Dark Matter of the Mind

by Daniel L. Everett

  Bigixisitísi Dies

  Told by Itaíbigaí

  1.

  Bigixisitísi

  hi baábi.

  Kapío xiai.

  Bigixisitísi

  he is sick.

  Other is

  Bigixisitísi has a different kind of sickness.

  2.

  Hi baábioxoi.

  He sick interrogative

  What is his sickness?

  3.

  Hi aigía

  ko Xápaí.

  Xí kagi

  hi xaoabábai.

  He thus

  hey Xapai

  her husband

  He nearly died

  He thus. Hey Xapai. Her husband nearly died.

  4.

  Hi ábahíoxioxoihí.

  Unknown.sickness interrogative

  Did he have an unknown sickness?

  5.

  Hi aigía.

  Koaísiaihíai.

  He thus

  became dead

  Well then. He was dead.

  6.

  Soxóa

  ti

  kabáo.

  Koaíso.

  Xai

  Bigixisitísi.

  Already

  I

  finished

  dies

  (he) did.

  Bigixisitísi

  Bigixisitísi is already finished, affecting me. Bigixisitísi died. He did.

  7.

  Xabí

  Xioitábi

  not there

  Linda [Xioitábi is her name in Pirahã]

  Linda was not there.

  8.

  Hi xabaí.

  She not

  She was not here.

  9.

  Ti xaigía gáxai. Xai.

  Hi abikaáhaaga.

  I thus speak. [I] do.

  He not be.

  I thus spoke. I did. He is no more.

  10.

  Hi oaíxi.

  Pixái.

  He dead

  now

  He is dead now.

  We see that these Pirahã texts—like texts in any language—reveal cultural values and knowledge(s), and require a culturally based hermeneutics. To understand these, we need to review some of the crucial issues in understanding the relationship of grammar to culture; in particular, how culture and grammar shape each other through their evolved symbiotic relationship.

  When I say that culture and grammar are symbiotically related, I mean first that grammar (and language) is dependent on culture not only for its functions but also for the very forms it employs to carry out those functions. But I also mean that culture is codified, regulated, reinforced, and partially formed by grammar. Thus, though grammar and culture may be epistemologically and ontologically distinct, they are not independent in practice.

  My notion of a grammar-culture symbiosis is not to be confused with the idea that either is supervenient on the other. Supervenience is a relationship such that “a set of properties A supervenes upon another set of properties B just in case no two things can differ with respect to A-properties without also differing with respect to their B-properties” (McLaughlin and Bennet 2011). Further, by symbiosis I mean that grammar and culture are each causally implicated in and dependent upon the other in their respective historical developments, even though there is no one-to-one mapping between them.

  Linguists and anthropologists, as to be expected, hold different views on the connection between grammar and culture. There are those who argue that any interaction between the two is trivial, the total range of interactions not moving much farther than a few lexical choices and things such as polite vs. formal address forms. At the other end are those who think of grammar as little more than a cultural artifact. Neither extreme captures what I am after here.

  CULTURAL CONSTRAINTS ON “CORE GRAMMAR”

  For another example of culture affecting grammar, I want to revisit Pirahã’s apparent lack of recursion (D. Everett 2005a, 2008, 2012b; Futrell et al., forthcoming). Most languages use recursive operations in the construction of their syntactic structures. This is so common cross-linguistically that in 2002 Marc Hauser, Noam Chomsky, and Tecumseh Fitch (HCF) made the startling claim that the single innate cognitive ability underlying language—possessed exclusively by Homo sapiens—was the ability to construct grammars recursively. Subsequently, this bold claim has since been falsified for being both too weak and too strong. The proposal is too weak because there is abundant data that humans are not the only species that use recursive cognitive or communicative operations (Corballis 2007; Golani 2012; Pepperberg 1992; Gentner et al. 2006; Rey et al. 2011). Second, the proposal is too strong because there are languages that lack recursion (D. Everett 2005a; Gil 1994; Jackendoff and Wittenberg 2012). To see what HCF mean by recursion, here is a statement from the original paper:

