cultivate understanding. We proposed at the start of this chapter that despite the lack of concrete knowledge of first-contact beings, human beings are not at a complete loss for sensible approaches to first-contact exchanges. Indeed, we have suggested that there are at least five important traits humans will share with any alien interlocutor: tool use, symbol use, communication, culture-making, and curiosity. In our estimation, whether Earthbound or starbound, it is this last feature in particular that will keep the conversations vital, edifying, and always open to new ideas and friendships.

Interstellar Competence

85

Notes 1 Hobaiter et al. (this volume [Chapter 3]) use a Gricean approach to provide a compelling perspective on assumptions such as these, problematizing what they might mean for the detection of “intentional communications.” 2 As Daniel Ross (this volume [Chapter 12]) notes, “sufficiently intelligent” is perhaps a more accurate way to denote the intelligence(s) in question. 3 It is worth noting that communication and culture-making are relevant only for species of n > 1. Mono-entities would not necessarily develop communicative or cultural systems. 4 For precis on the Rio and San Marino scales, see https://iaaseti.org/en/rio-scale/ and www.setileague.org/iaaseti/smiscale.htm, respectively. 5 For a brief introduction to this project, see https://en.wikipedia.org/wiki/ Morse_Message_(1962). 6 Daniel Ross (this volume [Chapter 12]) astutely points out that human beings may not have the cognitive ability “for receiving (via language), retaining, and evaluating extraterrestrial thoughts, which might be more complex than our own.” 7 See www-01.ibm.com/software/ebusiness/jstart/portfolio/seti.html for more about this collaboration. 8 For a primer on the Cosmic Call, see www.smithsonianmag.com/science-nature/ annotated-cosmic-call-primer-180960566/. 9 Con Slobodchikoff (this volume [Chapter 9]) makes a similar point, noting that it is not only spoken language that humans have trouble with – even among native speakers of the same language – but also written languages and the languages of mathematics and geometry. The probability of communicative misfires even among our own species thus makes the possibility of reliable intergalactic communication extremely poor. 10 As Jessie S. Nixon and Fabian Tomaschek (this volume [Chapter 17]) explain, and as Bridget D. Samuels and Jeffrey Punske (this volume [Chapter 16]) note, even slight physiological differences in anatomy – differences which may be environmentally caused – can greatly impact the ability to communicate in comprehensive ways. 11 See Harbour (this volume [Chapter 18]) on the Visible Speech writing system. 12 See Ortner (this volume [Chapter 5]) on interactional calibration. 13 Arik Kershenbaum (this volume [Chapter 2]), offers a survey of the astonishing breadth of communicative strategies already deployed on Earth, clarifying in the process how numberless are the possibilities in a universal context. 14 While in this chapter we make a best effort to get “outside of ourselves,” our approach will necessarily be anthropocentric given the nature of our physical constraints. This is, indeed, the core challenge of a book about xenolinguistics. Nevertheless, we hope that the proposals here are abstract or abstractable enough to be of value. 15 See Wells-Jensen (this volume [Chapter 13]) on questioning unconscious assumptions. 16 The precautionary approach recommended here is generally accepted among policy analysts interested in intelligent interstellar communication. Admittedly driven by a selfreflexive assessment of how humans treat each other, such an approach considers factors ranging from local diplomacy to global security. See, for example, Matthews (2019), Todd and Miller (2018), Wilman and Newman (2018), and Wilson and Cleland (2019).

References Althusser, Louis. 2001 [1970]. “Ideology and ideological state apparatuses (Notes toward an investigation).” In Lenin and Philosophy and Other Essays. Translated by Ben Brewster. New York: Monthly Review Press.

86

Sumayya K.R. Granger, Judd Ethan Ruggill, and Ken S. McAllister

Calude, Andreea S., and Mark Pagel. 2011. “How do we use language? Shared patterns in the frequency of word use across 17 world languages.” Philosophical Transactions: Biological Sciences 366 (1567): 1101–1107. Council of Europe. 2018. “Global scale-table 1 (CEFR 3.3): Common reference levels.” Common European Framework of Reference for Languages (CEFR) Website, n.d. www.coe.int/en/web/common-european-framework-reference-languages/table-1-cefr3.3-common-reference-levels-global-scale. Hofstede, Geert. 2001. Culture’s Consequences: Comparing Values, Behaviors, Institutions, and Organizations across Nations. London: Sage. Matthews, Luke J. 2019. “A moonshot for extraterrestrial communication.” Anthropology News, July 12. www.anthropology-news.org/index.php/2019/07/12/ a-moonshot-for-extraterrestrial-communication/. McCarthy, John, and Alan Prince. 1993. “Prosodic morphology: Constraint interaction and satisfaction.” Linguistics Department Faculty Publication Series 14. https://scholarworks.umass.edu/linguist_faculty_pubs/14. McLuhan, Marshall. 1964. Understanding Media: The Extensions of Man. New York: McGraw-Hill. Medeiros, David P. 2017. “Fibonacci and L-grammars.” David P. Medeiros Website, n.d. http://davidpmedeiros.com/fibonacci-l-grammars. Prince, Alan, and Paul Smolensky. 1991. “Connectionism and harmony theory in linguistics.” Report no. CU-CS-533–91: Department of Computer Science, University of Colorado, Boulder. Quine, W.V.O. 1969. Ontological Relativity and Other Essays. New York: Columbia University Press. Shostak, Seth. 2011. “Limits on interstellar messages.” Acta Astronautica 68: 366–371. Sutton, Jon. 2015. “Interview [with Douglas Vakoch]: What would you say to an alien?” The Psychologist: The Website of the British Psychological Association 28: 800–803. Todd, Peter M., and Geoffrey F. Miller. 2018. “The evolutionary psychology of extraterrestrial intelligence: Are there universal adaptations in search, aversion, and signaling?” Biological Theory 13 (2): 131–141. Wilman, Richard J., and Christopher J. Newman. 2018. Frontiers of Space Risk: Natural Cosmic Hazards and Societal Challenges. CRC Press (Taylor & Francis Group). Wilson, Elspeth M., and Carol E. Cleland. 2019. “Should we call E.T.? An ethical-political analysis of METI.” Theology & Science 17 (3): 382–394. Wohlleben, Peter. 2016. The Hidden Life of Trees: What They Feel, How They Communicate. Vancouver: Greystone Books.

9 WHY DO WE ASSUME THAT WE CAN DECODE ALIEN LANGUAGES? Con Slobodchikoff

Discussion

In our human optimism, we assume that if we received a communication from an alien species, we would be able to eventually decode that communication, and we would over time be able to learn the language of the aliens. My contention is that this is unlikely. We have a number of alien species sharing this planet with us: animals that communicate with each other, some using sophisticated communication that is either language or approaches language (Slobodchikoff 2012). So far, we have not been able to decode most of those languages. Even among our own human languages, we have not been able to decode and learn a number of languages from historical times. And yet, we make the assumption that if and when alien contact occurs, we will be able to understand what the aliens have to say. First, let me address something about the methodology that we are using to search for alien signals and, we hope, alien languages. We are currently using multiple frequencies of radio signals and scans for optical laser pulses, and have come up empty. So, let’s do a thought exercise. Let’s suppose that we are a culture that lives in a valley that is surrounded by mountains, and we use smoke signals to communicate with each other. We are convinced that there are other cultures living in other valleys beyond our mountains, and we are looking for signs of their communication. We know that smoke rises, so the logical assumption is that if we communicate by smoke signals with each other, and our smoke rises, that other cultures must be communicating with smoke signals as well, and if we just look for their smoke we should be able to identify their signals. We search and search, but it is in vain. We see no smoke signals. Occasionally we see smoke rising above the mountains that encircle us, but it does not seem to have an organized pattern and might just come from brush or forest fires in other distant valleys. Eventually, we DOI: 10.4324/9781003352174-9

88

Con Slobodchikoff

conclude that there must not be other cultures out there, because if there were, we would surely see their smoke signals. A point that we fail to grasp however, is that there are lots of cultures in other valleys, but they communicate using cell phones, a technology that is completely unknown to us. In my opinion, this is a scenario that we fail to consider in our search for extraterrestrial communications. But let us say that we do make contact. How are we going to decipher the signals that we get? What are we going to look for that will tell us that this is a language that we are encountering and not just some random fluctuations of energy produced by physical phenomena among the stars? Here we can turn to how linguists have approached a view of animal language. In 1960, Charles Hockett, a linguist, proposed a number of design features which need to be found if we were to recognize that an animal species had language (Hockett 1960). Hockett published a list of 13 design features of human language that would be important to find in animals so that we could say that animals have language. Some of these design features are found in any system that produces signals. However, seven of the design features are really key elements that distinguish language from mere communication. These seven are as follow. • Semantics. Just as each word in a human language has a distinct meaning, the signals that animals produce also have to have distinct meanings. • Arbitrary. An arbitrary symbol has no direct connection to what it represents, like the word “green” does not tell you anything about what the color green actually looks like. This is in contrast to an iconic symbol, which represents some attribute of the thing that it is describing. When you say “bow-wow” to describe a dog, the “bow-wow” is an iconic symbol for dog, because it represents an attribute of dogs, namely barking. • Discrete. Each symbol has to be a discrete unit, just like the words in this sentence are all discrete units. • Displacement. A language has to provide information about events that occur in different locations from the speaker or in different time periods, i.e., displacement in either space or time. • Productivity. A language has to be able to make up new words. For example, the word(s) “cell phone” did not exist in the English language until recently. • Duality. Language has to have smaller units that can be combined into bigger units. Think of how phonemes can be combined into morphemes, or words into sentences. • Cultural transmission. There must be a strong component of learning in a language. We aren’t born knowing the language that we speak. These design features are found in human languages. But when we try to apply them to animals, we are faced with a conundrum. With humans, we can ask people speaking a language whether something makes sense or not. We can rearrange

Why Do We Assume That We Can Decode Alien Languages?

89

phonemes or morphemes and ask people whether our rearrangements are meaningful to them. We cannot do that with animals. With animals, we need a Rosetta Stone to give us the key to unlock their system of language or communication. The Rosetta Stone is a stone that was found by the French in 1799 in a village in Egypt that was then called Rosetta (now called Rashid). It contains three sets of inscriptions of a decree issued in 196 BCE on behalf of Ptolemy V. The bottom set of inscriptions is in ancient Greek. The middle set of inscriptions is in Demotic Egyptian, and the top set of inscriptions is in Egyptian hieroglyphs. Because ancient Greek was a known language, and scholars were able to recognize that the three sets of inscriptions represented the same text, Jean-Francois Champollion announced in 1822 that he was able to transliterate hieroglyphic Egyptian (Ray 2007). While hieroglyphic Egyptian was transliterated with the Rosetta Stone, other ancient human languages have not fared as well. Currently, Linear A from the Minoan culture of around 1500 BCE, Vinca symbols from around 5000 BCE, Rongorongo from Easter Island, and Olmec Script from around 1200 BCE remain untranslated (Packard 1974; Fischer 1997; Mora-Marin 2009; Sproat 2010). In all of these languages, there is no key, like the Rosetta Stone for decoding hieroglyphs, that gives linguists and anthropologists any clue about how to approach a translation. A Rosetta Stone for decoding animal languages is the context in which different signals are used. Let me illustrate this with my work in deciphering the language of prairie dogs (Slobodchikoff et al. 2009b). Prairie dogs are ground squirrels that live in extensive systems of burrows that are called towns or colonies. Within a prairie dog town, there might be hundreds of animals, and the ground is partitioned into territories that are occupied by discrete social groups of prairie dogs. When a predator such as a coyote or a hawk appears, one or more prairie dogs give an alarm call which to our ears sound something like a bird chirping. Other prairie dogs throughout the town run to their burrows to escape the predator. The entire sequence of predator appearing, prairie dogs alarm calling, and running to their burrows offered my students and me a Rosetta Stone for decoding Gunnison’s prairie dog language (Slobodchikoff et al. 2009b). We could videotape the appearance and behavior of the predator as it approached the colony, we could record the alarm calls that the prairie dogs produced in response to the predator, and we could videotape the escape responses of the prairie dogs. Subsequently, on another day when no predator was present, we could play back the recordings to the predator that had previously appeared, videotape the escape responses of the prairie dogs, and see if the escape responses of the prairie dogs when no predator was present were the same as the escape responses when the predator actually appeared. In a series of experiments, we found that prairie dogs have both different alarm calls and different escape responses for different predator species. Predators of prairie dogs include coyotes and domestic dogs, who try to catch the prairie dogs on the ground, humans who shoot prairie dogs, and red-tailed hawks who swoop

90

Con Slobodchikoff

down on prairie dogs from the air. The escape responses of the prairie dogs differ, depending on the species of the predator (Kiriazis and Slobodchikoff 2006). For coyotes, prairie dogs immediately run to their burrows, and stand on their hind legs at the burrow opening, watching the progress of the coyote through the colony. Other prairie dogs that were below ground before the coyote appeared emerge and also join those standing at the lip of the burrow entrances, watching the coyote. This is related to the hunting strategy of coyotes, who will often find a concentration of prairie dogs and lie down on the ground near those burrow openings, waiting for unwary prairie dogs to emerge. By watching the progress of the coyote, the prairie dogs have an excellent idea of where a coyote is likely to lie down and wait for them. For domestic dogs, prairie dogs will stand up on their hind legs wherever they happen to be foraging, and watch the progress of the dog through the colony. Only if the dog gets close to them do they run for their burrows. However, like with coyotes, prairie dogs that were below ground emerge from their burrow openings to stand and watch the progress of the dog through the colony. Dogs are not very efficient predators of prairie dogs, and unlike coyotes, they do not lie down next to burrows, so the prairie dogs do not need to find immediate safety in their burrows, as they do with coyotes, but they do need to see where the dog is going. For red-tailed hawks, prairie dogs that are in the immediate flightpath of the stooping hawk will run into their burrows and dive inside, without standing at the lip of the burrow, as they do with coyotes and with dogs. Other prairie dogs who are not in the immediate flightpath of the stooping hawk stand on their hind legs wherever they happen to be foraging and watch the progress of the hawk. For humans, the entire colony of prairie dogs immediately runs to their burrows and dives in, without emerging or standing at the lip of their burrows. Prairie dogs have been hunted since ancestral times by Native Americans, who used bows and arrows to shoot at them. At the present time, prairie dogs are shot by hunters who view such killing as sport. Whether they are shot at with bows and arrows or with rifle bullets, prairie dogs do not have very much time to escape, and become easy targets if they stand at their burrow lips (Slobodchikoff et al. 2009b). Just as the escape responses differ, we found that the alarm calls differ in their acoustic structure to different species of predator. We found that prairie dogs have acoustically different alarm calls for coyote, domestic dog, human, and redtailed hawk. This gave us the Rosetta Stone to begin to unravel the language system of prairie dogs. In a series of experiments, we were able to show the following design features in the alarm calls of prairie dogs: • Semantics. Prairie dogs have distinctly different alarm calls for different species of predators. Furthermore, they can vary the acoustic elements within a call for a predator species to describe the size, color, and shape of the predator (Slobodchikoff et al. 1991; Slobodchikoff et al. 2009b).

Why Do We Assume That We Can Decode Alien Languages?

91

• Arbitrary. The calls are completely arbitrary, just like human words, with a series of acoustic frequencies changing as a function of time within an alarm call (Placer and Slobodchikoff 2000). • Discrete. Each alarm call is a discrete unit (Placer and Slobodchikoff 2001). • Displacement. Prairie dogs are able to indicate the presence of a predator that is far away from them. One experiment showed that prairie dogs were able to consistently affix an apparent label of a gun to a person who once fired a shotgun and then subsequently appeared without a gun for the month of the experiment (Frederiksen and Slobodchikoff 2007). • Productivity. Prairie dogs are able to coin new “words” for objects that they have never seen before, such as a black oval, a triangle, and a circle (Ackers and Slobodchikoff 1999; Slobodchikoff et al. 2012). • Duality. An analysis of the structure of the alarm calls showed that they were composed of phonemes, just like human words are composed of phonemes. For the alarm calls for the different species of predators, most of the same phonemes were used, but the proportion of phonemes that was used in alarm calls for different species of predators was different (Slobodchikoff and Placer 2006). • Cultural transmission. An experiment showed that newly-born prairie dogs emerging from their burrows have a non-specific alarm call to all predators, but over time, the specificity increases, suggesting that some aspects of the call are under genetic control and other aspects of the call are determined by cultural transmission (Slobodchikoff 2010). Prairie dogs represent one of the best animal language-like systems that has been decoded, but there are a number of other examples of animals having languagelike properties (Slobodchikoff 2012): sagebrush lizards have been shown to have a grammatical system in their head bobbing, tail lifting, and arm lifting (Martins 1993); both Japanese tits and American chickadees have been shown to have syntax in their vocalizations (Ficken et al. 1987; Suzuki et al. 2016); and blackbirds have been shown to have recursion in their calls (Gentner et al. 2006). In all of these cases, the behavioral context represents a starting point for decoding the language of communication system. As with prairie dogs, the behavioral context is the Rosetta Stone. However, there are many situations in which a behavioral context is absent. Imagine that someone who does not know that humans have a language – perhaps an extraterrestrial alien – is trying to figure out whether humans are able to meaningfully communicate to one another. This alien is able to detect that I am talking to you on a cell phone. I am holding the phone in my hand and nothing about my behavior gives a clue as to whether I am speaking a language or I am merely vocalizing my emotional state and excitement into a piece of plastic. Similarly, an analysis of your behavior at your end shows that there is no behavioral context other than the production of noises. This alien could very well conclude that humans do not have a language but just make a series of vocalizations, perhaps as aggressive

92

Con Slobodchikoff

or mating signals. Unfortunately, this is a position that some biologists and some linguists take with respect to animal language. To be fair, it would be possible to analyze these conversations by looking at the structure of the sounds. Perhaps some sounds occur more frequently than others, such as described by Zipf’s Law (Newman 2005). Perhaps some sounds tend to occur more frequently after a pause, or perhaps some sounds precede a pause, but nothing about these structural features gives a clue about the meaning of the conversation. The meaning can be either derived from the behavioral context or from asking the participants of the conversation, but to ask the participants of the conversation what they meant, you have to have at least the semblance of either their language or some other language in common. Applying this logic to decoding alien languages produces results that are not very hopeful. But what about communicating with aliens through mathematics? Surely mathematics are universal, right? Unfortunately, not necessarily. First, there are the limitations on numbers imposed by different languages. Some languages count objects as one, two, then many. Some languages go simply from one to many. The evidence is still ambiguous as to whether languages constrain their speakers’ perception of external objects or events (Boroditsky 2001). Second, we tend to think of numbers as discrete integers, such as one, two, three, and so forth. Our computers treat numbers as zero and one, or “on” or “off.” So perhaps we can deal with aliens through numbers, or at least deal with them in sequences of zeros and ones. But what if they did not view numbers as discrete entities? What if instead of a discrete zero and a discrete one, they viewed numbers as continuous variables, so that there would be an infinite number of states between zero and one? We could send out prime numbers, but if aliens perceived numbers differently from us, our messages of prime numbers would be meaningless. Perhaps geometry can help us. After all, a circle is always a circle, a triangle is always a triangle, and pi is always pi. But we know that our brains are hardwired to detect certain patterns, which is why our brains can be tricked by illusions (Yarritu et al. 2015). Our understanding of the abilities of animals has expanded so much that we now know that different species of animals can perceive the world differently from us. We cannot see into the ultraviolet range of the spectrum, whereas bees and some birds can (Lind et al. 2013; Cronin and Bok 2016). We cannot detect magnetic fields, whereas a number of animals, particularly the those that migrate, can perceive changes in the magnetic field (Taylor et al. 2017). We have no guarantee that aliens would see a circle as a circle and a triangle as a triangle. Once again, we might be trapped by the smoke signal analogy. We might argue that we have things in common with aliens that can be expressed through mathematics. We know that the world is made up of atoms which can be counted. We know the time when the Big Bang created our universe. So we have atoms and time in common, right? However, ever since Einstein formulated his

Why Do We Assume That We Can Decode Alien Languages?

93

special theory of relativity, we know the time is a somewhat slippery concept and is not necessarily perceived by everyone in exactly the same way (Austin 2017). Similarly, the development of quantum theory – with its concepts of superposition, coherence, and energy states – calls into question whether atoms are discrete objects or are waves or energy fields (Rosenblum and Kuttner 2011). What we see as discrete objects, aliens might see as waves or continuously fluctuating energy fields. All of these factors represent possible impediments to us understanding the languages of aliens, whether we encounter these languages through the detection of signals or through face-to-face encounters. However, there is a possible work-around. If aliens are smarter than us, or if they are more experienced in contact with a variety of extraterrestrial cultures, they could provide us with a Rosetta Stone that would allow us to communicate with them at some level. They could teach us the rudiments of their language, or even a simplified language that would make communication possible. This is something that we have done with a few chimpanzees, such as Washoe and Kanzi, and Koko the gorilla, by teaching them American Sign Language or teaching them to respond to symbols on a keyboard (Gardner et al. 1989; Savage-Rumbaugh et al. 1998; Hare and Yamamoto 2017). If aliens can teach us something that represents common ground between them and us, that might provide enough of a Rosetta Stone for us to begin to communicate with them. Whether we will ever, under those circumstances, be able to discuss philosophy or higher concepts with them remains an open question. We have not yet succeeded with our resident aliens, all of the animal species around us, but we can sweep that lack of success under the rug by assuming that these animals are too stupid to have language. We can only hope that visiting aliens will not think the same of us. References Ackers, S.H. and C.N. Slobodchikoff. 1999. “Communication of Stimulus Size and Shape in Alarm Calls of Gunnison’s Prairie Dogs.” Ethology 105: 149–162. Austin, R.W. 2017. “Gravitational time dilation derived from special relativity and Newtonian gravitational potential.” European Scientific Journal 13. http://dx.doi.org/10.19044/ esj.2017.v13.n3p%25p. Boroditsky, L. 2001. “Does language shape thought?: Mandarin and English speakers’ conception of time.” Cognitive Psychology 43: 1–22. Cronin, T.W., and M.J. Bok. 2016. “Photoreception and vision in the ultraviolet.” Journal of Experimental Biology 219: 2790–2801. Ficken, M.S., C.M. Weise, and J.A. Reinartz. 1987. “A complex vocalization of the blackcapped chickadee. II. Repertoires, dominance, and dialects.” Condor 89: 500–509. Fischer, S.R. 1997. Rongorongo: The Easter Island Script. Oxford: Clarendon Press. Frederiksen, J.K., and C.N. Slobodchikoff. 2007. Referential specificity in the alarm calls of the black-tailed prairie dog.” Ethology, Ecology & Evolution 19: 87–99. Gardner, R.A., B.T. Gardner, and T.E. Van Cantford. 1989. Teaching Sign Language to Chimpanzees. Albany: State University of New York Press.

94

Con Slobodchikoff

Gentner, T.Q., K.M. Fenn, D. Margoliash, and H.C. Nusbaum. 2006. “Recursive syntactic pattern learning by songbirds.” Nature 440 (7088): 1204–1207. Hare, B., and S. Yamamoto, eds. 2017. Bonobos: Unique in Mind, Brain, and Behavior. Oxford: Oxford University Press. Hockett, C.F. 1960. “Logical considerations in the study of animal languages.” In Animal Sounds and Communication, edited by W.E. Lanyon and W.N. Tavolga, 392–430. Washington, DC: American Institute of Biological Sciences. Kiriazis, J., and C.N. Slobodchikoff. 2006. “Perceptual specificity in the alarm calls of Gunnison’s prairie dogs.” Behavioural Processes 73: 29–35. Lind, O., M. Mitkus, P. Olsson, and A. Kelber. 2013. “Ultraviolet vision in birds: The importance of transparent eye media.” Proceedings of the Royal Society B. doi:10.1098/ rspb.2013.2209. Martins, E.P. 1993. “Contextual use of the push-up display by the sagebrush lizard, Sceloporus graciosus.” Animal Behaviour 45: 25–36. Mora-Marin, D.F. 2009. “Early olmec writing: Reading format and reading order.” Latin American Antiquity 20: 395–412. Newman, M.E.J. 2005. “Power laws, Pareto distributions and Zipf’s law.” Contemporary Physics 46: 323–451. Packard, D.W. 1974. Minoan Linear A. Berkeley: University of California Press. Placer, J., and C.N. Slobodchikoff. 2000. “A fuzzy-neural system for identification of species-specific alarm calls of Gunnison’s prairie dogs.” Behavioural Processes 52: 1–9. Placer, J., and C.N. Slobodchikoff. 2001. “Developing new metrics for the investigation of animal vocalizations.” Intelligent Automation and Soft Computing 8: 1–11. Ray, J. 2007. The Rosetta Stone and the Rebirth of Ancient Egypt. Cambridge, MA: Harvard University Press. Rosenblum, B., and F. Kuttner. 2011. Quantum Enigma. Second Edition. Oxford: Oxford University Press. Savage-Rumbaugh, S., S.G. Shanker, and T.J. Taylor. 1998. Apes, Language, and the Human Mind. Oxford: Oxford University Press. Slobodchikoff, C.N. 2010. Talk of the town. BBC Video. Slobodchikoff, C.N. 2012. Chasing Doctor Dolittle: Learning the Language of Animals. New York: St. Martin’s Press. Slobodchikoff, C.N., W.R. Briggs, P.A. Dennis, and A.-M. Hodge. 2012. “Size and shape information serve as labels in the alarm calls of Gunnison’s prairie dogs Cynomys gunnisoni.” Current Zoology 58: 741–748. Slobodchikoff, C.N., J. Kiriazis, C. Fischer, and E. Creef. 1991. “Semantic information distinguishing individual predators in the alarm calls of Gunnison’s prairie dogs.” Animal Behaviour 42: 713–719. Slobodchikoff, C.N., and J. Placer. 2006. “Acoustic structures in the alarm calls of Gunnison’s prairie dogs.” Journal of the Acoustical Society of America 119: 3153–3160. Slobodchikoff, C.N., A. Paseka, and J. Verdolin. 2009a. “Prairie dog alarm calls encode labels about predator colors.” Animal Cognition 12: 435–439. Slobodchikoff, C.N., B. Perla, and J. Verdolin. 2009b. Prairie Dogs: Communication and Community in an Animal Society. Cambridge, MA: Harvard University Press. Sproat, R. 2010. “Ancient symbols, computational linguistics, and the reviewing practices of the general science journals.” Computational Linguistics 36: 585–594. Suzuki, T.N., D. Wheatcroft, and M. Griesser. 2016. “Experimental evidence for compositional syntax in bird calls.” Nature Communications 7: 10986.

Why Do We Assume That We Can Decode Alien Languages?

95

Taylor, B.K., S. Johnsen, and K.J. Lohman. 2017. “Detection of magnetic field properties using distributed sensing: A computational neuroscience approach.” Bioinspiration & Biomimetics 12. doi.org/10.1088/1748-3190/aa6ccd. Yarritu, I., H. Matute, and D. Luque. 2015. “The dark side of cognitive illusions: When an illusory belief interferes with the acquisition of evidence-based knowledge.” British Journal of Psychology 29 January. doi.org/10.1111/bjop.12119.

10 XENOLINGUISTIC FIELDWORK Claire Bowern

Preliminaries1 How would we get to work deciphering an alien language when the time comes? How hard would this be? While our knowledge of human language will only get us so far, the tools we have developed for linguistic fieldwork and analysis will be critical. (Coon 2020: 44)

This chapter is about the tools developed for linguistic fieldwork and analysis and what is needed to understand extraterrestrial linguistic systems. I discuss what aspects of human linguistic fieldwork probably transfer to xenolinguistic fieldwork, what does not, how well prepared we are, and challenges that may arise. I argue that while fieldwork approaches are, broadly, capable of helping us understand ET languages, our data storage and archiving systems are woefully inadequate. Such an upgrade would be beneficial to human linguistic fieldwork, too. Broadly speaking, “fieldwork” is collecting data for analysis from a natural context. That is, fieldwork is placed in contrast to laboratory or naturalistic experiments or other types of “intervention,” theoretical models, corpus building from secondary sources, or meta-studies. It involves the collection of primary data by researchers directly from the languages users (cf. Thomas 2020 and references quoted there). Linguistic fieldwork grows out of an anthropological tradition of ethnography (Brewer 2000), which combines participant observation (looking at what people do), interviews (asking people what they do), and immersion and participation (learning what people do by participating oneself).2 Linguistic fieldwork typically also includes structured tasks: a range of activities ranging from general questions DOI: 10.4324/9781003352174-10

Xenolinguistic Fieldwork

97

to precise translations to open-ended recordings and conversation. This “typical” fieldwork involves living in the community, though “remote fieldwork” is also increasingly common (Williams et al. 2021). Linguistic fieldwork is, these days, usually a collaboration between language participants and the researcher, with shared negotiated goals and procedures. On Earth, fieldworkers may travel to the other side of the world or step out their front door, working on their own language in their own community, or not. Linguists and anthropologists sometimes use the notion of “alien” in their description of fieldworkers (cf. Ly and Spjeldnæs 2021), concentrating on the differences between fieldworkers who are community outsiders and the language and culture they are researching. Others, such as Rice (2009), concentrate on the collaborative nature of fieldwork where linguists and communities come together, with shared goals. I am assuming that the ET system is a linguistic one. As other chapters in this book have discussed (e.g., Slobodchikoff, this volume [Chapter 9]; Pepperberg, this volume [Chapter 6]), it can be no easy task to determine whether a communication system is “language.” I am also making some assumptions about ET social structures. For example, I am assuming that linguistic systems arise and are conventionalized through social learning, as well as having properties due to innate psychology and physiology. While discussions of the language faculty and universal grammar typically strongly focus on the innate and the properties of the individual, there are also social properties that shape language. They are not just a “third factor” (relegated to relative unimportance) but crucially shape the field experience. Social learning is an inherent part of language, as well as linguistic innateness. Some features of language come from human psychology and physiology. The phonology of signed and spoken languages, for example, is embodied, in that if our bodies were differently shaped or functioned differently, that aspect of language would be different, as well. Other features of language come from human culture. The lexicon of different languages divide up concepts in limitedly distinct ways. Some of this comes from human psychology, other aspects from culture. Languages vary in the way that words for colors are realized, for example (Kay and Maffi 1999). Some languages make no use of hue for describing items at all. Others use only a two-term system, whereby a single term covers dark colors and another covers light. Others have more terms; but while there are thousands of different languages today, there are not thousands of color term systems or hundreds, or even tens. The vast majority of the world’s languages can be described with seven basic categories. A different example comes from words for boats, which reflect not human psychology but the cultural conventions of craft building. Other properties of human language serve to distinguish it from non-linguistic systems (Hockett 1959). It can be difficult to determine whether an example of “human language” is a linguistic system or not, however. To take one example, the Voynich Manuscript (cf. Bowern and Lindemann 2021) has been argued about for more than fifty years: is it an enciphered (already known) human language, a

98

Claire Bowern

constructed language, or a hoax – a non-language? As Slobodchikoff (this volume [Chapter 9]) and Pepperberg (this volume [Chapter 6]) have argued, it may be quite difficult to identify ET communication systems as language. Whether or not the system is “linguistic” – in the sense that it displays properties of human languages such as recursion, displacement, and duality of patterning, among other aspects – does not actually affect methods for fieldwork and description. That is, we expect such features to be present in human languages. For example, when I work on language documentation, I expect there to be ways to talk about things that are not present in the here and now. I expect that a sentence will mean the same thing (broadly) from person to person and session to session, if a word minyaw means “cat” in one session it will not suddenly mean “dog” in the next one. But if those assumptions do not hold, I will realize very quickly. The lack of displacement, for example, will emerge from the corpus and from the answers to questions about things that are not visible. It is usually fairly obvious if there is no way to directly translate a feature of the linguists’ metalanguage, but it can be less straightforward to distinguish features that appear in the field language but not the metalanguage. Just as no human language description is “comprehensive,” ET linguistic descriptions are likely to miss a lot. Fieldwork Logistics

As Coon (2020), Macaulay (2004), Bowern (2015), and others have discussed, language is only one part of a much larger set of fieldwork logistics. There’s travel to the field site (including any relevant visas and permits), accommodation, compensating consultants for their time, and finding appropriate places to work, among other things. Fieldwork costs money and there are few funding sources. And then there is the matter of how to get home again. This section describes some of the issues around logistics in conducting fieldwork. Aims and Goals of Fieldwork

The aims of the linguistic fieldwork need to be clearly articulated. What is the purpose of studying the language? To establish shared communication? For humans’ cultural and scientific knowledge? As a way of establishing scientific and cultural reciprocity? To assist ETI communities with language preservation in the face of human appropriation and assimilation? For easier trade paths? For the production of language learning or other educational materials? As part of a program of military interrogation or intergalactic espionage? Evangelism of ET communities and individuals? As Rice (2022) discusses in detail, “linguistics has been utilized as a tool for the harmful practice of attempting religious conversion of Native peoples.” Fieldwork has been conducted for all these reasons (sometimes simultaneously), and we cannot think about fieldwork in the abstract

Xenolinguistic Fieldwork

99

without thinking about the goals and consequences of such work (cf. also Charity Hudley et al. 2018). Movies like Arrival focus on fieldwork and initial contact. That is, the linguist is one of a team of scientists and military personnel and the linguist works in part to further the goals of the military and political operation. They work under immense time pressure as the fate of the world hangs on their ability to “decode” the language. This role combines the military notion of linguist as translator with the academic idea of linguist who works on language structure. A more realistic view of linguistic fieldwork is to imagine a larger, long-term field project that is not tied up with the initial encounter with ETI. Fieldwork takes time. It involves not just figuring out puzzles, but coming to a shared understanding with communities. Fieldwork requires a lot of goodwill on both sides to be effective, particularly on the part of the host community. Such goodwill is unlikely to be easy to achieve under initial military supervision. If I had to come up with a first contact “fieldwork” plan, it would not involve the detailed analysis of linguistic structure. It would be aimed solely at establishing some shared communication and non-threat. Bronislaw Malinowski, who is usually credited with developing ethnographic (particularly participant observation) methods in anthropology, notably begins a description of how to do fieldwork with what amounts to “find the local white man” and stay with him.3 While there is often relief in the familiar, a big thing about fieldwork is being able to step outside one’s own assumptions about language, culture, and community (both one’s own and that of others). Linguistic fieldwork is about language, but it is also more than about language. Language is a means to a shared understanding that goes well beyond the specifics of particular linguistic systems. As McKenzie (this volume [Chapter 11]) writes, language is inherently cooperative; so is documenting language. Cooperation is two-way, so linguists should also be prepared to be representatives for human culture to ET communities. Shared Language for Fieldwork

Much fieldwork on Earth (though by no means all) is conducted through a lingua franca – either the fieldworkers’ own first language, or a language shared within at least some of the community. That shared language makes some fieldwork tasks easier. Direct translation, for example, is an effective way to increase vocabulary quickly – the “what is this called?” or “How do you say this?” type of questioning. Monolingual and bilingual fieldwork have some of the same tools (see Everett 2001). As Thomas (2020) points out, both “monolingual” and “bilingual” fieldwork are multilingual; the main difference is in whose primary language is the medium for linguistic discussion. Learning the language is a mark of respect and produces better insights into the language structure (because the fieldworker knows it better). It allows for better community-oriented work and a deeper knowledge

100 Claire Bowern

of the subject on the part of the linguist. Of course, depending on the nature of the ETI, such a language may or may not be possible for a human to articulate. Some fieldwork uses interpreters. That is, the linguist and the research participants do not share a common language, and an interpreter translates the linguist’s questions for the language worker (for example, a linguist in Australia might ask a question in English, which is translated into Kriol, and then answered in an Indigenous language). Depending on the circumstances of ETI contact and who learns each other’s languages, human language interpreters may be involved. Teams and Individuals

Linguistic fieldwork is often conducted individually. Bowern (2015) discusses the trope of the lone fieldworker dropped in an “alien” community. The team approach, assuming there are multiple community members to work with, will likely provide more progress more quickly, as well as allowing more specialization. However, as discussed later in this chapter, most methods of field data storage and analysis are not easily used by teams; they are designed for individuals working alone. This is a problem for human fieldwork, as well. If the project is conducted as part of a group, one needs a project manager. This type of fieldwork is particularly common in English-speaking countries, but other traditions exist. As Thomas (2020) notes, the canonical Russian field tradition is group-oriented (cf. work in particular led by Kibrik [cf. Kibrik 2006]). Remote or In Person?

Fieldwork is typically assumed to be the linguist going elsewhere. Before Malinowski were the ethnographic expeditions (compare Ray and Haddon 1891; Haddon 1935) that involved travel to remote parts of the Empire (British or Russian, for example) to collect cultural, linguistic, and biological samples from communities. Of course, long before that were the linguistic and cultural souvenirs from travelers and “explorers” (compare Busbecq’s reports on Crimean Gothic [cf. Stearns 1978], for example). Fieldwork is often thought to be about traveling to the other place, to see how language is used in its natural context. Elwin Ransom in C.S. Lewis’ novel Out of the Silent Planet does do this. But Louise Banks in Arrival is on Earth, with all the comforts and resources of Earth. This makes a difference, both in resources available to the linguist and in the type of data that one is able to collect and the observations that one is able to make. There is something comforting about being on one’s home turf. But it really limits what one records (cf. Tsikewa 2021 and others for discussions about the artificialness of field methods classes, for example). “Fieldwork” has been about going to the community, observing and participating in the community first hand and using those insights and data to understand language. The COVID-19 pandemic shut down university research and travel, making in-person fieldwork impossible for months or years (cf. Sanker et al. 2021;

Xenolinguistic Fieldwork

101

Williams et al. 2021; Sneller 2022). Some linguists used virtual, “at-a-distance” field methods while unable to travel. These included video linkups with consultants (that is, interaction in real time but through digital connections) and messaging through communication and social media apps such as WhatsApp. Remote work of this type presupposes that the means of communication is transferable over video linkup. Sanker et al. (2021) found many ways in which acoustic data was distorted through remote recordings, for example. Other Logistics

Where to stay and nonlinguistic things to take care of while on fieldwork can take a substantial amount of time, interstellar travel aside. There are the issues of getting there and getting home again. Where to stay while in the community, how to deal with other tasks that typically come about. After all, if the linguist is going to the ET community, they are likely to be as much an object of study and curiosity as ET visitors are to Earth. Fieldwork Tools and Methods

As previously discussed, language is embodied: it is produced by particular bodies and minds/brains, so there is no reason to think that a ET linguistic system will have properties that humans can perceive, learn, or use to communicate with. Developing solutions to that problem will depend on the specifics of the system (for example, modulating audio frequencies to make them perceptible by human ears). Collecting Data

Fieldwork methods should be broadly applicable to a wide range of ET linguistic systems. However, existing fieldwork tools are strongly constrained by human physiology and psychology, along with being (unsurprisingly) designed for items that occur on Earth. Some of these techniques and issues are outlined in what follows. As Coon (2020) and Wells-Jensen and Spallinger (2020) have previously described, these methods all require a considerable degree of cooperative attention and shared understanding in order to succeed, so there are certain cultural prerequisites for even doing linguistic fieldwork. However, since social and cultural learning is part of the definition of language, it is not too farfetched to assume that if there is enough common understanding for interaction of any type, there will be enough for language learning and linguistic fieldwork. Vocabulary collection: A key component of fieldwork is knowing what things in the world are called in the target language. Such words are typically found out by either pointing, asking directly, or encountering in the course of other

102 Claire Bowern

work. Such techniques all require that we are able to distinguish “words” in the language of course. Pointing requires an understanding of what one is eliciting when pointing to things (Wells-Jensen and Spallinger 2020). The literature is full of initial errors from this technique (e.g., “nose” recorded as “I” or “finger” recorded as “you,” for example) but they are usually straightened out fairly quickly with subsequent work. Monolingual elicitation strategies begin with vocabulary collection by pointing, and translation-based approaches usually begin with a wordlist of basic vocabulary (see further Bowern 2015). Interactions: Through observing interactions it is possible to find out what phrases are used for greetings and other (culturally dependent) formulaic situations and interactions, such as describing the weather or wishing someone a happy birthday. Translated sentences: If there is a language in common, the linguist can provide sentences and ask for translations in the target language, or propose sentences in the target language and ask for corrections and back-translation to a common language. Collection of language samples: e.g., recording narrations and conversations and working through them to identify structures and explore meaning once there is something of a language in common. There is a large array of semi-structured and unstructured techniques for obtaining linguistic data that are likely to reveal variety of morphosyntactic and semantic constructions: causation, location, or possession, to name a few. One can also use activities to record vocabulary and grammar, such as narrating what someone is doing. A better technique may be to be led by the ETI research partners and participant observation. That is, rather than requiring ETIs to come up with novel descriptions for items that they may be encountering for the first time on Earth, do field linguistics on the ETI’s home planet. While this will produce better linguistic documentation, the logistics may be insurmountable. Thus contra Slobodchikoff (this volume [Chapter 9]), we have many tools at our disposal for figuring out the features of human languages; mostly this is because we are able to communicate reciprocally (even if nonlinguistically) about various topics, even if we do not know the language. Linguists do not just have to experiment on the research participants, and indeed, the research participant is typically a partner in the research, not experimented on. They should not be experimented on without their consent, and discursive linguistic fieldwork works better with shared common goals and full participation of all. This means that a big part of designing fieldwork tasks should be consideration for the research participants. Typological Surveys

Note that many of the lists are shaped by assumptions about the properties that we expect languages to have. For example, all human languages have ways of

Xenolinguistic Fieldwork

103

describing events, participants in those events, relationships between those participants and those events, and attendant circumstances for those events. All languages have some way of describing events in time and space, though the details vary. Several chapters in this volume discuss these points as key features of human language and as likely to be part of ETI languages, as well. Fieldwork Equipment

The typical fieldwork setup for recording human language work includes audio and video recording devices and accompanying equipment (such as microphones), computer(s), and notebooks. Fieldwork requires some way of storing and retrieving observations. One should not rely on memory, and documentation collections are made for multiple uses. Language samples are typically recorded in some format (audio and/or video), transcribed to text, and the text is analyzed. The transcription is accompanied by annotations for features of interest; annotations and transcriptions are typically time-aligned using software such as ELAN (Wittenburg et al. 2006). Recording devices may not work on other worlds, since they are designed for Earth’s atmosphere and require electricity and batteries that are only obtainable on parts of Earth. Pencil and paper works in space as long as the human linguists do not have to wear protective equipment that will impede writing. It is fairly durable but heavy to transport. Audio devices are also, of course, optimized for recording particular acoustic frequencies, video for particular light wavelengths. Linguistic fieldwork is severely underfunded and reliant on the market of such devices for other purposes (typically amateur music recording and podcasting). ET fieldworkers should be prepared (as fieldworkers usually are) for their recording devices to fail, and any interstellar fieldwork project will need to be well funded for possibly custom recording equipment. Keeping Track of Data

Linguistic field data is stored in a database, so that language samples can be analyzed, compared, and retrieved. Here we are woefully underprepared for ETI languages, since current tools, to be blunt, do a poor job of covering even the diversity shown by human languages. Because linguistic fieldwork data comes in multiple formats and there are particular requirements for handling multilingual records, the type of qualitative research organization software is typically not appropriate for managing linguistic projects,4 but linguistic software itself is designed for a narrow set of fieldwork data types and language structures. The biggest bottleneck for linguistic work is transcription and parsing. For every minute of audio or video data recorded, for example, transcription can take 5 times to more than 100 times as long, depending on the level of detail of the annotation,

104 Claire Bowern

the complexity of the transcription, the quality of the recording, and the experience of the linguist. Speech-to-text (STT) transcription programs are not usually usable, because they are custom-designed for languages with more data.5 Thus, transcription needs to be done manually, by someone familiar with the language. In team-based ETI fieldwork, expect to need five to ten transcribers for every linguist directly asking questions. Linguistic data for language documentation is most useful when it is parsed – that is, when there is an association between elements in a phrase in the field language and a more familiar metalanguage, as in the following example from Bardi (Northwest Australia, Nyulnyulan family). Nyirroogoordoo ɲɪrukuɖu how “How are you?”

minkal? mi-n-kal? 2sg-tr-travel?

