Dark Matter of the Mind

by Daniel L. Everett

  Communication is a diffuse area of investigation. At its simplest, communication is just the transfer of information (Shannon 1949). If this is all that communication is, however, then even a thermostat could be described as “communicating” with its environment, taking in information about the temperature and producing an appropriate response. It “knows” what is relevant to it. It uses concepts. It responds to a stimulus. Yet we resist this characterization because there is no intentionality involved in the thermostat’s response to its environment. Without this function of intent, the judgment that thermostats communicate does not hold (Searle 1983). Intentionality seems a reasonable threshold for labeling an act “communicative.” In this sense, many animals communicate, including dogs, cats, nonhuman primates, birds, frogs, and so on. Communication does not entail language, though language is the apex of communication systems. But what is it that makes the special subset of communication we call “language”?

  The first crucial invention that is required for language is the symbol. The importance of symbols for language has been recognized at least since the philosophical work of Peirce (1977) and the linguistic development of the sign by Saussure ([1916] 2012). In recent years, Terrence Deacon’s (1998) work has offered more insights into the development of symbols for human language. But perhaps the most significant advances in understanding the symbol’s centrality to human language come from the theoretical framework known as construction grammar (for the founding work, see Lakoff 1977 and Fillmore 1988, among others; Croft 2001 and Goldberg 2006 are two of the most important developments of the theory). Construction grammar rejects the strict dichotomy between signs and grammar that has played such an important role in formalist theoretical linguistics, especially of the Chomskyan variety.

  One type of symbol is what Saussure ([1916] 2012) refers to as “signs” (these are the linguistically crucial symbols without which no human languages would exist). Signs have two components: their form (physical instantiation) and the meaning (sense and reference; Frege 1980). A symbol is a form combined with a meaning in a culturally specific (i.e., non-logically required) way. So dog is a form associated with the meaning “canine,” though a priori there is no special logical or other connection between the three phonemes of dog and the meaning “canine.”

  However crucial symbols may be, the development of symbols is perhaps not as difficult a breakthrough for larger hominin minds as one might think. In fact, these two components are crucial to every entity that has relevance to hominins. A special path, a loved one, a drawing on a cave wall, marks on bone—each must have both a form and meaning. The form may or not be invented. It could be, for example, a tree, a stream, a rock. But the meaning must be created, largely through apperceptions and social awareness as individuals in a group learn that the stream is the “place of good hunting” or the rock is “the place of the eagle.”

  Other forms are created by creatures and given meaning by the same creatures or others. A vervet monkey makes a sound when a bird of prey is overhead and alerts other vervets. If a human utters “ugh” and means by that “no,” and another human is subsequently attacked for their failure to recognize the “ugh” symbol’s meaning, that symbol is more likely to be recognized the next time it is encountered (by witnesses, at least). The difference between the vervet’s and the human’s symbols is that the former requires natural selection for its spread, whereas for the human culture, it is sufficient for the origin and spread of a symbol (and perhaps for other species—see the literature discussed in chap. 10). But although what animals produce may be symbols in a general, nontechnical sense, perhaps no other animal produces signs other than humans. The distinction, of course, from our earlier discussion, is that the relation of form to meaning in signs is not as straightforward. It can range from highly indirect (the classic example of different forms in different languages, such as dog, perro, cão, and other words for the animal {dog})—to the more direct (such as sound-symbolic words like snap, crackle, pop).

  Unlike the case of most other animal species, the sign in human languages takes many forms. Here are some of them:

  Indexicals. We discussed these earlier as various forms of symbolizing particular cultural and individual values, from body language to word selection.

  Sound symbols. These include things such as onomatopoeic words, like zoom-zoom for the sound of a motorcycle or clippety-clop for the sound of a horse’s hooves as it walks. Such symbols also include the sound /h/ in related words such as hut, house, and hovel, or the /gl/ sequence in glow, glimmer, and glisten.