  FLN only includes recursion and is the only uniquely human component of the faculty of language. . . . In particular, animal communication systems lack the rich expressive and open-ended power of human language (based on humans’ capacity for recursion). (Hauser, Chomsky, and Fitch 2002, 1569–70)

  There are many potential senses of the term recursion, so it is vital to understand what HCF had in mind. Their paper leaves no doubt that they intended a process that applies to its own output without limit. This is clear in such statements as the following, where they claim that when a language has recursion, then “there is no longest sentence (any candidate sentence can be trumped by, for example, embedding it in ‘Mary thinks that . . .’), and there is no nonarbitrary upper bound to sentence length” (reference 9, 1571; emphasis mine).

  Although such quotes are straightforward, some Chomskyan syntacticians now claim for it a more esoteric meaning (as a reaction to my criticisms and empirical work [D. Everett 2005a, 2008, 2012a, inter alia]). According to this initiate exegesis, recursion means for HCF only a (singleton) subset of recursive operations internal to the program known as minimalism, what Chomsky (1995) calls “Merge.”

  This means that Merge is potentially falsified by any exocentric or non-binary (ternary, quaternary, etc.) branching structure—for example, a structure with flat syntax. Culicover and Jackendoff (2005) have argued—to my mind, convincingly—that ternary structures exist in the syntax of some languages, and I (D. Everett 1988) have argued that non-derivable ternary structures exist in the metrical structure of Pirahã phonology. Further, I (D. Everett 2005b; 2009b) have elsewhere demonstrated that the syntax of the Wari' language of Brazil makes widespread use of non-endocentric constructions. Yet even though counterexamples exist, the authors and their followers continue to insist that Merge is what they meant by recursion.4

  Nevertheless, the Merge interpretation has to strain to produce the “no longest sentence” clause of their earlier quotation, since that is a result of the more general notion of recursion. Even Chomsky (2010, 2014) allows that Merge itself may be blocked from repeating endlessly by language-specific stipulations. But such stipulations play no part in the mathematical notion of recursion.

  I have argued elsewhere (D. Everett 2012b) that this post hoc, theory-internal reasoning is unhelpful. First, it excludes an important empirical space—namely, the class of languages that lack Merge but have other forms of recursion, such as languages with ternary branching but no longest sentence. Second, it ignores the possibility that some language may lack any form of syntactical recursion, such as Pirahã. Third, it overlooks what is to my mind the most important consideration in understanding the role of recursion in natural language—natural conversations, narratives, and other discourses.

  Lobina and Garcia-Albea (2009) offer a helpful elucidation of various notions of recursion that have been employed in mathematics, computer science, linguistics, and cognitive sciences. They correctly observe that Merge itself need not be a recursive operation, since iteration does not properly fall within the standard mathematical or computational definitions of recursion. But if it is not recu
rsive, there must be ancillary hypotheses to guarantee this, further weakening the hypotheses.

  Languages with non-recursive grammars—such as Pirahã (D. Everett 2005a) and perhaps Riau (Gil 1994)—are far from irrelevant to the construction of syntactic theories. In HCF’s sense, recursion is the only item in the “linguistic toolbox.” Further, HCF claim that recursion is the only biological difference between humans and other animals that makes language possible. If they then concede that not all languages require recursion, their original claim loses any empirical force it might have had. If recursion is the fundamental component of universal grammar, how could it be lacking in any language? It would be a strange proposal indeed if the absence of the singular biological underpinning of language is treated as empirically irrelevant.5 In fact, treating Riau, Pirahã, and other languages that lack recursion as exceptions would be like saying that finding a black swan or a penguin does not falsify the claims that all swans are white or that all birds fly.