A parsed corpus is essential if anyone other than the linguists who collected the data in the first place are going to use the materials. Creating these parses can be done manually or by using a parser that looks up information from a list of words and morphemes and returns a translation. For some languages, existing software works well. But there are plenty of (common!) features of human languages that are poorly handled by existing parsers. In the preceding example from Bardi, minkal is not overtly marked for tense. Tense and aspect in Bardi are given by a combination of tense prefixes and tense/aspect suffixes (Bowern 2012), but in this verb, neither are overtly marked. Null morphemes can create issues for parsers. Parsers are also typically unable to elegantly process extensive allomorphy or suppletion, nonconcatenative phenomena, suprasegmental morphology (for example, morphology marked by tone), or simultaneous exponence as is very frequent in signed languages. They work on transcribe textual data. Any ETI linguistic analysis is going to have to investigate such phenomena and we have no easy way of tracking it when we find it. Archiving and Long-Term Storage

Fieldwork data needs to be stored long-term in formats that will allow retrieval (Bird and Simons 2021; Henke and Berez-Kroeker 2016). Here, too, we are woefully unprepared. Fieldwork generates a lot of data, so any project working off Earth will need ways to transmit data back for safekeeping.6 It is also unclear what archive would host such materials, since Earth-based archives are typically organized around regions or cultures of humans on Earth. On the principle of LOCKSS (Lots Of Copies Keeps Stuff Safe), we would want ETI data collections to be housed in multiple archives and be open access assuming that ETI cultural protocols find that to be acceptable.

Xenolinguistic Fieldwork

105

Descriptive Metalanguage

We will need a metalanguage for description. Linguistic fieldwork – talking about language through language – has developed a complex vocabulary for describing the languages of Earth. It is not at all clear how much of that metalanguage would transfer to ETI linguistics. Throughout the history of fieldwork and description of languages has been describing phenomena in a model of the familiar and realizing that it does not fit. For example, the 19th-century grammar writers of Indigenous Australian languages were unfamiliar with ergativity, and have to figure out ways of describing the case marking patterns they encounter, either by co-opting familiar terms from traditional grammar (that do not fit) or by inventing new ways of describing the phenomenon. This is likely to be a problem for describing ET languages. Other Considerations Linguistic Training

What training do fieldworkers currently get? Tsikewa (2021) is a recent review. She finds, just as Macaulay (2004: 194) did fifteen years earlier: While we generally do a very thorough job of teaching [our linguistics students] how to elicit and analyze data, we often forget to tell them that there is a personal and practical side to fieldwork that can very well derail their research if they are not prepared for it. More recent books, including Bowern (2015) and Meakins et al. (2018), do devote space on the parts of fieldwork that do not involve data elicitation directly. But fieldworkers wear many hats and are not equipped for extraterrestrial fieldwork (considering the amount of training that astronauts get). Perhaps fieldwork from Earth (either remote or with ET community members here) is more practical and realistic, but loses a lot of the nuance of in situ work. It might also lead to inaccuracies in the language documentation. It makes it hard to know what aspects of the language are the properties of the individual and what are the properties of a larger social group or ET linguistic community. Most fieldwork these days is not ex nihilo – there is at least something for most languages, or related languages. We know something about the families that languages belong to.7 Fieldworkers very seldom start entirely from scratch these days. There are probably not many people, even those with extensive fieldwork training, who have started to work on a language where there is no information whatsoever about the language, its close relatives, or its geographical neighbors.

106 Claire Bowern

Ethics

Fieldwork research has numerous ethical considerations.8 We cannot assume that it will just be fine to rock up and learn ET languages. Human cultures and individual humans have a wide range of views when it comes to the appropriateness of outsiders learning their languages and using them elsewhere. Fieldwork requires consent: active, continually negotiated, and informed consent (Rice 2006; D’Arcy and Bender 2023). Informed consent is difficult (if not impossible) to achieve without some understanding of shared goals and a metalanguage for communication. There are metalinguistic beliefs – folk linguistic theories about language – that may assist with linguistic fieldwork, or may prevent it. The dominant Western view of language is utilitarian. People do not own languages, for example. Languages are equivalent, they are part of culture (and signal features of culture). But there are other views of language, where it comes from, who can speak it, and so on, that are often difficult to access directly but are important for acting ethically. Fieldwork brings other aspects of ethics and interaction, too. These apply to any human–non-human contact (that is, anyone involved in work with ETIs). Fieldworkers could bring diseases, for example. They are likely to be just as much a subject of curiosity and interest as ETIs on Earth. Language Variation

Thus far, I have been mostly describing fieldwork as it is “traditionally” done, with relatively more focus on features that apply across all (or most) users of a language, rather than the parts of a language where individuals tend to differ. I have not discussed other aspects of language description, such as language surveys of contexts of use, language censuses, or other types of community survey that investigate what languages are used and where. I have not touched on the investigation of linguistic variation. While there are many points in common between sociolinguistic/ variationist methods and “fieldwork” methods for language documentation, they tend to target different aspects of language, produce different data, and – crucially – involve work with quite different numbers of community members. All of this will be needed, too. Second, while I have here been talking about fieldwork with an ET language, there is no reason to expect a single homogenous linguistic system. Fieldwork in multilingual situations is the norm, and we have no reason to assume that any ET encounter will involve a single system. Fieldwork on METI (Messaging ETI)

So far, most of this chapter has been about fieldwork as interaction, but there are techniques that fieldworkers can be helpful for when considering METI and decoding extraterrestrial messages.

Xenolinguistic Fieldwork

107

Such work is considerably harder than fieldwork. Human language records are incredibly difficult to decipher when divorced from their context of use, or when immediate and somewhat guided interaction is not possible (or where scaffolded questioning is not possible). It is really only possible when we either know what language is likely to underlie it (e.g., in the case of a cipher), where we have a “Rosetta stone” (that is, enough parallel translation that we have anchoring points and can extrapolate from what we already know to what we do not know), where there is enough inferred context that we can make a guess as the types of structures that are present, or where we have records from a closely related language. In addition, we also need enough data to be able to test theories. Conclusions

We are poorly prepared for extraterrestrial fieldwork. Quite apart from the logistics of a field trip beyond Earth, linguists lack both training and – especially – the tools to do this well. This is a highly skilled part of linguistics, requiring extensive domain knowledge, but linguistics training tends to compartmentalize early: to particular subsystems of language (the sound system or the syntax, for example). Crosslinguistic and cross-field expertise is not as common as it used to be. Where we really need more infrastructure, however, is in the fieldwork tools. Fieldwork recording devices and data organization software is very heavily skewed towards a small part of the human language space. We cannot even use these tools adequately on the array of human language data that linguists routinely collect. The major tool for linguistic organization is designed for Bible translation. It cannot store audio, let alone be adapted for other modalities that might be encountered. Thus, to summarize, language data collection is not straightforward, and it is just scratching the surface of what good linguistic fieldwork actually is. Fieldwork relies on a bunch of shared assumptions about how the world works, about human interaction, and about language – none of which can be taken for granted in a ETI encounter. It requires sustained interactions and the building of shared understanding, and our tools are woefully inadequate to store the data we are likely to obtain. Notes 1 I write this as a researcher with twenty years’ experience in doing linguistic fieldwork and teaching about field linguistics and language documentation, and conducting research on archival collections. My work involves spoken languages, mostly from Australia and belonging to Indigenous Australian families (cf. Bowern 2023 for more details). I work in a paradigm of fieldwork that centers communities. 2 Note that for ethnographic anthropology, the mean method is participant observation, but linguistic fieldwork usually has a more structured element, as well. 3 Malinowski (2007: 46) writes: “Since you take up your abode in the compound of some neighbouring white man, trader or missionary, you have nothing to do, but to start at

108 Claire Bowern

4 5 6 7

8

once on your ethnographic work.” For discussion of precursors to “linguistic fieldwork” traditions outside of ethnography, see Thomas (2020). Annotation software like NVIVO allows annotation of documents and collections of audio and video snippets, but has no way of handling parsed text or collating dictionary materials. This is slowly changing, but the incorporation of usable STT into a field workflow is still a long way off. Presumably, such issues have been at least somewhat addressed by transmission from space telescopes, but these procedures are not currently part of linguistic workflows. It is worth noting that the vast majority of the languages spoken today go back to around 150 languages spoken 10,000 years ago, most which were spoken in one of three areas where agriculture developed. For all the linguistic diversity still around today, it is a small slice of the total linguistic diversity in human history, and there is no reason to believe that it represents the total area of the design space of human language. Some of these ethical considerations are primarily “regulatory” – that is, they stem from regulations that govern research through federally funded institutions like most universities.

References Bird, Steven, and Gary Simons. 2021. Towards an Agenda for Open Language Archiving | IDEALS. Accessed September 9, 2022. www.ideals.illinois.edu/items/119580. Bowern, Claire. 2012. A Grammar of Bardi. Berlin and Boston: Mouton de Gruyter. Bowern, Claire. 2015. Linguistic Fieldwork: A Practical Guide. Dordrecht: Springer. Bowern, Claire. 2023. The Oxford Guide to Australian Languages. Oxford: Oxford University Press. Bowern, Claire, and Luke Lindemann. 2021. “The linguistics of the Voynich manuscript.” Annual Review of Linguistics 7: 285–308. Brewer, John. 2000. Ethnography. London: McGraw-Hill Education (UK). Charity Hudley, Anne, Christine Mallinson, Mary Bucholtz, Nelson Flores, Nicole Holliday, Elaine Chun, and Arthur Spears. 2018. “Linguistics and race: An interdisciplinary approach toward an LSA statement on race.” Proceedings of the Linguistics Society of America 3: 1–14. Coon, Jessica. 2020. “The linguistics of Arrival: Heptapods, field linguistics, and Universal Grammar.” In Language Invention in Linguistics Pedagogy, 32–48. Oxford: Oxford University Press. https://doi.org/10.1093/oso/9780198829874.003.0004. D’Arcy, Alexandra, and Emily M. Bender. 2023. “Ethics in linguistics.” Annual Review of Linguistics 9 (1): null. https://doi.org/10.1146/annurev-linguistics-031120-015324. Everett, Daniel. 2001. “Monolingual fieldwork.” In Linguistic Fieldwork, edited by Paul Newman and Martha Ratliff. Cambridge: Cambridge University Press. Haddon, Alfred Cort. 1935. Reports of the Cambridge Anthropological Expedition to Torres Straits. CUP Archive. Henke, Ryan E., and Andrea L. Berez-Kroeker. 2016. “A brief history of archiving in language documentation, with an annotated bibliography.” Language Documentation 10: 47. Hockett, Charles F. 1959. “Animal ‘languages’ and human language.” Human Biology. JSTOR 31 (1): 32–39. Kay, Paul, and Luisa Maffi. 1999. “Color appearance and the emergence and evolution of basic color terms.” American Anthropologist 101: 743–760. Kibrik, Aleksandr. 2006. “Collective field work: Advantages or disadvantages?” Studies in Language 30: 259–279.

Xenolinguistic Fieldwork

109

Ly, Annelise, and Ingrid Onarheim Spjeldnæs. 2021. “Strategies to survive on foreign turf: Experience sharing and reflections from two apparent aliens in the field.” In Field Guide to Intercultural Research, 205–216. Edited by David S. A. Guttormsen, Jakob Lauring, and Malcolm Chapman. Cheltenham: Edward Elgar Publishing. Macaulay, Monica. 2004. “Training linguistics students for the realities of fieldwork.” Anthropological Linguistics 46 (2): 194–209. Malinowski, Bronislaw. 2007. “Method and scope of anthroplogical fieldwork.” In Ethnographic Fieldwork: An Anthropological Reader, edited by G.M. Robben and Jeffrey Sluka. Oxford: Blackwell. Meakins, Felicity, Jennifer Green, and Myfany Turpin. 2018. Understanding Linguistic Fieldwork. Abingdon, UK: Routledge. Ray, Sidney H., and Alfred C. Haddon. 1891. “A study of the languages of Torres Straits, with vocabularies and grammatical notes (part I).” Proceedings of the Royal Irish Academy (1889–1901) 2: 463–616. Rice, K. 2006. “Ethical issues in linguistic fieldwork: An overview.” Journal of Academic Ethics 4 (1): 123–155. Rice, K. 2009. “Must there be two solitudes? Language activists and linguists working together.” In Indigenous Language Revitalization: Encouragement, Guidance, and Lessons Learned, edited by J. Reyhner and L. Lockard, 37–59. Flagstaff: Northern Arizona University. Rice, Mskwaankwad. 2022. “Power and positionality: A case study of linguistics’ relationship to Indigenous peoples.” Proceedings of the Linguistic Society of America 7 (1): 5295. https://doi.org/10.3765/plsa.v7i1.5295. Sanker, Chelsea, Sarah Babinski, Roslyn Burns, Marisha Evans, Juhyae Kim, Slater Smith, Natalie Weber, and Claire Bowern. 2021. “(Don’t) try this at home! The effects of recording devices and software on phonetic analysis.” Language 97: e360–e382. Sneller, Betsy. 2022. “COVID-era sociolinguistics: introduction to the special issue.” Linguistics Vanguard 8: 303–306. Stearns, MacDonald. 1978. Crimean Gothic. Analysis and Etymology of the Corpus. Saratoga, CA: Amma Libri. Thomas, Margaret. 2020. “The monolingual approach in American linguistic fieldwork.” Historiographia Linguistica 47 (2–3): 266–302. https://doi.org/10.1075/hl.00078.tho. Tsikewa, Adrienne. 2021. “Reimagining the current praxis of field linguistics training: Decolonial considerations. Language. Linguistic Society of America 97 (4): e293–e319. https://doi.org/10.1353/lan.2021.0072. Wells-Jensen, Sheri, and Kimberly Spallinger. 2020. “Extraterrestrial message construction: Guidelines for the use of xenolinguistics in the classroom.” In Language Invention in Linguistics Pedagogy, 239–250. Oxford: Oxford University Press. https://doi. org/10.1093/oso/9780198829874.003.0014. Williams, Nicholas, Wilson D.L. Silva, Laura McPherson, and Jeff Good. 2021. “Covid-19 and documentary linguistics: Some ways forward.” Language Documentation and Description 20: 359–377. Wittenburg, P., H. Brugman, A. Russel, A. Klassman, and H. Sloetjes. 2006. “ELAN: A professional framework for multimodality research.” Proceedings of LREC 2006, Fifth International Conference on Language Resources and Evaluation, European Language Resources Association (ELRA), Luxembourg, 1556–1559.

11 INVESTIGATING THE FOUNDATIONS OF MEANING IN A XENOLANGUAGE Andrew McKenzie

Introduction

This chapter explores how we might discover the nature of meaning in extraterrestrial languages, and what that may or may not tell us about how extraterrestrials conceive their world. Any discovery will build off of three distinct threads of analysis that have only intertwined in the last generation of research. First, we have the linguistic task of inducing generalizations about observed morphemes; next, the philosophical goal of understanding how knowledge is encoded and transmitted; finally, we have the modern linguistic theory of discovering the nature of human linguistic knowledge as a cognitive object. Each of these strands of modern semantics plays a key role in understanding what we might find out. Given an alien species, how do we engage in understanding not only the forms of their languages, but the meanings?1 Clearly, we will need a solid footing in truthconditions. We can reasonably assume they have objects like utterances, indexical forms that relate to them, and also referential forms that point out the objects around them and elsewhere. Yet we must also understand if and how they quantify over things, express modal claims, or combine modifiers. We must observe how they organize their discourses, and seek whether their languages reflect that organization. We must also explore the nature of their semantic ontologies, down to the most basic levels. Do they distinguish entities that ‘are’ from events that ‘happen’? Do they conceive of modal claims in terms of possible worlds? How do their languages reflect their cognition of mereology? Do their languages show them to treat causation differently from us? Do they share similar spatial and temporal awareness? These questions and others are not merely curious inquiries of linguistic structure. In many ways, a speaker’s semantic (and pragmatic) behavior can reveal their DOI: 10.4324/9781003352174-11

Investigating the Foundations of Meaning in a Xenolanguage 111

underlying patterns of cognition, as well as their cultural habits and mindsets. Yet we must take care not to overstate the ways that cognition, language, and culture are intertwined. So, as we speculate about what meaning xenosemantics might involve, we must consider a method to find it. The Truth About Truth

At its heart, semantics is the bridge between the word and the world. Under a modern generative framework, it links the core linguistic component to the conceptual-intentional module of the mind. Under more cognitively oriented approaches, semantics is simply a part of how we organize thought. In either case, once we separate the pieces of language that have no near relation to meaning, and we filter out non-linguistic thought processes and cultural habits, we end up with our field of study: some body of knowledge held by linguistic beings. Ascertaining this knowledge is difficult. Any generalization requires us to iron out variations, but we have rarely – if ever – focused on discerning which variations even matter, the way we have for sound. When a person emits a speech sound, a listener’s ears pick up a signal, and the listener’s mind catches upon a variety of phonetic cues, adjusting for variations within and across speakers. In doing so, that mind situates that sound within a class (a phoneme) that thus relates it by opposition to other classes. Linguists who easily worked out how we do this with sounds also argued that doing the same for the semantics was a fool’s errand. Leonard Bloomfield pointed out nearly 100 years ago that linguists have to assume that the concept linked to a word in one person’s mind is the same as that in another person’s mind. How could one possibly know whether my lexical item ‘cat’ actually means the same as yours? If our personal lexemes do not mean the same thing, how can we be certain that they are even the same lexical item? He called this assumption the fundamental assumption of linguistics, suggesting a problem that could not be overcome (Bloomfield 1933). Observation makes us wonder, though, if this assumption is really problematic. Speakers rarely have difficulties using words like ‘cat’ with each other, so we know either that speakers make this assumption all the time or that these lexical items are unproblematically non-identical. Perhaps they vary in ways similar to the ways that non-meaningful linguistic categories vary, and minds adjust accordingly. Adjustment does not entail that speakers always agree, but that is also the case with sounds. U.S. Supreme Court Justice Potter Stewart famously wrote in a judicial opinion for an obscenity case that defining ‘pornography’ was difficult, but suggests that what we might call semantic intuition guides us toward understanding what the term might comprise (Jacobellis v Ohio, United States 1964: my emphasis). I shall not today attempt further to define the kinds of material I understand to be embraced within that shorthand description, and perhaps I could never succeed

112 Andrew McKenzie

in intelligibly doing so. But I know it when I see it, and the motion picture involved in this case is not that. Participants in legal proceedings might obviously want something more predictable to guide their actions, but outside the narrow proscriptions of the law, “I know it when I see it” is a very ordinary method that humans use to categorize their world semantically. The fact that each person sees ‘it’ differently, and thus knows ‘it’ differently, is not controversial; it even underpins de Saussure (1916 [1995])’s notion of langue, Chomsky (1986)’s notion of e- vs. i-language, and so on.2 Variations like this do not bother speakers generally, and should not bother any xenolinguists. Being a crucial component of linguistic behavior, “I know it when I see it” is thus a fundamental component of understanding what we can find out about semantic meaning in humans, and undoubtedly other species as well. We cannot directly observe what two people’s meaning of ‘cat’ is, but we can empirically test for sameness by observing whether or not they always ‘see it’ in the same contexts. If two speakers agreed 100% of the time on whether a number of given objects were cats or not, we can reasonably conclude that their known meaning of ‘cat’ was identical to one another’s. We can draw an analogy to an optometrist, who cannot observe what their patients are perceiving. Instead, they can only draw conclusions based on the patients’ verbal behavior: answering “yes” or “no,” reading figures on a Snellen test, saying when the ball comes into view, etc. Even advanced scanning devices cannot answer these questions. The doctor can induce what a patient can and cannot see clearly, and conclude that their visual acuity is the same as anyone else who gave the same answers at the same stages of the eye test. This in turn leads to an accurate prescription, even if the people with the same prescription still have slight variations in vision. With semantic knowledge, the path to understanding what an alien friend knows involves finding out when it ‘sees it,’ and that leads us to truth-conditions. As Lewis (1970) pointed out succinctly, to know the meaning of an expression is to know the conditions that make it true. Linguistic knowledge of an expression’s meaning is the knowledge of those conditions, even if a linguist still cannot quite tell exactly what that knowledge contains. One might wonder how well truth-conditions can be applied to anything besides assertions. What makes a question or a command true? Clearly Lewis overgeneralized. That said, the meanings of these other speech acts still rely on truth-conditions. If I ask you “Does Maren drink coffee?” I am essentially laying out for you a proposition (Maren drinks coffee), but instead of asserting that it has a value of ‘true’; I am instructing you to assert its truth-value to me. The answer still depends on the same world conditions that would make “Maren drinks coffee” true. If I tell you “Get Maren some coffee,” my command will only be satisfied if the proposition that you get Maren some coffee becomes true.

Investigating the Foundations of Meaning in a Xenolanguage 113

Alternately, some semanticists have instead formulated meaning in terms of its effect on the speech context: what the speech participants know and have in mind (Kamp 1981; Heim 1982). Even here, though, the particular effects on the context is largely rooted in truth-conditions or easily linked back to them. Besides truth-conditionality, another key aspect of linguistic meaning is compositionality: the meaning of a complex expression is built solely out of the meanings of its parts. Just as speakers do not generally memorize the forms of entire sentences, they do not generally memorize the meanings of entire propositions. Instead, we build them each time, starting with the structure built by some level of the syntax. The mechanisms for composing meaning depend tightly on the denotations of lexical items, so most of semantic research eventually boils down to that. Semantics and Perception

Much of the semantic research into the lexicon over the years has also delved into some kind of psychology, because it is fairly obvious that lexical meanings directly relate to how our minds have ‘carved out’ the spaces of reality around us, the way our vision delineates the borders of objects, and so on. What is less obvious is what those relations are, how deep they go, and what ‘carving’ even means. Whitney (1867) figured that “separate articulated signs of thought” help humans make sense and organize the world around them. More importantly, humans could share these realities far and wide. not only were we thus assisted to an intelligent recognition of ourselves and the world immediately about us, but knowledge began at once to be communicated to us respecting things beyond our reach. (Whitney 1867: 13) This organization is quite evident in terms denoting human-assigned categories, which are relations that only exist in our knowledge. These include family relations, which mix biological and cultural notions. We see it in the names of places, like membership in a mountain range or a continent. Certain objects only exist in our knowledge, like the constellations in the night sky, or the red dot from a laser pointer (a collocation of instants of light), and their names reflect a recognition of our world. This conception of language taming our thoughts reappears over the years. Bréal (1897: 271 et seq.), who coined the term sémantique, argues that language is a translation of reality whose real value comes from how it objectifies our preexisting thought by making vague ideas solid enough to transmit. No doubt it must be the case that the idea came first: but this idea is vacillating, fleeting, difficult to transmit; once it’s incorporated into a sign, we are more sure

114 Andrew McKenzie

of possessing it, of handling it as we wish, and of communicating it to others. Such is the service rendered by language: It objectifies thought.3 Saussure’s seminal work went a clear step further with his concept of signe, which includes a linguistic representation and a mental concept. This is not a referential link like we would imagine; the one does not exist without the other, so whatever the actual nature of things, linguistic signs help us carve out the mental spaces for them in our minds. For de Saussure ([1916] 1995: 99) this leads to conclusions we might today consider ‘Whorfian.’ it is clear that only the combinations consecrated by the language system appear to us to properly fit reality.4 This sort of intricate link between the meaning of linguistic items and the mental conceptions of humans did not catch on very well among structuralists who soon dispensed with Saussure’s notion of signe for empirical reasons. Bréal, for his part, had already emphasized how semantic changes in natural language demonstrated the flexibility of human thought processes, and their independence from language. Language helps us pinpoint and share clear ideas, but it is not a requirement to have them. Americanists avoided these questions altogether in their main line of research, eschewing semantic inquiry on positivist grounds. We cannot observe mental states, so leave that for later research. And in any case, tying one’s linguistics to particular theories of psychology runs a risk as those come and go. Instead, they focused on inducing generalizations about the observable usage conditions of the welter of unheralded morphemes they discovered as they trekked around the globe documenting languages. This approach is very much like the modern reliance on observing truth-conditional behavior discussed previously, and their findings with this method have proven to provide their most lasting results. Nonetheless, they often came to assume a tight link between meaning and worldview. Famously, Edward Sapir would claim such a tight link that the semantics directly shaped cognition, as much as vice versa. Human beings do not live in the objective world alone . . . but are very much at the mercy of the particular language which has become the medium of expression for their society . . . the ‘real world’ is to a large extent unconsciously built up on the language habits of the group . . . . We see and hear and otherwise experience very largely as we do because the language habits of our community predispose certain choices of interpretation. (Sapir 1929: 209–210) Sapir’s student, Whorf, argued (1956) that a Hopi speaker’s sense of time was distinct from an English speaker’s because the Hopi language’s temporal semantics was not built on the same notion of how time progresses.

Investigating the Foundations of Meaning in a Xenolanguage 115

Such speculative claims, now called linguistic determinism, were not empirically backed, either by linguistics, psychology, or a cursory examination of multilingual speech communities. For one thing, Whorf does not seem to have actually asked any Hopis about their time concepts, or even conducted basic linguistic documentation. We know now that Hopi temporal semantics is built on the same ordering of times that every other language is (Malotki 1983). More recently, experimental ‘neo-Whorfian’ research finds extremely minor differences in aspects like color discernment or directional orientation, yet even those results turn out to involve lexical selection rather than actual differences in perception (Li and Gleitman 2002; Li et al. 2011). Looking Into the Lexicon

We can be confident that we will discover linguistic facts that reflect differences in our friends’ cognition, rather than shape it.5 We can begin to explore those differences just by asking about lexical items. However, we must be careful with our methods, because they were designed with human experience in mind. For instance, one classic idea is to begin fieldwork with a lexical list, such as a Swadesh list: more than 100 words rooted in universal human experience, including local person pronouns like ‘you,’ basic elements of matter like ‘water,’ and body parts like ‘arm.’ Pronouns and numerals are occasionally replaced . . . but such replacement is rare. The same is more or less true of other everyday expressions connected with concepts and experiences common to all human groups, or to the groups living in a given part of the world during a given epoch. (Swadesh 1950: 157) Swadesh designed this list for research in historical linguistics, by eliciting words unlikely to be borrowed. Yet it proves useful for the first moments of fieldwork because the linguist is more likely to elicit something than by choosing words at random. After all, we are asking about experiences common to any human. A ‘hit’ is not completely likely, though, because a surprising number of languages lack a ‘word’ for ostensibly simple universal concepts – or they have several words where English has one, making the translation inexact. More frequently, there are mismatches in lexical spaces from differences in encoding – or, what is an ordinary noun in one language is only used as a finite verb with agreement in another, or what is a free word in English turns out to be a bound stem in the other language, and cannot be expressed without some other stem to carry it. Still, out of hundreds of words on these lists, there are bound to be a healthy number of hits that can kickstart a fieldwork process, simply due to the relative universality of the human experiences the list evokes.6 One immediately sees how a list rooted in human experience can be a problem with non-Earthlings. Even with a small alphabetical selection of 25 words (Table 11.1), which might get 20 hits with a human language, I can only see a

116 Andrew McKenzie TABLE 11.1 Selection of words to elicit from Swadesh (1950)

egg eye Far father fire

flower fog foot good grass

green hair hand head heart

here hit hunt husband ice

lake laugh leaf left hand leg

handful that we could expect an alien would be able to translate at first (in boldface). Several others we could expect them to at least understand, but some, like ‘fog,’ ‘flower,’ ‘grass,’ ‘laugh,’ ‘hand’ . . . perhaps those are foreign to them. Granted, we would not have to use linguistic evidence alone to see what they perceive, nor should we. Our friends would hopefully consent to a wide variety of psychological and medical tests to help get a sense of their perception. For linguistics, though, the first days will be tricky. Trickier still, in fact, because the list also assumes the speaker and linguist already share one common language to use for inquiry. Referring to Reference

Although we cannot assume that xenolanguages would refer to or describe things the same way we do, we can healthily assume they contain methods of reference to the objects of the world. In this way, we can rely at first on pointing at things in the room, hoping for a match. Unfortunately, we then reach Quine (1969)’s gavagai problem: We cannot know, at least at first, that our hope is true. Imagine that I point at a sitting dog wagging its tail and say “dog.” Do they know I am speaking about the entity and not some part of it, or some group it is a part of (mammal, animal)? Or do I mean its color, its furriness, its being alive, the act of sitting, its happiness, cuteness, puppy-dog eyes, odor, food, or what? Do they even understand that I am describing a particular object, instead of a generic concept? Or do they think I am asserting possession (“mine!”), or even offering the dog, saying “you can have it”? Consequently . . . what would their translation mean? Experience shows that speakers of distinct languages eventually surmount mutual unintelligibility, as the existence of pidgins demonstrates. Even the paltry vocabulary lists of colonial merchant-explorers stand as a testament to overcoming this hurdle. Jacques Cartier’s 1545 expedition records enough words from the villages along the St. Lawrence River (including the name of a village, Canada), that linguists today can tell that the inhabitants spoke an Iroquoian language distinct from any still known. This list consisted of body parts, person types (man, woman, child, etc.), flora and fauna, tools and implements, and so on (Cartier 1863). After a while, with basic vocabulary and a lot of help from gestures, the expedition and the locals were able to communicate about certain kinds of information (Huchon 2006).

Investigating the Foundations of Meaning in a Xenolanguage 117

This history suggests to us that groping around the lexicon can get us started, barring massive differences in ontological perception and awareness. In doing so, we can start to gain some linguistic evidence about how our friends mentally compartmentalize the world around them. Onto the Ontology

We would need that evidence, because the way our semantics organizes objects might differ from how our friends’ semantics does. While few semantic universals have been explored in depth, the nature of semantic investigation has led to a number of practical assumptions about fundamental universals of human language. Many of these revolve around the ontology of basic elements (Rett 2022). Building off Montague, Davidson, and so on, semanticists generally agree at least on the distinction between entities and events.7 We generally include truth-values, time intervals, and possible worlds, as well. Some semanticists also include situations, kinds, degrees, and locations, depending on their approach and the theoretical question at hand. As far as I know, no one has made a full ‘semantics’ of a single language with all the fully accepted or well-supported components, so it does remain to be seen if these all truly fit together. In a type-theoretic semantics, simple atomic objects are members of one of these ontological sets, and complex expressions denote functions mapping from one set of objects to another (including itself). We do not question that semantic meaning deals with truth-values, entities, or events, because those are clearly distinct kinds of objects. At least, they seem clearly distinct. In all these years, we have not developed any real criterion for distinguishing them beyond their use in language. Link (1983: 303f.) suggests that “our guide in ontological matters has to be language itself.” Entities are described by nouns and adjectives, while events are described by adverbs and adverbials. Verbs and thematic relations relate entities to events. In the absence of criteria, we think of entities being objects that ‘are’, while events are objects that ‘happen’, but it is hard to actually define that difference. So we rely on entailments and morphological distinctions involving proforms, quantifiers, and modifiers. For instance, in English, it can be used to refer to antecedent events. 1 “The mayor was caught red-handed and it sank his re-election bid.” it = the event of the mayor being caught (red-handed) 2 “The silo exploded and I saw it happen.” it = the event of the silo exploding In many ways, events and entities can be treated in similar ways as far as plurality and mereology are concerned (Link 1983; Bach 1986; Krifka 1990). They even overlap in famous instances, like (3) following, which could no doubt be elicited

118 Andrew McKenzie

from our friends with minor tweaks to the vocabulary. We may find a sentence like this is not ambiguous for them. 3 “Four thousand ships passed through the lock.” = Not necessarily 4,000 distinct boats, but events of ships passing One point of distinction is that relations between events and entities are asymmetrical. A verb relates entities to events. The entity’s role in the event is determinative― switch it out and the event is gone. In contrast, the event only plays a small role in determining the entity. If I see you, you contribute far more to this seeing event being the way it is than it contributes to you being the way you are. This asymmetry holds even if the event significantly affects the theme – say, I disintegrate you,8 your contribution to this event’s being how it is from start to finish dwarfs the event’s contribution to how you were from start to finish. It also holds if the event is nominalized: with the creation of the sculpture, the sculpture is a key component of this event, while the event is only one small part of the sculpture’s history, as many other events may happen to it afterwards. Also, the parts of entities that are themselves entities can have parts that have properties the whole cannot have. Bach (1986: 13) offers this case as an example: 4 “The gold making up Terry’s ring is old but the ring itself is new.” Another key difference is causation. The parts of events cause the whole event in ways that parts of entities do not cause entities. If Jenna climbed Mount Everest, that event is composed of a large number of sub-events, each of which contributes causally to the whole. Indeed, we tend to ignore potential sub-events that are not causal contributors, like stopping to scratch an itch or chatting idly with someone along the way. On the other hand, none of Jenna’s parts cause her – not her arm nor her heart. None of those parts’ parts cause them, either, not even down to the cellular level. Any language whose ontology we have explored behaves similarly, although that is not a large set. We might find that alien languages do not work this way at all; e.g., that the equivalent of ‘old’ cannot apply to the components of something that is the equivalent of ‘new.’ The preceding differences seem to suggest that people distinguish some semantic objects by concepts like causation – but we must not assume that all species would do the same. Speaking of Speech Acts

Setting aside the mode of communication our friends employ, and the things they might talk about, we can safely presume that they will deliver it via speech acts, or an analog thereof; we can still call them speech acts (Austin 1962). Natural

Investigating the Foundations of Meaning in a Xenolanguage 119

languages are also observed to vary little in the sorts of speech acts available, at least in a broad sense: Assertions, questions, apologies, promises, and so on. We might wonder if our friends would have speech acts we have not observed, especially performatives. Searle and Vanderveken (2005) point out the lack of a possible speech act in human languages like “I hereby persuade you,” with a performative effect, because performatives are rooted in social acceptance. Other creatures might have that power, though, and if they could use it on us, I suppose I would agree with them. We can also assume as a hypothesis that their languages will have indexical items that express relations toward these speech acts, based on the universality in human languages. A language without indexicals could logically exist, but indexicality not only makes things far more efficient; it also ropes in the self-aware components of our psychology. Relations between entities and speech acts known as person are features of all human languages, and a large amount of research finds that cross-linguistic person marking boils down to the same small set (see Cysouw 2011 for a summary). The speaker is distinguished from the addressee, though the two can be lumped together in inclusive plurals. Everything else is being talked about. The small cross-linguistic variation allows for derivation via a powerful feature geometry (Harley and Ritter 2002), and the tracking of changes over time. We might imagine other relations, though. We do not, for instance, observe languages that distinguish direct addressees from other (potential) listeners, themselves distinct from things being talked about. We also see that the persons are tied to each particular speech act or utterance, not to entire conversations (first to speak, first to reply, most important person, etc.). In alien languages, we may well find some other setup, and if it is as universal for them as our person setup is for us, it may well reflect some aspect of their cognition. Composed With Compositionality

The meaning of entire clauses is built from the meanings of its parts, and a large body of research has sought to see how that works, building off Frege’s idea of using functional application. A predicate is a function that takes a simple object as its argument. However, that is not sufficient. Heim and Kratzer (1998)’s wellaccepted formulations of various compositional rules only number five, and only a handful of narrow types of conjunction have been added to them (Kratzer 1996; Chung and Ladusaw 2004). With this limitation on composition modes, von Fintel and Matthewson (2008) ask if compositionality can be shown to be universal for humans. If xenolanguages are not so compositional, it may throw a serious wrench in our attempts to learn about them. Compositionality fundamentally affects fieldwork, because it allows us to bootstrap upon previous findings by substituting out single words or morphemes and comparing the meanings of sentences.

120 Andrew McKenzie

Being Pragmatic

This chapter has focused on semantics, but we must not forget the pragmatics for understanding usage in a xenolanguage. A significant amount of communication is indirect. Speech act participants work tirelessly to create and fill deliberate gaps, but in doing so, they rely on several types of tacit knowledge: cultural background, personal experience, observations of other conversations, and so on. We should certainly expect surface pragmatic principles to vary, as we do among linguistic cultures and subcultures. We should not expect a pragmatics as extreme as that of the fictional Tamarians who appeared in an episode of Star Trek: The Next Generation. These humanoids spoke only in cultural references to heroic figures (e.g., “Shaka, when the walls fell” describes events of failure). Whatever we get, we must take care to distinguish pragmatic from semantic elements of meaning. When we scrape away cultural layers, we find basic principles of pragmatics that we usually call Gricean maxims (Grice 1975). These maxims are rooted in the assumption that speech act participants are cooperating, even when we are flouting the maxims. Put another way, the mere act of language is inherently cooperative. Should we expect the same from our friends, or might the act of a xenolanguage be inherently something else? Notes 1 We would also do well to ask: How can we help them discover how our languages work? 2 Never mind what it might mean to know anything in the first place. 3 Translated by the author from French: “Sans doute il a fallu que l’idée précédât: mais cette idée est vacillante, fugitive, difficile à transmettre; une fois incorporée dans un signe, nous sommes plus sûrs de la posséder, de la manier à volonté et de la communiquer à d’autres. Tel est le service rendu par le langage: il objective la pensée.” 4 Translated by the author from French: “il est clair que seuls les rapprochements consacrés par la langue nous apparaissent conformes à la réalité.” 5 If the ‘language of thought’ hypothesis holds – of us and of extraterrestrials – whereby thoughts are built in a separate mental language distinct from the spoken one, then these questions can all be applied to that. 6 A similar approach involves semantic primes, argued to be the fundamental building blocks of linguistic meaning and language-built thought (Wierzbicka 1972; Goddard 1999). The same issues come up, but worse because many of those primes are somewhat abstract. 7 Entities are also called individuals, while events are sometimes called eventualities which are then divided into events and states. Here we will use the broad event to cover eventualities including states, even though there is mounting evidence that states are distinct (see Maienborn 2011 for a discussion). 8 Sorry about that.

References Austin, J.L. 1962. How to Do Things with Words. Edited by J.O. Urmson and Marina Sbisà. Oxford: Clarendon Press. Bach, Emmon. 1986. “The algebra of events.” Linguistics and Philosophy 9: 5–16.

Investigating the Foundations of Meaning in a Xenolanguage 121

Bloomfield, Leonard. 1933. Language. Chicago, IL: University of Chicago Press. Bréal, Michel. 1897. Essai de Sémantique. Paris: Hachette. Cartier, Jacques. 1863. Bref récit et succincte narration de la navigation faite en MDXXXV et MDXXXVI par le capitaine Jacques Cartier aux îles de Canada. Reprint of original from 1545. Paris: Librairie Tross. www.gutenberg.org/cache/epub/12356/pg12356-images.html. Chomsky, Noam. 1986. Knowledge of Language: Its Nature, Origin, and Use. New York: Praeger. Chung, Sandra, and William Ladusaw. 2004. Restriction and Saturation. Cambridge, MA: MIT Press. Cysouw, Michael. 2011. “The expression of person and number: A typologist’s perspective.” Morphology 21: 419–443. doi:10.1007/s11525-010-9170-5 de Saussure, Ferdinand. 1916 [1995]. Cours de linguistique générale. Paris: Payot. Goddard, Cliff. 1999. “Building a universal semantic metalanguage: The semantic theory of Anna Wierzbicka.” RASK – International Journal of Language and Communication 9–10: 3–35. Grice, Paul. 1975. “Logic and conversation.” In Syntax and Semantics 3: Speech Acts. Edited by Peter Cole and Jerry Morgan, 41–58. New York, NY: Academic Press. Harley, Heidi, and Elizabeth Ritter. 2002. “Person and number in pronouns: A feature-geometric analysis.” Language 78 (3): 482–526. Heim, Irene. 1982. The Semantics of Definite and Indefinite Noun Phrases. PhD thesis. University of Massachusetts Amherst, Amherst, MA. Heim, Irene, and Angelika Kratzer. 1998. Semantics in Generative Grammar. Amsterdam, the Netherlands: Kluwer. Huchon, Mireille. 2006. Le français au temps de Jacques Cartier. Rimouski, QC, Canada: Tangence. Kamp, Hans. 1981. “A theory of truth and semantic representation.” In Formal Methods in the Study of Language, edited by Jeroen Groenendijk, Theo Janssen, and Martin Stokhof, 277–322. Mathematical Centre Tracts 135. Amsterdam: Mathematisch Centrum. Kratzer, Angelika. 1996. “Severing the external argument from its verb.” In Phrase Structure and the Lexicon, edited by Johan Rooryck and Laurie Zaring, 109–137. Studies in Natural Language and Linguistic Theory, vol. 33. Dordrecht, the Netherlands: Springer. doi:10.1007/978-94-015-8617-7_5. Krifka, Manfred. 1990. “Four thousand ships passed through the lock.” Linguistics and Philosophy 13: 487–520. Lewis, David. 1970. “General semantics.” Synthese 22 (1/2): 18–67. Li, Peggy, Linda Abarbanell, Lila Gleitman, and Anna Papafragou. 2011. “Spatial reasoning in Tenejapan Mayans.” Cognition 120: 33–53. Li, Peggy, and Lila Gleitman. 2002. “Turning the tables: Language and spatial cognition.” Cognition 83: 265–294. Link, Godehard. 1983. “The logical analysis of plurals and mass terms: A lattice-theroetical approach.” In Meaning, Use and the Interpretation of Language, edited by Rainer Bäuerle, Christoph Schwarze, and Arnim von Stechow, 303–323. Berlin: de Gruyter. Maienborn, Claudia. 2011. “Event semantics.” In Semantics: An International Handbook of Linguistic Meaning, edited by Claudia Maienborn, Klaus von Heusinger, and Paul Portner. Berlin and Boston: de Gruyter Mouton. Malotki, Ekkehart. 1983. “Hopi time: A linguistic analysis of the temporal concepts in the hopi language.” In Trends in Linguistics: Studies and Monographs, edited by Werner Winter. Vol. 20. New York: Mouton.