  Icons. Drawings, pictures, statues, and so on, are special kinds of signs (in Peirce’s sense) that are intended to convey a likeness of the meaning (referent).

  Indexes. Direct physical evidence of the referent, such as tracks (for animals), smoke (for fire), cookie crumbs (for cookie eating), and so on.

  Signs. Here the crucial notion is not merely that there is a form linked in the culture to a meaning, but that the form-meaning connection is almost exclusively cultural—that is, a decision made by the society, a type of social contract (Searle 2006)—and thus for the Martian linguist or the newborn baby, the form-meaning connection is arbitrary. Why is fiddle an instrument for bluegrass music, the violin an instrument for classical music, but guitar an instrument for either? Only culture, if anything at all can be found, can explain such differences, via history in this case (fiddle may originate from Old English fithele, deriving from the Latin vitulari, “to celebrate,” and at one time came to refer to any instrument used in a party or celebration. Or it may be an English corruption of Old French viele—a predecessor of the violin).

  To the question of how signs originate, part of the answer is that this is like asking who the first person was who told a particular joke. Neither the origination of a sign or a joke is the act of a single individual. All individuals think original thoughts and make up what are (at least, until they make them up and others use them) nonsense words, expressions, jokes, and such, that in potential could spread but do not. Chomsky’s “Colorless green ideas sleep furiously” could itself be the origin of a new symbol in the language that could eventually become a sign. But for an odd form-meaning pair to become a symbol in a language, it must be adopted by the society. “Grammar” is a symbol to all linguists, though its meaning varies significantly according to linguistic subculture. For some it means that form and meaning are unrelated; in particular, that a phrase can in principle be both meaningless and grammatically well formed (as in Chomsky’s example, though most linguists see this largely as a theoretically internal fact, based on unique concepts of meaning and form). For others, grammar is the way that the language conveys meanings and is formed principally by the meanings it conveys. For one, extreme grammar is a vast domain. For another it is a relatively small residue of history, cultural constructions, and so on.

  What constitutes a sign, however, can vary tremendously from theory to theory. For some, grammatical constructions are not signs, but epiphenomena that result from grammatical rules taking individual words and putting them together in some arrangement. According to the Chomskyan conceptualization of syntax, once syntactic phrases are built up from words to larger units, they can be interpreted by the hearer top-down, using inverse compositional or decompositional semantics (Van Valin 2006).

  However, as indicated in the title of Lakoff’s famous 1977 paper, “Linguistic Gestalts,” at least some (perhaps all) phrases and sentences have meanings that are properties of the constructions themselves, and not predictable from compositional semantics. An example might be the sentence Keep tabs on John while he is out on parole. The isolated word tabs has no literal meaning in this construction, but is crucial to the form of this idiom where keep tabs on means “supervise” or “watch.” A related construction might be Keep an eye on Mary while she is shopping, where keep an eye on is not meant literally. Once constructions are recognized in the theory of grammars, the distinction becomes blurred between the “
grammar proper” (the rules that generate grammatical forms without regard for meaning) and the “lexicon” (the speaker’s brain-stored dictionary, based on their culturing experiences; the lexicon in this sense is one of the largest portions of our dark matter). Since signs are constructed culturally, this entails a greater role for culture in what has been traditionally known as grammar. In fact, as we see directly, culture’s hand can be found even in the grammar itself—the way that phrases are formed, to the degree that they are, apart from constructions or meaning.

  Grammar

  Following intention and symbols, there must also be grammar for language to appear (D. Everett [2012a] discusses the various “platforms” that must be in place before language is possible.) Some linguists—most notably Chomsky—have a purposely narrow understanding of what constitutes grammar. And if we take grammar to be a strictly formal, computational system that at most filters meanings without being driven by meanings, then our understanding of language will be quite different from the views of those who see grammatical form as primarily a function of meaning. These varied views represent the central controversy of linguistics, arguably one unresolved at least since Sapir (1921) and Bloomfield (1933) presented very different pictures of language as sociocultural interactions with the mind, meaning, and structures vs. meaning-free and culture-free structure, respectively.