  Ironically, although I have repeatedly argued that Pirahã shows recursion in texts, texts lie outside the sentential syntax that has defined generative theory since its inception, where, again, the “start” symbol (Σ) for all syntactic operations has always been the sentence. (D. Everett [1994] offers a [Chomsky-inspired] discussion of the “sentential divide” in grammar and cognition.)

  Linguists have long resisted the idea that culture is causally implicated in the formation of grammars, at least insofar as what Chomsky calls “core grammar”—the state of the language faculty after language-specific parameters have been set. Here I want to underscore arguments I have made elsewhere (D. Everett 2005a, 2008, 2012b). I will do this by examining the relationship between the morphosemantic notion of evidentiality, Pirahã culture, and Pirahã syntax.

  Evidentiality—the semantic notion of evidence for an assertion—is found in all languages in one form or another. For example, if I say, “The man came in here,” the default assumption in English is that I have direct evidence for this assertion. Evidentiality is arguably found in the pragmatics of every language because it helps the hearer distinguish speculation from evidence-based declarations, something that could save a lot of time in deciding where to hunt, build a village, and the like. However, for some cultures, evidentiality is not only a semantic or pragmatic fact but a morphosyntactic fact as well, encoded in some way on words (usually verbs) of the language. At some point in the development of such languages, speaker usage turned this near-universal semantic category into an overt symbol in their grammar. This is to me a cultural development, even if no speaker(s) consciously invented the evidentiality morphemes for their language (how cultural innovation or innovation in language arises is a complex issue, known as the “actuation” problem—how changes spread through a culture or language; Weinreich, Labov, and Herzog 1968). I believe that the relative importance of evidentiality, like any other category, can be calibrated for a given language from its effects on the morphosyntax and its role in the culture. The greater the effects, the more important it is; the fewer, the less important. These are determinations about dark matter based on observable behavior. They do not necessarily result from conscious manipulation of specific morphemes by speakers. This, then, is evidentiality. The next step for us is to understand how Pirahã evidentiality follows from Pirahã culture.

  Everett (2005a) describes a range of unusual features of Pirahã culture and language, many of them never documented for other languages (though one would not be surprised if many other languages had similar features or lacked such features). These include: simplest kinship system known, lack of color words, lack of numbers and counting, no perfect tenses, no creation myths, no historical or fiction myths, being monolingual after more than three hundred years of regular contact with Brazilians, and no recursion. I proposed to account for all these facts by the immediacy of experience principal (IEP). This is a principle found in some degree of strength in many Amazonian languages. (See Gonçalves 2005 for a discussion of the pervasiveness of immediacy of experience as a cultural value throughout Amazonia.)

  Dark matter’s effects are far-reaching. In fact, the IEP affects Pirahã grammar profoundly. To see how, let’s begin by restating this principle:

  Immediacy of experience principle (IEP): Declarative Pirahã utterances contain assertions only related directly to the moment of speech, either experienced (i.e., seen, overheard, deduced, etc., as per the range of Pirahã evidentials, as in D. Everett [1986, 289]) by the speaker or as witnessed by someone alive during the lifetime of the speaker.

  D. Everett (2005a) offers a range of arguments for the IEP, based on the empirical points mentioned earlier, as well as (among other things) the culturally important notion of xibipíío, “experiential liminality,” discussed in D. Everett (2008). This word is further evidence that liminality is an important cultural and individual concept in Pirahã. It is used to describe things that go in and out of vision or hearing, from the flickering of a match to the disappearance or appearance of a canoe around a bend in the river.

  Moving from this initial cultural statement to the grammar, and later back to link them, the evidence that Pirahã lacks recursion (also discussed in Everett 2012a) is as follows (though see Perfors et al. [2010] for another type of approach to checking the grammars of languages):

  First, the lack of recursion correctly predicts that factive and epistemic verbs will be absent. This follows because if Pirahã lacks recursion, then there is no way to express factive verbs as independent verbs, since these would require a complement clause. That would in turn require embedding and thus, ceteris paribus (in some analyses), a recursive rule in Pirahã syntax. Pirahã expresses such notions via verbal suffixes, consistent with the “no recursion” hypothesis, not with complement clauses.