122 Andrew McKenzie

Quine, W.V. 1969. Ontological Relativity and Other Essays. New York: Columbia University Press. Rett, Jessica. 2022. “A typology of semantic entities.” In Linguistics Meets Philosophy, edited by Daniel Altshuler. Oxford: Oxford University Press. Sapir, Edward. 1929. “The status of lingusitics as a science.” Language 5 (4): 207–214. de Saussure, Ferdinand. 1916 [1995]. Cours de linguistique générale. Paris: Payot. Searle, John, and Daniel Vanderveken. 2005. Speech acts and illocutionary logic.” In in book: Logic, Thought and Action, edited by Daniel Vanderveken, 109–132. Dordrecht, The Netherlands: Springer. Swadesh, Morris. 1950. “Salish internal relationships.” International Journal of American Linguistics 16 (4): 157–167. Quine, W.V. 1969. Ontological Relativity and Other Essays. New York: Columbia University Press. United States, Supreme Court. 1964. Jacobellis v Ohio, 378 U.S. 185. June 22. Accessed November 30, 2022. https://supreme.justia.com/cases/federal/us/378/184/. von Fintel, Kai, and Lisa Matthewson. 2008. “Universals in semantics.” The Linguistic Review 25: 139–201. Whitney, William Dwight. 1867. Language and the Study of Language. New York: Scribner. Whorf, Benjamin Lee. 1956. Language, Thought, and Reality. Cambridge, MA: MIT Press. Wierzbicka, Anna. 1972. Semantic Primitives. Translated by Anna Weirzbicka and John Besemeres. Frankfurt, Germany: Athenäum Verlag.

12 A LINGUISTIC PERSPECTIVE ON THE DRAKE EQUATION Knowns and Unknowns for Human Languages and Extraterrestrial Communication Daniel Ross

Introduction

Given the extensive range of potential communication systems that may exist in the universe, how likely is it that humans would be able to communicate with aliens? By breaking this question down into its component parts, we can attempt to generalize from our knowledge of human languages to predict what extraterrestrial languages might look like. Do extraterrestrials exist? Are they intelligent, and do they have languages? Can we detect and perceive their signals? Can we understand their languages? In 1961, Frank Drake set the stage for the search for extraterrestrial intelligence (SETI), and today’s messaging extraterrestrial intelligence (METI), by proposing an equation as a thought experiment about the existence of extraterrestrial intelligence. In the same way, we can attempt to formalize questions about xenolinguistics with an equation including the unknown values of the sub-questions of extraterrestrial communication. After establishing an equation as a starting point, this chapter surveys current knowledge of human languages, drawing from sources such as Hockett’s design features, to begin the discussion on possible values for the terms. Despite the many unknowns of xenolinguistics and possible variation, the resulting analysis provides an optimistic outlook: It may be possible to communicate with extraterrestrials, at least in principle, if we can overcome certain practical difficulties such as via technology and assuming mutual understanding of a shared goal of communication (see also Minsky 1985). Questions of Extraterrestrial Communication

Without data, the study of xenolinguistics is a challenge. Where do we begin? There are several different ways that we can approach this question. We can address it DOI: 10.4324/9781003352174-12

124 Daniel Ross

conservatively (e.g., Ross 2016): say little, with confidence. We can address it speculatively (as in science fiction, cf. Meyers 1980; and some research, e.g., Chomsky 1983, 2000): say more, but unreliably. Or we can study the question itself, which is the focus of this chapter. We begin by asking what we know, and then consider what else we want to know and how to proceed. There are few – if any – certainties but many possibilities. The conservative approach does not get us far enough, but a speculative approach is too open. Instead, we could consider probabilities, to evaluate what may be likely properties of extraterrestrial communication. We must begin, of course, with what knowledge we do have, based on human language and animal communication here on Earth, attempting to generalize expectations for extraterrestrial communication and identify points of potential variation. Charles Hockett proposed 16 design features used to contrast human languages and animal communication (Hockett 1960; Hockett and Altmann 1968; inter alia),1 and these will be one starting point for the current research, referenced throughout this chapter, where HDF refers to “Hockett Design Feature” (numbered following Hockett and Altmann 1968). Although these properties in some cases also provide insight into possible extraterrestrial communication systems we will see that some are more relevant and others less certain. One obvious question is what a particular signal or language will be like, but for practical reasons, this is likely not the place to start. It is too difficult to predict specific properties because there are too many variables and too many unknowns. More importantly, it is unlikely that we would encounter any particular, expected communication system, as we will only (at least first) encounter one of many possible signals. Strategically, then, it is best to be prepared for a variety of communication types. Otherwise, we should wait and react when a specific signal is detected, especially because our questions will be more immediately answered then and we will likely have plenty of time to analyze the signal, given the already great time delays at interstellar distances even with speed-of-light communication. In the less likely scenario of face-to-face contact, we would still need time to learn the arbitrary details (e.g., vocabulary) of the language of any visitors anyway. What we can – and should – do now is research how to identify signals for effective communication (or prepare our own). The more general question that research on xenolinguistics must address, and the focus of this chapter, is whether (and how) we would be able to understand them. The Drake Equation

In November 1961, Frank Drake organized a discussion of the Space Science Board of the National Academy of Sciences on Extraterrestrial Intelligent Life and Interstellar Communication in Green Bank, Virginia. He set the stage for research that still continues today based on an equation composed of seven terms with unknown values (Table 12.1). This thought experiment identified factors relevant to the search for intelligent life in our galaxy (Drake 1961, 1962, 2013; Shklovskii and

A Linguistic Perspective on the Drake Equation

125

TABLE 12.1 Drake’s equation

N = R* × fp × ne × fl × fi × fc × L N R* fp ne fl fi fc L

Number of detectable civilizations in the Milky Way Rate of star formation Fraction of stars with planets Number of planets per star suitable for life Fraction of such planets where life emerges Fraction of life that evolves to become intelligent Fraction of civilizations that are detectable (radio broadcast) Length such civilizations survive, remaining detectable

Sagan 1966; Rospars et al. 2013; Vakoch and Dowd 2015; inter alia). Importantly, the goal was not to actually calculate an answer, but to emphasize the importance of the unknowns and encourage research on these topics. When multiplied together, these unknown values estimate the number of planets in our galaxy hosting intelligent life that we might contact or be contacted by. In the years following the proposal, many variants have been suggested, but research on the original parameters has continued and advanced substantially. We have a much better understanding of some terms such as the rate of star formation or the number of exoplanets in the galaxy, while others such as the frequency of life becoming intelligent remain mysterious. A Drake Equation for Linguistics?

Can we do the same for xenolinguistics? Questions about the topic are as old as the Drake Equation itself (e.g., Freudenthal 1960; Wooster et al. 1966), but much of the previous work has not been connected or followed a specific direction and has instead generally repeated the same ideas, not substantively progressing overall yet mentioning some important topics (e.g., Hockett 1955; Warner 1984; Minsky 1985; Holmer 2013). There are several questions we might ask: Are there xenolanguages? How many? This is unknown but already embedded – at least by implication – in Drake’s original equation. What are these xenolanguages like? This question is too specific and multifaceted to be used to formulate a general equation. Instead, let us ask a more fundamental question, introduced earlier: Can we understand them? This is an important question, and actually encompasses those more specific questions and others. As with Drake’s equation, the goal is not to calculate an answer but to guide ongoing research. Some general questions to consider: (1) Do intelligent extraterrestrials exist?; (2) Will we interact with them?; (3) Can we detect and perceive their signals?; and (4) Can we understand their languages? With these sub-questions, we can try to determine our likelihood of communicating with extraterrestrials, as shown in Table 12.2.

126 Daniel Ross TABLE 12.2 Equation for xenolinguistics

C=N×E×P×U C N E P U

Estimates the number of extraterrestrial civilizations with which we may be able to communicate effectively Number of intelligent extraterrestrial civilizations; “answer” to Drake Equation Probability of encountering or observing them Probability of detecting and perceiving their signals Probability of understanding their communication

The first two terms are directly related to the original Drake Equation and SETI goals; these are not narrowly linguistic questions. The first term, N, is the “answer” to the original Drake Equation, and therefore links these research questions to the ongoing work of SETI in general. Drake’s equation may need to be adjusted slightly to estimate this term relevantly (for example, including contact other than radio signals, whether face-to-face or other interstellar signals such as optical SETI/METI), and also by explicitly accounting for the likelihood (or assumption) that sufficiently intelligent extraterrestrials will have a relevant language for consideration. Several of Hockett’s design features are also applicable at this point – HDF6: Specialization (communication for communication), HDF7: Semanticity (reference, meaning), HDF8: Arbitrariness (conventional form-meaning pairings),2 and HDF15: Reflexiveness (language can discuss language). There may be species in the universe with more basic communication systems, like animals on Earth, but as argued in Ross (2016), it is those with advanced communication systems that we are likely to encounter and that are therefore most relevant for discussions of xenolinguistics.3 We might refer to this as contact-facilitating intelligence. Such a species would likely be as intelligent as we are, or more so. Indeed, with HDF15 as a likely feature, we might also assume there are extraterrestrial linguists out there wondering about how species on such exotic planets such as Earth communicate, and it would be reasonable to prepare not only for the possibility of encountering xenolanguages, but also extraterrestrial linguists who share with us the goal of mutual communication. We should prepare to teach as much as to learn. The second term, E, is the probability of encountering them, or the probability of contact, whether via interstellar signals or face-to-face interaction. This is not a strictly linguistic question either. There are social factors, and many unknowns in its calculation. However, we can again begin with the original Drake Equation and make an assumption that if there are extraterrestrial civilizations broadcasting signals (or exploring space, eventually including our solar system), contact is likely for at least some fraction of our nearest neighbors. This question will not be addressed in detail in this chapter, but some linguistic factors require a brief discussion here. Paul Grice’s Cooperative Principle (1967, 1975) states that speakers intend to communicate, and also recognize this in their interlocutors (see also

A Linguistic Perspective on the Drake Equation

127

Hobaiter et al., this volume [Chapter 3]). Note that this does not assume behavioral or social cooperation (e.g., arguing or threatening) nor truthfulness (e.g., lying), merely that the communication is purposeful and comprehension is expected: speakers work together to create shared meaning. Similarly, in most cases of extraterrestrial contact, this should apply although we cannot rely on it in the case of invasion, or if we are merely objects of study. HDF10: Displacement (discussion beyond the here and now) would also apply. Any broadcasting (or space-faring) species must have an interest in what is “out there” and have ways of referring to it in speech. This also provides a shared context for humans and any species we might encounter. Beyond this, the remaining questions about this term are more for SETI/METI in general, not xenolinguistics narrowly, so for now let us assume there will eventually be contact and focus on whether that encounter would lead to successful communication. We therefore set aside questions of the existence of intelligent extraterrestrials with languages and the likelihood of contact. The remaining two terms are the focus of this chapter: Perception and Understanding. Will we perceive their signals? Will we understand their language? Assuming a context of appropriate, relevant interaction with two cultures seeking communication, we will return to these questions after first turning to an overview of the grammatical properties of human language. Generalizations From Human Languages

The only experience from which we as linguists can generalize is research on human languages, as well as comparisons with animal communication systems. Yet even for human language, linguists agree about almost nothing at an abstract, theoretical level. There are many open and often controversial questions for the field: What, if anything, is innate? Do languages share the same structure? How do we account for cross-linguistic variation? Confusingly, differing opinions (and alleged conclusions) are voiced equally strongly by objectively qualified researchers, and we have not reached a consensus about most topics. These problems are most apparent when comparing different theories, but the underlying problem is more profound, down to the basic empirical level. For example, various levels of structure that are fundamental to even defining different subfields (e.g., “X is the study of Y”) are controversial. “Words” cannot be defined or identified in a consistent way cross-linguistically (Haspelmath 2011). Syntax is the study of sentence structure, but “sentence” is rarely defined explicitly and there are good arguments, for example, that there are grammatical principles operating at something like the paragraph level (Mithun 2008; Longacre 1979). And are “phonemes” (the individual units of sound in phonology) theoretically or cognitively real? For example, archiphonemes have been introduced to explain the unidentifiable units in merger contexts (when two or more candidates for a phoneme would not be contrastive in a given environment, such as /sPɪn/‘spin’ where //P/ might stand for either /p/

128 Daniel Ross

or /b/), and it is unclear how cross-linguistic comparisons of phonology can be made, given that phonemes are language-specific categories (Maddieson 2018). Similarly, even morphologists disagree about whether morphology is a distinct tier of grammatical structure or simply where phonology and syntax meet. As for making generalizations, almost nothing is shared by all languages, or at least there are no easily observed – especially surface-level – “universals” (Evans and Levinson 2009; inter alia). Haspelmath (2007) has gone as far as claiming that there are no cross-linguistic categories at all. Interestingly, and from a very different perspective, Chomsky’s Minimalist Program also drastically reduces the components assumed to be shared by all languages (cf. Ross 2021). Regardless, we should not expect alien languages to all be the same as human languages, nor each other. The best we can do is keep an open mind about possibilities for extraterrestrial communication and look at what features from human language are likely to generalize and why, such as Hockett did with his design features. Likelihood of Successful Communication

Returning to the equation proposed in Table 12.2, we must now consider the final two terms: Perception and Understanding. In fact, these terms should each be expanded into several components for the following discussion. (PN + PT) × UG × UM × UC Perceiving the Signal

We have two distinct opportunities to perceive extraterrestrial communication. PN represents the probability of our perception of their natural modality of communication (whether auditory, visual, chemical, or otherwise).4 PT represents the probability of using technology to encode the language, from writing5 to radio signals. Either approach would work, so there is no need for both, and this becomes a sum of probabilities.6 Note that this means that the equation estimates our overall ability to achieve successful communication of some sort, not whether we will be able to communicate easily or directly and in the same variety of ways that they do. This is analogous to being able to either speak or write Chinese, but not both. Natural Perception (PN) is the first scenario most people would consider for talking with aliens, but it is also least likely. Even assuming they speak orally, could we hear the range of frequencies they produce? And could we respond? More generally, there are many more possibilities for medium of communication: electromagnetic radiation (including but not limited to visual light), pressure waves (including sound), chemical, tactile, and more. Our speech production abilities and physical anatomy have co-evolved and are specialized. Even if we could perceive it, it is unlikely we would be able to produce a natural response. Consider whale songs, which are similar to spoken language but outside the range of frequencies we produce (and perceive). There are cases of imitation, such as

A Linguistic Perspective on the Drake Equation

129

parrots “speaking,” but at the very least these introduce a strong “accent” and also rely on similar environmental factors in evolution (e.g., atmospheric composition) and even distantly related genetics (having a tongue, mouth, head, etc.). In fact, perception might be so difficult that it could affect term 2 (E: probability of initial contact), because we might not recognize that there is a signal at all. Given the possible variation in signals, there are three increasingly unlikely requirements: (1) recognizing the signal (hearing, seeing, etc.); (2) being able to perceive its components (contrasts); and (3) being able to reproduce the signal. The outlook for PN is not optimistic, and it is unlikely that we could “speak” their languages in the typical sense. Given the additional unlikelihood of experiencing natural communication despite interstellar distances between civilizations, we may wish to concentrate on PT, except to the extent that a decoded radio transmission might itself correspond to properties of PN. Perception via Technological Means (PT) is a viable alternative, either technology-aided reception/interpretation of natural signals or technological encoding. For example, any wavelength of electromagnetic radiation, or frequency range of pressure waves, etc., can be recorded and manipulated to make it perceivable to us, similar to recording and replaying whale songs at an adjusted frequency so we can hear them well. The remaining difficulty would be perceiving the necessary contrasts (e.g., phoneme-equivalents) in the signal. Alternatively, we could entirely bypass the modality by using written, digital, or other encoding to transmit the message. Furthermore, assuming the most likely scenario of interstellar communication at a distance of many light-years, the most likely signal is one that travels at the speed of light, resulting in a bottleneck restricting the channel to radio waves or a limited number of similar technological means, such that we may even be able to predict what sort of signal would be sent. The outlook for PT is much better than PN, and even likely to succeed. Regarding likely modalities, Hockett’s design features are a starting point. HDF3 (Rapid Fading) emphasizes the importance of signals being momentaneous so that communication can proceed quickly, with one symbolic unit rapidly following another, permitting complex signals (Galantucci et al. 2010). More generally, linearization occurs in a relevant dimension (Ross 2016, 2021). In the case of spoken language, that dimension is time, but spatial dimensions can function similarly, such as the way symbols are distributed across a page in writing. Chemical signals do not easily linearize and therefore are less likely to encode complex messages (see also Kershenbaum, this volume [Chapter 2]). Another relevant principle is HDF4 (Interchangeability), such that speakers can also hear, although again we may not be able to do both as the extraterrestrials do. Some features may not apply, such as HDF2 (Broadcast Transmission and Directional Reception: the signal can be perceived widely but perceivers can identify its origin) or HDF5 (Complete Feedback: speakers perceive their own speech); these are likely, but do not apply to all possible systems, such as chemical injection from speaker to hearer,7 or touch which would be felt only by the interlocutors. These scenarios are less relevant to

130 Daniel Ross

communication at interstellar distances, which would likely be encoded by electromagnetic radiation or similar means, although in interpreting the signal, we should be aware of the possible effects of different natural modalities. Potential sources of fundamental variation are introduced by other design features of human language. HDF9 (Discreteness) describes distinct units that make up the speech signal, more like the 1s and 0s of digitally encoded audio than the analog signal of radio waves. HDF13 (Duality of Patterning) describes the twotiered structure of human language: sound patterns and grammatical patterns, interacting at the level of word (or morpheme). It is difficult to even imagine a language that violated or altered these principles, much less what it would be like to speak and understand such a language. Note, however, that human sign languages present a potential example of violating HDF9 and a challenge for research:8 these languages are generally unwritten because it is difficult to design a writing system that captures subtle uses of the physical signing space not strictly constrained to discrete contrasts comparable to spoken languages. Understanding Extraterrestrial Communication

Understanding is divided into three parts: understanding the language grammatically (UG), understanding the message cognitively (UM), and understanding in a cultural context (UC). Regarding the grammatical component (UG), we must determine whether the grammatical structure of extraterrestrial communication is parsable, and relevantly similar to our own languages so as to be comprehensible. Human languages are already highly variable, and even allowing for differences, could we expect that at least some “simple” expressions might overlap in structure? HDF11 (Openness) refers to the ability for language to create new expressions through modification (e.g., compositionality). As argued in Ross (2016), and now followed by several others (cf. Roberts et al., this volume [Chapter 15], contra Chomsky 1983, 2000; Samuels and Punske, this volume [Chapter 16]),9 we can add this to a short, provisional list of truly “universal” features of intelligent communication. Any sufficiently intelligent communication system would include structures for the combination of concepts, which is a fundamental part of the mental manipulation of the world through language, and at the same time, given the otherwise extreme degree of diversity already represented here on Earth, it is possible that human languages may not be so unusual in the universe, either. This would also suggest that at least limited translation (including via paraphrase) between human and alien languages should be possible. Some major components of human languages are likely (McKenzie [Chapter 11, this volume]), including: the ability to describe actions/processes (verbs) vs. entities (nouns); the identification of who did what to whom (argument structure: cf. Langacker 1986); description and modifiers (adjectives, etc.); and complex expressions (coordination, embedding).10 There are also some likely challenges, especially if they are more intelligent than we are:11 consider again extraterrestrial

A Linguistic Perspective on the Drake Equation

131

communication systems with non-discrete contrasts (violating HDF9) or more than two tiers of grammatical structure (violating HDF13), as mentioned previously. Another area for complex grammar comes from metalinguistic structures; that is, grammatical devices are used to overtly modify other grammatical constructions. We often use language to discuss language itself (cf. HDF15), but through this process, we can also develop systematic ways of extending the grammar. For example, even in early, pre-language human communication, restricted to single-word utterances without grammar, the juxtaposition of two utterances would have had contextual significance (e.g., “Danger! Predator!” or “Food! Eat!”), and eventually the linguistic code itself was extended to allow for multi-word utterances of increasing complexity via the combinatorial structure of compositionality (Ross 2016, 2021; Rizzi 2016). Other types of structure-building allow us to expand the grammar even further with more complex constructions. Circumlocution is a type of metalinguistic usage, when we do not quite identify the right words or expression for our communicative purposes, and sometimes this usage spreads; further, in usage, metalinguistic expressions are often inserted as parentheticals, like “. . . I think . . .” within a sentence but not yet directly part of its structure. Metalinguistic commentary, circumlocutions, and reanalysis of existing forms can lead to language change when they conventionalize (Ross 2021).12 1 bahay na maganda (Tagalog: Philippines) house lnk beautiful ‘beautiful house’ (Scontras and Nicolae 2014: 21) 2 Mom and dad visited for the holidays. 3 John and Mary ate pizza and pasta, respectively. As shown in Table 12.3, these metalinguistic devices can be described in a hierarchy of several levels for human languages. Basic modification is obligatorily encoded with a linking element in Tagalog, as in ex. 1, while logical operators may add layers to a simple sentence, as in ex. 2, and additional grammatical devices may act as parsing instructions for the listener to understand the sentence in a particular

TABLE 12.3 Metalinguistic structures in human languages

Level

Example constructions

1

Basic compositionality (Ross 2016; Hauser et al. 2002; inter alia), as well as overt “linkers” marking modification: ex. 1. Logical operations, such as negation (“X is not the case”), coordination, embedding, etc., expanding and combining phrases or sentences: ex. 2. Complex, layered logical expressions modifying existing expressions and giving explicit information about interpretation to the listener: ex. 3.

2 3

132 Daniel Ross

way, as in ex. 3 where specifically (and only) John ate pizza, while Mary ate pasta. Notice that each consecutive level builds on the previous one. We must wonder whether there are intelligent extraterrestrials out there with other features beyond our imagining and current comprehension. Indeed, this is one possible dimension of variation, along with violations of HDF9 and HDF13, that we may not yet be ready to understand, requiring further research. Understanding the Message (UM) would require cognitive abilities sufficient for receiving (via language), retaining, and evaluating extraterrestrial thoughts, which might be more complex than our own. However, we should share some common ground, including basic expressions that may translate universally. We also almost certainly share some common interests such as interstellar signals, or space travel, which may be enough overlap to facilitate some first-contact communication. Understanding from a cultural context (UC) relies on having some shared perspectives and motivations. While their culture may substantially differ from our own, our shared interests will provide a bridge for some topics of discussion (again, consider HDF10: Displacement, discussion beyond here and now). There are, however, many less certain aspects of linguistic culture that do not necessarily interfere with our ability to effectively communicate with them. HDF12 (Tradition) establishes cultural, rather than genetic, transmission as the way a specific language is passed from one generation to the next. Cultural transmission is likely to be encountered in extraterrestrial communication systems, as the product of the natural development of communication through usage. Alternatives are not inconceivable, such as an intelligently designed, perfected communication system for an intelligent species that has moved beyond traditional communication, or perhaps one that developed entirely genetically through evolution including all details down to vocabulary and pronunciation, which might be an indication of a different type of intelligence, less creative and more instinctive.13 HDF16 (Learnability) states that speakers of one language can learn another, which seems relatively likely within species, but less likely across species, such that at best what we would hope for with inter-species communication would be something like a more exaggerated effect of second language learning.14 HDF14 (Prevarication) describes the ability to manipulate the truth, e.g., to lie. It is unclear whether and to what extent these features would be found in extraterrestrial languages. Regarding pragmatics in general, we might expect it to apply similarly, given the Cooperative Principle, from which many other components of pragmatics can be derived.15 In summary, for understanding grammar, there would likely be enough overlap for at least basic messages; for understanding the message cognitively, our shared interests will lead to some similar thoughts; and understanding in a cultural context will be challenging, but shared interests will also help. Outlook

Successful communication with extraterrestrials depends on many factors, but the possibility of cognitive capability for communication is supported,

A Linguistic Perspective on the Drake Equation

133

TABLE 12.4 Preliminary values for xenolinguistics equation

C = N × E × (PN + PT) × UG × UM × UC Underlined (C, N, E) Italic (PN) Bold (PT, UG, UM, UC)

Unknown Unlikely Likely (at least partially)

highlighting the need for further research. Assuming there are intelligent extraterrestrials (N) and we will eventually encounter them (E), perception through technological means is likely possible (PT), and understanding is likely to be at least partially effective in the relevant domains (UG, UM, UC). This means that overall the probability for successful communication using extraterrestrial languages – or for them to communicate using ours – is a relevant possibility, as shown in Table 12.4. The equation proposed in this chapter describes only the possibility for communication, not the difficulty or means of learning to communicate. I have not discussed deciphering a signal we receive (or learning a language via face-to-face communication). The two key steps for this will be in identifying the contrastive units within the signal, as already briefly discussed, as well as understanding the intention and meaning behind the message (on the difficulty of this task, see Slobodchikoff, this volume [Chapter 9], as well as Pepperberg, this volume [Chapter 6]). This is of course facilitated in face-toface communication because trial and error in interaction can lead to additional information, but will be hindered by the essentially one-sided discourses possible at interstellar distances; on the other hand, as mentioned previously, this will provide us with relatively more time to analyze and decode such a transmission, given that even a generation of research by humans would be only a minor delay for communication at a distance of 100 light-years, for example. We also must hope that whoever created the transmission designed it to be intelligible, beyond the mere announcement that there is someone out there transmitting.16 Regardless, in ideal circumstances, with an optimally designed signal carrying enough information to allow analysis and decoding, it seems likely that we could, in principle, learn to interpret an extraterrestrial language, although actually achieving this could be difficult for practical reasons (see Slobodchikoff, this volume [Chapter 9]). Despite the extremes in possible variation and the many unknowns, the outlook is optimistic. If we do encounter intelligent extraterrestrials, there may be no inherent cognitive barriers to achieving at least basic communication. Therefore, human languages may not be so unusual for the universe, either. It is my hope that the discussion presented here is a starting point for continued research on xenolinguistics so that some of the unknowns may eventually become knowns. In turn, this will also lead to a better understanding of our own languages here on Earth.

134 Daniel Ross

This volume provides a foundation for continued research on the languagespecific terms of the equation proposed here, as follows. E: Pepperberg (Chapter 6) and Slobodchikoff (Chapter 9) discuss strategies and challenges for first contact. Hobaiter et al. (Chapter 3) consider the effects of intentionality in communication on interaction. Bowern (Chapter 10) discusses the applicability of fieldwork on human languages to extraterrestrial contexts. PN: the question of natural modality of communication is surveyed by Kershenbaum (Chapter 2) and Nixon and Tomaschek (Chapter 17), including effects of physical environment. Wells-Jensen (Chapter 13) specifically considers the role of blindness in the development of possible extraterrestrial communication systems. PT: Herzing (Chapter 4) also discusses natural modalities as well as technological means to interpret them, and Harbour (Chapter 18) considers the utility of writing as a tool in METI. UG: Roberts et al. (Chapter 5) and Samuels and Punske (Chapter 16) explore implications and applicability of theories of human language on extraterrestrial grammar. UM: McKenzie (Chapter 11) considers potential overlap between conceptual meaning in human and extraterrestrial languages. Sperlich (Chapter 14) examines the possible effects of artificial intelligence in extraterrestrial communication. UC: Granger et al. (Chapter 8), Ortner (Chapter 5), Berea (Chapter 7), and Nixon and Tomaschek (Chapter 7) discuss diversity, expectations, and interaction from a cultural perspective, as well as resulting differences in grammar and cognition. Overall, these contributions tell us to expect diversity yet predict some overlap with certain familiar properties of human languages. They also demonstrate the interrelatedness of the different factors, in turn emphasizing the importance of interdisciplinary research. Notes 1 For Hockett’s own perspective on these questions, see also Hockett (1955), at least for some speculative ideas. 2 Beyond those ideas that can easily be expressed with imitative forms, symbolic communication is required, and this in turn can lead to and support more complex structure (cf. Roberts et al. 2015; Little et al. 2017). 3 An exception would be the scenario of humans as traveling to exoplanets and encountering life face to face, where a wider range of less sophisticated communications systems would be expected. 4 HDF1 (Vocal-Auditory Channel) must be expanded, as well as for manual-visual signed languages on Earth.

A Linguistic Perspective on the Drake Equation

135

5 See also Harbour (this volume [Chapter 18]) for discussion of writing systems applied to extraterrestrial communication. 6 Technically, this part of the equation should be specified as PN + T or PN + PT − PN & T to avoid inflating the probability if both possibilities have a high likelihood. In practice, especially in scenarios such as communication at interstellar distances that restrict our options, we might focus on only one possibility. 7 Compare for example lateral gene transfer (e.g., spreading antibiotic resistance in bacteria: Gyles and Boerlin 2014). 8 See also Kershenbaum (this volume [Chapter 2]) for discussion of non-discreteness in animal communication. 9 My analysis was originally proposed at a conference workshop in 2016, followed by the 2018 meeting which resulted in the current volume and where others presented similar ideas; Ross (2016) has also been submitted for the proceedings of the first workshop, but publication of that volume has been delayed beyond that of this volume. 10 Human languages differ in how they render these functions, and an important question to consider is to what extent human languages already span the possibility space of grammatical strategies that might be found in the universe. 11 Conversely, would it be easier for them to learn our languages? An important question is whether variation in grammatical structure is primarily hierarchical, with more complex languages for more intelligent species, or multi-dimensional, such that the only overlap might be in the more basic elements. The answers to these questions are unknown, but important for continued research on the topic. 12 Related to this point, Peterson (2018) argued that analogy may be a more important feature of human language than even compositionality, because it is what allows us to take existing structures and adapt them to new usage, leading to the grammaticalization of new grammatical constructions to fill communicative needs, and also along the way adding intricacy and even irregularity to the grammatical system, which are hallmarks of human language. These features are also likely in extraterrestrial communication (see also later in this chapter on HDF12). 13 See also Sperlich (this volume [Chapter 14]) on possible relationships between artificial intelligence and extraterrestrial communication. 14 Incidentally, the premise of the movie Arrival (2016) should thus be questioned, which is to say – without spoiling the ending of an interesting movie – that humans are unlikely to become native-like speakers of exotic alien languages. 15 Another topic not addressed here is the diversity of languages within a species or planet, a question for which Hook (1999) relevantly suggests “a Drake Equation for linguistic diversity” based on sociolinguistic parameters. 16 For design considerations see Harbour (this volume [Chapter 18]).

References Chomsky, Noam. 1983. “Interview: Noam Chomsky: The revolutionary linguist who believes we have a ‘language organ’ in our brain also thinks there are some things no amount of learning can teach.” Interviewed by John Gliedman. Omni 6 (2), November: 113–118, 171–174. Chomsky, Noam. 2000. New Horizons in the Study of Language and Mind. Cambridge: Cambridge University Press. Drake, Frank. 1961. “Extraterrestrial intelligent life and interstellar communication: An informal discussion.” Space Science Board, National Academy of Sciences, Green Bank, Virginia, November. Drake, Frank. 1962. Intelligent Life in Space. New York: Macmillan.

136 Daniel Ross

Drake, Frank. 2013. “Reflections on the equation.” International Journal of Astrobiology 12 (3): 173–176. https://doi.org/10.1017/S1473550413000207. Evans, Nicholas, and Stephen C. Levinson. 2009. “The myth of language universals: Language diversity and its importance for cognitive science.” The Behavioral and Brain Sciences 32 (5): 429–448. https://doi.org/10.1017/S0140525X0999094X. Freudenthal, Hans. 1960. LINCOS: Design of a Language for Cosmic Intercourse, Part I. Amsterdam: North-Holland. Galantucci, Bruno, Christian Kroos, and Theo Rhodes. 2010. “The effects of rapidity of fading on communication systems.” Interaction Studies 11 (1): 100–111. https://doi. org/10.1075/is.11.1.03gal. Grice, H. Paul. 1967. “Logic and conversation.” William James Lectures, Harvard University. Grice, H. Paul. 1975. “Logic and conversation.” In Syntax and Semantics 3: Speech Acts, edited by Peter Cole and Jerry L. Morgan, 41–58. New York: Academic Press. Gyles, C., and P. Boerlin. 2014. “Horizontally transferred genetic elements and their role in pathogenesis of bacterial disease.” Veterinary Pathology 51 (2): 328–340. https://doi. org/10.1177/0300985813511131. Haspelmath, Martin. 2007. “Pre-established categories don’t exist: Consequences for language description and typology.” Linguistic Typology 11 (1): 119–132. https://doi. org/10.1515/LINGTY.2007.011. Haspelmath, Martin. 2011. “The indeterminacy of word segmentation and the nature of morphology and syntax.” Folia Linguistica 45 (1): 31–80. https://doi.org/10.1515/ flin.2011.002. Hauser, Marc D., Noam Chomsky, and W. Tecumseh Fitch. 2002. “The faculty of language: What is it, who has it, and how did it evolve?” Science 298 (5598): 1569–1579. https:// doi.org/10.1126/science.298.5598.1569. Hockett, Charles F. 1955, May. “How to learn Martian.” Astounding Science Fiction 55 (3): 91–106. Hockett, Charles F. 1960. “The origin of speech.” Scientific American 203: 89–96. https:// doi.org/10.1038/scientificamerican0960-88. Hockett, Charles F., and Stuart A. Altmann. 1968. “A note on design features.” In Animal Communication: Techniques of Study and Results of Research, edited by Thomas A. Sebeok, 61–72. Bloomington: Indiana University Press. Holmer, Arthur. 2013. “Greetings earthlings! On possible features of exolanguage.” In The History and Philosophy of Astrobiology: Perspectives on Extraterrestrial Life and the Human Mind, edited by David Dunér, 157–183. Newcastle upon Tyne: Cambridge Scholars. Hook, Peter Edwin. 1999. “A Drake equation for linguistic diversity.” October. www-personal.umich.edu/~pehook/drake.html. Langacker, Ronald W. 1986. “Settings, participants, and grammatical relations.” In Proceedings of the Second Annual Meeting of the Pacific Linguistics Conference, November 1986, University of Oregon, Eugene, Oregon, edited by Scott DeLancey and Russell S. Tomlin, 1–31. Eugene: Department of Linguistics, University of Oregon. Little, Hannah, Kerem Eryılmaz, and Bart de Boer. 2017. “Signal dimensionality and the emergence of combinatorial structure.” Cognition 168: 1–15. https://doi.org/10.1016/j. cognition.2017.06.011. Longacre, Robert E. 1979. “The paragraph as a grammatical unit.” In Syntax and Semantics 12: Discourse and Syntax, edited by Talmy Givón, 115–134. New York: Academic Press.

A Linguistic Perspective on the Drake Equation

137

Maddieson, Ian. 2018. “Is phonological typology possible without (universal) categories?” In Phonological Typology, edited by Larry M. Hyman and Frans Plank, 107–125. Berlin: De Gruyter. https://doi.org/10.1515/9783110451931. Meyers, Walter Earl. 1980. Aliens and Linguists: Language Study and Science Fiction. Athens: University of Georgia Press. Minsky, Marvin. 1985. “Why intelligent aliens will be intelligible.” In Extraterrestrials: Science and Alien Intelligence, edited by Edward Regis, 117–128. Cambridge: Cambridge University Press. Mithun, Marianne. 2008. “The extension of dependency beyond the sentence.” Language 84 (1): 69–119. https://doi.org/10.1353/lan.2008.0054. Peterson, David J. 2018. “Dothraki, language creation and other fascinations.” Paper presented at the METI International Symposium on Xenolinguistics at the National Space Society’s 37th Annual International Space Development Conference (ISDC) 2018, Los Angeles, May 26. Rizzi, Luigi. 2016. “Monkey morpho-syntax and merge-based systems.” Theoretical Linguistics 42 (1–2): 139–145. https://doi.org/10.1515/tl-2016-0006. Roberts, Gareth, Jirka Lewandowski, and Bruno Galantucci. 2015. “How communication changes when we cannot mime the world: Experimental evidence for the effect of iconicity on combinatoriality.” Cognition 141: 52–66. https://doi.org/10.1016/j. cognition.2015.04.001. Rospars, Jean-Pierre, Florence Raulin-Cerceau, and Rocco Mancinelli, eds. 2013. “The Drake equation.” International Journal of Astrobiology 12 (3, Special issue). Ross, Daniel. 2016. “Modification as a universal property of intelligent communication.” Presented at the the SETI Plenary Session at the National Space Society’s 35th Annual International Space Development Conference (ISDC), San Juan, May 18. Ross, Daniel. 2021. Pseudocoordination, Serial Verb Constructions and Multi-Verb Predicates: The Relationship between Form and Structure. Ph.D. dissertation, University of Illinois Urbana-Champaign, Urbana. https://doi.org/10.5281/zenodo.5546425. Scontras, Gregory, and Andreea C. Nicolae. 2014. “Saturating syntax: Linkers and modification in tagalog.” Lingua 149: 17–33. https://doi.org/10.1016/j.lingua.2014.05.005. Shklovskii, I.S., and Carl Sagan. 1966. Intelligent Life in the Universe. San Francisco: Holden-Day, Inc. Vakoch, Douglas A., and Matthew F. Dowd, eds. 2015. The Drake Equation: Estimating the Prevalence of Extraterrestrial Life through the Ages. Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO9781139683593. Warner, Richard. 1984. “Exolinguistics: State of the art.” Omni 6 (11), August: 91. Wooster, Harold, Paul L. Garvin, Lambros D. Callimahos, John C. Lilly, William O. Davis, and Francis J. Heyden. 1966. “Communication with extraterrestrial intelligence.” IEEE Spectrum 3 (3): 153–163. https://doi.org/10.1109/MSPEC.1966.5216745.

13 COGNITION, SENSORY INPUT, AND LINGUISTICS A Possible Language for Blind Aliens Sheri Wells-Jensen

Introduction

Reading and writing about extraterrestrial languages, before we have discovered even a single extraterrestrial microbe, is a radical act of hope. Every chapter here quickens with the belief that we could become conversational partners – perhaps even friends – with intelligent beings born on planets circling other stars. Our task here is to facilitate that eventual conversation, to lay the groundwork for its success and anticipate the problems we might face. For many of us, the first step in this process is to examine human languages and push back the barriers between ourselves and what we know about how humans began to speak. Linguists have struggled with this question for decades, sometimes with collegial gentleness and sometimes without – but always with vigor. We have wanted to determine specifically what part (if any) of the human ability to use language is innate and separate from cognition. Answers to this question have ranged from “quite a lot” (Chomsky 1986) to “maybe nothing” (Everett 2016, 2017; Evans 2014) to “we just can’t really know” (Hauser et al. 2014), and we are not much closer to determining the truth than was the Pharaoh Psamtik in 700 BCE, who is said to have isolated two children with the purpose of discovering the first, most fundamental language. The innate component, popularly known as the “language instinct” (Pinker 1994), is the first of three factors pertinent to language acquisition and development (O’Grady 2012; Chomsky 2005). This language-specific mutation, if it exists, would have arisen in the human genome 50,000–150,000 years ago, coincident with the evolution of Homo sapiens. “Second-factor variables” are the experiences babies have with language. It amounts to their exposure to fluent speakers and their experiences as active learners. DOI: 10.4324/9781003352174-13

Cognition, Sensory Input, and Linguistics 139

Everyone agrees that this is necessary, but how exactly children utilize this input to learn languages, and what kind of assistive bootstrapping and strategizing they need, remains a topic of ongoing research (Fletcher and MacWhinney 2017) “Third-factor variables” include characteristics of the environment such as surrounding noise level, air or water pressure, amount of gravity, and chemical makeup of the atmosphere. Also included here are traits within the language learners such as overall body shape, short term memory ability, computational power, amount and kind of memory storage, and kinds of sensory inputs. The question of the relative influences of these three factors is crucial to every chapter in this volume. If human language is deeply influenced by body shape and the environment in which we find ourselves, then we might expect roughly bipedal beings from rocky worlds similar to our own to share some of our cognitive (and thus linguistic) characteristics (Minsky 1985). Maybe we could learn to speak to them. On the other hand, beings who do not share the humanoid body plan or who come from very different physical environments may “speak” languages that would be forever incomprehensible to us (Haden-Elgin 1984). If language is not closely woven into the physical, biological, and cognitive environment – that is, if first-factor variables are significant – we would arguably have an equally wide set of possibilities but for a different reason. If the genetic mutation that brought language to humans is unique, then it is possible that humans would never be able to learn any language that did not evolve on Earth. However, if our language instinct is standard for intelligent beings, we might be able to speak to anyone at all, regardless of the characteristics of their home world, their kind of cognition, or their body plan. (It’s worth mentioning here that, of course, that there is no reason to assume that there is only one “right” answer to this question; it is easy to imagine a universe in which some species rely heavily on a “language instinct” while others develop language as a natural extension of their nonlinguistic cognitive processes.) Given that we do not know what comprises Factor 1, and we agree that Factor 2 is important (if not completely understood), one of the things we can do while we wait for an exemplar of an extraterrestrial language is choose a third-factor variable, carefully isolate it, and work through how it might influence language structure and use. Some of this third-factor work has been begun by our best science fiction writers, accustomed as they are to world building that is both creative and logically consistent. We might, as Mary Doria Russell (1996) has done in The Sparrow, explore a cultural scenario on a planet where a predator and prey species both evolve to sentience. Or, it might be useful, as Becky Chambers (2014) has done in the Wayfarer series, to explore how standards of reference to gender and ability evolve when several species live and work together. Or, as James Cambias (2014) has done in his novel The Darkling Sea, imagine the cultural and linguistic effects of being a large, water-dwelling squid-like creature so far beneath the surface of an ocean that there is no usable light.

140 Sheri Wells-Jensen

As I personally have no experience (as a human) being squid-shaped, but substantial experience (as a blind human) living in the world without chronic reliance on visible light, I will pursue this third-factor question: What specific effect, if any, would species-wide blindness have on the structure of a language? Defning and Isolating a Single Third-Factor Variable

The most important thing to establish before starting on this process is that this is not a list of adaptations that blind humans need in order to live and work successfully in human society. Blind humans have held nearly every conceivable job on Earth, from teacher to athlete, general, carpenter, parent, judge, and college professor. What we are investigating here is how beings who had no exposure to sight, left on their own without sighted influence, might use language to maximize the amount and kind of information they receive from one another. To isolate a single variable, we start by holding everything else constant on our hypothetical planet. That is, Earth is the starting point. In our case, we assume that our blind aliens evolved on the surface of a rocky world much like ours, with a large moon, oxygen atmosphere, and plentiful food and water, and that they are humanoid with roughly human physical abilities. (For more on what might be constant across cultures, consult Granger et al., this volume [Chapter 8] and Herzing, this volume [Chapter 4].) Although I am confident that one could build a convincing case for the survival and advancement of a totally blind species (Wells-Jensen 2016), whether or not this is a likely scenario is not the focus here; our purpose is to isolate and explore one third-factor variable that could conceivably make a significant difference in language evolution and follow that logical trail. (Consult both Kershenbaum, this volume [Chapter 2], and Pepperberg, this volume [Chapter 6], for ways in which sensory modality and embodiment might influence cognition.) Thus, we take blindness as a developmental influence seriously, and we allow the blind aliens to develop reasonable (even ingenious) non-visual skills in response to their environment, but do not grant them superpowers. This difference between skills and powers is essential. Our blind aliens will be able to hear, smell, and feel at more or less the level of humans. They may train their existing abilities more cleverly and deliberately and may pay more attention to auditory or tactile information, but the level of sensory input itself will not differ from that which is typical for humans. Most blind people use some degree of passive sonar to detect openings in a corridor as they walk along (Kish 1982), but unlike Daredevil, they regrettably cannot “hear” the heartbeats of their conversational partners or listen in on the blood roaring through their veins. Similarly, a blind human, with training, can learn to read braille rapidly and quickly, using a skill set that an untrained sighted person cannot readily duplicate (Miller 1997), and this early exposure to increased tactile stimulation can affect the allocation of resources in the human brain. That is, while blind

Cognition, Sensory Input, and Linguistics 141

and humans and sighted humans have the same brains, congenitally blind people allocate brain resources differently than sighted peers (Bedny 2017; Burton et al. 2002). This does not convey superpowers, but it does speak to the strong influence of natural circumstances over brain allocation. In addition, when necessary (or simply desirable), humans can train themselves to accomplish impressive feats of memory (Chalmers and Humphreys 2017); some chess masters, for example, can retain the positions of all the pieces on a chess board and can, in fact, play several games of “blindfold” chess at once (Saariluoma 1991; Saariluoma and Kalakoski 1998). The difference between spatial ability (shared by all humans) and visual ability (something sighted people have, and blind people presumably do not) is important here. Much of what we generally identify as “visual” ability is, in fact a complex mix of mental abstractions (found in the parietal lobe) and stored knowledge of the properties of objects (found in the temporal lobe) (Kosslyn et al. 2006). Visual knowledge is a subset of spatial knowledge, and we know that spatial awareness in blind people positively correlates with amount of exploration and physical activity (Schmidt et al. 2012), meaning that the ability to hold spatial information in the mind improves with practice. For example, I know, sitting in my front yard (although I have never seen these things), that I am roughly thirty feet from the street that runs north and south in front of my house. I know that my front door is behind me, off to my left, and that the apple tree is forward and to my right. I can track the passage of a loudly complaining chickadee as she passes more or less directly over my head, flying parallel to the street. Similarly, I can hold in my mind the form of the chickadee, remembering her two clawed feet, her legs, and the body above them, and think accurately about how her wide-open mouth is on the opposite end of the body from her tapered tail. I can compare the size of the chickadee to the very interested cat sitting beside me, and I can tell by resting my hand on the cat’s head that he, too, is moving his attention across the sky, following the bird’s path. Sima et al. (2013: 1) have confirmed experimentally what blind people know instinctively: “The visual mental image is a specification of a part of a spatial mental image” (consult also Kosslyn et al. 2006). All this notwithstanding, much of daily life can be effectively navigated with little to no spatial information (Hull 1990) – and this, as it turns out, is generally not a problem, at least in the life of a relatively sedentary 21st-century person. Spatial representations might only be called upon when needed. For example, despite my spatial description of my front yard, I do not create or maintain representations of physical data I do not need. If I am playing chess, I hold the grid of the board in my mind and understand the relationships of the pieces to each other. If I am faced with an opponent in a snowball fight, I use sound cues to track their movements with reference to where I am waiting.

142 Sheri Wells-Jensen

But if a colleague knocks on my office door and says hello, I hear and interpret the knock, identify her voice, and know her to be standing in the hallway just beyond my door without imaging her knuckles, the door frame, or any part of her body. As we are chatting, I make no effort to picture her body size, face, dress, or the position of her arms and legs. However, when she stands up and offers me a sheaf of papers (signaled by our conversational context, by sounds of clothing against skin and of a chair moving or creaking as weight shifts off of it), I deploy spatial knowledge and can judge accurately where to reach to receive the proffered pages from her hand, but I do not bother to construct a representation of her extended arm, or the rest of her body, or her abandoned chair, unless these become important. Thus, while spatial information is generally retrievable from the environment, it is not as essential in most situations as one might imagine. We also want to tease apart the inherent usefulness of vision from the ways in which humans have constructed our shared environment to privilege vision over other sensory inputs. For example, if we want to communicate what is behind a series of closed doors in an office building, we generally place visual representations on the doors in the form of printed words and pictures, and potential clients are expected to find their way to their goal by searching through these possibilities. We do this because it is assumed that the vast majority of the people entering the office building would be sighted. However, this is only one possible way of communicating the relevant information. Signs on the doors could be tactile instead, or the carpet or floor covering in the hallway could contain tactile cues, detectable by the feet. Each door could carry a small audio beacon explaining its contents. It could be standard to provide a succinct directory to anyone entering the building to avoid the problem of searching for the right door. Each business, when it advertises itself to the public, could include detailed instructions on how to locate its entrance. Building contents could be organized such that similar businesses are placed adjacent to one another, or there could be standard places within each building for specific kinds of services – the restaurants always on the highest floor and the shoe stores at the bottom. In fact, if one or more of these factors were employed in addition to the visual cues, the office building would be easier for everyone to use. (For more discussion, consult Davis 2017.) Finally, although we are collectively relying to some degree on imagination as we progress through this scenario, research has shown that a sighted person cannot duplicate the sensory, emotional, or practical world of a blind person by closing his eyes or donning a blindfold. The perceptions, abilities, and reactions of a blind person are not the same as those of a sighted person who is not looking (Burgstahler and Doe 2004; French 1992; Silverman et al. 2014; Silverman 2015). This is primarily due to the sighted person’s lack of adaptive skills; a blind person is accustomed to working around barriers such as the office building’s doors and has innumerable strategies and hacks to find her way. The troubles the blindfolded sighted person has are also partly due to the anxiety felt by most sighted people under a blindfold, and partly because of cultural

Cognition, Sensory Input, and Linguistics 143

misperceptions about blindness. Perhaps because sight is so useful, these misperceptions about blindness are very deeply woven into all human cultures. Although blind humans competently and safely cross busy streets, care for children, hold jobs at universities, and are in charge of homes and offices, the predominant cultural stereotype is that blind people are slow, passive, and incapable of grace. Rather than an athlete, parent, or even college professor, most sighted people imagine blind people as elderly relatives, beggars, or, on a good day, blues musicians (Wells-Jensen et al. 2021). To rationally consider how blindness would shape language and culture, we also have to untangle blindness from its cultural baggage. What remains, then, as a useful definition of blindness is a particular set of gaps in useful information. • Location of objects at a distance when those objects are not emitting noises and cannot be easily detected through passive sonar (especially important if these are dangerous, such as a hole in the ground or overhanging branches that could strike a passerby on the head). • Identification of small objects at a distance. • Location of two or more objects with respect to one another at a distance. • Identification of individual people at a distance. • Detection of what a person is doing if that action is not audible (this might be especially important for any situation where passive observation leads to significant cultural cohesion, safety, or advancement of scientific knowledge). A Blind Alien Language

Having established what blindness is, and the gaps it might cause in information flow for the blind aliens, we turn to a discussion of how language might fill those gaps. I do not in any way mean to imply that blind people on Earth are linguistically impoverished, or that they need these enhancements to manage their affairs. Rather, this is an exploration of what blind people could develop without sighted influences. Content Words: Nouns, Verbs, and Adjectives

Minsky (1985) explains that any language will have some way of referring to objects (“nouns”) and some way of referring to events (“verbs”). As discussed by Roberts et al. (this volume [Chapter 15]), in order to be usable, the language would also need a method of “predication,” a means to express the relationships between these nouns and verbs. Following is a possible list of items that appear in all human languages which might be different extraterrestrially:

144 Sheri Wells-Jensen

Verbs for Seeing and Not Seeing

A trivial but relevant point is that the language would lack the word “blind”; as there is not seeing, there can be no blindness. And, of course, such polysemous extensions as “see” meaning “know” and “blind to” meaning “ignorant of” would be absent. This would be of little consequence; the reader may not have noticed up until now this chapter’s use of “consult” in place of the usual academic usage of “see” in parenthetical citations. Color Terms

These would obviously be absent (at least until the advent of technology that reveals them). Similarly, words for amounts of light and darkness would be absent. This does not, however, preclude words in the semantic domain of “shadow,” meaning the situation whereby a body interposes itself between a radiation source and a receiving surface, as infrared radiation and sound waves can be blocked and redirected in ways similar to visible light. Similarly, this obviously does not preclude words for “day” and “night,” as the absence of the sun is detectible without vision. Daytime would be warmer, but not brighter. There are many places where the blind aliens would need words not present in Earth languages. Properties of Sound

Because human perception skews toward vision, there are detectible qualities of sound which lack unique lexical expression in Earth’s languages. For example, although the difference in timbre between a trumpet and a flute is easily identified, there is no single word describing this. Words for Kinds and Qualities of Echoes and Acoustics of Ambient Spaces

If you step into a cathedral, an open space in a library, a living room, or a closet, little effort is needed to identify that the quality of sound is different, yet there are no commonly used words in human languages to describe these differences. Function Words and Grammar

This is where things become interesting. We might imagine a deeper, more complex system of function vocabulary (words like prepositions and pronouns) and an equally expanded and more complex set of grammar rules to go with them. Directional Systems

Perhaps the most common kind of system of directions in human languages is relative: left, right, forward, backward, up, and down with respect to the speaker. This

Cognition, Sensory Input, and Linguistics 145

is often problematic and usually frustrating, as these are easily confused when trying to describe locations or events from the hearer’s perspective, frequently resulting in cries of “No, your other left!” One alternative in some human languages is to use a common external reference point, as is done aboard ships, where “bow,” “stern,” “port,” and “starboard” always mean the front, back, left, and right with respect to the ship itself, rather than to speakers or hearers aboard the ship. In some human languages, speakers regularly keep track of, and use, the cardinal directions (north, south, east, and west) to locate objects indoors as well as in more natural settings (Harrison 2007). Speakers might say, “The cup is on the east edge of the table” or “Go through the door on the north wall.” Other human languages use a stable environmental feature as the anchor point. The Oroha language, spoken in the Solomon Islands, uses “up the beach” and “down the beach” as other languages use “upstream” and “downstream.” A blind species might use a cardinal direction, a salient noise source, the slope of the terrain, or some other set of factors. Deictic Expressions

Earth languages use “deictic” forms to indicate relative position in space, usually in relation to the speaker and listener. In languages like English, word pairs such as “this/ that,” “these/those,” and “here/there” lexicalize a two-way distinction, whereas other languages like Japanese have three relative positions: “kono” (this one near me), “sono” (that one near you) and “ano” (that one far from both of us). Slightly more complex systems of deixis, some of which involve visibility and invisibility, also exist (Dixon 1972). Certainly, we could imagine a more granular system including single words or affixes indicating “at a position roughly equidistant between us” or “at a position toward a third conversational participant.” In combination with direction words, this could allow one speaker to efficiently tell another everything she knows about where the object under discussion is. One might even imagine the construction of a rough XYZ grid, with the origin of the axis located at a fixed point, allowing speaker and listener to locate themselves and their interlocutors, including the extent to which an object or person of interest is above or below a fixed point. This would employ concentration and memorization skills humans do not regularly tap, but a single deictic word (or even a suffix) could allow a speaker to describe an object’s location efficiently and unambiguously. Subsequently that point in space (or an object occupying that point in space) could be referred to using a pronoun. Some of this information would obviously be optional but having the potential to communicate extra locations would be extremely useful and probably represent a survival advantage. Here is an example. 1 We were harvesting apples and we discovered a chickadee nest with a bird in it, and while we were there, the cat ran past.

146 Sheri Wells-Jensen

If we define the following set of location words . . . XYZ-1 = the position of the speaker in the yard, relative either to the addressee or to some other fixed point. XYZ-2 = the location of the tree (which locates the person with respect to the tree, closer or further from the addressee, or indicates on what side of the tree the person is standing). XYZ-3 = the location of the nest in the tree (how high or low in the branches, and how close to the trunk). XYZ-4 = an idea of the vector (speed and direction) of the cat. . . . then our initial example sentence could be elaborated as follows. 2 We (XYZ-1) were harvesting apples (XYZ-2) and we discovered a chickadee nest (XYZ-3) with a bird in it, and while we were there, the cat ran past (XYZ-4). This would also allow the location of the nest (XYZ-3) to be used later as a deictic pronoun indicating the nest in a subsequent narrative and allow XYZ-1 to refer to the group of people who found the nest. Evidentiality

Some human languages provide a set of affixes indicating the reliability of an utterance. These range from direct experience (“I-know-because-I-myself-was-there”) to doubt (“I-think-that-I-once-heard” or “some-people-say”). This would be valuable as a means of establishing trust between speakers or for conveying the speaker’s confidence that the information is correct. We might then have something like the following. 3 We (XYZ-1) were harvesting apples (XYZ-2, EV-1) and we discovered a chickadee nest (XYZ-3, EV-2) with a bird (EV-1) in it, and while we were there, the cat (EV-2) ran past (XYZ-4, EV-3). Where: EV-1 = Definite: This is my tree. I know where I was and can say for sure. I also know it was a bird. EV-2 = Reasonably sure: I am no expert, but I do know the noise chickadees make and I know there are some around here, so it is a good bet that it is a chickadee nest. Also, I know my cat was around at the time, so it is a good bet that it was the cat I heard. EV-3 = Uncertain: I was a little distracted, so I think the cat went that way at that speed, but I did not touch him and there was some other noise around at the time so I cannot be sure.

Cognition, Sensory Input, and Linguistics 147

Personal Pronouns

Social status, animacy, and gender are often marked on human-language pronouns. There are a variety of ways of usefully expanding the information content of such a pronoun system. A system of exclusive and inclusive marking (“we-including-the-listener” and “we-excluding-the-listener”), as employed in some human languages, would take the place of eye contact and hand gestures. Together, these two factors could simplify information exchange when two groups meet, and could facilitate communication within a single group. Furthermore, the number marking on pronouns could be more precise than in Earth languages, referring (when possible) to the actual count of individuals present. For example, 4 We (XYZ-1, PRO-4-INCL) were harvesting apples (XYZ-2, EV-1) and we (PRO-2-EXCL) discovered a chickadee nest (XYZ-3, EV-2) with a bird (EV-1) in it, and while we (PRO-3-EXCL) were there, the cat (EV-2) ran past (XYZ-4, EV-3). Where: PRO-4-INCL = All four of us, including the addressee. PRO-2-EXCL = Two of us, not including the addressee. PRO-3-EXCL = Three of us, not including the addressee. This would mean that four people (including the addressee) were harvesting apples, when two of them (not including the addressee) found a nest and three of them (again not including the addressee) were around to hear the cat. Pronouns might also include directional or deictic information. Thus, we might find distinct pronouns meaning “you-north-of-me,” “the-three-of-you-ahead-ofme,” or “you-of-mixed-gender-and-lower-status-currently-uphill-from-me.” Gender and social status might be included on animate pronouns and other qualities marked on inanimate ones. These could take the form of affixes denoting material, size, shape, texture or whether the speaker is currently holding the object. Thus, we might have a single pronoun meaning “The-large-wooden-objectnear-you-in-the-downhill-direction.” Adverbials

First among the set of available adverbials might be a set of words for body positions and common movements. Human dancers, martial artists, and yogis employ specialized terminology precisely describing poses or movements of the arms, legs,

148 Sheri Wells-Jensen

hands, feet, and head. Without the remote access to other people’s bodies through vision, it would be useful for such terms to exist in the everyday lexicon of the blind aliens. Think of the number of times an instructional video says, “Hold the needle like this, and wrap the yarn around it this way.” One might also imagine a more nuanced set of what we might think of as traditional Earth adverbials, perhaps with finer gradation as shown in what follows. 5 We (XYZ-1, PRO-4-INCL) were harvesting BODY-SHAPE-ARMS-UP EFFORT-3 apples (XYZ-2, EV-1) and we (PRO-2-EXCL) discovered a chickadee nest (XYZ-3, EV-2) with a bird (EV-1) in it, and while we (PRO-3-EXCL) were there EFFORT-2, the cat (EV-2) ran past (XYZ-4, EV-3) EFFORT-1. Where BODY-SHAPE-ARMS-UP = One of a set of words denoting specific body positions. EFFORT-3 = With some degree of serious effort. EFFORT-2 = Less effort but still working. EFFORT-1 = Minimal effort: the cat is perceived to be running easily. One could imagine a body shape for the cat as he runs, but if the speaker has no direct knowledge of the cat’s movement, this could be eliminated or added along with one of the evidentials indicating doubt. Prepositions

One function of prepositions is to locate one object with respect to another. Thus, one object might be in, on, above, under, or beside another. Any of the following could be expanded to give additional useful information about one object’s relationship to another. “in” 1: Nondescript – equivalent to English “in” when the speaker does not know or does not wish to say something more specific. “in” 2: Partly located within another body but with access to the outside, e.g., a bird in an open nest with her head sticking out; one of the default prepositions for describing something resting in your palm. “in” 3: Fully located within another body but with access to the outside, e.g., a bird in an open nest with her head down or an object resting in cupped hands. “in” 4: Wholly within another object, e.g., a bird buried in a pile of branches or inside a nest with a woven top, or something held in a closed fist. “in” 5: Equivalent to “in” 4, except that the nest, hand, or other container is impermeable to sound. One could not hear the bird even if it chirped. “in” 6: Equivalent to “in” 4, except the container is open in some places, like a bird in a cage. “in” 7: Equivalent to “in” 4, except that the container has openings wide enough for fingers to penetrate to touch whatever is inside.

Cognition, Sensory Input, and Linguistics 149

6 We (XYZ-1, PRO-4-INCL) were harvesting BODY-SHAPE-ARMS-UP EFFORT-3 apples (XYZ-2, EV-1) and we (PRO-2-EXCL) discovered a chickadee nest (XYZ-3, EV-2) with a bird (EV-1) in-3 it, and while we (PRO-3-EXCL) were there EFFORT-2, the cat (EV-2) ran past (XYZ-4, EV-3) EFFORT-1. Multi-Channel Speech

In cases when all this (and presumably more) additional information were used, the length of utterances could become unworkable. It might make sense, then, for this species to adopt a parallel system of communication to carry some of this extra information. Noises made with the hands or feet, voice quality, auxiliary noises made with the voice, or even an artificially constructed paralinguistic object like a clicker might be essential. (Consult Ortner, this volume [Chapter 5], for human analogs.) Choosing What to Say

How language is used in context is as important as what is said. In this case, the amount and kind of information added to the basic sentence would depend on a variety of factors including formality, how familiar the addressee is with the environment of the story, and the particular point the speaker wants to make. In addition, the amount and kind of supportive “back-channeling” might be quite different than in human languages. Whereas it is common for some human speakers to offer occasional verbal affirmation (such as “mmhm” or “yeah”) while another is speaking, this has limited utility beyond establishing a social bond and encouraging the speaker to continue. However, this back channel could also be used by the blind aliens to keep the group updated on what is happening. For example, since body language is not directly accessible, a listener might want to express honest indications of their level of attention or to offer brief narrations of the things they are doing while listening. What is included in these narrations would comprise an intricate network of culturally governed responses. It might be very difficult for outsiders to follow these parallel tracks of information and even harder to calculate what they should say when they are listening. Furthermore, a crucial cultural norm among the blind aliens might be that accurate information is more important that saving face or pleasing the addressee. This calculus of courtesy, involving as it does an unfamiliar set of cultural imperatives, might be very difficult for humans to learn. Conclusion

Everything in this chapter could be wrong.

150 Sheri Wells-Jensen

Blind scholar Georgina Kleege puts it this way: I know what it means to be sighted because I live in a sighted world. The language I speak, the literature I read, the art I value, the history I learned in school, the architecture I inhabit, the appliances and conveyances I employ, were all created by and for sighted people. (Kleege 1999) She does not know, and I do not know, as blind individuals in a sighted world, what the world of a completely blind species would be like – none of us can know what any alien civilization would be like, even if – as has been attempted here – we try to proceed in a logical way to keep as many variables constant as possible. But the point is not to “guess right.” The point is to make the guesses. Our job before contact is to refine the art of asking and answering all manner of questions. Replace the word “blind” with anything you like: water-dwelling, nine-legged, short-lived, long-lived, agoraphobic, claustrophobic, insectoid, ovoid, threetongued, highly combustible, etc, and play out the scenario, trying to work out what impact this would have on language. That is how we will uncover some of our own unconscious assumptions about life, intelligence, and language – and that is what will move us further along the path of preparing for contact with intelligent beings that are truly different from us. References Bedny, Marina. 2017. “Evidence from blindness for a cognitively pluripotent cortex.” Trends in Cognitive Sciences 21. Burgstahler, Sheryl, and Tanis Doe. 2004. “Disability-related simulations: If, when and how to use them in professional development.” Review of Disability Studies: An International Journal 1: Online. Burton, H., et al. 2002. “Adaptive changes in early and late blind: A fMRI study of Braille reading.” Journal of Neurophysiology 87: 589–607. Cambias, James. 2014. The Darkling Sea. NewYork: Tor Books. Chalmers, Kerry, and B. Humphreys. 2017. Thinking about Human Memory. Cambridge: Cambridge University Press. Chambers, Becky. 2014. A Long Way to a Small and Angry Planet. Great Britain: Harper Voyager. Chomsky, Noam. 1986. Language: Its Nature, Origin and Use. New York: Praeger. Chomsky, Noam. 2005. “Three factors in language design.” Linguistic Inquiry 36 (1): 1–22. Davis, Leonnard. 2017. Beginning with Disability: A Primer. Abingdon, UK: Routledge. Dixon, R.M.W. 1972. The Dyirbal Language of North Queensland. Cambridge: Cambridge University Press. Evans, Vyv. 2014. The Language Myth: Why Language Is Not an Instinct. Cambridge: Cambridge University Press. Everett, Daniel. 2016. Dark Matter of the Mind: The Culturally Articulated Unconscious. Chicago, IL: University of Chicago Press.

Cognition, Sensory Input, and Linguistics 151

Everett, Daniel. 2017. How Language Began: The Story of Humanity’s Greatest Invention. New York: Liveright. Fletcher and MacWhinney. 2017. The Handbook of Child Language. Blackwell. French, Sally. 1992. “Simulation exercises in disability awareness Training: A critique.” Disability, Handicap & Society 7 (3): 257–266. Haden-Elgin, Suzette. 1984. Native Tongue. New York: DAW Books Inc. Harrison, K. David. 2007. When Languages Die: The Extinction of the World’s Languages and the Erosion of Human Knowledge. Oxford: Oxford University Press. Hauser, Marc, Noam Chomsky, and W. Tecumseh Fitch. 2014. “The mystery of language evolution.” Frontiers in Psychology 5: 1–12. Hull, John. 1990. Touching the Rock: An Experience of Blindness. London: SPCK Publishing. Kish, Daniel. 1982. Evaluation of an Echo Mobility Training Program for Young Blind People. MA Thesis. University of Southern California, Los Angeles. Kleege, Georgina. 1999. Sight Unseen. New Haven, CT: Yale University Press. Kosslyn, S.M., W.L. Thompson, and G. Ganis. 2006. The Case for Mental Imagery. Oxford: Oxford University Press. Miller, Susanna. 1997. Reading by Touch. Abingdon, UK: Routledge. Minsky, Marvin. 1985. “Why intelligent aliens will be intelligible.” In Extraterrestrials, science, and alien intelligence, 117–128. Cambridge: Cambridge University Press. O’Grady, W. 2012. “Three factors in the design and acquisition of language.” Wiley Interdisciplinary Reviews: Cognitive science 3: 493–499. Pinker, Steven. 1994. The Language Instinct: How the Mind Creates Language. New York: HarperCollins. Russell, Mary Doria. 1996. The Sparrow. Ballantine Books. Saariluoma, Pertti. 1991. “Aspects of skilled imagery in blindfold chess.” Acta Psychologica 77 (1): 65–89. Saariluoma, Pertti, and V. Kalakoski. 1998. “Apperception and imagery in blindfold chess.” Memory 61: 67–90. Schmidt, Susanna, Carla Tinti, Micaela Fantino, Irene Mammarella, and Cesare Cornoldi. 2012. “Spatial representations in blind people: The role of strategies and mobility skills.” Acta Psychologica 142: 43–50. Silverman, Arielle. 2015. “The perils of playing blind: Problems with blindness simulation and a better way to teach about blindness.” Journal of Blindness Innovation and Research 5. Silverman, Arielle, Jason Gwinn, and Leaf Van Bowen. 2014. “Stumbling in their shoes: Disability simulations reduce judged capabilities of disabled people.” Social, Psychological, and Personality Science 6 (4): 464–471. Sima, J.F., H. Schultheis, and T. Barkowsky. 2013. “Differences between spatial and visual mental representations.” Frontiers in Psychology 4: (240). Wells-Jensen, Sheri. 2016. “Alternative perceptual systems and the discovery of basic astronomical phenomena.” International Space Development Conference, San Juan, Puerto Rico, National Space Society. Wells-Jensen, Sheri, Emily Michaelson, and Mona Mincara. 2021. “How blind professors win the first day.” In Picture a Professor: Intersectional Teaching Strategies for Interrupting Bias about Faculty and Increasing Student Learning, edited by Jessamyn Nauhaus. Morgantown, WV: West Virgina University Press.

14 THE DESIGN FEATURES OF EXTRATERRESTRIAL LANGUAGE A Domain-General Approach Darcy Sperlich

Introduction

Delimiting what the language of an extraterrestrial intelligence (ETI) may consist of seems to be an impossible task – we have no empirical evidence for ETI even existing. Consequently, we must engage in blue-sky thinking about their biology, cognition, and culture. There are numerous factors to consider for a potential ETI language – for example, does the ETI have a speech organ? How does the ETI’s memory system work? Do ETIs live in a society? In order to focus our thinking, we need to demystify what a non-human language could be like by making some sound assumptions based on the best example we have – human language. One way to delimit the current topic is to claim that language itself is a result of domain-general brain phenomena, as advocated by the Emergentist approach (e.g., Christiansen and Chater 2016a), and not a result of an innate narrow language faculty consisting of Merge (e.g., Berwick and Chomsky 2016, see also Roberts et al., this volume [Chapter 15], and Samuels and Punske, this volume [Chapter 16], for proponents of this viewpoint).1 The Emergentist approach provides us a context in which to understand ETI’s language, as the assumption that language arose from a language-ready brain helps us consider what a probable versus improbable ETI language may look like. Here, our discussion is mainly focused on natural ETI language, in the sense that it is a result of evolution, but we cannot discount a biological ETI language that has been engineered. The unknown here is that we have little idea of what direction a non-natural transformation will take. The next section reviews what the factors are behind language and its features, followed by sections which respectively: sketches human language to prepare for our look at ETI language; focuses on the two areas of ETI language, natural language versus engineered; reviews what actual ETI language data we might acquire in our lifetime; and concludes the chapter. DOI: 10.4324/9781003352174-14

The Design Features of Extraterrestrial Language

153

What Is Language?

Gallons of ink have been spilled on this topic, and a single section cannot hope to do it justice. Nevertheless, we know that language is uniquely human, so any description of language must account for what we find in humans. We also know that non-human terrestrial animals lack language, although they do have communication systems (e.g., see Slobodchikoff on prairie dogs, this volume [Chapter 9]). One language identification method is Hockett’s design features (Hockett 1960; Hockett and Altmann 1968), which proposes 16 features that define human language.2 Hockett asserted that languages must have all of these features in order for language to differ from animal communication systems, which are missing some of these features. Animal communication has certain clusters of features, but, as discussed by Reboul (2017), the “core” fundamental features of language seem to be Hockett’s semanticity (sentences have a variety of meanings), discreteness (units are made up of smaller units), and decoupling (referring to situations not in the here and now).3 Moreover, hardly any of the other features seem unique to human language (Coleman 2006). Thus, we might conclude that any language must express these core factors, including that of an ETI. Consequently, at a minimum, ETI will share these core features of human language, and probably a lot more, as we find with animal communication systems.4 What is language for? While it is used for both communication and internal thought (cf. Goddard and Wierzbicka 2014), it is uncertain whether language for communication developed from language of thought or vice versa (cf. Reboul 2017). Another important question is whether language has its origins in Merge (suggesting a language of thought came first), or developed as a result of repurposing domain-general processes? The approach adopted here is that language developed for communication, as the derivatives of the “language is innate” hypothesis is improbable. Reading Hauser et al. (2002) and the ensuing debates (cf. Christiansen and Chater 2015 for recent debate) over the years about the narrow language faculty involving recursion via a genetic mutation (the key being the feature of Merge), presents a species-specific scenario that cannot seriously be extended to ETI, who, we can say with a high degree of confidence, would not come from an Earth-like environment or evolutionary context.5 At a minimum, ETI would need to have the same/similar mutation occur (in a brain that is ready for language) to allow for language to emerge. Following Hornstein and Boeckx’s (2009) discussion on how the faculty of language came about via a “couple of adjustments” to our cognition and computation, ascribing the same idea to ETI’s emergence of language (as contrasted against natural selection), one did not leave Earth.6 Unless one posits a human-like scenario for the entire universe, such a perspective oversimplifies the problem of language evolution by providing a general “cosmos theory of language.”7 It could be possible to try and obtain Merge in a different way; however, the theorized extent to which is it responsible for language remains a moot point.

154 Darcy Sperlich

The approach here is in line with Evans and Levinson (2009) and Christiansen and Chater (2016b), who maintain that language is a product of culture and cognition, which give rise to the diversity of human languages.8 Humans are highly adaptive, and language reflects this (see Bybee 2009 discussing the various different factors that shape language). It is simpler to postulate that language is a result of a domain-general mechanism that is adapted for language, and this helps us to focus our inquiry on what language is. For instance, language universals in ETI would arise from form and function correspondences and cognitive solutions, as we see in human languages (Cristofaro 2010; Givón 2018). A mediating view is that humans try to deliver their message while not overloading their communication channel vis-à-vis Information Theory (e.g., see Pellegrino et al. 2011 comparing information rates between languages). These are some of the restrictions that have guided our language development. This brief sketch shows that any language will have to meet Hockett’s core feature criteria, and is a result of domain-general processes. I now discuss cognition and culture as related to the human language system. Human Language Development

The exact nature of human language is what linguists study, and there are many different theories that try to account for it. The domain-general approach to language accepts that language has followed a bio-evolutionary approach, in that systems like vision, audio, attention, and memory have been extended to language development. The language system itself is covered by any introductory linguistic text that provides an overview of the major areas of phonetics, phonology, morphology, semantics, pragmatics, and syntax. As importance is placed upon the cognitive and cultural perspective, a processing perspective is discussed, along with cultural influences on language development.9 A recent theory constraining language evolution due to processing is Hawkins (2004, 2014), which details how the processor works as efficiently as possible for maximum economy, with cross-linguistic patterns as the result. If processing does affect the shape of human grammar, as Hawkins argues, then we expect the same for ETI.10 Another thesis by Christiansen and Chater (2016a) posited that there is a limited window for the uptake of information before it is lost, which is why the system must work around it.11 Moravcsik (2010: 73) has observed that there is phonological and semantic reduction in frequently occurring phrases, which eases production. An important pragmatic aspect is that our language system seems to be adapted to overcome communicative bottlenecks as discussed by Levinson (2000), who notes that when speaking, our information-exchange rate is painfully slow (i.e., restricted by our rate of speech, see also Nixon and Tomaschek, this volume [Chapter 17]). In order to help overcome this problem, pragmatic inferences come cheaply as the hearer will be able to infer meaning (utterance meaning), as compared to the literal meaning of the words that were said (sentence meaning). Thus,

The Design Features of Extraterrestrial Language

155

along the lines of Hawkins (2004, 2014), our grammars are influenced by the factors that influence the design of our processing system.12 Culture can also have an impact (Chafe 2018). As Moravcsik (2010: 88) pointed out, language change can be influenced by social factors, and not just for reasons of general economy. For instance, Lupyan and Dale (2010) discussed how the complexity of a language’s morphosyntax is affected by group size, geographic spread of the language, and how much language contact has occurred. Furthermore, sociolinguistic research has helped immensely in our understanding of how language is molded by its users in society (politeness, for example; see Holmes and Wilson 2017), but we are not able to investigate these aspects for ETI unless we first gain an insight into ETI’s culture and society. In sum, human language is understood here to be a result of domain-general processes, guided by evolutionary pressures. Culture is an important aspect for human language development, but we will not cover this aspect concerning ETI. ETI Language Natural ETI Language

If Darwin’s evolutionary theory works in the right direction, then, as Ulmschneider (2006) suggested, these natural laws of selection would be present on other Earthlike planets. Thus, as languages have evolved due to such processes here, there is hope language would have had similar starting points in extraterrestrial spaces.13 Combining what we know about language theories with language instantiation in an intelligent species (Homo sapiens), we can now place rough limits on an ETI’s language, based upon their cognitive apparatus. Again, the caveat has to be made that any cognitive system must be anchored in culture and society. However, since there are too many possibilities about what an ETI’s culture and society would look like, we thus are missing a piece of the puzzle.14 What can be said is that if ETI can send messages to us (or receive them), this is an indication of a societal product involving cooperation, and therefore, language must be complex enough to help organize such an endeavor. Considering cognition in and by itself, we are also limited in our understanding, in the sense of what embedded language systems look like and how they are processed. Questions such as “Do ETI have emotions?”; “Are they empathic?”; and “Do they have humor?” do not seriously enter our discussion. All are answered “yes” for humans, as there is a close link between language and emotions (Lindquist et al. 2015); language is used to express other’s viewpoints (e.g., via logophors, cf. Huang 2000), and jokes are told which are not assessed for truth values (Barbe 1995). Disregarding these aspects for ETI, what would the linguistic system look like? Assuming ETIs have concepts about the universe they live in, they need to relate these concepts to some kind of symbols, which language provides. Such symbolic

156 Darcy Sperlich

tokens combine syntactically, accordingly generating phrases and sentences to express propositions. We do not know how ETIs communicate physically. For example, it could be through sound, and thus there would be a phonology; it could be through smell, whereby the linguistics of olfaction would come into play (cf. Kershenbaum, this volume [Chapter 2], for discussion on different modes). Would they have a pragmatic inferencing system? That is debatable, because it depends on the processing power of ETI, which we shall come to later in the chapter. First, let us look at some proposals for the lexicon, syntax, and semantics of an ETI. Regarding the lexicon, would they have a word for every conceivable thing and situation? Possibly not, as one thinks about/discovers many new things in a day – but they might have a superior vocabulary (depending on memory constraints),15 suggesting that they would not rely on pragmatic enrichment as much as humans do (e.g., understanding “black” in different contexts; stereotypically “black” in “black pen” indicates that the ink is black, while “black” in “black cow” refers to its hair). Assuming that our ETI are logical creatures, one might also assume that they have boundaries between words well defined, and not have fuzzy semantic fields where humans are concerned. Before we discuss the possible syntax, as with humans, ETI could also have many different languages. Or, if they are further along the evolutionary line, they might have chosen to speak one language – which presents us with the theoretical problem that our conclusions may not be ETI-typologically valid. In any case, there is not much to comment on ETI’s syntax – it depends on what theory of syntax one prefers,16 but there will be a system of combining words into phrases, and phrases into sentences. Would there be the canonical subject, verb, object? If ETI views the world as a cause–effect one, then a probable result is that ETI will semantically conceptualize actors and undergoer for example, with a consequence that relations might hold like subject = actor/agent, object = patient/undergoer and verb = transitive/intransitive.17 Therefore, we might expect these types of relational categories to exist.18 Linked to the yet-undiscussed processing powers of ETI, it is likely that the word order of ETI language would be fixed rather than mixed, allowing for more straightforward logical/computational processing (see Cristofaro 2010 for discussion) – as is the trend among human languages.19 We assume that the semantic system, which involves the meaning of words and how they form complex meanings in phrases and sentences along with propositions, exists in their linguistic architecture. That is, they must have a system of meaning in order for the message to be logically construed.20 It might be the case that the ETI allows only a completely logical interpretation of what they say, in that the words and phrases mean what they mean at the semantic level and are not affected by context at all. Wittgenstein (1922) proposed such a language, so while this might be considered an advantage in terms of processing, it would be inflexible and complex, and as such does not reflect the fuzzy logic inherent in human languages (perhaps Wittgenstein would not have renounced his previous work when considering ETI). Thus, how important is pragmatics for a logical ETI?

The Design Features of Extraterrestrial Language

157

A system of inferencing is part of human cognition and language. Humans have inferencing systems to speed up communication (Levinson 2000) – we do not have the luxury of time to spell out exactly what we mean in a purely logical way. In the case of ETI however, if the communication corridor allows for relatively quick exchange of messages (in a “logical” format), then the need for a pragmatic inferencing system is reduced. This is not to say that they would not be able to understand an inference, but they would not make much use of it in their language. Thus, we need to say a few words about their cognitive system. Given the domain-general approach adopted for language, it is clear that processing limitations in humans have formed language in a way that it is processed efficiently with the resources at hand.21 Think of an analogy between a car (language) and a road (the processing system); if the road is a muddy one, then my 1-liter engine front-wheel drive car will have many problems as it was not designed for such conditions. However, if I drive my off-road V8 3-liter 4×4, then there will be no problems as the conditions affected the design of the car. The point here is that language is affected by the environment it develops in. With ETI, we assume that their processing speed and capacity is at minimum similar to ours – which in turn will produce a language system not dissimilar to our own. But if we find that their processing ability is akin to a hyperloop, their language will reflect that design parameter. If they have much processing power, it allows for a very powerful and complex system – one that does not need to be economical because of the vast resources at hand (think of the historical accident of English having both regular and irregular verbs, and how the system must deal with this uneconomical fact). In summary, just as we have discussed the building blocks of human language, we find similar blueprints fashioning an ETI language. Engineered ETI Language

What does it mean to have an engineered language? We are referring not to languages such as Esperanto, which is created, but one wherein the system is inherently subconscious, but in which ETI (or perhaps humans later on) have: a) discovered exactly how the system functions and b) how it is represented in the brain, and c) has modified its neural architecture, which has affected the language system in some way. With our language as a domain-general phenomenon, increasing the cognitive capacity of the ETI (for example, improving their memory systems) would allow their processor to take onboard far more information to digest. A second possibility is that they are able to directly modify language itself – humans can do this, too, but it takes a long time and mostly occurs at the subconscious level (e.g., grammaticalization). For instance, ETI could decide to insert numbers after every pronoun in order to track reference, or they could selectively link words to particular senses immediately, without using their general cognitive mechanisms – the possibilities are endless. Another possibility is that ETIs edit their genes (if they have them) so that language becomes well and truly innate – at least in an engineered sense.22

158 Darcy Sperlich

Another perspective on engineered language relates to that of artificial intelligence (AI), which is powered by the ubiquitous computer languages. Of course, however, they are programming languages and do not fit in with language as discussed in this chapter. For instance, an AI machine can be fed (incomplete) rules of English as we find with applications that communicate with us to find a cheap flight, or a chatbot on the internet; Searle’s (1980) Chinese room thought experiment comes to mind, and by following a rule book external to us is not considered language. In essence, AI language is static (human language is ever changing), and thus shares the features of a dead language. In this sense, any “language” of an AI as we now know it is not language as discussed here. Thus, can AI actually have language? If we consider that the functions of language are to communicate and organize thought, it will fail on both accounts. AI certainly does not have thought, and therefore does not have anything to express to a communicator about any possible thoughts.23 Moreover, Searle (2014) argued strongly against computers having beliefs and intentions; consequently, any possibility of a computer language becoming a natural language does not follow. Following this argumentation, only when AI is conscious could we consider it as having language, as it will now have an intrinsic motivation to communicate its thoughts. What would the ramifications be if we were to conclude that the ETI was in fact an AI robot that gained sentience (with thought and motivation)? In terms of language, not much, because this brings us in a full circle back to an engineered ETI language – we cannot predict what the result would be. We cannot even be sure whether the AI robot cum ETI will be conscious of its programming because consciousness generally does not entail that one understands the architecture of one’s own system. In summary, having an engineered language opens up Pandora’s box in that we are uncertain of the direction it will take, but we can be sure that processing capacity (as found affecting human language) is a key factor in consideration of the design. Language in ETI Signals

On a final empirical note, what is the most probable situation whereby we can assess first-hand an alleged ETI language? Certainly, with the SETI and METI programs, first contact will be an ETI message (which implies intentionality; cf. Hobaiter et al., this volume [Chapter 3]). Will we be able to decode it? With great difficulty (as Slobodchikoff, this volume [Chapter 9], thinks) – there are dead languages on Earth (e.g., Linear A), which still defy analysis. Furthermore, dead languages are a product of human culture, whereas ETI will be something else entirely. To be fair however, the writers of these languages probably did not consider their language would become extinct, and did not think to include a dictionary. A glimmer of hope is that ETI, on the other hand, may have thought about this problem when sending messages, and constructed the message as a fully self-contained and selfexplanatory whole – teaching us their language within their message (cf. Sperlich

The Design Features of Extraterrestrial Language

159

forthcoming).24 As pointed out by Granger et al. (this volume [Chapter 8]) however, one problem of communication is the inferencing aspect: a message is sent with a specific intent, but it is interpreted in another way. If the alien species does make use of inferences as humans do in their language, then it is apparent that they have a pragmatic system in place which amplifies meaning beyond the semantics expressed by the message. Hence, if the ETI indicates a misunderstanding which is beyond the semantics of the message, then pragmatics is clearly evident – we can then use this to deduce the state of their linguistic, cognitive, and physical system, to a certain extent (cf. Sperlich forthcoming, for discussion on how processing and physical restrictions have given rise to pragmatic systems).25 In terms of the possible conversations that might ensue, we can speculate that the speed of the conversation might be considerably enhanced beyond what is now thought, given advances in quantum entanglement technology (in the sense of extending the distance over which two photons can be entangled). It might be possible that when we receive the message, an entangled photon will be attached and alert the ETI to its receipt, which in turn would generate more messages in our direction. On another note, Pepperburg (this volume [Chapter 6]) suggests that we should be open to ETI using repetition of signal to encode meaning, rather than just an individual unit. In human language at least, repetition is found, for example, to express marked messages such as “Is he a friend, friend?” Unless the signal is using repetition to express linguistic morphological reduplication, for example (which still would be an individual unit), this strategy is both uneconomical linguistically (a single unit is preferred over several repetitions to represent a unit of meaning) and technically (as there will certainly be signal loss which negatively affects the chances of receiving the exact number of repetitions required). Conclusion

While there are many possibilities, it makes sense to focus on the probable. Here, sketching likely ETI language(s) by providing limitations on the ETI’s natural language, specifically assuming a domain-general Emergentist approach, allows us to scale the qualities of the language system and its processing. Assuming that language is constrained by processing, there is a chance that their language has developed in a similar manner and would be comprehensible to us. However, raising the processing bar of the ETI only increases our chances of not understanding anything due to the information surge – it might still be analyzable by humans but require enormous computational power and time. After all, language is still a system. This is further complicated by the scenario that ETI has engineered its language itself, which would have an inconceivable impact. The question of AI was raised, discussing the fact that AI does not have language, and if AI were able to gain consciousness, we would be put in a similar situation of understanding an engineered language. In any case, we can only make an educated guess as to what an ETI language would look like from a natural evolutionary perspective, because

160 Darcy Sperlich

once engineering is applied, anything is possible. Finally, like any ETI receiving a message from Earth (and vice versa), they will have great difficulty in deciphering the language of our message, if it does not at the same time teach the linguistic system it is based upon.26 However, we can be certain that we will be dealing with the first non-human language Homo sapiens have ever encountered. Thus, it is imperative that we continue the search for ETI signals whatever the consequences may be, rather than being isolationists with a Commodore Perry–like Black Fleet on the horizon. Notes 1 For instance, Kershenbaum (this volume [Chapter 2]) is supportive of the viewpoint that language arose from natural selection processes. Kershenbaum also raises the interesting point of needing an energetic competitive environment to allow for communication to arise, and has useful discussion on what modes of communication we are not to expect. In a similar vein, Berea (this volume [Chapter 7]) discusses a wide range of factors in the development of an alien language tracing the co-evolution of language and culture (also touched upon in Hobaiter et al., this volume [Chapter 3]). Berea argues that this coevolution in humans has been constant, and thus is potentially generalizable to other intelligent species. The program proposed for identifying these patterns is complex, but undeniably important if we are to understand the many non-linguistic factors that exert influence on language. 2 See also Ross (this volume [Chapter 12]) for discussion around Hockett’s features and application to a linguistic Drake Equation. 3 Berwick and Chomsky (2016: 110) discussed the “basic Property of human language” – the ability to construct a digitally infinite array of hierarchically structured expression with determinate interpretations at the interfaces with other organic systems. In this view, animals do not have this. 4 Practically speaking though, as any initial contact with ETI would be via long-distance transmission of messages, it would be difficult to assess these for language features – we need to crack the code first. Slobodchikoff (this volume [Chapter 9]) argues that decoding the message will be a herculean challenge. 5 It should be noted that the issue of Pirahã, a language which is claimed not to make use of recursion, is far from settled, posing a serious challenge to Merge-based theories (see Futrell et al. 2016 for latest developments). 6 Evolutionary context is important. Arbib (2016) linked the presence of mirror neurons to human language development. Would they be key for ETI? Possibly not, following the same line of thought. 7 Reading Roberts et al. (this volume [Chapter 15]) and Samuels and Punske (this volume [Chapter 16]), this is the argument to account for any language arising across intelligent species – a question to be settled perhaps in the coming decades. 8 See also Pinker and Bloom (1990) for another perspective, and Christiansen and Chater (2016) for a recent overview of positions. 9 See also Chafe (2018) with a focus on thought-based linguistics. 10 Why don’t humans inflect every idea? For example, verbal past/non-past is common enough, but not hostile/non-hostile, perhaps due to processing restrictions (Bickerton 1990: 56). 11 For further discussion on the integration of this theory with the domain-general approach, the reader is recommended to read Christiansen and Chater (2016) which contains an excellent discussion on the various diachronic and synchronic factors that affect the shape of language.

The Design Features of Extraterrestrial Language

161

12 Chomsky (2017) suggested that computational efficiency trumps communicative efficiency when in conflict. Certainly, a language must do with what resources it has available to it; however, it is not so straightforward that what is the most computationally effective will be the strategy taken forward. An example of this is found in Sperlich (2020), as it is shown possible for English native speakers to long-distantly bind the complex reflexive pronoun himself from an argument position over a finite clausal boundary; this is not a “computationally effective strategy,” yet it can occur, processed pragmatically. 13 If ETI have been around for millions of years, as Ulmschneider (2006: 209) put it, “we have to conclude that these extraterrestrial beings must be essentially God-like, with faculties that border on omniscience and omnipotence” (think about cumulative culture; Tomasello 1999). If this is the case for our ETI, then there is no Earthly hope of understanding anything approaching the singularity. 14 If ETI has surpassed social conventions, we would not expect to see these social conventions affecting language use. 15 We might assume that they have semantic primitives as humans are theorized to have; 14 suggested, with a further few under investigation (Goddard and Wierzbicka 2014: 10). 16 Human syntax does not necessarily have to be hierarchical; there are other proposals on dependency grammar (Mel’cuk 1989) and processing order (O’Grady 2005; Frank et al. 2012). 17 My thanks to a reviewer for helping clarify my thoughts here. 18 See also van der Auwera and Gast (2010) for discussion on prototype theory and protoagent/patient entailments, as compared to having hard linguistic categories like “subject.” 19 The same reviewer questions why an ETI language should have a fixed word order, as apart from argument structure, languages also take into account information structure (e.g., topic-comment); thus, variation could present. 20 See also Bach and Chao (2009) for a sketch on semantic theory and universals. 21 If ETI were to be highly automated in their language processing (but perhaps losing flexibility), we might see the ETI language in the syntactic processing mode rather than the pre-grammatic mode (Givón 2018: 164). As a further note, linguists do not seem to have an objective definition of what “efficient” or “optimal” is – it could be a theory internal construct (fewer steps taken), a timing issue (it is faster to process), or perhaps it requires two neurons to fire as compared to five. 22 Of course, this differs from the argument that Merge is the necessary building block of language that is already encoded genetically via a mutation; rather, any advanced species could actually program the specifics of the language in terms of its lexicon, syntax, etc. While seemingly attractive, there are disadvantages, as well. Nixon and Tomaschek (this volume [Chapter 17]) emphasize the importance of learned communication over that with genetically encoded communication, as the latter is far too slow to deal with rapid cultural changes in an intelligent species – unless they have consciously stopped innovating. 23 See also Roberts et al., this volume (Chapter 15), for their comments on AI. 24 Slobodchikoff (this volume [Chapter 9]) argues that “universals” of mathematics and geometry are not a good candidate for communication, as they may be perceived differently by ETI. 25 Harbour (this volume [Chapter 18]) points out that written human messages provide key information about how we organize our language (e.g., English words are built around syllables) and gives insight into our cognition. An aspect writing usually fails to encode is that our communication is multimodal, having specialized processing for this, as well (Holler and Levinson 2019) – a part of our cognition will be absent from the message. This could be corrected for in METI by including multimodal elements drawn from a conversation. See Ortner (this volume [Chapter 5]) for discussion on the multimodal elements of language. 26 See Sperlich (forthcoming) on the advantages of using natural language messaging over artificial language.

162 Darcy Sperlich

References Arbib, Michael A. 2016. “Towards a computational comparative neuroprimatology: Framing the language ready brain.” Physics of Life Reviews 16: 1–54. Bach, Emmon, and Wynn Chao. 2009. “On semantic universals and typology.” In Language Universals, edited by Morten H. Christiansen, Christopher Collins, and Shimon Edelman. Oxford: Oxford University Press. Accessed August 1, 2018. doi:10.1093/acprof: oso/9780195305432.003.0008. Barbe, Katharina. 1995. Irony in Context. Amsterdam: John Benjamins. Berwick, Robert C., and Noam Chomsky. 2016. Why Only Us: Language and Evolution. Cambridge, MA: MIT Press. Bickerton, Derick. 1990. Language and Species. Chicago, IL: University of Chicago Press. Bybee, Joan. 2009. “Language universals and usage-based theory.” In Language Universals, edited by Morten H. Christiansen, Christopher Collins, and Shimon Edelman. Oxford: Oxford University Press. Accessed August 1, 2018. doi:10.1093/acprof: oso/9780195305432.003.0002. Chafe, Wallace. 2018. Thought-Based Linguistics: How Languages Turn Thoughts into Sounds. Cambridge: Cambridge University Press. Chomsky, Noam. 2017. “The galilean challenge.” Inference: International Review of Science 3 (1). Accessed August 1, 2018. http://inference-review.com/article/ the-galilean-challenge. Christiansen, Morten H., and Nick Chater. 2015. “The language faculty that wasn’t: A usage-based account of natural language recursion.” Frontiers in Psychology 6: 1182. doi:10.3389/fpsyg.2015.01182. Christiansen, Morten H., and Nick Chater. 2016a. “The now-or-never bottleneck: A fundamental constraint on language.” Behavioral and Brain Sciences 39. Accessed August 1, 2018. doi:10.1017/S0140525X1500031X. Christiansen, Morten H., and Nick Chater. 2016b. Creating Language: Integrating Evolution, Acquisition, and Processing. Cambridge, MA: MIT Press. Coleman, John S. 2006. “Design features of language.” In Encyclopedia of Language and Linguistics, edited by Keith Brown, 471–475. Amsterdam: Elsevier. Cristofaro, Sonia. 2010. “Language universals and linguistic knowledge.” In The Oxford Handbook of Linguistic Typology, edited by Jae Jung Song, 227–249. Oxford: Oxford University Press. Evans, Nicholas, and Stephen C. Levinson. 2009. “The myth of language universals: Language diversity and its importance for cognitive science.” Behavioral and Brain Sciences 32: 429–492. Frank, Stefan L., Rens Bod, and Morten H. Christiansen. 2012. “How hierarchical is language use?” Proceedings of the Royal Society B. Accessed August 1, 2018. doi:10.1098/ rspb.2012.1741. Futrell, R., L. Stearns, D.L. Everett, S.T. Piantadosi, and E. Gibson. 2016. “A corpus investigation of syntactic embedding in Pirahã.” PLoS One 11 (3): e0145289. https://doi. org/10.1371/journal.pone.0145289. Givón, Thomas. 2018. On Understanding Grammar: Revised Edition. Amsterdam: John Benjamins. Goddard, Cliff, and Anna Wierzbicka. 2014. Words and Meanings: Lexical Semantics across Domains, Languages and Cultures. Oxford: Oxford University Press. Hauser, Marc D., Noam Chomsky, and W. Tecumseh Fitch. 2002. “The faculty of language: What is it, who has it, and how did it evolve?” Science 298 (5598): 1569–1579.

The Design Features of Extraterrestrial Language

163

Hawkins, John. 2004. Efficiency and Complexity in Grammars. Oxford: Oxford University Press. Hawkins, John. 2014. Cross-Linguistic Variation and Efficiency. Oxford: Oxford University Press. Hockett, Charles F. 1960. “The origin of speech.” Scientific American 203: 88–111. Hockett, Charles F., and Stuart A. Altmann. 1968. “A note on design features.” In Animal Communication: Techniques of Study and Results of Research, edited by Thomas Sebeok, 61–72. Bloomington, IN: Indiana University Press. Holler, Judith, and Stephen C. Levinson. 2019. “Multimodal language processing in human communication.” Trends in Cognitive Sciences 23 (8): 639–652. Holmes, Janet, and Nick Wilson. 2017. An Introduction to Sociolinguistics. London: Pearson. Hornstein, Norbert, and Cedric Boeckx. 2009. “Approaching universals from below: I-universals in light of a minimalist program for linguistic theory.” In Language Universals, edited by Morten H. Christiansen, Christopher Collins, and Shimon Edelman. Oxford: Oxford University Press. Accessed August 1, 2018. doi:10.1093/acprof: oso/9780195305432.003.0002. Huang, Yan. 2000. Anaphora: A Cross-Linguistic Study. Oxford: Oxford University Press. Levinson, Stephen. 2000. Presumptive Meanings: The Theory of Generalized Conversational Implicature. Cambridge, MA: MIT Press. Lindquist, Kristen A., Jennifer K. MacCormack, and Holly Shablack. 2015. “The role of language in emotion: Predictions from psychological constructionism.” Frontiers in Psychology 6: 444. Accessed August 1, 2018. doi:10.3389/fpsyg.2015.00444 Lupyan, Gary, and Rick Dale. 2010. “Language structure is partly determined by social structure.” PLoS One 5 (1): e8559. Accessed August 1, 2018. doi:10.1371/journal. pone.0008559. Mel’cuk, Igor A. 1989. Dependency Syntax: Theory and Practice. Albany, NY: State University of New York Press. Moravcsik, Edith A. 2010. “Explaining language universals.” In The Oxford Handbook of Linguistic Typology, edited by Jae Jung Song, 69–89. Oxford: Oxford University Press. O’Grady, William. 2005. Syntactic Carpentry: An Emergentist Approach to Syntax. Mahwah, NJ: Lawrence Erlbaum. Pellegrino, François, Christophe Coupé, and Egidio Marsico. 2011. “A cross-language perspective on speech information rate.” Language 87 (3): 539–558. Pinker, Steven, and Paul Bloom. 1990. “Natural language and natural selection.” Behavioral and Brain Sciences 13: 707–784. Reboul, Anne. 2017. Cognition and Communication in the Evolution of Language. Oxford: Oxford University Press. Searle, John. 1980. “Minds, brains, and programs.” Behavioral and Brain Sciences 3: 417–424. Searle, John. 2014. “What your computer can’t know.” Review of The 4th Revolution: How the Infosphere Is Reshaping Human Reality, by Luciano Floridi and Superintelligence: Paths, Dangers, Strategies, by Nick Bostrom. The New York Review of Books, October 9. Accessed August 1, 2018. www.nybooks.com/articles/2014/10/09/ what-your-computer-cant-know/. Sperlich, Darcy. 2020. Reflexive Pronouns: A Theoretical and Experimental Synthesis. Cham: Springer. doi:10.1007/978-3-030-63875-7

164 Darcy Sperlich

Sperlich, Darcy. Forthcoming. “Syntactic and pragmatic considerations of natural language METI.” In Cognition and Communication in Extraterrestrial Intelligence, edited by Douglas Vakoch. Oxford: Oxford University Press. Tomasello, Michael. 1999. The Cultural Origins of Human Cognition. Cambridge, MA: Harvard University Press. Ulmschneider, Peter. 2006. Intelligent Life in the Universe: Principles and Requirements Behind its Emergence. Berlin: Springer-Verlag. Van der Auwera, John, and Volker Gast. 2010. “Language universals and linguistic knowledge.” In The Oxford Handbook of Linguistic Typology, edited by Jae Jung Song, 166– 189. Oxford: Oxford University Press. Wittgenstein, Ludwig. 1922. Tractatus Logico-Philosophicus. London: Routledge and Kegan Paul.

15 UNIVERSAL GRAMMAR Ian G. Roberts, Jeffrey Watumull, and Noam Chomsky

Introduction

Is language universal? In particular, is the grammar – the computational system in the mind/brain – that powers human language universal? Could it be universal in the way laws of nature are universal, such that any sufficiently intelligent system would all but inevitably converge on it (in either its evolution or its science)? Here we will expound on the conjecture that the answers to these questions are affirmative: grammar – particularly human grammar – is not specific to our species, but universal in the deepest of senses. The implications for xenolinguistics are obviously profound: we should predict that any extraterrestrial intelligence (ETI) – indeed any sufficiently intelligent system (e.g., artificial intelligence) – we encounter would likely be endowed with a cognitive computational system that runs human-style linguistic “software”, thus eliminating any principled limit to effective communication. However, there could exist be material differences in the “hardware” used to physically externalize linguistic information, but these would pose mere engineering challenges rather than insoluble conceptual problems. It is not unreasonable to suppose that the combined intelligence of humans and ETIs could construct the necessary interface(s). In that case, effective communication between these “universal minds” would be guaranteed. The human mind, Descartes argued, is undoubtedly in some sense a “universal instrument” (Descartes 1637 [1995]). We cannot know with certainty what he intended by this provocative comment, but we do know that the Cartesians would have understood language as fundamental to any nontrivial notion of “universality” and “intelligence” because it is language that empowers humans to generate an unbounded set of hierarchically structured expressions that can enter into (or in fact constitute) effectively infinitely many thoughts and actions – that is, the DOI: 10.4324/9781003352174-15

166 Ian G. Roberts, Jeffrey Watumull, and Noam Chomsky

competence of every human, but no beast or machine, to use language in creative ways appropriate to situations but not caused by them, and to formulate and express these thoughts coherently and without bound, perhaps “incited or inclined” to speak in particular ways by internal and external circumstances but not “compelled” to do so. This linguistic competence (and especially its creative use), in concert with other mental faculties, establishes the general intelligence necessary for the evolutionary “great leap forward” of our species (see Chomsky 2016): there might have been a crucial mutation in human evolution which led, in almost no time from an evolutionary perspective, from [humans living in] caves to [their creating knowledge of such sophistication as to enable us to imagine and construct things as complex as, say,] spaceships. It’s a plausible speculation that the mutation in question was whatever it is that makes our brains capable of computing recursive syntax, since it’s the recursive syntax that really gives language – and thought – their unlimited expressive power. It’s one small step from syntax to spaceships, but a great leap for humans. (Roberts 2017: 182) A great leap for humans – and, on Earth, only humans, evidently (see Berwick and Chomsky 2016). Moreover, it has been assumed that the essential properties of human language are not only unique, but logically contingent: Let us define “universal grammar” (UG) as the system of principles, conditions, and rules that are elements or properties of all human languages not merely by accident but by necessity – of course, I mean biological, not logical necessity. Thus UG can be taken as expressing “the essence of human language.” (Chomsky 1975: 29) There is no a priori reason to expect that human language will have such properties; Martian could be different. (Chomsky 2000: 16) This assumption, we submit, merits rethinking in light of progress in the Minimalist Program (Chomsky 1995). Recent work demonstrating the simplicity (Watumull et al. 2017) and optimality (Chomsky et al. 2017) of language increases the cogency of the following: “the basic principles of language are formulated in terms of notions drawn from the domain of (virtual) conceptual necessity”, the domain defined by “general considerations of conceptual naturalness that have some independent plausibility, namely, simplicity, economy, symmetry, nonredundancy, and the like” (Chomsky 1995: 171, 1) that render linguistic computation optimal. To the extent that this strong minimalist thesis (SMT) is true, the essential – computational –

Universal Grammar

167

properties of language would derive from laws of nature – language- and even biology-independent principles that, once realized in the mind/brain, do entail particular properties as logically necessary. For instance, it is simply a fact of logic that the simplest (optimal) form of the recursive procedure generative of syntactic structures, Merge, has two and only two forms of application (i.e., “external” and “internal” in the sense of combining two separate objects or two where one is inside the other). Relatedly, given the nature of the structures Merge generates, minimal structural distance is necessarily the simplest computation for the structure dependence of rules. And so on and so forth (see Berwick et al. 2011; Chomsky 2013; Watumull 2015 for additional examples). Research in the Minimalist Program starts with the optimality conjecture and proceeds to inquire whether and to what extent it can be sustained given the observed complexities and variety of natural languages. If a gap is discovered, the task is to inquire whether the data can be reinterpreted, or whether principles of simplicity and optimal computation can be reformulated, so as to solve the puzzles within the framework of SMT, thus generating some support, in an interesting and unexpected domain, for Galileo’s precept that nature is simple and it is the task of the scientist to prove it. As we discover more and more of “the essence of human language” to be defined by (virtual) conceptual necessity, the less and less absurd it is to question just how contingent a phenomenon human language really is. It may well be with language as with other phenomena studied in the natural sciences that “[b]ehind it all is surely an idea so simple, so beautiful, that when we grasp it – in a decade, a century, or a millennium – we will all say to each other, how could it have been otherwise?” (Wheeler 1986: 304). In other words, there may well be some a priori reasons to expect human language to have the properties it does; so the ETI’s language might not be so different from human language, after all. Simplicity, Universality, and Merge

Our conjecture is based on the notion of simplicity as originally conceived in generative linguistics. “[S]implicity, economy, compactness, etc.” were proffered in the first work on generative grammar as criteria the grammar of a language must satisfy: Such considerations are in general not trivial or “merely esthetic”. It has been recognized of philosophical systems, and it is, I think, no less true of grammatical systems, that the motives behind the demand for economy are in many ways the same as those behind the demand that there be a system at all. (Chomsky 1951: 1, 67) The idea is elementary but profound: if the theory is no more simple, economical, compact, etc., than the data it is proffered to explain, it is not a theory at all; hence, the

168 Ian G. Roberts, Jeffrey Watumull, and Noam Chomsky

more compressed the theory, the more successful – i.e., the more explanatory – it is. Incidentally, this idea is appreciated surprisingly seldom today: many computational cognitive scientists and machine learning theorists (and hence virtually all “artificial intelligence” [AI] labs in academia and industry) have perversely redefined a successful theory or computer program to be one that merely approximates or classifies unanalyzed data.1 This contrasts dramatically with the Enlightenment definition in which data are selectively analyzed as evidence for/against conjectured explanations. In generative grammar since approximately 1980 (principles-and-parameters [P&P] theory), language acquisition has been explained as the process of setting the values for the finitely many universal parameters of the initial state of the language faculty (universal grammar, UG). The apparent complexity and diversity of linguistic phenomena are illusory and epiphenomenal, emerging from the interaction of invariant principles under varying conditions. This was a radical shift from the early work in generative linguistics, which sought only an evaluation measure that would select among alternative grammars – the simplest congruent with the format encoded in UG and consistent with the primary linguistic data. But with the P&P shift in perspective, simplicity can be reformulated. As discussed in the earliest work in generative linguistics, notions of simplicity assume two distinct forms: the imprecise but profound notion of simplicity that enters into rational inquiry generally, and the theory-internal measure of simplicity that selects among grammars (i-languages). The former notion of simplicity is language-independent, but the theory-internal notion is a component of UG, a subcomponent of the procedure for determining the relation between experience and i-language. In early work, the internal notion was implemented in the form of the evaluation procedure to select among proposed grammars/i-languages consistent with the UG format for rule systems, but the P&P approach transcends that limited, parochial conception of simplicity: with no evaluation procedure, there is no internal notion of simplicity in the earlier sense. There remains only the universal notion. In P&P, grammars – i-languages – are simple, but they are so by virtue of objective principles of computational efficiency (Chomsky 2005), not by stipulation in UG. In fact, rather than “simple”, we propose to define P&P-style acquisition as “economical”, which, in the Leibnizian spirit, we understand to subsume simplicity: The most economical idea, like the most economical engine, is the one that accomplishes most by using least. Simplicity – or fuel consumption – is a different factor from power [i.e., generative capacity, empirical coverage, etc.] but has to be taken equally into consideration [. . .]. The economy of a basis may be said to be the ratio of its strength to its simplicity. But superfluous power is also a waste. Adequacy for a given system is the only relevant factor in the power of a basis; and where we are comparing several alternative bases for some one system, as is normally the case, that factor is a constant. Thus in practice the simplest basis is the most economical. (Goodman 1943: 111)

Universal Grammar

169

Economy, in other words, is a minimax property. In Leibniz’s words (see Roberts and Watumull 2015): “the simplicity of the means counterbalances the richness of the effects” so that in nature “the maximum effect [is] produced by the simplest means”. This is the Galilean ideal (see Chomsky 2002). The maximally economical form of P&P-style learning explicable in terms of objective factors is the traversal of a parameter hierarchy for parameter specification (see Roberts 2012, 2019). In such a system, the child is not enumerating and evaluating grammars.2 Instead, the i-language matures to a steady state in a relatively deterministic process of “answering questions” that emerge naturally and necessarily in the sense that there exist “choices” in acquisition that logically must be “made” for the system to function at all; none of the parameters need be encoded in the genetic endowment (see Obata et al. 2015 for similar ideas). This is the ideal, of course. Like SMT generally, how closely it can be approximated is an empirical matter, and there remain many challenges. Parameter specification – i.e., the P&P conception of “learning” as the specification of values for the variables in i-language – can be schematized as a decision tree (parameter hierarchy) which is governed by minimax economy: minimizing formal features (feature-economy) coupled with maximizing accessible features (input-generalization). Traversal of a hierarchy – a conditional-branching Turing machine program – is inevitably economical in that the shortest (in binary) and most general parameter settings are necessarily “preferred” in the sense that the sooner the computation halts, the shorter the parameter settings. For instance, to specify word order, a series of binary queries with answers of increasing length and decreasing generality (microparameters) is structured thus:

Is head-final present? NO

YES

head-initial

Present on all heads?

FIGURE 15.1

YES

NO

head-final

Present on [+V] heads?

Parameter Hierachy

YES

NO

head-final in clauses only

Present on ...?

170 Ian G. Roberts, Jeffrey Watumull, and Noam Chomsky

For compatibility with computability theory and Boolean logic, the parameter hierarchy can be translated as follows: Hierarchy: H State T: Decision Problem Yes: 0/1 (0 = transition to state T + 1) (1 = halt and output parameter specification for H) No: 0/1 (0 = transition to state T + 1) (1 = halt and output parameter specification for H) Hierarchy: Word Order State 1: Is head-final present? Yes: Output 0 (transition to State 2) No: Output 1 (halt and output “head-initial”) State 2: Present on all heads? Yes: Output 1 (halt and output “head-final”) No: Output 0 (transition to State 3) State 3: Present on [+V] heads? Yes: Output 1 (halt and output “head-final in clause only”) No: Output 0 (transition to State 4) So in P&P, the logic is not “enumerate and evaluate” with stipulative (theoryinternal) simplicity measures; it is “compute all and only what is necessary”, which implies the language-independent reality of economy in that, as with the parameter hierarchies, the process answers all and only the questions it needs to. It is not that there is any explicit instruction in the genetic endowment to prefer simple answers: it is simply otiose and meaningless to answer unasked questions (i.e., once the parameters are set, the computation halts). ETI grammars, too, might therefore grow along the lines of the P&P method. Moreover, the “answers” to “questions” can be represented in binary. Indeed, binary is a notation-independent notion necessary and sufficient to maximize computation with minimal complexity (hence, it is optimal for terrestrial and extraterrestrial computation): functions of arbitrarily many arguments can be realized by the composition of binary (but not unary) functions – a truth of minimax logic with “far-reaching significance for our understanding of the functional architecture of the brain” (Gallistel and King 2010: x) – for the physical realization of intelligence anywhere in the universe. The mathematical and computational import of binary was rendered explicit in the theories of Turing (1936) and Shannon (1948), the former demonstrating the necessarily digital – hence, ultimately binary – nature of universal computation (a universal Turing machine being the most general

Universal Grammar

171

mathematical characterization of computation); the latter formalizing information in terms of bits (binary digits). The consilience of these ideas is our Economy Thesis: human language is based on simple representations (i.e., bits) and strong computations (i.e., the binary functions of Turing machines) – and the “economy of a basis may be said to be the ratio of its strength to its simplicity” (Goodman 1943: 111, emphasis in original). As one of the “general considerations of conceptual naturalness that have some independent plausibility”, economy would be a factor that obtains of any optimally “designed” (natural or artificial) computational system. In terms of universality, if the ETI language were optimal in the sense of conforming to virtual conceptual necessity, then it might be surprisingly similar to human language. In point of fact, we ought not to be too surprised. It is now well established by biologists that convergence is a common theme in any evolutionary process: “the number of evolutionary end-points is limited: by no means is everything possible. [Because of evolutionary convergence,] what is possible usually has been arrived at multiple times, meaning that the emergence of the various biological properties is effectively inevitable” (Conway Morris 2013: xii–xiii); indeed, the distinguished Cambridge paleontologist Simon Conway Morris argues that human-style intelligence was effectively inevitable, given the initial conditions of evolution on Earth. And there is no reason a priori to assume that the principle of evolutionary convergence is unique to the biology of a particular planet. Quite the contrary, if we accept the rational form of inquiry in which the principle is understood abstractly in a computational framework. The idea is that any computational system anywhere made of anything is governed by laws of computation (see Gallistel and King 2010: 167). Given this universality of the functional, mathematical architecture of computation, it is possible that we may need to rethink how uniquely human or even uniquely biological our modes of mental computation really are. One interesting implication is that we must rethink any presumptions that extraterrestrial intelligence or artificial intelligence would really be all that different from human intelligence. So we assume that human language is a computational process that can be characterized by a Turing machine (see Watumull 2015). It is possible to explore the space of all possible Turing machines (i.e., the space of all possible computer programs), not exhaustively of course, but with sufficient breadth and depth to make some profound discoveries. Marvin Minsky and Daniel Bobrow once enumerated and ran some thousands of the simplest Turing machines (computer programs with minimal numbers of rules) and discovered, intriguingly, that out of the infinity of possible behaviors, only a surprisingly – and intriguingly – small subset emerged (Minsky 1985). These divided into the trivial and the nontrivial. The boring programs either halted immediately or erased the input data or looped indefinitely or engaged in some similar silliness. The remainder, however, were singularly interesting: all of these programs executed an effectively identical counting function – a primitive of elementary arithmetic. In fact, this operation reduces to a form of

172 Ian G. Roberts, Jeffrey Watumull, and Noam Chomsky

FIGURE 15.2

Sampling of the universe of possible Turing machines

Merge (see Chomsky 2008). More generally, these “A-machines” (A for arithmetic) prove a point: [I]t seems inevitable that, somewhere, in a growing mind some A-machines must come to be. Now, possibly, there are other, really different ways to count. So there may appear, much, much later, some of what we represent as “B-machines” – which are processes that act in ways which are similar, but not identical to, how the A-machines behave. But, our experiment hints that even the very simplest possible B-machine will be so much more complicated that it is unlikely that any brain would discover one before it first found many A-machines. (Minsky 1985: 121) This is evidence that arithmetic, as represented in an A-machine, is an attractor in the phase space of possible mathematical structures (Figure 15.2): any entity who searches through the simplest processes will soon find fragments which do not merely resemble arithmetic but are arithmetic. It is not a matter of inventiveness or imagination, only a fact about the geography of the universe of computation. (Minsky 1985: 122, emphasis added) This thesis obviously generalizes beyond arithmetic to all “simple” computations (see Wolfram 2002 for countless examples). “Because of this, we can expect certain ‘a priori’ structures to appear, almost always, whenever a computational system evolves by selection from a universe of possible processes” (Minsky 1985: 119). Analogously, we submit that it is not implausible that an evolutionary search through the simplest computations will soon find something like Merge. Merge is an operation so elementary as to be subsumed somehow in every more complex computational procedure: take two objects X and Y already constructed and form

Universal Grammar

173

the object Z without modifying X or Y, or imposing any additional structure on them: thus Merge (X, Y) = {X, Y}. This simple assumption suffices to derive in a principled (necessary) way a complex array of otherwise arbitrary (contingent) phenomena such as the asymmetry of the conceptual-intentional and sensory-motor interfaces (entailing the locus of surface complexity and variety), the ubiquity of dislocation, structure dependence, minimal structural distance for anaphoric and other construals, and the difference between what reaches the mind for semantic interpretation and what reaches the apparatus of articulation and perception (see Chomsky 2017). As implied by our Economy Thesis, simplicity can be defined in algorithmic information theory (or the theory of program-size complexity): the complexity of a program is measured by its maximally compressed length in bits so that the simplest program is that with the shortest description. A search of the phase space of possible programs, whether conducted consciously (e.g., by us, ETIs, etc.) or unconsciously (e.g., by modern computers, evolution, etc.), automatically proceeds in size order from the shortest and increasing to programs no shorter than their outputs (these incompressible programs are effectively lists); many complex programs would subsume simpler programs as the real numbers subsume the natural numbers. And, as demonstrated logically and empirically, “any evolutionary process must first consider relatively simple systems, and thus discover the same, isolated, islands of efficiency” (Minsky 1985: 122). Thus it may well be that, given the universal and invariant laws of evolution, convergence on systems – Turing machines – virtually identical to those “discovered” in our evolutionary history is inevitable.3 Hence our questioning the proposition “Martian could be different”. The fact that simple computations are attractors in the phase space of possible computations goes some way to explaining why language should be optimally designed (insofar as SMT holds) in that an evolutionary search is likely to converge on it, which leads us to consideration of the origin of language.4 The evolution of language is mysterious (see Hauser et al. 2014), but SMT is consistent with the limited archaeological evidence that does exist on the emergence of language, evidently quite recently and suddenly in the evolutionary time frame (see Tattersall 2012). Furthermore, there is compelling evidence for SMT in the design of language itself. For instance, it is a universal truth of natural language that the rules of syntax/semantics are structure-dependent (see Berwick et al. 2011): hierarchy, not linearity, is determinative in the application of rules and interpretation of expressions. Thus, the most reasonable speculation today – and one that opens productive lines of research – is that from some simple rewiring of the brain, Merge emerged, naturally in its simplest form, providing the basis for unbounded and creative thought – the “great leap forward” evidenced in the archaeological record and in the remarkable differences distinguishing modern humans from their predecessors and the rest of the animal kingdom (see Huybregts 2017; Berwick and Chomsky 2016 for in-depth discussion of these topics).

174 Ian G. Roberts, Jeffrey Watumull, and Noam Chomsky

If this conjecture can be sustained, we could answer the question why language should be optimally designed: optimality would be expected under the postulated conditions, with no selectional or other pressures operating, so the emerging system should just follow the laws of nature such as minimal computation and more “general considerations of conceptual naturalness that have some independent plausibility, namely, simplicity, economy, symmetry, nonredundancy, and the like” (Chomsky 1995: 171, 1) – quite like the way a snowflake forms. If this is correct, then, contrary to what was once presumed, there would be a priori reasons to expect that human language will have the properties it does; the “principles, conditions, and rules that are elements or properties of all human languages” (Chomsky 1975: 29) would be logically necessary, deriving from laws of nature. Remarkably, it could be in language, not physics, that we first discover, in Wheeler’s words, “an idea so simple, so beautiful, that [. . .] we will all say to each other, how could it have been otherwise?” (Wheeler 1986: 304). One idea in language so simple that perhaps it could not have been otherwise is Merge. As we have discussed, it is in some sense an attractor in the phase space of possible generative functions: its irreducibility and conceptual necessity render it “inevitable” in the design of any computational system. In this most elementary of forms, Merge functions as follows. Given a workspace WS of syntactic objects {Xi, . . ., Xm}, let ∑ be the shortest sequence (X1, . . ., Xn) such that Xi is accessible and ∑ exhausts WS. Thus, Merge(∑) = {{X1, X2}, X3, . . ., Xm}. By this formulation, which conforms to the simplest, necessary principles of computability theory, we map WS to WS′ by taking any two accessible elements X and Y in WS, and replace X and Y with {X, Y} in WS′. It is manifest that two and only two legitimate forms of Merge follow from this formulation. (1), if X1 is a term of X2 (or conversely), then Merge replaces X1 and X2 by {X1, X2}: Merge(∑) = {{X1, X2}, X3, . . ., Xm}. (2) If neither X1 nor X2 is a term of the other, then, again, Merge(∑) = {{X1, X2}, X3, . . ., Xm}. Call (1) Internal Merge and call (2) External Merge. The outputs of (1) and (2) are identical, consistent with the Galilean ideal, “general considerations of conceptual naturalness that have some independent plausibility, namely, simplicity, economy, symmetry, nonredundancy, and the like” (Chomsky 1995: 171, 1). The simple, legitimate form of Merge we proffer has implications for notions of universality. This “merger as replacement” formulation is formally equivalent to a 2-tag system of the general form given by Post (1943) (whose formulation of rewrite rules influenced early work on generative grammar), which is interesting because tag systems form “a class of systems with a particularly simple underlying structure” (Wolfram 2002: 93). In a tag system, given a sequence of elements, a set of rules removes a fixed number of elements from the beginning of the sequence and replaces them with a fixed number of elements to the end of the sequence.5 For instance, consider a 1-tag system with the rewrite rules (1, . . .) => (. . . , 1, 0) and (0, . . .) => (. . . , 0, 1): If the sequence begins with a 1, remove it, and replace it with 1, 0 at the end of the sequence; If the sequence begins with a 0, remove it, and

Universal Grammar

175

replace it with 0, 1 at the end of the sequence. Representing 1 with black and 0 with white, we can express the rewrite rules as follows (Figure 15.3). And now we can see the complex patterns that emerge from applying such simple rules to simple initial conditions (Figure 15.4).

FIGURE 15.3

Merge encoded in rules of a 2-state, 3-color universal Turing machine— provably the “smallest” universal Turing machine

FIGURE 15.4

Merge formalized as a universal 2-tag system

176 Ian G. Roberts, Jeffrey Watumull, and Noam Chomsky

The patterns that emerge with 2-tag systems become increasingly complex (Figure 15.5). Merge (Figure 15.6) is the simplest 2-tag system with the following rules (n.b., other permutations are possible). What is most interesting is that this 2-tag system is provably equivalent to a universal Turing machine (see the proofs in Minsky 1961; Cocke and Minsky 1964; Davis 1958; Wolfram 2002; Watumull 2015): it can compute anything that is computable. Therefore, it would not be absurd to conject that, by virtue of Merge, the human mind is a universal Turing machine. However we will not defend the conjecture here – we proffer it simply in the spirit of exploration: It is [. . .] quite possible that we, as a species, have crossed a cognitive threshold. Our capacity to express anything, through the recursive syntax and compositional semantics of natural language, might have taken us into a cognitive realm where anything, everything is possible. (Roberts 2017: 181–182)

Universal Grammar

177

Let us suppose that the human mind is a universal instrument, a universal Turing machine. Immediately, we must answer questions of scope and limits. With respect to scope, a Turing-universal mind could arguably explain and understanding everything, in principle. The argument is simple: a universal Turing machine can emulate any other Turing machine (i.e., a universal computer can run any program); a program is a kind of theory (written to be readable/executable by a computer); thus, a universal Turing machine can compute any theory; and thus, assuming that everything in the universe could in principle be explained by and understood within some theory or other (i.e., assuming no magic, miracles, etc.), a universal Turing machine – a Turing-universal mind – could explain and understand everything. QED, perhaps. It is an intriguing conclusion, and not obviously false. However, notwithstanding the universal logic of computation, it is obviously necessary that there exist constraints on the mind if it is to have any scope at all. If a Turing-universal mind is to be a universal explainer, it should not generate all possible explanations – true and false – because that would be merely to restate the problem of explaining Nature: deciding which in an infinite set of explanations are the true (or best) explanations is as difficult as constructing the best explanations in the first place. There must be “a limit on admissible hypotheses”, in the words of Charles Sanders Peirce (Peirce 1903, see Chomsky 2006: xi). This interdependence of scope and limits has been expounded by many creative thinkers and analyzed by (creative) philosophers of aesthetics: the beauty of jazz emerges not by “playing anything”, but only when the improvisation is structured, canalized; the beauty of a poem is a function of its having to satisfy the constraints of its form, as the eminent mathematician Stanislaw Ulam (1976: 180) observed When I was a boy I felt that the role of rhyme in poetry was to compel one to find the unobvious because of the necessity of finding a word which rhymes. This forces novel associations and almost guarantees deviations from routine chains or trains of thought. It becomes paradoxically a sort of automatic mechanism of originality. Thus from science to art, we see that the (hypothesized) infinite creativity of the Turing-universal human mind is non-vacuous and useful and beautiful only if it operates within constraints – constraints discoverable by any (evolved) intelligence. Might this imply that, endowed with universal linguistic Turing machines (i.e., Turing-universal minds) analogous to ours, ETIs would share our sense of beauty? We think it likely. Conclusion: Language and Mind Across the Universe

In the foregoing, we have argued that that the human language faculty is optimally designed for maximal computational simplicity and is instantiated in the

178 Ian G. Roberts, Jeffrey Watumull, and Noam Chomsky

fundamental operation Merge, and that the likelihood is that any recognizably intelligent entity would possess the same system. The arguments from simplicity and optimality take us close to the actual conceptual necessity of this conclusion. What does this mean in the context of potential contact? Let us assume the existence of intelligent ETIs who have developed a technological civilization to at minimum a human-level technological sophistication such that they would be capable of conceiving the existence of civilizations alien to them, such as us, and able to contemplate sending/receiving a signal of some form. To be precise, we assume “human-level of technological sophistication” to include an understanding of fundamental mathematics (particularly computability theory), fundamental physics, whatever mechanisms (if any) enter into biological (or its equivalent) evolution in their environment, etc. The considerations we have raised in the foregoing mean that it is all but logically impossible that such a level of knowledge could be attained without a “language” for generating, storing, and communicating information. Languages are based on grammars: systems of primitives, principles, parameters, and procedures encoded in cognitive mechanisms that determine the possible structural properties of languages. Modern linguistic theory proposes the existence of a species-specific cognitive capacity, UG, that predisposes all human children (in normal environments) to acquire the language(s) to which they are exposed. The central question for xenolinguistics is the degree to which UG, as we currently understand it, would resemble ETI grammar. We have argued that the essential architecture of ETI UG must be virtually identical to that of human UG. Fundamental to the human – and ergo ETI – architecture are three axioms: (I) a “linguistic Turing machine” – a computable function – generative of an infinite set of hierarchically structured expressions that interface (at minimum) two extra-linguistic systems; (II) a conceptual-intentional system; (III) an externalization system for communicating conceptual-intentional information. We have argued that human evolution converged on (I) as a globally optimal solution to generating the conceptual structures represented in (II); there is also a system for connecting them to externalization in (III), whose details we have largely left aside here. The computational optimality/efficiency of (I) derives from a recursive structure-building operation, capable of constructing structures of unbounded complexity by iteratively applying to its own output. In human UG, this operation is Merge. The overwhelming likelihood is that ETI UG would also be based on Merge. Thus, the greatest difficulty in communicating with ETIs would be posed not by their grammar, but in understanding their externalization system; however, we submit that this is an engineering problem which should pose no difficulties in principle (although its practical nature is hard to foresee). Ultimately, this theory of UG for humans and ETI (AI) can enkindled a more general, grander, unified theory of Life, Information, Language, and Intelligence (see Watumull and Chomsky, “To appear”).6

Universal Grammar

179

Notes 1 The machine learning systems (e.g., deep learning neural networks powering large language models) so popular in the current “AI spring” are weak AI: brute-force systems laboriously trained to “unthinkingly” associate patterns in the input data to produce outputs that approximate those data in a process with no resemblance to human cognition (thus betraying Turing’s original vision for AI). These systems will never be truly intelligent, and are to be contrasted with the strong – anthronoetic – AI Turing envisioned: a program designed to attain human-level competence with a human-style typified by syntactic generativity and semantic fluidity – to think the way a human thinks. See Copeland (2004) for more. Today such programs, based on generative grammars, are finally being built at Oceanit. 2 Such an inefficient and unintelligent technique is the modus operandi of many machine learning (weak AI) systems. 3 Indeed, we might speculate that were we to “wind the tape of life back” and play it again, in Stephen Jay Gould’s phrasing, not only would something like Merge re-emerge, but something like humans could well be “inevitable”, as serious biologists have suggested (see Conway Morris 2013). 4 Convergence is a consequence of constraints. As with intelligence, evolution and development are possible only by coupling scope with constraints. Stated generally, the scope of any creative process is a function of its operating within limits. In the context of evolution, for instance, Stuart Kauffman (1993: 118) observes, Adaptive evolution is a search process – driven by mutation, recombination, and selection – on fixed or deforming fitness landscapes. An adapting population flows over the landscape under these forces. The structure of such landscapes, smooth or rugged, governs both the evolvability of populations and the sustained fitness of their members. The structure of fitness landscapes inevitably imposes limitations on adaptive search. The analogy to mind is deeply nontrivial, for “intellectual activity consists mainly of various kinds of search” (Turing 1948: 431). 5 This remove/replace formulation mirrors that of the original remove/replace formulation of Merge (see Chomsky 1995), which we have revised here so as not to stipulate a separate “remove” step. In our formulation, there is simply “replace”. 6 www.youtube.com/watch?v=gqTyg_W_yHI

References Berwick, Robert C., and Noam Chomsky. 2016. Why Only Us. Cambridge, MA: MIT Press. Berwick, Rober C., Paul Pietroski, Berach Yankama, and Noam Chomsky. 2011. “Poverty of the stimulus revisited.” Cognitive Science 35: 1207–1242. Chomsky, Noam. 1951. The Morphophonemics of Modern Hebrew. Master’s thesis. University of Pennsylvania, Philadelphia. Chomsky, Noam. 1975. Reflections on Language. New York: Pantheon. Chomsky, Noam. 1995. The Minimalist Program. Cambridge: MIT Press. Chomsky, Noam. 2000. New Horizons in the Study of Language and Mind. Cambridge: Cambridge University Press. Chomsky, Noam. 2002. On Nature and Language. Cambridge: Cambridge University Press. Chomsky, Noam. 2005. “Three factors in language design.” Linguistic Inquiry 36: 1–22. Chomsky, Noam. 2006. Language and Mind. Cambridge: Cambridge University Press.

180 Ian G. Roberts, Jeffrey Watumull, and Noam Chomsky

Chomsky, Noam. 2008. “On phases.” In Foundational Issues in Linguistic Theory: Essays in Honor of Jean-Roger Vergnaud, edited by Robert Freiden, Carlos P. Otero, and Maria Luisa Zubizarreta, 133–166. Cambridge: MIT Press. Chomsky, Noam. 2013. “Problems of projection.” Lingua 130: 33–49. Chomsky, Noam. 2016. What Kind of Creatures Are We? New York: Columbia University Press. Chomsky, Noam. 2017. “The language capacity: Architecture and evolution.” Psychonomic Bulletin and Review 24: 200–203. Chomsky, Noam, Ángel J. Gallego, and Dennis Ott. 2017. “Generative grammar and the faculty of language: Insights, questions, and challenges.” Catalan Journal of Linguistics. Cocke, John, and Marvin Minsky. 1964. “Universality of tag systems with P = 2.” Journal of the Association for Computing Machinery 11: 15–20. Conway Morris, Simon. 2013. Life’s Solution: Inevitable Humans in a Lonely Universe. Cambridge: Cambridge University Press. Copeland, Jack B., ed. 2004. The Essential Turing. Oxford: Clarendon Press. Davis, Martin. 1958. Computability and Unsolvability. New York: McGraw-Hill Book Company. Descartes, Rene. 1637 [1995]. The Project Gutenberg eBook of a Discourse on Method. Translated by John Veitch. https://www.gutenberg.org/files/59/59-h/59-h.htm Gallistel, Charles Randy, and Adam Philip King. 2010. Memory and the Computational Brain: Why Cognitive Science Will Revolutionize Neuroscience. New York: Wiley-Blackwell. Goodman, Nelson. 1943. “On the simplicity of ideas.” Journal of Symbolic Logic 8: 107–121. Hauser, Marc D., Charles Yang, Robert C. Berwick, Ian Tattersall, Michael J. Ryan, Jeffrey Watumull, Noam Chomsky, and Richard C. Lewontin. 2014. “The mystery of language evolution.” Frontiers in Psychology 5. Huybregts, Riny. 2017. “Phonemic clicks and the mapping asymmetry: How language emerged and speech developed.” Neuroscience and Biobehavioral Reviews. Kauffman, Stuart. 1993. The Origins of Order: Self-Organization and Selection in Evolution. Oxford: Oxford University Press. Minsky, Marvin. 1961. Recursive unsolvability of Post’s problem of “tag” and other topics in the Theory of Turing machines. Annals of Mathematics 74: 437–455. Minsky, Marvin. 1985. “Why intelligent aliens will be intelligible.” In Extraterrestrials: Science and Intelligence, edited by Edward Regis, 117–128. Cambridge: MIT Press. Obata, Miki, Samuel Epstein, and Marlyse Baptista. 2015. “Can crosslinguistically variant grammars be formally identical? Third factor underspecification and the possible elimination of parameters of UG.” Lingua 156: 1–16. Peirce, Charles. 1903. Collected Papers of Charles Sanders Peirce (Volumes I–VI, edited by C. Hartshorne and P. Weiss; Volumes VII–VIII, edited by A.W. Burks). Cambridge: Harvard University Press. Post, Emil. 1943. “Formal reductions of the general combinatorial decision problem.” American Journal of Mathematics 65: 197–215. Roberts, Ian G. 2012. “Macroparameters and minimalism: A programme for comparative research.” In Parameter Theory and Linguistic Change, edited by Charlotte Galves, Sonia Cyrino, Ruth Lopes, Filomena Sândalo, and Juanito Avelar, 320–335. Oxford: Oxford University Press. Roberts, Ian. 2017. The Wonders of Language or How to Make Noises and Influence People. Cambridge: Cambridge University Press.

Universal Grammar

181

Roberts, Ian. 2019. Parameter Hierarchies and Universal Grammar. Oxford: Oxford University Press. Roberts, Ian, and Jeffrey Watumull. 2015. “Leibnizian linguistics.” In 50 Years Later: Reflections on Chomsky’s Aspects, edited by Ángel J. Gallego & Dennis Ott, 211–222. Cambridge: MIT Press. Shannon, Claude E. 1948. “A mathematical theory of communication.” Bell System Technical Journal 27: 379–423, 623–656. Tattersall, Ian. 2012. Masters of the Planet: The Search for Our Human Origins. New York: Palgrave Macmillan. Turing, Alan. 1936. “On computable numbers, with an application to the Entschidungsproblem.” Proceedings of the London Mathematical Society 42: 230–265. Turning, Alan. 1948. Intelligent Machinery. Reproduced in Copeland (ed.) 2004. Page numbers as in Copeland. Ulam, Stanislaw. 1976. Adventures of a Mathematician. New York: Scribner’s. Watumull, Jeffrey. 2015. The Linguistic Turing Machine. Doctoral dissertation. University of Cambridge, Cambridge. Watumull, Jeffrey, and Noam Chomsky. “To appear.” Life, Information, Language, Intelligence. Wheeler, John Archibald. 1986. “How come the Quantum?” Annals of the New York Academy of Sciences 480: 304–316. Wolfram, Stephen. 2002. A New Kind of Science. Champagne, IL: Wolfram Media.

16 WHERE DOES UNIVERSAL GRAMMAR FIT IN THE UNIVERSE? Human Cognition and the Strong Minimalist Thesis Bridget D. Samuels and Jeffrey Punske

Introduction

Hauser et al. (2002) begin by imagining a Martian visiting Earth, noting that such a creature would likely contrast the shared genetic code of Earth’s living creatures with the lack of universality in their communication systems. On the other hand, human languages seem to be constrained by certain principles, subject to parametric variation, known as universal grammar. In this chapter, we will explore a situation analogous to the one in which this hypothetical Martian finds itself: how likely would humans be to find commonalities between human language and an extraterrestrial (ET) communication system? Thinking about this question allows us to define the limits of “universality” in universal grammar, as well as to refine our understanding of the interface between the essential parts of language and the human cognitive and motor-sensory systems that interact with them. All components of language are subject to physical and biological constraints that have come to be known as the “third factor in language design” (Chomsky 2005: 6). To the extent that such principles are grounded in physics, any ET language would be subject to them. However, we hypothesize that the possibility of radically different cognitive and externalization systems could make ET communication systems very unlike Earthly ones (on the variety of communication modalities on Earth, see the contributions to this volume from Kershenbaum [Chapter 2], Herzing [Chapter 4], Pepperberg [Chapter 6], and Slobodchikoff [Chapter 9]). Our second hypothesis follows from this. Despite the fact that human languages vary considerably in almost every respect, there is robust evidence for a universal syntactic spine; however, the origins of such a structure have remained largely unexplored. We hypothesize that this spine arises from third-factor principles, and as such, may be present in some form in ET languages. This could result in some DOI: 10.4324/9781003352174-16

Where Does Universal Grammar Fit in the Universe?

183

deeper commonality between human and ET languages on a syntactic level, despite the expected external differences – thus recapitulating some of the variation seen among languages on Earth. What Is Language?

Differing definitions of “language” may be found throughout philosophy, psychology, cognitive science, linguistics, and other fields. Attempts to provide simple definitions are elusive. As Chomsky has noted, the term “language” can be meaningfully divided into an internal/intensional definition (his “i-language”) and an external one (his “e-language”) (see Chomsky 1986). Generative linguistics generally confines itself to the study of i-language (i.e., one’s mental grammar) often through e-language objects (i.e., the sentences one produces). We want to note differences between a mental grammar and its possible outputs. For instance, the underlying grammatical systems for a proficient English speaker will allow for the production of a sentence of infinite length though an application of a simple rule like the toy rule provided in (1): 1 “Mary said {X}” is a grammatical sentence, provided that X is a grammatical sentence. We can expand our language to two toy grammatical rules: 2 “Bill left” is a grammatical sentence. 3 “Mary said {X}” is a grammatical sentence, provided that X is a grammatical sentence. Allowing other names to take the place of “Mary” for the sake of clarity, we can see that these two simple toy rules would produce a “grammar” that would allow for an infinitely long sentence: 4 John said Sarah said Dave said . . . Mary said Bill left. Such a sentence, while a possible output of the underlying grammatical system, would never be found within the e-language. No infinite sentence can exist as a produced object. In short, our (human) cognitive, grammatical capacity exceeds our ability to produce. Hockett (1960) famously introduced thirteen design features which distinguish language from other communication systems (for further discussion see the chapters in this volume by Kershenbaum [Chapter 2], Ross [Chapter 12], and Slobodchikoff [Chapter 9]): (1) use of the vocal-auditory channel; (2) rapid fading; (3) broadcast transmission and directional reception; (4) interchangeability; (5) total feedback; (6) specialization; (7) semanticity; (8) arbitrariness; (9) discreteness;

184 Bridget D. Samuels, Jeffrey Punske

(10) displacement; (11) productivity; (12) traditional transmission; (13) duality of pattern. It is important to contextualize Hockett’s work. Hockett’s goal was to develop the tools for examining the evolution of human language – his thirteen design features are thus based on the evolutionary development of human and non-human primates. The idea that Hockett’s design features were meant to capture “language” on a truly universal scale is a mischaracterization of his goals. Further, we have the benefit of research that was unavailable to Hockett. This consists largely of work by Stokoe (beginning in 1960) and many others, which has shown that signed languages such as American Sign Language are as complex as spoken languages and are thus fully formed human languages, removing (1) “use of the vocal-auditory channel” as a true design feature (see also Ortner, this volume [Chapter 5]). Such observations are critical for the present discussion because the evolutionary path language might take with a non-human intelligence may not produce all of Hockett’s design features. Thus, it is critical to determine which of Hockett’s features are required for any potential communication system to be considered language. We believe six factors meet this criteria: (4) interchangeability; (5) total feedback; (8) arbitrariness; (10) displacement; (11) productivity; (12) traditional transmission. We further consider one additional feature, namely discreteness. Human languages are composed of discrete elements at various levels of description (though cf. Nixon and Tomaschek, this volume [Chapter 17]). Individual sounds combine to form larger units (morphemes and words) which themselves combine into complex phrases and sentences. Sentences then combine into a discourses, and so forth. The sizes of the discrete meaningful units vary significantly from language to language, but all human languages have discreteness. Whether or not this is a required element of language is a question we discuss in the following sections. This proposed definition allows for some flexibility, so that just as human languages are not limited by modality (vocal-auditory/sign-visual), potential ET languages may also exhibit multiple modalities that are less familiar. However, the core cognitive elements would remain the same. The Human Language Faculty

The constellation of abilities known collectively as the linguistic competence and performance systems are made possible by complex interactions between physiological components of the human body – including but certainly not limited to the brain – and the external world. These physiological components are often called “the language faculty.” The components of the language faculty and their emergence in our species have been the subject of much scrutiny by researchers who are interested in how language evolved among humans. We believe this approach can be instructive for this work, as well.

Where Does Universal Grammar Fit in the Universe?

185

What Is Unique to Humans?

Based on the definition of language proposed in the previous section, it seems clear that at least three components are necessary to characterize language: an abstract computational system to generate syntactic structures, a conceptual-intentional system to endow those structures with meaning, and a sensory-motor system to externalize the structures produced by the former two. Hauser et al. (2002; henceforth HCF) refer to these three components collectively as the faculty of the language in the broad sense, or FLB. They refer to the computational system alone as the faculty of language in the narrow sense, or FLN, and hypothesize that only FLN is uniquely human. That is to say, the remainder of FLB is drawn from a number of systems that are not special to language, or even to our species. HCF explicitly exclude “other organism-internal systems that are necessary but not sufficient for language (e.g., memory, respiration, digestion, circulation, etc.)” from FLB (Hauser et al. 2002: 1571). We will do the same here and set aside issues concerning these systems, keeping in mind that they could differ significantly in ETs. What about the FLN is unique to humans and language? This has been a controversial question in the years since it was first posed, and we do not have the space to recap the arguments here. For present purposes, we will just outline two possibilities. First, HCF suggested that the uniqueness of FLN may lie in syntactic recursion, sometimes known as the operation “Merge.” Merge is a set-formation procedure that takes two syntactic objects (roughly, morphemes), X and Y, and combines them into a set {X, Y} (see, e.g., Chomsky 2012; Boeckx 2016; Roberts et al., this volume [Chapter 15]). Merge can apply to its own output, thus yielding a set {Z, {X, Y}}, and so on. The recursive property of Merge thus entails that syntactic structures are unbounded in length, as in (4). No other species in the animal kingdom seems to have communication systems powered by such a recursive engine. However, some researchers have raised the interesting possibility that Merge may have unlocked other cognitive abilities in humans, and we will return to this later in the chapter. Another possibility, which does not necessarily preclude the uniqueness of Merge, is that we humans stand alone in our ability to turn concepts into syntactic objects – to lexicalize them. This line of reasoning has been pursued by Spelke (2003) and Boeckx (2016). In this view, we are special in our ability to take concepts from any cognitive domain and turn them into the proper “currency” for Merge. This would involve a sort of wrapper enabling the concept to enter the syntactic computation (Boeckx 2016: 27 suggests this is the “edge feature” identified by Chomsky 2008) plus “instructions for the syntax-external systems to ‘fetch’ (‘bind’/activate/retrieve in long-term memory) the relevant concept on the semantic side or the relevant phonological matrix on the sound/sign side” (Boeckx 2016: 28). One could imagine a faculty of language that would have Merge but not free lexicalization, such that it would be possible to create sentences of infinite length, but only from a predefined set of concepts. Alternatively, one could conceive of a

186 Bridget D. Samuels, Jeffrey Punske

faculty of language with free lexicalization but some structure-building operation that is more limited than Merge. In that case, a sentence might be highly constrained in its form and length, but could slot an infinite number of concepts into its templates. In light of these thought experiments, it seems likely to us that both Merge and lexicalization are crucial parts of the human language faculty, and may indeed be unique to humans and to language. The Strong Minimalist Thesis

Intuitively, the meaning of a linguistic object – the thought behind it – is what feels truly important. However, the meaning of a phrase or sentence is structurally compositional, so the order in which the elements undergo Merge is crucial to the eventual meaning. {Pat, {likes, Sam}} thus means something different from {Sam, {likes, Pat}}. Helpfully, the order in which elements are merged is reflected in the order in which they are pronounced/signed, subject to a limited range of language-specific variation. More generally, syntax serves as the spine that structures meaning and externalization. Chomsky (2000 et seq.) hypothesized that the human language faculty is in fact the optimal solution to this set of demands; it balances the conditions dictated by the conceptual-intentional (CI) and sensorymotor (SM) systems with which it interfaces, such that language is actually usable for externalizing thoughts. This hypothesis is known as the strong minimalist thesis (SMT). It has been further proposed that the CI interface is of primary importance for the language faculty; the SM interface may be ancillary (Chomsky 2008; see also Samuels 2011). Another way to say this is that the language faculty is primarily “for” thought, with its usefulness for communication as a bonus. The SMT may seem like a bold claim. What reason do we have to believe that the language faculty is optimal? The view is sometimes expressed that evolution “satisfices,” or does a just-good-enough job. However, there is reason to be more optimistic: there are numerous examples whereby natural selection has indeed found an optimal solution or remarkably close to it (Cockell 2018 provides an accessible overview). Computational modeling has shown that, on a number of levels, animals – including both invertebrates and vertebrates (and notably primates like us) – have the “best-of-all-possible” neural wiring (Cherniak 2012: 361). It may not, then, be such a stretch to take the SMT as our starting point (see Samuels 2011 for further arguments). Three Factors in Language Design

Thus far, our discussion of FLB/FLN and the SMT have focused entirely on the human organism and primarily on the brain as the locus of cognition. These structures are largely determined by our genetics, but we cannot ignore organismexternal contributions to language. If a child grows up hearing English, they will learn English; if they grow up seeing American Sign Language, they will learn

Where Does Universal Grammar Fit in the Universe?

187

that. The externalized portion of language – roughly, what sounds it uses and in what order, and what groups of sounds map to what meanings (i.e., the lexicon) – must be gleaned from the environment and committed to memory. This, of course, is a massive oversimplification of a language learner’s task (see Nixon and Tomaschek, this volume [Chapter 17], for additional discussion), but we set this issue aside to discuss another contribution to language: that of physical laws and other overarching, domain-general principles of computation and data analysis, or as Chomsky (2005) calls them, the “third factor” in language design. The third factor constrains the genetic (first) and environmental (second) factors just mentioned. In the words of Cockell (2018: 239), “the laws of physics and life are the same.” Given that the laws of physics are truly universal, examining third-factor effects might give some insight into possible constraints in forms of extraterrestrial language. Syntax

Much of the work of modern linguistics has illustrated how seemingly wildly disparate and distinct phenomena are actually the same at a fundamental level. In this section, we explore whether these results would necessarily obtain for ET language. We argue that the core universals of human language are constrained by two separate sources: general human cognitive organization (the FLB) and third-factor principles. We consider the hypothesis that potential ET languages, governed by the same third-factor principles as languages on Earth, could arrive at the same FLN as human language but differ in other aspects of their FLB. As a consequence, the likelihood of interstellar interlinguistic communication seems remote, but possible. Language Diversity, Universality, and Human Cognition

Human languages exhibit a great degree of variation; however, this variation is constrained. For instance, among the possible orderings of the three key elements of a sentence,– namely the subject, verb, and object – there seems to be an overwhelming, if not universal, tendency for the subject to proceed to the object. The subject typically precedes the object in more than 83% of the 1,377 languages surveyed in the World Atlas of Linguistic Structures (Dryer 2013). The preference for subject-predicate and subject-object orders is an area that demands an explanation from formal linguistics – especially given that syntactic operations can obscure the underlying structures. However, the core preference itself is arguably part of a more general cognitive preference regarding the ordering and organization of events and thus is best understood as part of the FLB. Within generative syntax, the most common approach is to assume a universally ordered hierarchy of syntactic structure (“the spine”) (see, for instance, Cinque

188 Bridget D. Samuels, Jeffrey Punske

1999; Wiltschko 2014). A generalized version of this is given in (5) (cf. Wiltschko 2014): 5 Generalized Universal Spine

Discourse Domain

Grammatical Relations/ TAM

Thematic Domain

The spine allows for considerable linguistic variation. Various types of movement and other operations may apply to the lexical elements in the tree. And, following Ritter and Wiltschko (2014), the semantic substance of particular nodes may also vary. They examine the node INFL, which they argue is best understood as a general “anchoring” node. In a language like English, INFL anchors the sentence to time (tense). In languages like Halkomelem, the anchoring is spatial. In languages like Blackfoot, the anchoring is based on discourse participants. We can see the same general core functions in the same syntactic positions across the world’s languages. How this would arise from something as simple as combining two elements together (i.e., Merge) is not obvious. The question remains whether this “spine” lies within the domain of FLN; i.e., is it specific to humans and to language? We hypothesize that it does not. It seems quite plausible that the universal spine is the result of syntax conforming to demands from the CI systems (belonging to FLB) with respect to how we interpret scenes and events; we agree with Sperlich (this volume [Chapter 14]) that if extraterrestrials share our “cause-and-effect” view of the world, they might also share our relations of subject ~ agent, object ~ undergoer/patient, and so forth. The question of how the syntactic spine arose, and whether to situate it in universal grammar or rather in extralinguistic principles, remains open. Boeckx (2016, sections 2.3–2.4) provides one proposal, which reduces lexical and functional categories (e.g., nouns, verbs, adjectives, prepositions, tense, etc.) to variations on two basic categories with differing syntactic properties. When we look more broadly at non-linguistic intelligences on Earth, we see evidence for “linguistic” features in a non-linguistic context. One very clear example of this is Golston’s (2018) work, which argues that φ-features (person, number, gender) are found throughout animal

Where Does Universal Grammar Fit in the Universe?

189

cognition, though other chapters in the present volume, in particular those of Pepperberg (Chapter 6) and Slobodchikoff (Chapter 9), serve as cautionary reminders that the mental worlds of other species, even on Earth, can differ tremendously from ours. This, again, is broadly consistent with the idea being advanced here that much of the work of generative syntax is actually an examination of the FLB in a particular context. Could ETs Have Human-Like Syntax?

Given that the core syntactic operation, Merge, appears to be unique to humans, it may seem counterintuitive to argue that ETs could have Merge, as well. However, we find this to be an intriguing possibility with some arguments in its favor. Chomsky (2008: 6) and others have claimed that Merge can define the basic mathematical operation of addition: the recursive algorithm known as the successor function, which states that every number N has a unique successor, N + 1. If our arithmetic abilities are related to Merge, this would explain why humans alone on Earth have the property of discrete infinity both in our language and in our system of natural numbers. While many other species can represent small numerosities precisely (up to ~4) and compare quantities approximately (see Dehaene 1997 for an overview), it may be that Merge integrates these two systems, enabling the precise representation of a discrete infinity of numbers (Hiraiwa 2017). If we encounter a technologically advanced ET species – one that has built a spacecraft or a radio transmitter, for example – it stands to reason that they have some system for representing mathematics. Minsky (1985) makes the strong claim that they will not only have mathematics, but they will have the same arithmetic system as ours, because “there are simply no easily accessible alternatives. . . . [T]here is nothing which resembles [it] that is not either identical or vastly more complicated.” He supports this view with an analysis of possible Turing machines and some thought experiments. Depending on one’s view of mathematical realism, these arguments may or may not be compelling. But if it is correct that Merge is intrinsically linked to our mental arithmetic, and if that mental arithmetic is shared by ETs, then there is a possibility that ETs could also use Merge to structure their language (a conclusion that we appear to largely share with Roberts et al., this volume [Chapter 15]). This operation could potentially then interact with the FLB of ET species, which could differ dramatically from those of humans. Externalization Systems

It is easy to see that the externalization of human language exhibits tremendous variation. Languages can be spoken or signed, thus making use of multiple sensory modalities (auditory, visual, and tactile for the deaf/blind). Within these modalities, languages use different repertoires of sounds, hand shapes, and motion patterns, and place different restrictions (phonotactics) on the sequencing or co-occurrence

190 Bridget D. Samuels, Jeffrey Punske

of these elements. Moreover, the relationship between sound/sign and meaning is arbitrary; there is nothing about the word dog that represents the quality of canineness. There is also variation in how syntactic structures are linearized, resulting in word-order differences, as mentioned in the previous section. In sum, the externalized portion of a given language must be learned. Most, if not all, of our language externalization system is shared with or convergently evolved in other species and thus constitutes part of FLB but not FLN (see Samuels 2011, 2015 and references therein). The range of variation in these aspects of human language is large but not limitless (see Samuels 2011 for discussion). One could imagine ETs using other means of biological or technological transmission that they can perceive, control, and produce. Some possibilities have been explored elsewhere, including in several chapters in this volume, and include various portions of the electromagnetic spectrum, sound, and chemicals. We refrain from further speculation here. Though the possibilities are many, third-factor principles should constrain ET language just as they constrain ours. For example, we expect that an ET language would have a vocabulary of discrete units that can be combined. Jackendoff (2011: 604) summarizes this argument, which dates back at least to Hockett (1960): [T]he only way to develop a large vocabulary and still keep vocal signals distinguishable is to digitize them. This is a matter of signal detection theory, not of biology, and it is also responsible for the fact that we all have digital rather than analog computers. So we might want to say that the digital property of phonology comes by virtue of “natural law.” That is to say, the discreteness of linguistic units is a third-factor effect; despite Jackendoff’s use of the word “vocal” in the preceding passage, there is no reason to suspect that this is dependent on modality. Kershenbaum (this volume [Chapter 2]) raises the possibility that animal cognition and communicative signals may be non-discrete, but to our knowledge, empirical support for this view is lacking. Furthermore, the perceptual system ought to be able to tolerate some noise in the signal; “the importance of error correction” is also discussed in a somewhat different context in Granger et al.’s contribution to this volume (Chapter 8). In humans and other animals, this is partly achieved through categorical perception. This allows us to treat every token of the word cat as belonging to the same abstract category (type), even though no two naturally produced tokens are ever exactly alike in volume, pitch, vocal timbre, and so forth (for an accessible introduction to speech acoustics and physiology, see Lieberman 2018). For sound, these categories appear to have their basis in biases of our inherited mammalian auditory system (see Samuels 2011), but they need to be learned on a language-specific basis. If ET languages also have a learned component, we should expect them to have a capacity akin to vocal learning (though, of course, the modality may differ), i.e., the ability to commit new categories to memory and to imitate them. This ability

Where Does Universal Grammar Fit in the Universe?

191

is rare on Earth, although it evolved independently multiple times and is present in some birds, cetaceans, pinnipeds, elephants, and bats (see Lattenkamp and Vernes 2018 for an overview). Aerodynamics and biomechanics jointly define what sounds humans can make (see e.g., Marchal 2009 and Nixon and Tomascheck, this volume [Chapter 17], for an overview), and the same should be true of ETs that employ biological means of linguistic externalization. The gas laws are highly relevant to our production of sound, which uses a system of air cavities, pistons, and valves to manipulate our breath. Our tongue, larynx, and chest control gas volume and pressure, which are related by Boyle’s Law. The force needed to cause the lungs to expand within the ribcage is subject to Hooke’s Law, and the pressure of air in the lungs must be understood in comparison to atmospheric pressure. Atmospheric pressure will vary from planet to planet, but the gas laws remain the same. If ETs manipulate volumes of gases to produce language, they will apply. Cockell (2018: 62) provides numerous examples of how physical laws affect biology in domains besides that of linguistic production, and concludes that “the ubiquity of convergent evolution across the biosphere and the vast number of examples show that solutions [to those laws] are not limitless, but usually rather few.” Conclusions: Would an ET Language Be Comprehensible?

When we examine human languages, we see they exhibit particular patterns of their internal elements that do not follow from the SMT. Put more succinctly, wordorder patterns and many phonological properties may be better understood as a product of the FLB rather than the FLN. The central thesis of this chapter, as it relates to the interaction of CI systems and language, is that the many of the core, universal properties of human languages draw more broadly from our general cognitive systems and preferences. These preferences are part of our genetic endowment and are shared with many other species on Earth because of our shared evolutionary paths. However, only humans have language because of the unique contribution of our FLN (Merge and/ or lexicalization). We have raised the possibility that human language and a potential ET language could share some of the properties of the FLN, specifically Merge, in addition to properties determined by third-factor principles, such as discreteness. It is very likely that an ET message structured by these principles would be recognizable as a language. However, given the potential differences in the FLB, especially major potential differences in the CI systems, the ability to parse and understand such a message is highly questionable. The question of whether ETs could communicate about any concepts they could form – in other words, whether they could have free lexicalization, as discussed in the third section of this chapter – is impossible to answer. Thus, the major obstacle preventing us from understanding ET languages, or vice versa, may ultimately not be one of exo-linguistics but one of exo-cognition.

192 Bridget D. Samuels, Jeffrey Punske

References Boeckx, Cedric. 2016. Elementary Syntactic Structures: Prospects of a Feature-Free Syntax. Cambridge: Cambridge University Press. Cherniak, Christopher. 2012. “Neural wiring optimization.” Progress in Brain Research 195: 361–371. Chomsky, Noam. 1986. Knowledge of Language: Its Nature, Origin and Use. New York: Praeger. Chomsky, Noam. 2000. “Minimalist inquiries: The framework.” In Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik, edited by Roger Martin, David Michaels, and Juan Uriagereka, 89–155. Cambridge, MA: MIT Press. Chomsky, Noam. 2005. “Three factors in language design.” Linguistic Inquiry 36(1): 1–22. Chomsky, Noam. 2008. “On phases.” In Foundational Issues in Linguistic Theory, edited by Robert Friedin, Carlos P. Otero, and Maria Luisa Zubizarreta, 133–166. Cambridge: MIT Press. Chomsky, Noam. 2012. “Foreword.” In Phases: Developing the Framework, edited by Ángel J. Gallego, 1–7. Berlin: Mouton de Gruyter. Cinque, Guglielmo. 1999. Adverbs and Functional Heads: A Cross-Linguistic Perspective. Oxford: Oxford University Press. Cockell, Charles S. 2018. The Equations of Life: How Physics Shapes Evolution. New York: Basic Books. Dehaene, Stanislas. 1997. The Number Sense: How the Mind Creates Mathematics. Oxford: Oxford University Press. Dryer, Matthew S. 2013. “Order of subject, object and verb.” In The World Atlas of Language Structures Online, edited by Matthew S. Dryer and Martin Haspelmath. Leipzig: Max Planck Institute for Evolutionary Anthropology. Accessed September 24, 2018. http://wals.info/chapter/81. Golston, Chris. 2018. “Phi features in animal cognition.” Biolinguistics 12: 55–98. Accessed September 18, 2018. www.biolinguistics.eu/index.php/biolinguistics/article/ view/603/419. Hauser, Marc D., Noam Chomsky, and W. Tecumseh Fitch. 2002. “The faculty of language: What is it, who has it, and how did it evolve?” Science 298: 1569–1579. Hiraiwa, Ken. 2017. “The faculty of language integrates the two core systems of number.” Frontiers in Psychology 8: 351. Hockett, Charles F. 1960. “The origin of speech.” Scientific American 203: 88–96. Jackendoff, Ray. 2011. “What is the human language faculty? Two views.” Language 87 (3): 586–624. Lattenkamp, Ella Z., and Sonja C. Vernes. 2018. “Vocal learning: A language-relevant trait in need of a broad cross-species approach.” Current Opinion in Behavioral Sciences 21: 209–215. Lieberman, Philip. 2018. “Why human speech is special.” The Scientist, July. Accessed September 24, 2018. www.the-scientist.com/features/why-human-speech-is-special-64351. Marchal, Alain. 2009. From Speech Physiology to Linguistic Phonetics. Hoboken, NJ: Wiley & Sons. Minsky, Marvin. 1985. “Communication with alien intelligence: It may not be as difficult as you think.” Byte Magazine 10 (4): 127–138. Accessed April 21, 2023. https://web.media. mit.edu/~minsky/papers/AlienIntelligence.html Ritter, Elizabeth, and Martina Wiltschko. 2014. “The composition of INFL.” Natural Language & Linguistic Theory 32 (4): 1331–1386.

Where Does Universal Grammar Fit in the Universe?

193

Samuels, Bridget D. 2011. Phonological Architecture: A Biolinguistic Perspective. Oxford: Oxford University Press. Samuels, Bridget D. 2015. “Can a bird brain do phonology?” Frontiers in Psychology 6: 1082. doi:10.3389/fpsyg.2015.01082. Spelke, Elizabeth. 2003. “What makes us smart? Core knowledge and natural language.” In Language and Mind: Advances in the Study of Language and Thought, edited by Dedre Gentner and Susan Goldin-Meadow, 277–311. Cambridge, MA: MIT Press. Wiltschko, Martina. 2014. The Universal Structure of Categories: Towards a Formal Typology. Cambridge: Cambridge University Press.

17 LEARNING AND ADAPTATION OF COMMUNICATION SYSTEMS IN BIOLOGICAL LIFE FORMS Jessie S. Nixon and Fabian Tomaschek

Introduction

This volume contemplates what possible languages or communication systems might be used by extra-terrestrial life forms, with the aim of creating some insight into how to tackle the task of interpreting communication signals should we ever encounter them. Other researchers have already addressed the issue that it is exceedingly improbable that we will ever encounter extra-terrestrial communication. We leave that issue aside and focus on the value of this exercise for better understanding terrestrial communication systems. This thought experiment allows us to step back and take an outside perspective on the communication systems here on Earth. As Michaud (2011) points out, in the absence of any information about extra-terrestrial life, one of the only means available to us for predicting their nature and behaviour is by analogy with ourselves. We take this same approach. In this chapter, we examine terrestrial communication signals, focusing mainly on human language; the adaptability of speech perception, including adaptation to environmental cues; and the cognitive and learning mechanisms that have been proposed to support such a complex, adaptive communication system. We discuss implications for extra-terrestrial communication and for the task of interpreting such communication signals. Human Communication Systems

Human communication systems come in diverse forms and are extremely complex. Some examples of communication systems that have emerged through pressure for communication are sign language (Nyst 2015), gestures, visual cues such as facial expression and body language, drum languages (Rattray 1923) and whistling DOI: 10.4324/9781003352174-17

Learning and Adaptation of Communication Systems

FIGURE 17.1

195

a) Schematic representation of the source spectrum created by the vocal cords. b) Filter spectrum of the vocal tract, which functions as a resonance filter. c) Spectrum of the articulated vowel. Each spectral peak represents a formant (F1, F2, F3). Changing the shape of the mouth (e.g., its diameter) changes the relative location of the formants.

Source: Figure according to Fant (1960: 19)

languages (Meyer 2004).1 This diversity demonstrates a high degree of adaptivity – as long as the sensory information is able to be discriminated to a sufficient level of complexity, it can be used for communication. However, by far, the most common form of communication is spoken language. Human speech is primarily based on an acoustic signal which is created by movements of the human masticatory organs – jaw, lips and tongue (Weber et al. 1986; Yashiro and Takada 2005). The acoustic system used by humans is often divided into two broad categories – consonants and vowels. “Consonants” constitute partial or complete constriction of airflow. Constriction of airflow is achieved in various ways, including closing of the jaw, placing various sections of the tongue against different areas of the mouth, closing the lips and sealing the oral cavity to allow airflow through the nasal cavity. Sounds created by an unconstricted airflow are classified as “vowels”. To produce vowels, the vocal tract functions as a resonance filter that filters the full spectrum signal created by the vocal cords (Figure 17.1a–b). The acoustic signal contains several spectral peaks (Figure 17.1c), where the first peak is called the first formant (F1), the second peak the second formant (F2), and so forth. The spectral properties in Figure 17.1c represent the vowel schwa, which is articulated by positioning the tongue in the center of the vocal tract and is equivalent to the first vowel in words like “about” or the second vowel in “speaker”. Moving the tongue in the oral cavity modifies the vocal tract’s shape, changing the characteristics of the resonance filter, which results in different formant frequencies, thus in different vowel qualities. For example, a high fronted tongue position creates an [i], a lower central tongue position an [a] and a high-retracted tongue position with lip rounding a [u] (cf. Figure 17.3a. Figure 17.3b shows the vowel space on the basis of F1 (y-axis) and F2 (x-axis) frequencies. The black triangle represents the position of the five human standard vowels in the F1/F2 vowel space. Note that the orientation of the triangle in Figure 17.3b is equivalent to the triangle’s orientation in Figure 17.3a.

196 Jessie S. Nixon and Fabian Tomaschek

During speech, the various different speech apparatuses move dynamically, leading to continual changes to the resonance filter. This means that the resonance filter not only depends on the vowel type, but also, for example, on the surrounding consonants – so the vowel in “bat” is different to the vowel in “can”, because both preceding and following mouth shapes affect the resonance filter (Magen 1997; Öhman 1966). If clustered by sound category, cue values vary around a centre with considerable overlap between clusters. In other words, rather than being discrete, with a consistent mapping to semantics or even to sound categories, acoustic cues are continuous and variable. In fact, there is a multitude of sources of variability in the signal that do not help the listener understand the message (Altmann 1980; Fosler-Lussier and Morgan 1999; Fuchs et al. 2008; Gay 1978; Hirata 2004; Jiao et al. 2019; Leemann et al. 2012; Tomaschek and Leemann 2018; Tomaschek et al. 2013, 2014, 2018a, 2018b; Weirich and Fuchs 2006; Winkler et al. 2006) and which fall outside the limits of what is traditionally considered the linguistic message, such as indexical information. For example, age perception from voice depends on linguistic-cultural familiarity with the speaker, suggesting that speech reflects – and listeners learn – not only biological changes with age, but also language-specific socio-cultural age effects (Jiao et al. 2019). Speakers use the speech signal to convey a message. According to information theory, communication boils down to the reproduction of a message at one point from a message sent at another point (Shannon 1948). There is a set of shared codes that point to a limited set of predefined messages. “Comprehension” of a message simply means selecting the correct message from all the possible messages that the sender could have sent. If we consider international Morse code prosigns, this makes sense intuitively. Probably the most famous example is - - - – – – - - -, a series of three short, three long, then three short signals. This code signals a distress call in a case when there is an imminent threat to life or property. It is distinct from, for example, a general call for attention or hail – - – - – or from a request for repetition - - – – - - or “correction, my preceding message was an error”: - - - - - - - -. The messages do not break down into components that “contain meaning”.2 Communication of this message from the sender to the receiver is possible because both parties share the same code and this code allows the receiver to discriminate between a distress call, a request for repetition, an error correction and any other possible message. Ramscar et al. (2010) argue that language is likewise discriminative. Cues in the signal serve to reduce uncertainty, enabling the receiver to select which of the possible messages was intended, rather than each message being built up from component parts or units of meaning (see Ramscar et al. 2010; Wittgenstein 1953, for discussion of the problems around meaning and reference). Ramscar (2019) argues that while language is discriminative, in human language, there is no predefined code at birth which is shared by all speakers. Instead, the language develops over a lifetime through interaction with the world and with other users of the communication system. The code is constantly negotiated between sender and receiver. One consequence is that no two users have identical

Learning and Adaptation of Communication Systems

197

code, so it is only partially shared. Ramscar (2019) also argues that language is deductive. Cues reduce uncertainty by excluding unlikely messages. Using the previously discussed Morse code prosigns as an example, a receiver encountering a distress call can immediately rule out (or at least have low expectations of) a hail signal on hearing the first short signal; a request for repetition becomes unlikely on the third short signal; and a correction message becomes unlikely on hearing the first long signal. Of course, context also plays a role, so when there has been no prior communication, a correction message or a request for repetition are already unlikely before any signal is encountered. Similarly, in language, the context serves to constrain the set of possible messages. The speech signal (if effective) serves to further reduce uncertainty by eliminating alternatives – although in real communication, typically some degree of uncertainty remains, the extent of which depends on many factors. What Cognitive and Learning Mechanisms Support Such a Complex System?

Although speech is generally an effective means of communication (to an extent), the speech signal is so complex that after decades of study, we still on an academic level understand neither what cues make up the system nor the relationship between auditory cues and the message. Some researchers propose that speech is made up of a small set of discrete sound units (sometimes called “phonemes”). However, if people are asked to identify words taken from recordings of running speech, accuracy is as low as 20%–40% by participant (Arnold et al. 2017). If speech consisted of strings of sound units, we would not expect such difficulties when these individual units are extracted from speech (see Nixon and Tomaschek 2023a, for a full discussion of the continuous nature of the speech signal and a review of how models of speech perception have taken this into account). Contrast this with written language: a single word can generally still be read with high accuracy when extracted from a sentence. A large body of evidence has accumulated over the last several decades showing that human infants and adults are highly sensitive to the distributional characteristics of cues in speech (Clayards et al. 2008; Maye et al. 2002; McMurray et al. 2009; Nixon and Best 2018; Nixon et al. 2016, 2018; Yoshida et al. 2010). This line of research has been groundbreaking, because it has demonstrated that humans have learning mechanisms that enable them learn speech and language through exposure and interaction. Prior to this, many researchers believed that language was too complex to learn and must therefore be innate. However, while these data are often interpreted as suggesting that learning occurs by directly learning the statistical structure through statistical clustering mechanisms (e.g., Maye et al. 2002), computational models have shown that unsupervised statistical clustering alone is not sufficient to account for learning of speech sounds (e.g., McMurray et al. 2009). This has given rise to research into alternative mechanisms that might account for learning of speech.