  This “meaning movement” began in the modern era in the US with “generative semantics” in the 1970s. This was a theory developed by several of Chomsky’s former champions and, ironically, was based on Chomsky’s notion of “deep structure,” the level of syntax in which the lexical and basic phrasal meanings were once thought to reside. Many linguists find it humorous that non-linguists frequently confuse deep structure with universal grammar. However, this is unfair; the mistake is in fact quite a natural one to make. Part of the reason is that in the literature on UG, the definition is usually unclear. Perhaps a main reason, though, for confusing UG and deep structure stems from Chomsky’s competitor theory of the 1970s, generative semantics’ universal base hypothesis (UBH). According to the UBH, all languages share the same semantic (deep) structure that serves as the foundation for their sentences—clearly a notion similar to UG.

  Other meaning-based theories that emerged from generative semantics included functionalism (in its myriad forms), construction grammar (already mentioned), and role and reference grammar. Interestingly, however, even these meaning-based theories that rejected Chomskyan linguistics for ignoring the central role of meaning in grammatical structure failed to reach deeper for the source of meaning, appealing in Chomskyan fashion to universals of human meaning as if those are preordained in some way.

  If I am correct, however, the meaning that shapes human grammars emerges at least partially from culture. This seems to be what Sapir (1929, 2) is getting at when he says: “Speech is a non-instinctive, acquired, ‘cultural’ function.” Therefore, I want to look more closely at the idea that language is significantly shaped by culture. One way to interpret Sapir’s statement is to say that grammars arise within particular systems of cultural values and that the values and patterns of conversation, telling stories, uttering phrases, and so on, themselves constrain the grammatical structure of a language.

  Before entering in to a detailed discussion of dark matter, culture, and grammar (phonology, morphology, discourse, and syntax), however, we need to review some linguistic history in order to better understand the issues. It is also important to understand how linguistics has been led away from its traditional concern with language as partially constructed by culture. Principally, this shift occurred due to a reification of the field that began with Chomsky’s earliest work, continuing through to the present state of formal linguistics more broadly.

  From there, I argue that the child learns its culture and other forms of dark matter at least as early as—perhaps even before—it begins to learn its language. This non-linguistic learning affects the child in many ways, including its conception of how language is used for communication, a conception that can in turn affect her grammar.

  Next, I want to look at Pirahã’s segmental phonology. I argue that, contrary to most linguists’ expectations, even the sound system of Pirahã is architectonically affected by culture. I conclude the discussion of Pirahã with a suggested methodology for establishing culture-grammar connections.

  We then examine two additional Pirahã texts to exemplify how culture and grammar interact in discourse. From this we turn to a consideration of how Pirahã culture affects its syntax, via the overarching cultural value IMMEDIACY OF EXPERIENCE (IEP) (D. Everett 2005a, 2008, 2012b), and the reflex of this value in the evidentiality system of the grammar, through the POTENTIAL EVIDENTIALITY DOMAIN. This potential evidentiality domain (PED) is a culturally motivated principle of Pirahã grammar. A side effect of the PED is to bar recursion from the morphosyntax entirely.

  The discussion leads us, I argue, to the formula

  COGNITION, CULTURE, AND COMMUNICATION→GRAMMAR

  arguing that Pirahã is not a unique case and that all languages will show culture-language connections if we look. But of course, it is hard to find such connections if we do not look for them.

  REIFICATION OF LINGUISTICS

  Until the 1950s, the common professional classifications and departmental homes for linguists were anthropology and foreign language departments. The idea that language was itself a component of culture, society, folklore, and so on, was shared on both sides of the Atlantic. Both Sapir (1921) and, later, Roman Jakobson (1990) wrote widely about language’s various manifestations in discourse, poetry, conversation, sound systems, and so on, underscoring the symbiotic relationship between language and culture.

  But Chomskyan linguistics turned away from interest in human culture in the late 1950s, leading to a marked reification, ignoring various intersections of culture and grammar (e.g., discourse structure, idioms, sound symbolism, and field research) in favor of what it intended to be a deeper understanding of the forms of language, as manifested in phonology, morphology, and syntax.

  Chomsky famously designated the sentence as the “start symbol” (Σ) of grammar, ignoring the possibility that sentences might be constituents of discourses and conversations. Chomsky (1956, 1959) argues that a grammar will generate “all and only the sentences” of a language L. Thus not only does he make the (now apparent) error of making the sentence the foundation of any grammar, but then, as a necessary consequence of the initial assumption, all that grammars can generate are sentences—completely omitting discourse and conversations, arguably the most interesting units of any language. This in effect says that there is nothing of interest for the grammarian above the sentences.

  And discourses and conversations are not simply arrangements of sentences, as a myriad of work shows.1 The failure to look at discourse (and culture) in studying sentences is on par with the now-outdated position of earlier linguists who avoided incorporating morphological phenomena into the analysis of phonemic structures (Pike 1952). As many linguists have shown, beginning in particular with Givón (1983)—one cannot understand sentence structures well without understanding the discourses they are embedded in. Though generativists might insist on looking only at sentence-level phenomena, the rest of the (psycho)linguistic enterprise has long moved on from this artificial, self-imposed limitation.

  Looking beyond the work of Sapir, Jakobson, and Pike, inter alia, recent research by Levinson (2006), Enfield (2002), Silverstein (2003) and many others has advanced our understanding of sentences as units of interaction. However, formalists respond that such research is orthogonal to the enterprise of generative linguistics, because it fails to explicitly focus on the basic grammatical architecture of languages. To address this objection, therefore, I discuss in what follows precisely the effects of culture on this so-called core grammar.

  Mainstream generative studies of core grammar have focused on the forms of sentences, phrases, and words—thus offering si
mply a continuation of Bloomfieldian structuralism with the ideas of the mind and nativism added. Starting from the assumption of UG, the analyst proposes a deductive set of categories. Subsequent analyses in the same tradition then apply and tweak these categories or processes, with the aim of showing that they fit all languages in some way. Generative studies are said to differ from Bloomfieldian structuralism by paying greater attention to mental representations. But this claim can be dismissed at once because the mental is never causally implicated in any analysis of the theory, and UG is orthogonal to the analyses (D. Everett 2012b).

  In this new Chomskyan structuralism, several assumptions have come to dominate thinking about syntax in theoretical linguistics: (i) all grammars are hierarchically organized by recursive procedures; (ii) all grammars involve derivations; (iii) all syntactic structures are formed by combining two units at a time to produce endocentric, binary branching (and hierarchical) structures; (iv) all grammars derive from a genetic endowment common to humans, called universal grammar; and (v) the domain of grammar is the sentence. Many linguists, however, would argue that all of these points have been falsified. Jackendoff and Wittenberg (forthcoming) have argued that Riau and Pirahã have nonrecursive syntax (see also Futrell et al., forthcoming). Robert Van Valin (2005) and others have argued that derivations are never necessary in any grammar. S. Frank, Bod, and Christiansen (2012) have even argued that hierarchy and recursion are unnecessary for the proper analysis of any natural language. Lieberman (2013) has developed a formidable case to the effect that there is no neurological support for the idea that grammars derive from language-specific innate principles. Culicover and Jackendoff (2005) have argued convincingly against (i)–(iii). And I myself have offered analyses of various languages—especially Pirahã (D. Everett 2005a) and Wari' (D. Everett 2005b, 2009b)—that appear to falsify (i)–(iv). (For a slightly different though still compatible perspective, see Corballis 2007.)