How Language Began

by Daniel L. Everett

  As they stabilise by conventionalisation, gestures become sign languages. But sign languages are formed when gestures replace all speech functions. The idea that speech develops from gestures thus makes little sense either functionally or logically. The ‘gesture-first’ theory gets the direction of evolution backwards.

  However, in spite of my overall positive view of McNeill’s reasoning about the absence of gesture-first languages, there seems to be something missing. If he were correct in his additional assertion or speculation that two now-extinct species of hominin had used either a gesture-first or a gesture-only language, and that this is the first stage in the development of modern language, then why would it be so surprising to think that Homo sapiens had also used gesture-first initially? I see no reason to believe that the path to language would have been different for any hominin species. In fact, I doubt seriously that pre-sapiens species of Homo would have followed a different path, since there are significant advantages to vocal vs gesture communication.

  There are still other types of gesture important in human communication. These include iconic gestures, metaphoric gestures and beats. Each reveals a distinct facet of the gesture–speech relationship and its relationship to cognition and culture. There is no need to discuss these here, other than to mention them as evidence for the complexity of the relationship between gestures and speech and that they each contribute to our progress along the semiotic progression.

  Nevertheless, we have yet to see anything in grammar, gestures, or other aspects of language that would lead us to believe that anything needs to be attributed to the genome of Homo sapiens that is specific to language. Cultural learning, statistical learning and individual apperceptional learning complemented with human episodic memory seem up to the task. The literature is rife with claims to the contrary, namely that there are phenomena that can be explained only if language is acquired at least partially based on language-specific biases in the newborn learner.

  It is claimed that there is a spontaneous emergence of language features in communities that are claimed to otherwise lack language, such as Nicaraguan Sign Language and Al-Sayyid Bedouin Sign Language. These languages are purported to come into existence suddenly as a population needs language but otherwise lacks one. The problem with this claim is that all of these languages begin with very simple structures and then become more complex over time with more social interaction. Often it takes at least three generations to develop a complexity roughly equal to better-established languages. But this is just what we would expect if they were not derived from innate knowledge but invented and improved over time as they were learned by subsequent generations. For this reason, even if such examples provided some evidence for an innate predisposition to language, the knowledge would be very limited.‡

  Susan Goldin-Meadow’s work argues that homesigners develop symbols for objects, principles for ordering these and constituents of distinct gestures. She also suggests that these newly minted gestures can fill slots in larger sentence-like structures, structures represented by the kind of tree diagrams we saw earlier. She also discusses a number of other characteristics of homesigns. Her conclusion is that all of this knowledge must be innate or how else could it appear so quickly among a group of speakers?

  But none of these characteristics is specific to language. Indexes and icons, in all probability early forms of gestures, are used in one way or another by several species. There is no reason to believe that homesigns cannot be learned easily by humans. In fact, on one interpretation, that is all that Goldin-Meadow’s results on symbols show us, namely that children readily learn and adopt symbols. The object is a form with a meaning. As the child learns the object and desires to communicate, then – perhaps the most striking characteristic of our species – whether due to an interactional instinct or an emotional urge, the child will represent the object and the meaning of the object in the particular culture comes along for the ride. Children participate in their parents’ lives, even if without language, and try to communicate, as Helen Keller’s remarkable odyssey shows us. With an ability to see or hear or feel, the child can receive input from the environment, from its caregivers, and in fact will do so with most caregivers and in most environments. Learning the use of the object, the salience of the object to its parents and environment, it is unsurprising that children communicate about objects, as most other species (at least mammal species) do. Whole objects, as perceivable in a particular space in time, are most salient and learned relatively easily by dogs, humans and other creatures. Humans try to represent their objects, unlike other animals, because humans strive to communicate.

  The fact that some features of the objects stand out to children is likewise unsurprising, though the particular reason that shape and size win out over many other features, if Goldin-Meadow is correct, is unclear. She ascribes it to the child’s native endowment. But I would suggest looking first at the way that objects are used, presented, structured and valued in the examples of the child’s caregivers. Furniture, dishes, houses, tools and so on are far more easily arranged and far more prevalent in the environment of US caregivers’ salient objects than other features. At least that could be tested and there is no suggestion that any such tests were contemplated.

  With regard to the claim that homesigners’ speech is organised hierarchically, there are two caveats. First, structure vs simple juxtapositioning of words like beads on a string are very difficult to distinguish in practice. Are three objects related as in diagram (a) or (b) of Figure 34? Either might be the case, and the reasons for choosing one over the other are highly theoretical. For example, in Pirahã utterances, we might say, ‘The man is here. He is tall.’ Or, ‘I spoke. You are coming.’ And these could be interpreted as ‘The man who is tall is here,’ or ‘I said that you are coming.’ But the analysis is quite possibly much simpler, with the syntax lacking hierarchical structure.

  In none of Goldin-Meadow’s examples purporting to show hierarchical structure in homesigners’ utterances is there convincing evidence for structures like (b). The second caveat is that some configurations provide a natural solution to presenting information, independent of language, and thus if one finds them in some languages, this is not evidence that there is an innate linguistic bias. Again, if a structure provides a useful solution to communicating information then nothing else needs to be said about why it is found in languages around the world. As information demands grow due to increasing societal complexity, hierarchy is the most efficient solution to information organisation, across many domains. Computers, atoms, universes and many other complex objects of nature are organised this way. It is a naturally occurring and observable solution. In fact, for any action that involves ordering, such that ‘you must do x before you do y’, there is structure. Such solutions are used in automobiles, canine behaviours and computer filing systems. There is absolutely nothing special about them when they also appear in language.

  The ordering that homesigners are claimed to impose on their structures is mundane. First, they have no alternative but to put their symbols in some order. And since the main ingredients of any utterance are the thing being reported about and what happened to it, sentences tend to be organised in terms of topic and comment. The topic of a sentence (as opposed to the topic of a larger story) is the old information that is either being discussed or the information that the speaker assumes that the hearer will know. The comment is the new information about the topic. Very often, but not always, the topic is the same as the subject and the comment is the same as the predicate or verb phrase:

  Figure 34: ‘The big boy’

  ‘John is a nice guy.’

  Here the old information is ‘John’. The speaker offers nothing more than a name and assumes that the hearer knows who is being mentioned. The new information is ‘is a nice guy’. In other words, the speaker is saying something that he or she thinks might be new information for the hearer about ‘John’. A paraphrase might be something like, ‘I k
now you know John, but you might not know that he is a nice guy.’

  The topic in most languages of the world precedes the comment. In other words, languages prefer to begin their sentences with shared or old information before giving new information. This order may aid our short-term memory. Within the comment, where the new information is placed, a large number of languages prefer to place the object before the verb. So if someone somewhere is eating a fruit, this can either be described as John fruit eats (majority of languages) or John eats fruit (English and many other languages). In the Subject Object Verb group are languages like German, Japanese and Pirahã. Languages like French and English, on the other hand, use the order Subject Verb Object. In fact, English used to belong to the former group (it is closely related to German), but after the Norman invasion of 1066 English switched to the French order of Subject Verb Object.

  It is claimed occasionally that word orders may be heavily influenced by different communication strategies for dealing with problems in the communicative process. One problem is noise corrupting the signal. As people talk they can be distracted by background noise, children tugging on them, animals in the distance and new people arriving during the conversation. Therefore, linguistic strategies must be able to overcome the effect of ‘noise’ on the hearer’s ability to perceive the speaker’s words. These researchers claim that this is why Subject Object Verb order is most common – it helps avoid confusing the subject and the object and avoids confusing which is topic and which is comment, by having the phrase before the verb ‘to be’ in the comment section. On the other hand, since there are thousands of languages that use different orders, such as Subject Verb Object, Object Verb Subject, Object Subject Verb, Verb Subject Object and Verb Object Subject, no one of these orders is far superior to the others. The order that a language adopts results from cultural pressures that affect the history of a particular society. This is why that English changed in its history from Subject Object Verb to Subject Verb Object, as a result of the influence of the French spoken by the Norman conquerors.

  Homesigning follows largely the same principles. The basic problem that interlocutors must solve is keeping novel information distinct from already known information. Thus the fact that homesigners follow a common order is simply not a big deal. It is just the way that communication works most efficiently.

  Nor should it be surprising that once a basic sequence of words is made conventional, it is easier for speakers to remain consistent with that choice than to use different strategies in different parts of the language. Therefore, if your language chooses Subject + Object + Verb, it will, ceteris paribus, also choose to put possessors before nouns (as in ‘John’s book’ vs ‘Book John’s’). This is because, just as the verb is the semantic core or nucleus of the sentence, the possessed noun is the nucleus of its own noun phrase. The general rule in such a language would be ‘put the nucleus last’. Yet because this decision is based only on efficiency, and because languages often do things inefficiently, this type of rule is frequently violated all over the world and we observe a good deal of variation in word order principles in different languages for this reason. Nor is there anything specific in the genetic endowment of Homo sapiens that would be expected to be linked to information structure. Topic-comment is a natural communicative arrangement. But many who talk about the implications of homesigns, supposedly revealing innate linguistic capacity, neglect to discuss the way that information should ordered for most effective communication. Failure to refer to this, however, misses the most natural explanation for the facts about homesigns and appeals instead to the highly implausible idea that language is innate.

  Homesigning clearly illustrates the desire of all members of sapiens to communicate. And it shows solutions to this communication problem can be straightforward and easy to ‘invent’ and understand. Unfortunately for the claims of homesign researchers that language is innate, there is not even a convincing analysis of the facts of the grammar of these languages. But evidence suggests that gestures are sufficiently motivated by communicational needs that sign languages and gestures of all kinds simply emerge because they are so useful. Utility explains ubiquity.

  * To say that this also means that we use gestures without learning them, as McNeill seems to suggest, is unwarranted.

  † As stated, many researchers have speculated that gestures might have preceded speech in the evolution of human language. It is possible that gestures were first in some way, preceding language proper. Though I should have first expected yells along with gestures. Even McNeill does not disagree entirely with this position.

  ‡ More importantly, what marks the work of Goldin-Meadow and many others is what I consider to be an over-charitable interpretation of the linguistic aspects of the signs and a less charitable view of the cultural input the child receives as well as the nature of the task the child is facing. Absent a serious consideration of either the task or the input, such claims are severely weakened.

  11

  Just Good Enough

  Le mieux est l’ennemi du bien.

  Voltaire

  Not everything worth doing is worth doing well.

  Kenneth L. Pike

  NOBEL-PRIZE-WINNING economist Herbert Simon introduced the concept of ‘satisficing’ into the science of problem solving. His point was that the solutions preferred by business, throughout human endeavours and by the mind itself are not usually the best ones but the good-enough ones – the ones that ‘satisfice’ the need rather than perfectly satisfy it. The same principle applies to evolution. With respect to language, this means that human grammars and sound systems are not required to be optimal – in fact, they never are. Language gets the job done just well enough, never perfectly. Herbert Simon has echoed Voltaire, claiming: ‘The good is the eternal enemy of the best.’

  It is this aspect of language that so strongly supports the idea that it is an ancient invention, tinkered with throughout human history. There are basic strategies of communication that usually work, but frequently fail – such as when one omits information that is assumed to be shared by one’s interlocutor. Or communication can fail due to the failure to remember or the very lack of a word or, more importantly, a sentence, to translate a concept in one culture or person or language into one’s native language. Human languages leak. They are not mathematical, perfectly logical codes.

  If someone yells, ‘Stop the car at the stop sign!’ they assume that the person they are instructing knows what ‘stop’ means, how to drive a ‘car’, what a stop sign is and what it means to stop at a stop sign as opposed to stopping on the edge of a cliff. (One may stop at a sign with the front wheels slightly beyond the sign. That wouldn’t work out well at a cliff.) Those assumptions are just built into the choice of the words. I imagine that most readers of this book, as members of a largely uniform driving culture, know that one has not ‘stopped at the stop sign’ if they stop 200 yards in front of it or ten feet in front of it or even if it is level with the rear seats. Proper stoppage at the signage brings the vehicle to a complete stop roughly 1–5 feet before the front of the car passes the stop sign. This is part cultural knowledge, part lexical (word) knowledge. ‘Common’ sense is just experience and acquired cultural information.

  For a more involved example, consider a 2016 opinion piece published in the New York Times on the purported hacking of a presidential candidate’s emails by Russia. Most of the values embedded in this opinion piece are easy to discover, though some are worth pointing out for discussion because one often reads over such unspoken information without noticing it. A few comments are offered, in square brackets in order to reveal unspoken cultural and contextual information and values that would be expected to be implicitly understood by every culturally North American reader. Imagine the potential effect of seeing such editorials in a newspaper one takes on a regular basis over time, reading them passively, perhaps, while sipping coffee at the breakfast table or on the train or bus to work. Their effect could be larg
ely subliminal, reinforced with every story that assumes similar values, which is the case of particular newspapers, with the New York Times usually speaking for liberalism.