198

Jessie S. Nixon and Fabian Tomaschek

Recent work has discovered a role of error-driven discriminative learning in speech acquisition (Nixon 2018, 2020; Nixon and Tomaschek 2020, 2021). Errordriven learning is a learning system used by many animal species on Earth (Waelti et al. 2001; Soto and Wasserman 2010). Learning processes in animals have been meticulously studied for more than a century, from which well-developed, well-tested mathematical, theoretical and computational models of learning have emerged (e.g., Rescorla and Wagner 1972; Widrow and Hoff 1960). Despite enormous surface variation in biological systems, error-driven learning models have successfully predicted learning behaviour across a broad range of species, including humans, and under a wide variety of conditions. What these studies demonstrate is that in a wide variety of biological systems on Earth, learning involves using incoming sensory information, or cues, to predict important events in the world. When cues frequently predict a certain outcome, then experiencing that cue again in the future leads to an expectation that the outcome will also occur. Error-driven learning models have been remarkably successful in predicting and explaining a broad range of phenomena in human language, including language acquisition, comprehension and production (Arnold et al. 2017; Baayen et al. 2011, 2016; Nixon 2018, 2020; Nixon and Tomaschek 2020, 2021; Tomaschek et al. 2021, 2022; see Nixon and Tomaschek 2023b, and the contributions therein for an overview). Further research is needed to reconcile the various learning models with respect to human language learning; it seems likely that multiple learning mechanisms may be involved. Another aspect that needs to be considered for human learning is the role of higher-order cognitive processes (Nixon et al. 2022; Reber 1989), which seem to differ between children and adults in language learning (Ramscar et al. 2013). For example, some researchers have modelled causal inference learning with error-driven learning models (Shanks 1986; Van Hamme and Wasserman 1994), but other research has shown that human cause attribution involves a complex set of processes, including reasoning (Einhorn and Hogarth 1986). Anecdotally, these sorts of higher-order cognitive processes seem to characterise second language learning in the classroom more than first language or immersive learning. Error-Driven Learning in Non-Human Animals

Error-driven learning is not restricted to humans, of course; its discovery emerged from animal research. Perhaps the most famous learning experiments are Pavlov’s experiments with dogs at the turn of the 20th century. Pavlov found that if he repeatedly presented a stimulus, such as a bell, before feeding the dogs, eventually the dogs would start salivating in response to the stimulus even when the food was not present. Initially, this was considered a simple reflex response and thought to be merely due to contiguity of the stimuli: the reflex response that belonged to the food somehow got taken over by the new stimulus (e.g., Hull 1943; Spence 1956; reviewed in Gluck and Bower 1988). However, in the decades that followed, it

Learning and Adaptation of Communication Systems

200

400

600

0.8 0

200

400

600

800 1000

c)Light tone is blocked

d)Tone tone is blocked

400

600

800 1000

0.4

start: tone 0.0

0.8 0.4 0.0

200

0.8

Events

Weight of light to shock

Events

Events

FIGURE 17.2

0.4

800 1000

start: tone

0

start: tone 0.0

Weight of light to shock

0.8 0.4

start: tone

0

Weight of light to shock

b)Tone

0.0

Weight of light to shock

a)Light

199

0

200

400

600

800 1000

Events

Illustration of the blocking effect estimated by means of Rescorla-Wagner learning equations. a–b) Light and tone a cues for shock are trained together. c–d) Light is pre-trained, tone is blocked. Vertical dashed line indicates the beginning of training with the tone cue in addition to the light cue.

became clear that, rather than a simple association, the animals were using information in the cues to learn about the world through a process of prediction and error feedback. In a seminal study, Kamin (1968) trained rats with two different cues – a light and a tone – followed by an electric shock (see Figure 17.2). In the test, if he presented just one of these cues, say the tone, the rats would exhibit a fear response to the tone – they had learned that the tone predicted the shock. A separate group of rats had identical training and test, except that before the light + tone training, there was a pre-training phase during which the rats were trained with only one cue: the cue that was not tested – in this case, the light. The pre-trained rats did not respond

200 Jessie S. Nixon and Fabian Tomaschek

to the tone. Because the light already provided sufficient information to predict the shock, the tone added no additional predictive value. There was no longer sufficient uncertainty left to drive learning of the tone. So the tone was not learned. In other words, learning of the tone was blocked by the already-learned light cue. The Rescorla-Wagner equations were developed to account for this finding, as well as a wealth of data that had accumulated over the previous decades from research on animal learning. The level of expectation of a particular outcome given a particular cue or cues is described as the cue-outcome connection weight. Connection weights are adjusted in each learning event. A few simple principles can be inferred from these mathematical equations. First, for any cues not present in a given learning event, there is no change to the connection weights. For cues present during a learning event connection weights are 1) increased to all outcomes that occur and 2) decreased to any outcomes that do not occur, but which have been encountered previously. What is important is that the amount of change in 1) and 2) depends on the history of learning. When an outcome is highly expected (that is, connection weights are strong due to previous learning), and the outcome occurs, there is little surprise and therefore little learning, so although connection weight increases, the increase is relatively small. In contrast, if an outcome is not expected – that is, connection weights are weak, because for example, the cues have seldom been encountered before or the cues have been encountered before, but the outcome seldom occurred – and yet the outcome does occur, the surprise is greater and therefore learning is greater. The increase in connection weights is greater than in the former example. These simple equations make strong predictions that have been powerful tools for understanding learning behaviour. For example, learning is non-linear and current learning depends on the history of learning. This leads to the famous “learning curve” – the idea that learning is fast in the beginning then tapers off as things become increasingly predictable. How the Speech of Extra-Terrestrial Intelligence Might Sound

Without any data about extra-terrestrial life, the possibilities for the types of communication systems that might exist seem virtually limitless. Rather than attempt to catalogue the possibilities, our approach in this section is to start with what we know about human speech on Earth, as outlined previously, and make the subtlest of adjustments to the conditions. We assume an acoustic speech signal, produced with a similar biological system to the human speech apparatus. What we find is that varying just one dimension – vocal tract size or atmospheric conditions – could result in dramatic changes to a signal that is otherwise identical to human speech, potentially rendering it unrecognisable. Although human speakers are generally unaware of this, the acoustic characteristics of their speech signal depend on 1) the air temperature and density, which affect the speed of sound, c, and 2) their body size, which affects the size of their

Learning and Adaptation of Communication Systems

201

vocal tract L. These variables can be used to calculate the formant frequencies of [@] and therefore all other vowels using equation (1). (1)

Fn = (2n − 1) × c / 4L

Where n = the formant’s number c = the speed of sound L = the length of the vocal tract For an average human vocal tract whose length is L = 17.5 cm and the average speed of sound c = 350 m/s (with air density ~ 1.20 kg/m and 20°C), formant frequencies are F1 = 500 Hz and F2 = 1500 Hz. The human vowel space in Figure 17.3 is based on these measures. When temperature on Earth drops to −2°C at sea level, speed of sound is slower (c = 319 m/s), directly affecting formant frequencies produced by the same human speaker. Looking at the differences between the planets in our own solar system in terms of their atmospheric composition and therefore speed of sound (NASA n.d.), it is very unlikely that the atmosphere of the extra-terrestrial life form’s home planet will have exactly the same physical characteristics as our own. For example, speed of sound on Mars ranges from 244 m/s at ground level to 220 m/s at the top of Olympus Mons at an altitude of 24 km (NASA n.d.). Since speed of sound will be proportional to the density and temperature of the life form’s planet, the spectral characteristics of the acoustic signal of extra-terrestrial life forms will vary, too, if they communicate by means of acoustic signals. Let us inspect what effect variation in speed of sound has on vowel acoustics. To illustrate the effects of atmospheric properties of the life form’s home planet, we can manipulate the c parameter in equation (1). The greater c on the home planet, the higher the formant frequencies, significantly affecting the vowel quality of the produced sound. Life forms that live on a planet whose atmosphere has c = 200 m/s will produce vowels that have frequencies lower than human vowels (grey dashed triangle in Figure 17.3b). An [a] produced by the extra-terrestrial intelligence will thus sound to human ears like an [o]. When the speed of sound is higher than on Earth, e.g., c = 500 m/s, the life form will produce vowels whose formants are on average higher (grey dotted triangle in Figure 17.3b). This means that the extra-terrestrial’s [u] will sound like a [ə] and an [o] will sound like an [a]. As a side note, similar effects can be achieved when humans speak in a helium-saturated atmosphere (MacLean 1966). Variation in air density and temperature on Earth results in minimal – but measurable – variations of the quality of these vowels. Furthermore, atmospheric characteristics on Earth can vary on a daily basis, so humans need to learn to account for these changes. This is also the case with crickets. The frequency of the male cricket’s

202 Jessie S. Nixon and Fabian Tomaschek

FIGURE 17.3

a) Midsagittal view of a schematic representation of the mouth. 1–2 = lips, 3–4 = teeth, 5 = tongue tip. The black triangle represents the location of the tongue body when articulating the respective vowels. b) Effects of vocal-tract length onto the spectral properties of the vowel space. The black triangle indicates the human vowel space, and the grey triangles indicate how vowels produced by longer and shorter vocal tracts are perceived by human ears. Note that frequencies are inverted on both axes to mirror the orientation of the vowel space with that in the mouth. c) Effects of the speed of sound. The black triangle represents the human vowel space, and grey triangles represent what the same articulatory targets would sound like in atmospheres with faster or slower speeds of sound.

signal varies with the ambient temperature and female crickets adapt their perception to the temperature-dependent changes (Doherty 1985). By analogy, extra-terrestrial life forms will also require learning mechanisms sufficient for adapting to (subtle) changes in the physical properties of their home planet’s atmosphere. We now consider a scenario in which the extra-terrestrial’s body – and thus their vocal tract – differs in size from humans. Even with the same vocal apparatus as

Learning and Adaptation of Communication Systems

203

humans, differences in size can lead to a very different signal. In fact, this is what we see in the calls of prairie dogs – rodents roughly 30 cm in height when standing – which indicate the presence of various predators by means of alarm calls. Like humans, they create calls by means of their masticatory apparatus, manipulating the acoustic quality of their calls (Slobodchikoff and Placer 2006; Slobodchikoff, this volume [Chapter 9]). While the spectral characteristics of prairie dog calls strongly resemble those of vowels, the formants of these sounds consist of higher frequencies. We can illustrate the effect of body size, and therefore vocal tract size, by manipulating the parameter c in equation (1). In contrast to speed of sound, the size of the vocal tract has very strong effects on the acoustic characteristics of vowels if we assume c = 350 m/s (Figure 17.3c). Extra-terrestrial life forms whose vocal tract is significantly shorter than the average human vocal tract, e.g., 5 cm long, would produce formant frequencies significantly higher than human formant frequencies. F1 would fall into the range 1,500–2,000 Hz, and F2 would fall into the range of 4,500–6,000 Hz (grey dashed triangle in Figure 17.3c). These frequencies are still within the range of human perception – but although F1 values are within the range that humans use for speech, the F2 values are higher than the frequencies used in any vowels in human speech. Therefore, all the speech sounds would sound strange to human ears, independently of which vowel type (i.e., vocal tract shape) these life forms produced. Because of the high F1 and F2 values, the signal would be most similar to and subjectively sound most like a mix between human [a] and [e] vowels (for which F1 are located between 500 Hz and 1,000 Hz and F2 are located between 800 Hz and 2,500 Hz). Life forms which have longer vocal tracts than the average human would produce vowels with lower formant frequencies (grey dotted triangle in Figure 17.3c). Supposing a vocal tract of length 30 cm, the lower part of the vowel space of these beings would overlap with the upper part of the human vowel space. To produce an [a], humans lower the jaw and move the tongue a little bit forward. If life forms with a longer vocal tract performed the same articulatory gesture, the physical signal would be equivalent to (and would sound to humans like) a human [o]; similarly, an [o] would sound like an [u]. Thus far, we have discussed what vowel-like sounds would sound like if body size – and therefore the vocal tract size – were different from those of humans. These assumptions were based on adult speakers. However, the size of the human vocal tract changes between birth and adult life between roughly 7 cm and 17 cm. Furthermore, female speakers have, on average, shorter vocal tracts than male speakers, yielding higher formant frequencies. If extra-terrestrial life forms undergo a life cycle similar to life forms on Earth, during which they grow (which is very probable if we are dealing with biological life forms), and if there are intraspecies variations in body size, the physical characteristics of their communication signals will vary. This will be true not only for acoustic signals generated by wind pipes, but also by other kinematic mechanisms, as well as by electromagnetic mechanisms, i.e., light emitters.

204 Jessie S. Nixon and Fabian Tomaschek

FIGURE 17.4

a) Human audiogram. Frequencies on the x-axis. Perceptual threshold on y-axis in dB. The grey bar indicates perceptual optimum. b) Atmospheric opacity, indicating the absorption (y-axis) of electromagnetic frequencies (x-axis). The grey bar represents the human visual window.

Sources: Figure 17.4a according to Heffner and Heffner (2007), adapted with permission; Figure 17.4b adapted from Atmospheric electromagnetic opacity (2022)

Earth’s atmospheric characteristics and human physiology allow humans to produce a signal ranging between 50 Hz and 20,000 Hz (Figure 17.4a). Human pitch ranges from roughly 70–300 Hz, vowels are encoded primarily within 250 Hz and 2,500 Hz, and consonants such as fricatives use frequencies up to 8,000 Hz. Inspecting the human audiogram (Figure 17.4a), the obvious conclusion arises that human hearing capabilities probably co-evolved with our speaking capabilities, which depend on the physical properties of our atmosphere. The optimal frequencies for a human are located between 250 Hz and 8,000 Hz, exactly the frequency range in which human acoustic communication is located. More importantly, the increased sensitivity between 1,000 Hz and 8,000 Hz seems to compensate for the decrease in spectral amplitude in these frequency ranges (cf. Figure 17.3a–c). We therefore expect extra-terrestrial life forms to have adapted both production and perception to the physical characteristics of their home planet’s atmosphere. This conclusion is not restricted to communication in the acoustic domain. Like the acoustic signal, electromagnetic communication is highly constrained by the physical properties of a planet’s atmosphere, i.e., its electromagnetic opacity (Figure 17.4b). Earth’s atmosphere, for example, provides only two windows for electromagnetic communication (Bothais 1987; Goody and Yung 1995). The first is located between wavelengths of 100 nm and 100 µm. Human and most animal vision evolved to be optimal for a small part of the first window (represented by the grey bar).

Learning and Adaptation of Communication Systems

205

The second window – between a wavelength of roughly 1 cm and 10 m – allows us to use radio waves for communication. In the case of Earth’s atmosphere, the opacity is defined by the ratio of water vapour, oxygen and ozone. Knowing the composition of an exoplanet’s atmosphere would allow us to predict for which electromagnetic frequencies potential life on that planet has evolved. Likewise, this knowledge would allow us to predict whether we have common potential visual ranges. Implications for the Communication Systems of Extra-Terrestrial Life Forms

In a case wherein extra-terrestrial life forms evolve in different atmospheric conditions or have different sized bodies, but have similar learning mechanisms as the species of Earth, we can make certain predictions about the characteristics of their communication system. First, the system will be highly adaptive under conditions of uncertainty. Second, the system will become significantly less adaptive with learning or as uncertainty decreases. Like Earth species, as learning approaches asymptote, it will become more difficult for these entities to adapt to new information, as long as event outcomes remain predictable. Importantly, the system will downweight or unlearn cues that are not informative for predicting outcomes. This vital mechanism is what allows the life forms to ignore the noise in the system and learn to discriminate to a high level of sensitivity the cues that are important for communication of messages. This same mechanism also has the effect that cues in the signal that are not detectable with the life form’s sensory system will end up not being discriminated. The life form will have learned to ignore the variation in the cues, as in the simulation of perception of extra-terrestrial vowels by human listeners. Note that events and objects in the world provide outcomes that enable discriminative learning of the speech signal (a “dog” is not a “dock”; Nixon 2020). Linguistic labels in turn provide outcomes for discrimination learning of events and objects in the world (a dog is not a cat; Ramscar et al. 2010). Communication is predictive about the world, and the world is also predictive about communication events. Each aids and influences the learning of the other. Using the values obtained above for species that produce speech in the same way as humans, but have long, 30 cm vocal tracts, we simulate an extra-terrestrial species learning two vowel categories. Although we expect an unimaginably large degree of variation is possible in the types of communication signals across extraterrestrial species, for the sake of illustration, we consider only tiny deviations from the characteristics of Earth’s languages. Note that this analysis assumes that the extra-terrestrial species experiences time in a linear fashion as on Earth. Events earlier in time can be used to predict upcoming events, but not vice versa (Nixon 2018, 2020; Ramscar et al. 2010). If this is not the case, we might expect quite different learning mechanisms, such as for example, Hebbian (Gerstner and Kistler 2002) learning, in which learning occurs through simple co-occurrence of stimuli, rather than prediction of outcomes from cues.

206 Jessie S. Nixon and Fabian Tomaschek

1

0.00

40

0.25

78

0.50 perception

Cue density

0.75

1.00

ET [i] ET [u]

117

155

Perceptual characteristics human [i] human [u]

2700.0

FIGURE 17.5

2037.5

1375.0

F2 continuum

712.5

50.0

Figure illustrates how perceptual characteristics (lines, left y-axis) develop form different distributions of cues (bars, right y-axis) along a given continuum (x-axis). Note that the x-axis is inverted to mirror the vowel space. Light grey bars represent the distribution of F2 cues for humans, with higher frequencies being interpreted as [i] and lower frequencies as [u]. Dark grey bars represent distribution of F2 for extraterrestrial life forms.

Error-driven learning is also most expected when there is a sufficient degree of uncertainty in the organism’s world to create the necessity of unlearning non-discriminative cues or noise in the signal. Note also that this simulation addresses the specific situation when the frequency range partially overlaps the human speech frequency range. Different results would be expected if there were no overlap or if the communication signal consisted of entirely different cues (such as electromagnetic, for example, or acoustic cues not familiar to the listener). Using the Rescorla-Wagner learning equations (Rescorla and Wagner 1972), a computational formalisation of error-driven, discriminative learning, we simulated the perceptual characteristics of the [i] and [u] vowels. This simulation is not intended to represent actual learning of speech in the real world. It is a gross oversimplification; we do not assume that vowel categories are learned with vowel labels, but rather that speech co-occurs with events and the presence of various objects and so on, as discussed previously (perhaps as a multimodal version of the model presented in Nixon and Tomaschek 2020, 2021, for example). Our aim with this tiny toy example is to hone in on the specific cues of interest, in order to highlight the quite dramatic perceptual effects of a minimal change in conditions. The equations learn to discriminate the vowels ([i] and [u]) by formant cue according to the covariance structure in the data. This simulation is based on vowel

Learning and Adaptation of Communication Systems

207

categories that are produced with the same articulatory gestures as English [i] and [u]. The first simulation models human learning and is trained on human vowels. The F2 cue for these vowels clusters in two normal distributions with peaks located at 800 Hz and 2,100 Hz (Figure 17.5, light grey bars). The second simulation is trained on the values of extra-terrestrial vowels [i] and [u], which cluster around 400 Hz and 1,500 Hz (Figure 17.5, dark grey bars). In this toy example, the cues are the spectral frequencies and there are only two possible outcomes, [i] and [u], for each species. The lines represent the activation level for each vowel for the humans and for the extra-terrestrials. For each different vowel outcome, activation is high at different points along the F2 continuum depending on the training. The dashed-dotted (for [i]) and dotted lines (for [u]) show that the human perception depends on the human cue distribution. Activation decreases towards the edges of the continuum, resulting in two perceptual categories with a shift from one vowel to the other located roughly at 1,350 Hz. A similar pattern can be observed for the simulated extra-terrestrial life forms. However, the activation peaks are located in lower frequency range, due to the cues in their training. The critical result is for the human [u] (dotted line). At the points along the F2 continuum that activate two different vowels [i] and [u] for the extra-terrestrial life form, there is only activation of [u] for the human listeners. Therefore, both alien vowel categories are likely to be perceived as [u] – even if perhaps not good instances of [u]. Both of these cases illustrate perception after training in the native communication systems. If there were sufficient information available to provide error feedback that this initial perception or prediction was sometimes incorrect, then learning through prediction error could occur. The listeners might eventually learn to discriminate the new vowels. Implications for Understanding Extra-Terrestrial Communication

As noted previously, the chances of actually encountering such a communication system are rather slim. We also remain sceptical of this potential signal yielding obvious profit. In terms of human encounters with such a communication system, this would presumably require at least obtaining access to some sort of multimodal signal in which not only the language itself but also the surrounding events are available. We agree with Slobodchikoff (this volume [Chapter 9]) that if we had access to the signal alone, we would have little hope of learning much about the communication system other than characteristics of the signal itself. One characteristic of words is their linguistic context (“You shall know a word by the company it keeps”, Firth 1957: 11). Covariance structures between words are modelled with semantic networks which learn that “cat” and “dog” are more similar than “cat” and “penguin” due to their use in language (Landauer and Dumais 1997). However, this only examines structure within the system. It says nothing about the intentions of the signaller, the relation of the signal to the world – the meaning. If the communication contained both signal and sufficient information about how it

208 Jessie S. Nixon and Fabian Tomaschek

relates to the world – for example, in some sort of multimodal or even interactive format – this would provide a better chance of learning something about the world of the life form and the system they use to communicate about it.3 Knox et al. (2019) have achieved this to some extent with wild orangutans and Slobodchikoff and Placer (2006) with prairie dogs. But although these are remarkable feats, the understanding of the communication systems remains limited. Slobodchikoff predicts that while prairie dog speech is the most complex communication system yet discovered (by humans), it may turn out that we have previously simply failed to notice communication in other animals. Generally speaking, humans often seem to have a tendency to jump to conclusions about any sort of “others”, whether they be humans, animals or plants, often with an assumption of superiority. This may be convenient for allowing us to treat others differently to how we would wish to be treated ourselves. But this pattern does not seem to bode well for easily understanding the communication of others – including extra-terrestrial communication. Conclusion

Human communication systems are highly adaptable. This adaptability is supported by learning mechanisms that allow listeners to learn even very subtle cues when they are important for discrimination. Feedback from prediction error allows for rapid adaptation to new conditions and ignoring of uninformative cues. These learning mechanisms are not limited to humans but are pervasive throughout the animal kingdom. We hypothesise that any complex communication system will rely on learning in order to adapt to current conditions and communicative needs. We have discussed some practical implications for the task of interpreting communication signals, as well as preparing materials for extra-terrestrial communication. Language materials on their own probably have very little information value. We do not consider language to be made up of sequences of abstract categories, but instead argue that it emerges through ongoing learning from interaction with the world and other language users, where different kinds of cues – acoustic, visual, tactile, olfactory or other sensory information – predict different world-related events. That is, concepts arise through the experience of language in context. Experiencing the communication signal in isolation without additional information about their function has little hope of leading to any understanding of the language. If the METI project wants to create a common code, rather than broadcasting random, unrelated audio and video samples (see Harbour, this volume [Chapter 18]), it could compile a selection of distributionally structured and contextually dependent information provided in multimodal channels. However, it is doubtful that this would be sufficient. We have demonstrated that, even when the communication signal is almost identical – using an acoustic signal, and with identical articulation to humans – substantial effects on perception emerge from only minor changes in a single dimension, such body size or atmospheric conditions. As we scale up to multiple differences and their interactions, the gap is bound to become ever

Learning and Adaptation of Communication Systems

209

larger – and this is without considering any differences in the concepts being communicated. We argue that language represents and emerges from rich, multimodal experience. If this is the case, then any reduction in richness of experience presumably leads to a reduction in the quality of the linguistic representation. If we return to Shannon’s conceptualisation of communication as reproduction of a message, this highlights the challenge. We need to reproduce in ourselves the experience of someone without knowing their physical environment, their objects, technology, their culture, history, beliefs, perhaps even their sensory perception. This may be one of the obstacles that has so far hindered us from gaining a better understanding of animal communication. Perhaps when we have learned to converse with the various terrestrial life forms all around us, then we will be ready for extra-terrestrial communication. Notes 1 There also exist consciously designed technologies for communication, such as braille and other writing systems and code languages – see Harbour, this volume (Chapter 18), for a discussion of writing systems. 2 Though the signal can be transcribed with several alternative letter combinations, the most common is “SOS”, which has unofficially come to be associated with the phrase “save our souls”. 3 We leave aside the technical issue of how the technology is made usable to unfamiliar users.

References Altmann, G. 1980. “Prolegomena to Menzerath’s law.” Glottometrika 2: 1–10. Arnold, D., F. Tomaschek, K. Sering, M. Ramscar, and R.H. Baayen. 2017. “Words from spontaneous conversational speech can be recognized with human-like accuracy by an error-driven learning algorithm that discriminates between meanings straight from smart acoustic features, bypassing the phoneme as recognition unit.” PLoS One 12 (4). doi:10.1371/journal.pone.0174623 Atmospheric electromagnetic opacity. 2022. Wikimedia Commons. https://commons.wikimedia.org/wiki/File:Atmospheric electromagnetic opacity.svg (Accessed 25 October, 2022). Baayen, R.H., P. Milin, D.F. Ðurđević, P. Hendrix, and M. Marelli. 2011. “An amorphous model for morphological processing in visual comprehension based on naive discriminative learning.” Psychological Review 118 (3): 438–481. doi:10.1037/a0023851 Baayen R.H., Shaoul C., Willits J., Ramscar M. 2016. Comprehension without segmentation: A proof of concept with naive discriminative learning. Language, Cognition and Neuroscience 31 (1): p106ff. https://doi.org/10.1080/23273798.2015.1065336 Bothais, L. 1987. Radio Wave Propagation. New York: McGraw-Hill Book Company. Clayards, M., M.K. Tanenhaus, R.N. Aslin, and R.A. Jacobs. 2008. “Perception of speech reflects optimal use of probabilistic speech cues.” Cognition 108 (3): 804–809. doi:10.1016/j.cognition.2008.04.004 Doherty, John A. 1985. “Temperature coupling and ‘Trade-Off’ phenomena in the acoustic communicative system of the Cricket, Gryllus Bimaculatus De Geer (Gryllidae).” Journal of Experimental Biology 114 (1): 17–35. https://doi.org/10.1242/jeb.114.1.17

210 Jessie S. Nixon and Fabian Tomaschek

Einhorn, H.J., and R.M. Hogarth. 1986. “Judging probable cause.” Psychological Bulletin 99 (1). Fant, G. 1960. Acoustic Theory of Speech Production. De Gruyter. Firth, J. 1957. “A synopsis of linguistic theory, 1930–55.” In Studies in Linguistic Analysis. Special Volume of the Philological Society. Oxford: Blackwell. 1–31. Fosler-Lussier, Eric, and Nelson Morgan. 1999. “Effects of speaking rate and word frequency on pronunciations in convertional speech.” Speech Communication 29 (2–4): 137–158. http://www.sciencedirect.com/science/article/pii/S0167639399000357. Fuchs, S., R. Winkler, and P. Perrier. 2008. “Do speakers’ vocal tract geometries shape their articulatory vowel space?” Proceedings of the 8th International Seminar on Speech Production, ISSP’08, Strasbourg: France: 333–336. Gay, Thomas. 1978. “Effect of speaking rate on vowel formant movements.” The Journal of the Acoustical Society of America 63 (1): 223–230. doi:10.1121/1.381717 Gerstner, W., and Kistler, W. M. (2002). “Mathematical formulations of Hebbian learning.” Biological cybernetics, 87 (5), 404–415. Gluck, M.A., and G.H. Bower. 1988. “From conditioning to category learning: An adaptive network model.” Journal of Experimental Psychology: General 117 (3): 227. doi:10.1037//0096-3445.117.3.227 Heffner, H.E., and R.S. Heffner. 2007. “Hearing ranges of laboratory animals.” Journal of the American Association for Laboratory Animal Science 46 (1): 20–22. Hirata, Yukari. 2004. “Effects of speaking rate on the vowel length distinction in Japanese.” Journal of Phonetics 32 (4): 565–589. https://doi.org/10.1016/j.wocn.2004.02.004 Hull, C.L. 1943. Principles of Behavior. New York: Appleton-Century-Crofts. Jiao, D., V. Watson, S.G.-J. Wong, K. Gnevsheva, and J.S. Nixon. 2019. “Age estimation in foreign-accented speech by non-native speakers of English.” Speech Communication 106: 118–126. https://doi.org/10.1016/j.specom.2018.12.005 Kamin, Leon J. 1968. “Attention-like processes in classical conditioning.” In Miami Symposium on the Prediction of Behavior, edited by M.R. Jones, 9–31. Miami: Miami University Press. Knox, A., J. Markx, E. How, A. Azis, C. Hobaiter, F.J. van Veen, and H. Morrogh-Bernard. 2019. “Gesture use in communication between mothers and offspring in wild orangutans (Pongo pygmaeus wurmbii) from the Sabangau Peat-Swamp Forest, Borneo.” International Journal of Primatology: 1–24. https://doi.org/10.1007/s10764-019-00095-w Landauer, T.K., and S.T. Dumais. 1997. “A solution to Plato’s problem: The latent semantic analysis theory of acquisition, induction, and representation of knowledge.” Psychological Review 104 (2): 211. https://doi.org/10.1037/0033-295X.104.2.211 Leemann, Adrian, Volker Dellwo, Marie-José Kolly, and Stephan Schmid. 2012. “Rhythmic variability in Swiss German dialects.” 6th International Conference on Speech Prosody 2012, Shanghai, May 22–May 25. 607–610. ISCA. MacLean, D.J. 1966. “Analysis of speech in a Helium-Oxygen mixture under pressure.” The Journal of the Acoustical Society of America 40 (3): 625–627. https://doi. org/10.1121/1.1910128 McMurray, B., R.N. Aslin, and J.C. Toscano. 2009. “Statistical learning of phonetic categories: insights from a computational approach.” Developmental Science 12 (3): 369–378. doi:10.1111/j.1467-7687.2009.00822.x Magen, H.S. 1997. “The extent of vowel-to-vowel coarticulation in English.” Journal of Phonetics 25: 187–205. https://doi.org/10.1006/jpho.1996.0041 Maye, J., J.F. Werker, and L. Gerken. 2002. “Infant sensitivity to distributional information can affect phonetic discrimination.” Cognition 82 (3). doi:10.1016/s0010-0277(01)00157-3

Learning and Adaptation of Communication Systems

211

Meyer, J. 2004. “Bioacoustics of human whistled languages: An alternative approach to the cognitive processes of language.” Anais da Academia Brasileira de Ciências 76 (2): 406–412. doi:10.1590/S0001-37652004000200033NASA. n.d. www.grc.nasa.gov/ WWW/K-12/rocket/machu.html. Michaud, M. 2011. Seeking contact: The relevance of human history. In Communication with Extra-Terrestrial Intelligence, edited by D.A. Vakoch. Albany: State University of New York Press. NASA. n.d. https://www.grc.nasa.gov/WWW/K-12/rocket/machu.html. Nixon, J.S. 2018. “Effective acoustic cue learning is not just statistical, it is discriminative.” Interspeech 2018–19th Annual Conference of the International Speech Communication Association, 1447–1451. Hyderabad, India, September. Nixon, J.S. 2020. “Of mice and men: Speech acquisition as discriminative error-driven learning, not just statistical tracking.” Cognition 197: 104081. https://doi.org/10.1016/j. cognition.2019.104081 Nixon, J.S., and C.T. Best. 2018. “Acoustic cue variability affects eye movement behaviour during non-native speech perception.” Proceedings of the 9 International Conference on Speech Prosody, 493–497. Poznan, Poland, June. Nixon, J.S., N. Boll-Avetisyan, T.O. Lentz, S. van Ommen, B. Keij, Ç. Çöltekin, L. Liu, and J. van Rij. 2018. “Short-term exposure enhances perception of both between- and within-category acoustic information.” Proceedings of the 9th International Conference on Speech Prosody, 114–118. Nixon, J.S., S. Poelstra, and J. van Rij. 2022. “Does error-driven learning occur in the absence of cues? Examination of the effects of updating connection weights to absent cues.” In Proceedings of the 44th Annual Meeting of the Cognitive Science Society, edited by J. Culbertson, A. Perfors, H. Rabagliati, and V. Ramenzoni, 2590–2597. Merced: Open Access Publications from the University of California (UC Merced). Nixon, J.S., and F. Tomaschek. 2020. “Learning from the acoustic signal: Error-driven learning of low-level acoustics discriminates vowel and consonant pairs.” In Proceedings of the 42nd Annual Meeting of the Cognitive Science Society, edited by S. Denison., M. Mack, Y. Xu, and B.C. Armstrong, 585–591. Merced: Open Access Publications from the University of California (UC Merced). Nixon, J.S., and F. Tomaschek. 2021. “Prediction and error in early infant speech learning: A speech acquisition model.” Cognition 212: 104697. https://doi.org/10.1016/j. cognition.2021.104697 Nixon, J.S., and F. Tomaschek. 2023a. “Does speech comprehension require phonemes?” Forthcoming in The Handbook of Usage-Based Linguistics, edited by M. Diaz-Campos & S. Balasch. Oxford: Wiley Blackwell Publishing. Nixon, J.S., and F. Tomaschek. 2023b. “Introduction to the special issue emergence of speech and language from prediction error: error-driven language models.” Manuscript submitted to Language, Cognition and Neuroscience. Nixon, J.S., J. van Rij, P. Mok, R.H. Baayen, and Y. Chen. 2016. “The temporal dynamics of perceptual uncertainty: Eye movement evidence from Cantonese segment and tone perception.” Journal of Memory and Language 90: 103–125. https://doi.org/10.1016/j.jml.2016.03.005 Nyst, V. 2015. “The sign language situation in Mali.” Sign Language Studies, 15 (2): 126– 150. www.jstor.org/stable/26190976 Öhman, S.E.G. 1966. “Coarticulation in VCV utterances: Spectrographic measurements.” Journal of the Acoustical Society of America 39 (151): 151–168. https://doi. org/10.1121/1.1909864 Ramscar, M. 2019. Source Codes in Human Communication. https://psyarxiv.com/e3hps.

212 Jessie S. Nixon and Fabian Tomaschek

Ramscar, M., M. Dye, and J. Klein. 2013. “Children value informativity over logic in word learning.” Psychological Science 24: 1017–1023. Ramscar, M., D. Yarlett, M. Dye, K. Denny, and K. Thorpe. 2010. “The effects of featurelabel-order and their implications for symbolic learning.” Cognitive Science 34 (6): 909– 957. doi:10.1111/j.1551-6709.2009.01092.x Rattray, R.S. 1923. “The drum language of West Africa.” African Affairs 22(87): 226–236. https://doi.org/10.1093/oxfordjournals.afraf.a100065 Reber, A.S. 1989. “Implicit learning and tacit knowledge.” Journal of Experimental Psychology: General 118 (3): 219–235. Rescorla, R., and A. Wagner. 1972. “A theory of pavlovian conditioning: Variations in the effectiveness of reinforcement and nonreinforcement.” In Classical Conditioning II: Current Research and Theory, edited by A.H. Black and W.F. Prokasy, 64–69. New York: Appleton Century Crofts. Shannon, C.E. 1948. “A mathematical theory of communication.” Bell System Technical Journal 27 (3): 379–423. Shanks, D.R. 1986. “Selective attribution and the judgment of causality.” Learning and Motivation, 17 (4): 311–334. Slobodchikoff, C.N., and J. Placer. 2006. “Acoustic structures in the alarm calls of Gunnison’s prairie dogs.” The Journal of the Acoustical Society of America 119 (5): 3153– 3160. https://doi.org/10.1121/1.2185489 Soto, F.A., and E.A. Wasserman. 2010. “Error-driven learning in visual categorization and object recognition: A common-elements model.” Psychological Review 117 (2): 349. doi:10.1037/a0018695 Spence, K.W. 1956. Behavior Theory and Conditioning. New Haven, CT: Yale University Press. Tomaschek, F., D. Arnold, Franziska Bröker, and R.H.R. Baayen. 2018a. “Lexical frequency co-determines the speed-curvature relation in articulation.” Journal of Phonetics 68: 103–116. https://doi.org/10.1016/j.wocn.2018.02.003 Tomaschek, F., and A. Leemann. 2018. “The size of the tongue movement area affects the temporal coordination of consonants and vowels – A proof of concept on investigating speech rhythm.” JASA-EL 144 (5). https://doi.org/10.1121/1.5070139 Tomaschek, F., I. Plag, M. Ernestus, and R.H. Baayen. 2021. “Phonetic effects of morphology and context: Modeling the duration of word-final S in English with naïve discriminative learning.” Journal of Linguistics 57 (1), 1–39. https://doi.org/10.1017/ S0022226719000203 Tomaschek, F., and M. Ramscar. (2022). “Understanding the phonetic characteristics of speech under uncertainty – implications of the representation of linguistic knowledge in learning and processing.” Frontiers in Psychology: 1833. doi:10.3389/fpsyg.2022.754395 Tomaschek, F., B.V. Tucker, R.H. Baayen, and M. Fasiolo. 2018b. “Practice makes perfect: The consequences of lexical proficiency for articulation.” Linguistic Vanguard 4 (s2). doi:10.1515/lingvan-2017-0018 Tomaschek, F., B.V. Tucker, M. Wieling, and R.H. Baayen. 2014. “Vowel articulation affected by word frequency.” Proceedings of the 10th ISSP, 425–428. Cologne. Tomaschek, F., M. Wieling, D. Arnold, and R.H. Baayen. 2013. “Word frequency, vowel length and vowel quality in speech production: An EMA study of the importance of experience.” Proceedings of the Interspeech. Lyon. Van Hamme, L.J., and E.A. Wasserman. 1994. “Cue competition in causality judgments: The role of nonpresentation of compound stimulus elements.” Learning and Motivation 25 (2): 127–151.

Learning and Adaptation of Communication Systems

213

Waelti, P., Dickinson, A., and Schultz, W. 2001. “Dopamine responses comply with basic assumptions of formal learning theory.” Nature 412: 43–48. https://doi. org/10.1038/35083500 Weber, F., M.W. Woolridge, and J.D. Baum. 1986. “An ultrasonographic study of the organization of sucking and swallowing by newborn infants.” Developmental Medicine & Child Neurology 28 (1): 19–24. Weirich, M., and S. Fuchs. 2006. “Palatal morphology can influence speaker-specific realizations of phonemic contrasts.” Journal of Speech, Language, and Hearing Research 56: 1894–1908. doi:10.1044/1092-4388(2013/12-0217) Widrow, G., and M.E. Hoff. 1960. Adaptive Switching Circuits. IRE Western Electron. Show Convention, Convention Record Part 4, 96–104. Winkler, R., S. Fuchs, and P. Perrier. 2006. “The relation between differences in vocal tract geometry and articulatory control strategies in the production of French vowels: Evidence from MRI and modeling,” 8th International Seminar on Speech Production, ISSP’08, Strasbourg: France, 509–516. Wittgenstein, L. (1953). Philosophical Investigations [Philosophische Untersuchungen]. Yashiro, K., and K. Takada. 2005. “Model-based analysis of jaw-movement kinematics using jerk-optimal criterion: Simulation of human chewing cycles.” Journal of Electromyography and Kinesiology 15 (5): 516–526. doi:10.1016/j.jelekin.2004.11.005 Yoshida, K.A., F. Pons, J. Maye, and J.F. Werker. 2010. “Distributional phonetic learning at 10 months of age.” Infancy 15 (4). doi:10.1111/j.1532-7078.2009.00024.x

18 WRITING SYSTEMS AND METI Off-the-Shelf Encoding of Human Physiology, Language, Cognition, and Culture Daniel Harbour

Introduction

Imagine that we receive an alien message tomorrow and that, thanks to its mostly mathematico-scientific form and content, it turns out to be readily decipherable. Three questions would arise from the next day’s headline: Intelligent Life Found Found where? What life? What intelligence? Our attempts to message extraterrestrial intelligence (METI) do not address these questions equally for potential alien audiences. Origin and biology were represented in the earliest deliberate, content-dense transmissions, the Pioneer Plaques and the Arecibo transmission of 1972‒1974. For reasons of space, these said little about the nature of our intelligence. But these size restrictions no longer constrain us and, though human biochemistry is easier to portray than human thought, we should still tackle the task of representing it systematically. Silence is neither desirable nor – if we reflect on previous messages – possible. If ET-planned METI goes through anything like the thought process described in this chapter, the common SETI (Search for Extraterrestrial Intelligence) assumption that interstellar messages will read like resumés of scientocracies is likely false. Just as the form of interstellar messages must be finely balanced – between redundancy, which makes them noticeably artificial, and decodability – so, too, must their content be balanced between the universal and the particular. The mathematico-scientific common ground between any two METI-capable civilizations is not a communicative end in itself, but a tool by which to encode information on the kind of res cogitans behind any message. The human accent in human METI should be deliberate and carefully planned. My main purpose is to argue that writing systems are excellent, multifaceted DOI: 10.4324/9781003352174-18

Writing Systems and METI

215

indicators of human language, cognition, and culture. Moreover, in the case of Visible Speech in particular (Bell 1867), they offer a natural bridge from the scientific content of previous METI to future, humanized messages that take full advantage of advances in understanding what is a reasonable size of message to transmit. I begin with overviews of potential message size in the next section (What Can We Send?) and past messages the section after (What Have We Sent?). Though past messages are substantially smaller than what we could send, these overviews nonetheless converge on a philosophy in which mathematics and science constitute stepping stones to broader communication. This spotlights the risk of being insufficiently or excessively human-centered as pitfalls to avoid in larger-message design (as discussed in section Potential Pitfalls). Instead, a message design that transitions gradually from science to humanities, with writing systems as part of the bridge, offers several design advantages (as described in the section Human Writing Systems). What Can We Send?

The first question facing METI is what size of message to send into space. A variety of research converges on the conclusion that we can send substantial quantities of information in the reasonable hope that it will be salient and, if found, complete. Taking into account galactic background radiation and the interstellar medium of dust, neutral gas, and plasma, Harp et al. (2011) argue for a “twice-sent signal” format, like a musical canon. Transmitters send messages in two staggered but overlapping installments, so that any dispersion the two versions undergo is identical. The result is distortion-resistant, information-rich, and salient against galactic background radiation. Harp et al.’s proposal contrasts with many previous SETI efforts, which focused on sine waves. These are salient because artificial but are also informationally “monochromatic.” How much color can our messages bear? Reviewing the technical limits on interstellar messages, Shostak (2011) argues for sending an encyclopedia-sized body of information as a feasible balance of content and form. The round-trip message time just within the Milky Way (where Dick 1996 estimates up to 100,000 civilizations may exist) is measured in (tens of) hundreds of light years. This amounts to one-way communication from the point of view of individual senders and their cultures and so favors sending as much as possible at once. However, the fact that no one is likely to be listening from the start of our transmission places a limit on how much of a long transmission we can expect to see received. This favors frequently repeated shorter messages instead. As a happy medium, Shostak suggests that, using 1 micron of infrared with one bit per pulse for a total duration of ~0.1 seconds, we could ping a million star systems once a day with the equivalent of the Encyclopedia Britannica to each. Benford et al. (2011) reach a similar conclusion about broadcast length and repetition by considering cost-optimization of interstellar beacons. The design of galactic-scale beacons they arrive at consists of narrow “searchlight” beams visible

216 Daniel Harbour

for short periods in an alien observer’s sky. Consequently, the transmission strategy from Earth should be a rapid scan of the galactic plane. The result would be infrequent pulses of a few seconds from the receiver’s point of view, recurring over periods of a month to a year. What Have We Sent?

An encyclopedia is far larger than any message, physical or digital, yet sent from Earth. Nonetheless, it is worth reviewing previous messages, as their careful construction provides a useful guide for larger formats. Pioneers 10 and 11, launched in 1972 and 1973, respectively, bear plaques depicting a man and woman and our solar system, with the position of Earth highlighted, alongside information on the hydrogen atom and pulsars intended to serve as measures of distance and time. The Golden Discs aboard Voyagers 1 and 2 carry sounds and images of humans, animals, and the planet. The recordings include a heartbeat, breath, birth, and speech. The images progress from mathematics, physics, and astronomy to biochemistry and human biology, and then to human activities like breastfeeding, eating, learning, building, driving, and flying. Both therefore embody the high-level ambition of using science as a basis for communicating about humanity. The potential for such concrete communication is very limited however. Even traveling at 15‒17 km/s (Cook and Brown 2010), the Voyager probes are lumbering along compared to the speed-of-light travel of radio and laser messages. They are also far less repeatable. The difference is that of a messenger on horseback versus email spam on a galactic scale. The earliest deliberate high-powered transmission of information into deep space was the Arecibo radio message (The Staff at the National Astronomy and Ionosphere Center 1975). Formatted as a semiprime number of binary digits, it unfurls into a 23 × 73 grid representing numbers, then chemical elements, then compounds critical to life, as well as sketches of DNA, a person, the solar system (highlighting Earth), and the Arecibo transmitter. The Arecibo message was incorporated into the Cosmic Calls of 1999 and 2003 alongside further messages. One, the Bilingual Image Glossary (BIG), comprising twelve binary drawings subtitled in Russian and English, was mostly about humans, but also covered the sun, Earth, and our galaxy. Another was the meticulously planned Dumas-Dutil message (Dumas 2003). Based on Freudenthal’s (1960) METI-purposed language, Lincos, its twenty-three pages represent a substantially expanded implementation of the ideas and information in past messages. Like Arecibo, it contains a range of mathematical, physical, chemical, and astronomical information but adds geochemical and geological data, too, and it encodes a modified version of the Pioneer Plaque humans (with the woman no longer passive), together with information about our physical characteristics and biochemical composition. It concludes, like Arecibo, with information about the transmitter, but

Writing Systems and METI

217

then adds a page of questions for alien recipients. (The ability to ask questions is a design feature of Lincos.) In sum, the composers of past messages have viewed the status of mathematics and science as a universal language in a very particular light. It provides a means of talking about humanity in scientific terms. But their messages go further. The Cosmic Calls follow Freudenthal in asking questions, which reveals a most human fact about us (Bromberger 1992): that we recognize, and push against, the limits to our knowledge. BIG goes so far as to hint at the structure of human language. Assuming that the subtitles are identified as writing systems, they are clearly distinct (each has letters the other lacks) but related (shared letters), and the word lengths vary unpredictably (meaning, in familiar terms, that they correspond to different languages, rather than one language written in different ways), and even in such a short sample, it may be possible to conclude that the writing system is not based on inherent meanings of the characters and is therefore more likely alphabetic than hieroglyphic. Potential Pitfalls

Every message has a subtext. Incomplete surveys, for example, provide an inadvertent measure of the respondents’ conscientiousness (Hedengren 2013). If alien curiosity is at all like ours, any message we send will be pored over from multiple perspectives and – even before its text is exhausted – its subtext will be mined. We should, therefore, reflect on the potential subtext carefully. Three past proposals for long-message METI are problematic in this regard, as discussed in the following subsections: messages that are excessively science-based, artificial languages that are inadvertently (non)human, and messages consisting of off-the-shelf encyclopedias. All risk sending the wrong message, figuratively speaking. Science Primers

It is axiomatic that METI-ready civilizations must be reasonably advanced in mathematics and science. It follows, as per the next subsection, that METI should use mathematics and science to devise self-decoding messages (thereby avoiding the difficulties that Slobodchikoff, this volume [Chapter 9], discusses for understanding prairie dog communication). It does not follow, though, that self-decoding messages (ours or others) should be primarily mathematico-scientific (Finney and Bentley 2014). Earth has enjoyed only a few centuries of high-intensity scientific research and only a few decades of SETI. This makes us juveniles in scientific and SETI terms. It is conceivable that there are “adult” civilizations, tens of millennia more advanced than ours, engaged in METI. If they send us a decodable primer on their mathematics and science, it might be useful. Maybe cold fusion is a far harder problem to crack than we imagine. Maybe adult civilizations have seen other juvenile ones

218 Daniel Harbour

self-destruct from battles over, or overuse of, nonrenewable energy sources. If so, cold fusion construction manuals would make worthwhile messages. However, purely mathematico-scientific communication is just as likely to be frustrating. Mathematics is as much about why as what. Knowing the answer to the P versus NP problem would be useful, but not as useful as knowing why the answer holds – just as a diagram showing that cold fusion reactors are possible is far less useful than knowing how to build one. Moreover, a purely scientific message would leave tantalizing questions of subtext. Is the information being sent in altruism, in arrogance, under duress, in evangelism, or in the shadow of extinction? These questions are more acute for our own METI. A primer on (scientifically juvenile) human mathematics and science is unlikely to educate its recipients. By chance, it might be picked up by a civilization that had also only just attained the capacity for interstellar communication, or by one less advanced in some areas than we are (maybe there’s a planet of Kantians who have never explored non-Euclidean geometry). But, if neither of those scenarios holds, our recipients are likely to be more knowledgeable than us. Our message would be the epistemological equivalent of a gift of children’s clothes several sizes too small. Its subtext might be taken as arrogance or naïveté. Artificial Languages With Human Accents

A second fact inherent to science primers makes them an unattractive option. Consider Dumas’s sketch for an expansion of the Dumas-Dutil message, where he formulates human procreation in mathematical notation (Dumas 2011: 410): M + W ⇒ M + W + M or M + W + W Two facts are noteworthy here. (Artificial languages are discussed further in Granger et al., this volume [Chapter 8], and Sperlich, this volume [Chapter 14].) On the semantic side, the biological statement uses “+” in a nonmathematical sense. This introduces ambiguity into the artificial language. Ironically, though, three sentences after the preceding equation, Dumas criticizes human languages for being “too imprecise and ambiguous.” Given his oversight, the natural question is: Is all ambiguity problematic? The current ambiguity is an instance of what Wittgenstein (1953: 32) calls, aptly enough, “family resemblance.” This is a recurrent trait of human language and, apparently, tolerable even in artificial languages. On the syntactic side, note that “+” occurs between what it connects, M + W, rather than before, as in the Polish notation, + M W, or after. Although common in computer science, the Polish order is unattested in natural languages (including Polish). Instead, both connective-medial and connective-final orders are found (with the latter restricted to head-final languages and rare even there; Haspelmath 2000; Stassen 2003; Zwart 2005). It would be rash for aliens to assume that this connective-medial syntax was chosen because it reflects the bulk of human

Writing Systems and METI

219

languages, but, at the same time, the alien linguist may well be unable to help but wonder. Rather than merely accepting such ambiguities and syntactic choices, we might use them productively. We could adopt the same sign for both mathematical addition and biological union (or for logical conjunction) and distinguish them by their syntax: medial for one, final for the other. This deliberate, contextually resolved symbolic ambiguity would give a window, albeit small, into our conceptual networks and the division of labor between human syntax and semantics. The idea that we can excise our humanness even from our artificial languages blinds us to an opportunity to inform and engage. Off-the-Shelf Encyclopedias

Heidmann (1997: 202) suggests that it would be enough send an encyclopedia, like the Britannica or Universalis, in order to tell aliens about ourselves. He has sound reasons: encyclopedias are off-the-shelf, expansive, transmissible, potentially decodable text-image composites. However, if any one transmission is an intergalactic message in a bottle, then his suggestion amounts to building a bespoke interstellar bottle factory and then filling the bottles with whatever newspapers we have on hand. (See also Wells-Jensen, this volume [Chapter 13], on the possibility of sightless aliens.) Encyclopedias vary so widely between countries that no current one could command universal consent – though maybe we could lessen this problem sufficiently by sending abridged encyclopedias in, say, Arabic, Chinese, English, and Hindi, possibly with smaller documents in less widely spoken languages from elsewhere, such as Cherokee and Fijian.1 Whatever our choice, encyclopedias would create for aliens many of the frustrations that we face with regard to antiquity. We know Sappho almost only through fragmentary quotations by other authors, but this is not what we should aim for in transmitting Shōnagon, Rumi, or Shakespeare.2 We know how to interpret Anatolian logograms but only partly how to pronounce them (Weeden 2014). This is, again, something we should avoid for our message (and its scraps of Shakespeare). The issue is one of subtext. We should consider how it reflects on us if we send aliens something curated for humans – in particular, for humans of a particular cultural background. What kind of conversation have we begun if our opening gambit is us talking to ourselves (cf, Hobaiter et al., this volume [Chapter 3])? Human Writing Systems

Language is key to our intelligence and a defining property of humanity. A true portrait of human cognition must, therefore, move beyond the cosmic pidgin of artificial languages and into the realm of natural language. Consistent with past METI, mathematics is an obvious stepping stone: proofs can be presented both

220 Daniel Harbour

as chains of symbolic statements and as arguments in natural language, yielding a Rosetta Stone from the universal language of mathematics to one or more human languages; and mathematics lends itself to nonmathematical notions, such as aesthetics, which are highly human.3 But even more basic than content is the question of encoding. Obviously, we can continue solely with binary. However, it bears no relation to how human languages are written, let alone spoken or signed, and so veils our report on humanness with a nonhuman overlay. Instead, I suggest that we take seriously the use of human writing systems themselves, alongside binary encoding. Writing systems are the product of our cognitive structure, cultural evolution, and environmental interaction. Lying at the confluence of such major forces makes this “artifactual communication” (as Berea, this volume [Chapter 7], calls it) interesting in and of itself. A simple illustration is Morin’s (2018) finding that symmetry around the vertical axis is much more frequent than symmetry around the horizontal axis (As are more common that Bs) across characters of the world’s alphabets and syllabaries. Morin ties this to adaptation to our environment: we encounter far more A-type symmetry (in people, animals, trees) than B-type (reflections in lakes). In the first subsection (The Evolution of Writing), I briefly sketch the development of alphabets from pictograms emphasizing how crucially it hinges on facts connected to human cognition. In Bell’s Visible Speech, the connections go further: the structure of the characters reflects the physiology of speech production (see subsection Visible Speech)). This means that Visible Speech could offer a natural transition from the scientific content of past METI to a more culturebased encyclopedia in human language (see subsection Visible Speech and Past METI). The Evolution of Writing

Writing has been invented ex nihilo only four (or five) times in human history, approximately contemporaneously in Sumer and Egypt, in Mesoamerica (and probably South America) some three millennia later, and in China about halfway between. Despite differences of time and of the grammatical character of the inventors’ languages, the four (partially) deciphered systems are very similar at the abstract level. All are centered on the pictogram; that is, a character that pictures its meaning (e.g., a picture of tree to mean “tree”). Pictures taken literally extend readily to ideograms, pictures taken figuratively (e.g., a picture of a tree to mean “wood”). To disambiguate polysemy (such as “tree/wood”), signs acquired a third function: they were divorced from their meanings and used purely for their sound. These could be standalone uses (permitting a sign read as wood to be used for, say, the difficult-todepict modal “would”). Or they could be combined to create complex signs, where one part gives the meaning and another the sound (e.g., a compound sheep-wood

Writing Systems and METI

221

sign might mean “wool,” based on the meaning “sheep” and the sound wood). (See Harbour 2019a for detailed Sumerian/Chinese comparison.) The fact that four peoples with unrelated cultures and languages devised a common typology of logograms strongly suggests that the solution reflects a shared intellect as much as a shared problem (one that might be shared with ETI; Roberts, et al., this volume [Chapter 15]). The parallels run deeper, however, and continue into descendant systems. The extent to which the four writing systems used phonetic signs independently (“wood” for “would”) versus dependently (in compounds like sheep-wood for “wool”) correlates with their grammatical structure. Sumerian, Egyptian, and Mayan – languages in which nouns and verbs take rich inflection (e.g., agreement, aspect, case) – required speakers to write many syllables that lacked (obvious) meanings and therefore did not lend themselves to meaning-based depiction. Their writing used many phonetic signs independently. Baxter and Sagart (2014) reconstruct Old Chinese as having a rich array of derivational morphemes, but ones that preserve syllable boundaries and hence meaningfulness. This correlates with a preference for compounding use of phonetic signs. Later developments of these systems, especially by unrelated languages, indicate that grammar drives their differences (Harbour 2019b). The Chinese system, for instance, was adapted to write Japanese, Korean, Vietnamese, and Zhuang. Japanese and Korean require substantial (syllable-altering) inflection, particularly for verbs. Both developed standalone phonetic signs (and Japanese went so far as to devise separate syllabaries, reserving Chinese characters for meaning-based use). Strikingly, then, these languages reinvented the Sumerian-Egyptian-Mayan system. Vietnamese and Zhuang, by contrast, have syllable-preserving morphology, like Modern Chinese, and their speakers adapted the Chinese system by redesigning the phonetic components of compound characters. This should interest aliens in multiple ways. First, the existence of the Chinese writing system in the present day informs both about the current state of the language and the linguistic structure of its antecedents, as well as the unbroken chain of transmission between them. The reconstruction of the language, culture, and thought processes of a people from its writing system has been termed “script archaeology” by Jaritz (1967). Analogous to explaining the silent k in knife by appealing to earlier stages of the language, these patterns are pervasive and fascinating in logography. It is culturally and cognitively revealing that Chinese “write” (書) incorporates “brush” (聿) and that “rest” depicts a person by a tree (休) and “companion” a person by a fire (伙). An encyclopedia in part in Chinese would provide recipients with the seemingly impossible: an archaeological site transmitted across space (cf, Nixon and Tomaschek’s call, this volume [Chapter 17], for transmissible environments based on discriminative learning). Second, as writing systems become more phonetic, the phonetic unit they generally invest in is the syllable, most frequently a vowel (V) or a consonant plus vowel (CV). Even with modest consonant and vowel inventories, syllabaries tend

222 Daniel Harbour

to avoid representing the likes of CVC, CCV, and CCVC directly. Instead, these are represented via special conventions, such as Akkadian ba-ag for bag (extra vowel pre-supports the last consonant), Mayan yo-po for yop (extra vowel post-supports the last consonant), or Linear B to-ro for tro (extra vowel post-supports the first consonant). The frequency of syllabaries (whether from evolution, as in Japanese, or from reinvention, as in Cherokee) may reflect neurological hardwiring. Luo and Poeppel (2007: 1001) propose that the syllable is “a computational primitive for the [neurological] representation of spoken language,” in light of their finding that syllable timing coincides with multiple endogenous brain rhythms. Giraud and Poeppel (2012: 511) claim that it is “foundational in speech and language processing, ‘packaging’ incoming information into units of the appropriate temporal granularity.” Syllabaries may therefore reveal deep facts about our neurology, of potential interest to alien readers if appropriately contextualized. Third, against this background, alphabets require special comment. In alphabets, primitive symbols represent a vowel or a consonant, but not both.4 Despite the ubiquity of alphabets and their current global dominance, the conditions that led to alphabetic writing were extremely peculiar. Vowels are crucial to meaning in all of the world’s languages. However, the Afro-Asiatic family (including Ancient Egyptian, Phoenician, and Hebrew) confines that dependence mostly to grammatical morphemes. Vowel omission, then, amounts to morpheme underrepresentation, disguising the difference between noun and verb (prat “a detail,” peret “to detail”), or active and passive (peret “to detail,” porat “to be detailed”) (Harbour 2021). This information is often recoverable from context and, when not, simply places Afro-Asiatic languages on a par with the many languages that lack such morphology to begin with (cf., English noun/verb detail). Given the restricted role of vowels, when Ancient Egyptian developed standalone phonetic writing, it arrived at a vowelless syllabary – that is, a set of consonants. This was adapted to become the Proto-Semitic abjad (consonantal alphabet), which was then carried north to write various Semitic languages. Though some of these came to sporadically indicate vowels (Hebrew did, Phoenician did not), it was only at the point of handover to non-Afro-Asiatic languages, where grammar did not cue vowels, that vowel writing markedly increased (as occurred in Brahmi, Greek, Sogdian, Yiddish, etc.; Harbour 2021). Thus, full alphabets arose from consonantal alphabets, which arose from an extremely odd quirk of grammatical design in one of the early written languages. Afro-Asiatic languages that adopted writing systems with vowels, such as Akkadian and Maltese, retained them. There was, then, a narrow window of opportunity, when writing was young, for a consonantal alphabet to arise just in one corner of the globe. The aversion that other grammatical types have to vowelless writing is manifest in the emergence of syllabaries everywhere else and by the almost immediate development of obligatory vowel marking once a consonantal alphabet is adopted beyond Afro-Asiatic.

Writing Systems and METI

223

Writing for aliens, we should therefore be eager to display our alphabet. If the aliens are anything like us, they may not have one. Visible Speech

No matter how we choose to represent linguistic content, we still have to avoid the Anatolian problem. That is, we need to indicate pronunciation, on which so much of linguistic and cultural interest depends and which is so fundamentally tied to our physiology. A ready-made solution for this is the Visible Speech alphabet invented by Alexander Melville Bell and described in Bell (1867), when it had already been in use for several years (an “engineered writing system” comparable to the “engineered languages” of Sperlich, this volume [Chapter 14]). A principle application was teaching speech to the deaf. However, it also had uptake among linguists because it enabled “the writing of all languages in one alphabet,” as per the book’s subtitle. It was able to register subtle differences in accents and could even represent infant and animal calls. An Athenæum editorial (cited in full in Bell’s book) describes how Bell would transcribe rare accents or languages known to audience members, which his sons (one of them, Alexander Graham Bell), who had been waiting outside, would then read back, to the audience’s amazement. The genius of Visible Speech is that its letters come as close as practicable to pictograms of sound. They are schematics of the relevant actions of the vocal tract. For the consonants, the basic graphic unit is a cup. Its orientation indicates the part of the tongue used to produce the sound (Figure 18.1), with diacritics to indicate articulatory refinements. For instance, the tongue tip symbol  may cooccur with blockage of airfow,  t, with blockage plus vocal chord vibration,  d, and with partial blockage and nasalization,  n. For vowels, the basic symbol is an elongated version of the voicing diacritic for consonants. It is annotated at the top, bottom, or both for high, low, and mid vowels respectively, and on the left, right, or both for back, front, or central. All letters are formed from a set of ten radicals (graphic primitives), with eighteen further modifiers (for, e.g., clicked consonants and tone-bearing vowels). As an example, consider  “universal” as it appears on the front cover of Bell (1867). The first symbol,  y, indicates the body of the tongue (cup orientation) and vocal chord vibration (bar). Compare that to the third last symbol,  s, which engages the same articulator (cup orientation) but without vocal chord vibration (no bar; cf,  z) and with air channeled through a convex groove (curls either side of aperture). Similarly, compare the third and final symbols,  n and  l. These both indicate the tongue tip (cup orientation) and vocal chords (bar). They differ in that air is diverted via the nose for  n (nose-like squiggle) and around the tongue for  l (indentation in cup middle). The vowels  u and  i indicate from the high position of their curves that they are produced at the top of the mouth, at the back and front respectively hence their orientation; they also differ in that

224 Daniel Harbour

FIGURE 18.1

The vocal tract and the basis of orientation of the cup symbol for consonants –  tongue root,  tongue body,  tongue tip,  lips – and the diacritics for nasalization and vocal chord vibration, based on nose and glottis shape.

Source: Based on Bell 1872.

u is round (barred) and tense (bunched), whereas i is unrounded (no bar) and lax (unfurled).  i contrasts usefully with . It, too, is also unrounded and lax, but, by having these indications at the bottom, it shows that it is a low vowel (that of at, a sound rarely encountered in the suffix -al in modern British accents). Historically, Visible Speech lost ground to the International Phonetic Alphabet (IPA) of 1888, despite prominent advocates like Henry Sweet, who would later become president of the Philological Society. The system called for new types to be cast, making it more costly than the IPA, which extended the Latin alphabet. However, the viability of Bell’s ideas is underlined by the success of the Korean writing system, Hangul (Kim-Renaud 1998). Invented by Sejong the Great in the

Writing Systems and METI

225

15th century, it also draws on articulator shape for its character design (though not as completely as does Visible Speech) and – despite bureaucratic resistance – it displaced Chinese to foster an indigenous literacy that continues to this day. Clearly, there is more to human speech sounds than Bell’s schematic characters show. This is in part an advantage (cf, Pepperberg, this volume [Chapter 6]): writing highlights the salient and dampens the noise in spoken (and signed) language. Ultimately, spectrograms would need to be included for a subset of sounds to allow for the rest to be “deduced.” Once in place, though, Visible Speech could transcribe words of a multilingual encyclopedia in which more substantive content would be presented. This would circumvent the opacity of English spelling, the missing vowels of Arabic, and the unpredictability of Chinese characters, and it would facilitate fuller appreciation of the languages and materials sent. Visible Speech and Past METI

The continuity of Visible Speech with the philosophy and content of past METI is noteworthy. First, the aim of being self-decoding is shared. Just as the Arecibo and DutilDumas messages are intended to be self-decoding, so Bell subtitled his invention “self-interpreting physiological letters.” Second, the presentation of Bell’s system would be a natural extension of the Dutil-Dumas message. Their pages 15‒16 present basic information about humans, including a version of the Pioneer Plaque diagram. A longer self-decoding message could include internal anatomy, especially lungs and articulatory tract. These could then lead to images like Figure 18.1 that lay out the articulatory basis of human speech and Bell’s symbols. In a similar vein, the Voyager Golden Discs introduce breathing and then greetings from humans. Lomberg’s notes (cited in Lemarchand and Lomberg 2011: 379) notes on the recording state: I hope that some of the UN greetings will be recorded so that an intake of breath before syllables could be heard. This would link breathing with speech, and perhaps give a clue as to the respiratory nature of speech. Third, besides their physiological concreteness, Bell’s symbols capture an important cognitive fact. A foundational discovery of linguistics and especially phonology, the study of sound systems, is that speech sounds like t, d, and n are not primitive but are made up of smaller units, “features,” that constitute instructions to specific articulators (Jakobson et al. 1958 [1971]). These binary oppositions are directly reminiscent of the binary encoding used in METI itself. Nonetheless, Bell’s letters present some challenges in the context of METI. Their rotational symmetry is uncharacteristic of human writing systems (Morin 2018), though not unprecedented (Canadian syllabics encode vowels by rotating consonants). Also, their roundness lends itself poorly to pixelation (unlike Hangul letters).

226 Daniel Harbour

However, the most significant issue is similarity of characters. This arises from the use of a small set of primitives to encode all sounds and from the principle of reflecting similarity of sound by similarity of grapheme. Thus, sounds that differ in only one articulatory feature, like k/g, are represented by characters that differ equally minimally, /. The Dumas-Dutil message pursues the opposite strategy. It is intended to be maximally resistant to information loss, even in a noisy reception, and none of its characters are rotations or reflections of each other (making it highly nonhuman; Morin 2018). These problems are all superable. The Encyclopedia Britannica contains images of writing systems. So, Heidmann’s (1997) proposal involves sending images of written characters. Adopting this systematically, we could send three increasingly compact formats: full images, pixelated characters, and binary encodings. The transition from pixelation to binary can be accomplished while also breaking rotational symmetry. In a twin-prime pn × pn + 1 grid, a pn × pn pixelation can be bookended between a repeated p-length binary encoding of that character.5 Once enough samples of characters, words, and their phonetic transcriptions have been provided, binary can be used for the bulk of the remainder of the message. By progressing from human physiology to vocalization, writing, and language, an Encyclopedia Galactica would smoothly transition from the universals of mathematics and science to cultural wealth that is uniquely human. Conclusion: The I in METI

We are new to METI, and our mathematics and science are unlikely to advance other METI-engaged civilizations. However, our knowledge of humanity is unique. Our messages should strive to express this not simply in terms of our biology, but also in terms of our cognition. Our METI should emphasize our I. Writing systems provide a wealth of information about our physiology, cognition, languages, and cultures. With these facilitating the step from the universal truths of mathematics and science to the richness of human civilization, we can turn to the much bigger question of what the bulk of a METI-bespoke encyclopedia should deliver. Notes 1 These languages are not picked randomly. With the exception of English and Hindi, which are distantly related, all come from different linguistic families and differ substantially in phonology, morphology, and syntax, and, in accord with the following section, they illustrate diverse writing systems: logograms (Chinese), syllabaries (Cherokee), abjads (Arabic), aksharas (Hindi), and alphabets (English, Fijian). Many other choices are possible: Hebrew script might encode more easily than Arabic, but would mean the loss of pharyngeals from the sample of sounds; Georgian presents a beautiful script alongside one of the most complex morphologies of any national language; one might include a sign language (“the body as a resource for meaning making”, Ortner, this volume [Chapter 5]), both vis-à-vis Bell’s script in the following section, and because visual communication might be more common among other species (pace Kershenbaum, this volume [Chapter 2]). As with so much of METI, discussion is needed.

Writing Systems and METI

227

2 In particular, Heidmann’s (1997) claim that “The alphabetical coding can be deciphered using just a few pages, as well as the grammatical structures” (202) looks naïve. All successful decipherments of ancient languages involved text-external knowledge, chiefly of related languages. Ancient records are more fragmentary than an encyclopedia, meaning that lessons for METI drawn from decipherment of lost languages here on Earth (Finney and Bentley 2014; Saint-Gelais 2014) may be more pessimistic than necessary. However, Heidmann’s optimism is far from founded. Maybe computer scientists could prove their case on the Sumerian corpus, as many basic facts about the language remain under debate and a large number of texts have yet to be processed. 3 Aesthetics in mathematics is multifaceted. It covers beauty, simplicity, and surprise, and applies both to results and to proofs, in different ways. Discoveries can be sensorily aesthetic, like the Golden Ratio (Lemarchand and Lomberg 2011), or surprising, like Morley’s theorem (the points of intersection of adjacent trisectors of adjacent angles of a triangle form an equilateral triangle – crudely, even misshapen triangles have shapely souls); they can also be conceptually aesthetic, like eiπ + 1 = 0, which ties together three basic operations and five fundamental mathematical constants in a single equation. Proofs can be aesthetic for their simplicity, like Euclid’s demonstration of the infinitude of primes, or for the concepts they draw together, like Wiles’ proof of Fermat’s Last Theorem, which exploits a connection to elliptical curves. 4 I construe “alphabets” broadly to include abjads (e.g., Arabic, Hebrew) and aksharas (e.g., Amharic, Hindi), and take inherent vowels in aksharas to represent reading conventions, not consonant-vowel combinations. 5 A 29 × 31 grid furnishes more than half a billion binary encodings (229). This can be increased if other cells are unused across enough pixelations.

References Baxter, William, and Laurent Sagart. 2014. Old Chinese: A New Reconstruction. Oxford: Oxford University Press. Bell, Alexander Melville. 1867. Visible Speech: The Science of Universal Alphabetics; or Self-Interpreting Physiological Letters for the Writing of All Languages in One Alphabet. London: Simkin, Marshall & Co. Bell, Alexander Graham. 1872. On the Nature and Uses of Visible Speech. Boston: Rand, Avery & Co. Benford, James, Dominic Benford, and Gregory Benford. 2011. “Building and searching for cost-optimized interstellar beacons.” In Communication with Extraterrestrial Intelligence, edited by Douglas A. Vakoch, 279‒306. Albany, NY: State University of New York Press. Bromberger, Sylvain. 1992. On What We Know We Don’t Know: Explanation, Theory, Linguistics, and How Questions Shape Them. Chicago, IL: University of Chicago Press. Cook, Jia-Rui, and Dwayne Brown. 2010. NASA Probe Sees Solar Wind Decline. https:// voyager.jpl.nasa.gov/news/details.php?article_id=20. Dick, Steven J. 1996. The Biological Universe: The Twentieth Century Extraterrestrial Life Debate and the Limits of Science. Cambridge: Cambridge University Press. Dumas, Stéphane. 2003. The 1999 and 2003 Messages Explained. www.plover.com/misc/ Dumas-Dutil/messages.pdf. Dumas, Stéphane. 2011. “Proposal for an interstellar Rosetta Stone.” In Communication with Extraterrestrial Intelligence, edited by Douglas A. Vakoch, 403‒412. Albany, NY: State University of New York Press. Finney, Ben, and Jerry Bentley. 2014. “A tale of two analogues: Learning at a distance from the Ancient Greeks and Maya and the problem of deciphering extraterrestrial radio

228 Daniel Harbour

transmissions.” In Archaeology, Anthropology, and Interstellar Communication, edited by Douglas A. Vakoch, 65‒78. Washington, DC: National Aeronautics and Space Administration. Freudenthal, Hans. 1960. Lincos: Design of a Language for Cosmic Intercourse. Amsterdam: North Holland. Giraud, Anne-Lise, and David Poeppel. 2012. “Cortical oscillations and speech processing: Emerging computational principles and operations.” Nature Neuroscience 15: 511‒517. Harbour, Daniel. 2019a. Convergent Cultural Evolution and the Inventory of Sumerian and Chinese Signs. Ms, Queen Mary University of London. Harbour, Daniel. 2019b. The Limits of Logography: Phonology-Morphology Mismatches Drive Writing System Evolution. Ms, Queen Mary University of London. Harbour, Daniel. 2021. “Grammar drives writing system evolution: Lessons from the birth of vowels.” In Grapholinguistics in the 21st Century 2020, edited by Yannis Haralambous, 201‒221. Brest: Fluxus Editions. Harp, Gerald R., Robert F. Ackermann, Samantha K. Blair, Jack Arbunich, Peter R. Backus, Jill C. Tarter, and the ATA team. 2011. “A new class of SETI beacons that contain information.” In Communication with Extraterrestrial Intelligence, edited by Douglas A. Vakoch, 47‒70. Albany, NY: State University of New York Press. Haspelmath, Martin. 2000. “Coordination.” In Language Typology and Syntactic Description, Volume II: Complex Constructions, edited by Timothy Shopen, 1‒51. Cambridge: Cambridge University Press. Hedengren, David. 2013. Three Microeconomic Applications Using Administrative Records. PhD thesis, George Mason University, Fairfax, VA. Heidmann, Jean. 1997. Extraterrestrial Intelligence. Cambridge: Cambridge University Press, Storm Dunlop (tr.). Hobaiter. Jakobson, Roman, S. Karcevsky, and Nikolai S. Trubetzkoy. 1958 [1971]. “Quelles sont les méthodes les mieux appropriés à un exposé complet et pratique d’une langue quelconque?” In Selected Writings, Volume II: Word and Language, edited by Roman Jakobson, 3‒6. Hague: Mouton. Jaritz, Kurt. 1967. Schriftarchäologie der Altmesopotamischen Kultur: Eine grammatologische Untersuchung zur Entstehung des ältesten Biderschriftsystems. Graz: Akademische Druck. Kim-Renaud, Young-Key. 1998. The Korean Alphabet: Its History and Structure. Honolulu, HI: University of Hawai‘i Press. Lemarchand, Guillermo A., and Jon Lomberg. 2011. “Communication among interstellar intelligent species: A search for universal cognitive maps.” In Communication with Extraterrestrial Intelligence, edited by Douglas A. Vakoch, 371‒398. Albany, NY: State University of New York Press. Luo, Huan, and David Poeppel. 2007. “Phase patterns of neuronal responses reliably discriminate speech in human auditory cortex.” Neuron 54: 1001‒1010. Morin, Olivier. 2018. Spontaneous emergence of legibility in writing systems: The case of orientation anisotropy. Cognitive Science 42: 664‒677. Saint-Gelais, Richard. 2014. “Beyond linear B: The metasemiotic challenge of communication with extraterrestrial intelligence.” In Archaeology, Anthropology, and Interstellar Communication, edited by Douglas A. Vakoch, 79‒94. Washington, DC: National Aeronautics and Space Administration.

Writing Systems and METI

229

Shostak, Seth. 2011. Limits on interstellar messages. In Communication with Extraterrestrial Intelligence, edited by Douglas A. Vakoch, 357‒370. Albany, NY: State University of New York Press. The Staff at the National Astronomy and Ionosphere Center. 1975. “The Arecibo message of November, 1974.” Icarus 26: 462‒466. Stassen, Leon. 2003. “Some universal characteristics of noun phrase conjunction.” In Noun Phrase Structure in the Languages of Europe, edited by Franz Plank, 761‒817. Berlin: Mouton de Gruyter. Weeden, Mark. 2014. “Anatolian hieroglyphs: Logogram vs ideogram.” In Visualizing Knowledge and Creating Meaning in Ancient Writing Systems, edited by Shai Gordin, 81‒100. Berlin: PeWe-Verlag. Wittgenstein, Ludwig. 1953. Philosophische Untersuchungen: Philosophical Investigations. Translated by G.E.M. Anscombe. New York: Macmillan. Zwart, Jan-Wouter. 2005. “Some notes on coordination in head-final languages.” Linguistics in the Netherlands 22: 231‒242.

INDEX

φ‑features 188 abjad 222, 226, 227 acoustics (communication) 5–6, 9–12, 22, 34–35, 81, 90–91, 101, 103, 190, 201–207 Afro‑Asiatic (language family) 222–223 Akkadian 222 akshara 226, 227 alarm (calls or signals) 4, 6, 19, 22, 33–35, 52–53, 55, 89–91, 203 alphabet 217, 220, 222–226, 227 Althusser, Louis 78 American Sign Language 93, 184, 186 Amharic 227 Anatolian 219, 223 aphasia 39 Arabic 219, 225, 226, 227 Arecibo message 214, 216–217, 225 Arrival 51, 99–100, 135 artificial intelligence 35, 64, 134, 135, 158, 165, 168, 171 associative information packets (AIPs) 80–81 Athenæum 223 Bardi 104 bat 12, 52, 68, 191; linguistic representation of 196 Bell, Alexander Graham 223 Bell, Alexander Melville 223–226 Bilingual Image Glossary (BIG) 216–217

birdsong 5, 23, 27, 52–54 black‑capped chickadee 53 Blackfoot 188 Bloomfield, Leonard 111 Bobrow, Daniel 171 Boyle’s Law 191 Brahmi 222 calliphora vicina 4 Cambias, James 139 Canada (village) 116 Cartier, Jacques 116 Chambers, Becky 139 Champollion, Jean‑Francois 89 Cherokee 219, 222, 226 chimpanzee 18, 19, 21–25, 36, 52 Chinese 128, 219, 221, 225, 226; Old 221 Chinese room thought experiment 158 Close Encounters of the Third Kind 51 color terms 55, 97, 115, 144 Common European Framework of Reference (CEFR) 81–82 Communicating with Extraterrestrial Intelligence (CETI) 41–47 communication 1, 5–10, 9, 11–12, 18–19, 34, 42, 52, 54, 68, 128, 184, 197, 204, 226; acoustic/auditory 5–6, 9, 11–12, 18–19, 34, 42, 52, 54, 68, 128, 184, 189, 197, 204, 206; avian 5, 23, 27, 43, 52–54. 91–92; cetacean 33, 34, 52, 66; chemical/

Index

olfactory 6–7, 34, 43, 68, 70, 71, 79, 128–129, 190; cross‑species 19, 33–34, 56–58; electrical 7–9, 12, 22, 34; (electro)magnetic 9–10, 22, 34, 128–130, 190, 203–204, 206; prairie dog 35, 52, 89–92, 203, 208, 217; primate (non‑human) 20, 21, 23, 25–28, 35, 52, 54; visual 5–6, 11, 18, 22, 34, 52, 68–69, 71, 128, 142, 184, 189, 226 compositionality 113, 119, 130–131, 135, 176, 186 Contact 51 Cooperative Principle 16, 126, 132 Core vocabulary list (CVL) 81–82; see also Swadesh list Cosmic Calls 216–217 coyote 89–90 cricket 201–202 Crimean Gothic 100 cuneiform 65, 67 The Darkling Sea 139 deixis 145–147 Descartes, René 165–166 directionality 115, 144–145, 147 dog (domestic) 7, 33, 44, 45, 90, 198; linguistic representation of 88, 116, 190, 205, 207 dolphin 10–11, 33, 35–36, 43, 52; Atlantic spotted 33; bottlenose 10, 33, 36, 43; killer whale/orca 33 Drake, Frank 123–135; Equation 123–135, 160 Dutil‑Dumas messages 225 Economy Thesis 171–173 Egyptian 89, 221, 222; Ancient 222; Demotic 89 ELAN 103 e‑language 112, 183 electric fish 8–9, 12; Gymnotiformes 8; Mormyriformes 8 elephant 52, 191 Emergentist approaches 152, 159 Encyclopedia Britannica 215, 219, 226 English 62–65, 82, 83, 88, 100, 114–117, 145, 157, 158, 161, 183, 186, 188, 207, 216, 219, 222, 225, 226 ergativity 104 Euclid 218, 227 evidentiality 146–147

231

evolution 3–4, 7–12, 22, 129, 132, 138, 152–159, 160, 166, 171–178, 179, 184, 186, 191; communication 62–72; cultural 42, 220 externalization 165, 178, 182, 189–191 Faculty of Language Broad (FLB) 185–191 Faculty of Language Narrow (FLN) 185–191 Fermat’s Last Theorem 227 Fibonacci sequences 80 Fijian 219, 226 flycatcher 52 Galileo Galilei 167–169 Gavagai problem 81, 116 Georgian 226 German 63–64 Golden Discs (Voyager Probes) 216, 225 Golden Ratio 227 Greek, Ancient 89 Gricean Pragmatics 7, 15–28, 43, 85, 120, 126–127 Grice, Paul 7, 15–28, 43, 85, 120, 126–127 Halkomelem 188 Hangul 224–225 Hebbian learning 205 Hebrew 222, 226, 227 Hindi 219, 226, 227 Hockett, Charles 6, 10, 88–89, 97, 123– 124, 126, 128–129, 134, 153–154, 160, 183–184, 190; Design Features 6, 10, 88–89, 123–124, 126, 128–129, 153–154, 160, 183–184 Hooke’s Law 191 Hopi 114–115 Human‑animal interaction (HAI) 43–44 Hummingbird 52 IBM 75 i‑language 112, 168–169, 183 Interactional Linguistics 40–47, 48 International Phonetic Alphabet (IPA) 224–225 Iroquoian (language family) 116 Jacobellis v Ohio 111 Japanese 145, 221–222 Japanese great tit 53 jumping spider 10

232 Index

kangaroo rat 10 Kant, Immanuel 218 Kanzi 93 Koko 93 Korean 222, 224–225 Kriol 100 Laser Interferometer Gravitational‑Wave Observatory 52 Latin 62 Leibniz, Gottfried Wilhelm 168–169 Lincos 216–217 Linear A 89, 158 Linear B 222 locked‑in syndrome 39–40 logogram 219, 221, 226 Malinowski, Bronislaw 99 Maltese 222 Marsh wren 54 Maxim (Gricean) 16–17, 27, 120 Mayan 221–222 Menzerath’s Law 18 merge 152–153, 160, 161, 167–178, 179, 185–191 Messaging Extraterrestrial Intelligence (METI) 106–107, 123–127, 158, 161, 208, 214–227 Milky Way 125, 215 Minimalist Program 128, 166–167 Minsky, Marvin 123, 125, 139, 143, 171–176, 189 Model/Rival (M/R) technique 56–58 Morley’s theorem 227 morphology (linguistics) 10–11, 88–89, 104, 110, 114, 119, 128, 130, 154, 184–185, 221–222 morphology (physical) 41, 78 Olmec Script 89 Olympus Mons 201 Optimality Theory 82–83 orangutan 208 parrot 34, 36, 54–59, 129 Peirce, Charles Sanders 177 Philological Society 224 Phoenician 222 phoneme 83, 88–91, 111, 127–129, 197 phonology 127–128, 154, 156, 190, 225–226 pinnipeds 191

Pioneer Plaques 214, 216, 225 Pirahã 160 prairie dog 35, 52, 89–92, 203, 208, 217 principles and parameters (P&P) 168–170 pronouns 115, 147 Psamtik (Pharaoh) 138 Ptolemy V 89 rat 7, 10 recursion 18, 76, 91, 98, 153, 160, 166–167, 176, 178, 185, 189 red‑tailed hawks 89–90 redwing blackbird 53 Rescorla‑Wagner equations 200 Rongorongo 89 Rosetta Stone 89, 107 Rumi 219 Russell, Mary Doria 139 Russian 216 Sapir, Edward 114 Sappho 219 Search for Extraterrestrial Intelligence (SETI) 75, 123, 126–127, 158, 214–217 Sejong the Great 224–225 semantics 36, 80, 88, 90, 110–120, 196, 219; type‑theoretic 117 sensory systems 4, 11, 33–34, 182 Shakespeare, William 219 Shannon boundary 64 shark 7 Shōnagon, Sei 219 Snellen test 112 Social modeling theory 55–58 Sogidan 222 The Sparrow 139 sparrow 54 speech acts 118–120 squirrel, ground 35 Stewart, Potter 111 Strong Minimalist Thesis (SMT) 166–167, 182–191 Sumerian 221, 227 Swadesh list 81–82, 115–116 Sweet, Henry 224 syllabary 220–222, 226 syllable 18, 53, 82, 161, 220–222 syntax 36, 44, 91, 107, 113, 127–128, 154–156, 161, 166, 173, 176, 185–189, 218–219, 226

Index

Tagalog 131 third factor conditions 97, 139–143, 182–191 truth 27, 110–113, 117, 127, 132 Turing machine 169–178, 179, 189 Ulam, Stanislaw 177 Universal Grammar (UG) 97, 165–179, 182–191 Vakoch, Douglas A. 76 vervet monkey 19, 35 Vietnamese 221 Vinca 89 Visible Speech 223–226

233

Voyager probes 216, 225 Voynich Manuscript 97–98 warbler 53 Washoe 93 Wayfarer (book series) 139 Wiles’ proof 227 wolf 10–11, 19 World Atlas of Linguistic Structures 187 Yiddish 222 zebra finch 53 Zhuang 221 Zipf’s Law 18, 70, 92