  Second, Pirahã has no marker of subordination. This is also predicted by my hypothesis, because if Pirahã lacks recursion, there is no subordination to mark.

  Third, Pirahã has no coordinating disjunctive particles (e.g., or). The absence of explicit markers of disjunction is predicted by my hypothesis, since disjunction entails recursion.

  Fourth, Pirahã has no coordinating conjunctive particle (e.g., and). There is only a more general particle, píaii, which may appear preverbal or sentence final and which means “is thus/simultaneous” (vague meaning), which never works like proper conjunction, but only supplies the information that these two things were simultaneous. Again, this is predicted by my analysis, since coordination also entails recursion.

  Fifth, Pirahã has no syntactic complement clauses. If Pirahã has recursion, where is the unambiguous data? I have claimed that it lacks embedded clauses. Others claim, based on my own data and my own earlier analysis, that it has them (Nevins, Pesetsky, and Rodrigues 2009).6 But although quotatives could be embedding, there are no multiple levels of embedding, which would be expected if Pirahã has recursion.

  Sixth, Pirahã does not allow recursive possession. The point of Pirahã possessives that I have made is not simply that it lacks prenominal possessor recursion, but that it lacks recursion of possessors anywhere in the noun phrase. Nevins, Pesetsky, and Rodrigues (2009) might be correct to suggest that German, like Pirahã, lacks prenominal possessor recursion. But German does have postnominal possessor recursion, while Pirahã has none. The facts are therefore exactly as my analysis predicts them to be.

  Seventh, Pirahã prohibits multiple modifications in the same phrase. As I have discussed above and in Everett (2008) and (2009a), there can at most be one modifier per word. You cannot say in Pirahã “many big dirty Brazil nuts.” You’d need to say, “There are big Brazil nuts. There are many. They are dirty.” This paratactic strategy is predicted by my analysis since multiple adjectives, as in English, would entail recursion. But the paratactic strategy does not.

  Eighth, Pirahã semantics shows no scope from one clause into another (e.g., no “NEG–raising”). Pirahã lacks examples such as John does not believe you left (where not can negat
e believe or left, as in It is not the case that John believes that you left vs. It is the case that John believes that you did not leave). In this example, not can take scope over believe or left. That is not possible without recursion, so my analysis predicts the absence of such scope relations. This is also predicted, correctly, to be impossible in Pirahã under my account, since it would entail recursion.

  Ninth, Pirahã shows no long-distance dependencies except between independent sentences—that is, discourse. The kinds of examples that are standardly adduced for long-distance dependencies include:

  “Who do you think John believes ___ (that Bill saw ___)?”

  “Ann, I think he told me he tried to like ___.”

  We have stated the IEP and rehearsed the evidence against syntactic recursion in Pirahã. It remains now to show how these fit together causally. It turns out that they engage like the teeth in cogs, via evidentiality. Pirahã, like many other languages (see, inter alia, Aikhenvald 2003 and Faller 2007), encodes evidential markers in its verbal morphology as affixes: -híai, “hearsay”; -sibiga, “deduction”; -ha, “complete certainty”; and -0 (zero affix), “assumption of direct knowledge.” The Pirahã IEP, in conjunction with its requirement that evidence be provided for all assertions, produces a narrow domain in which assertions and their constituents need to be warranted. Reminiscent of the potential focus domain developed by Van Valin (2005, 70ff), I label this domain in Pirahã (and presumably some version of this will exist in all languages—at least, those with evidentiality morphology) the potential evidentiality domain (PED); that is, the range of structures where the actual evidentiality domain could in principle fall. The actual domain of evidentiality in a given utterance will be as follows: