Biology & Philosophy

, Volume 23, Issue 3, pp 347–361

The trait of human language: lessons from the canal boat children of England


    • Speech-Language-Hearing Sciences, Lehman CollegeCity University of New York

DOI: 10.1007/s10539-007-9104-8

Cite this article as:
Locke, J.L. Biol Philos (2008) 23: 347. doi:10.1007/s10539-007-9104-8


To fully understand human language, an evolved trait that develops in the young without formal instruction, it must be possible to observe language that has not been influenced by instruction. But in modern societies, much of the language that is used, and most of the language that is measured, is confounded by literacy and academic training. This diverts empirical attention from natural habits of speech, causing theorists to miss critical features of linguistic practice. To dramatize this point, I examine data from a special population––the canal boat children of early twentieth century England––whose language developed without academic influence, but was evaluated using instruments designed primarily for academic use. These data, taken together with related research, suggest that formal instruction can convert language from a purely biological trait that was selected, to a talent that was instructed, while altering the users of language themselves. I then review research indicating that formal instruction can also mask or distort inter-sexual differences in the social applications of language, a significant handicap to evolutionary theorizing. I conclude that if biological theories of language are to succeed, they must explain the spontaneous speaking practices of naturally behaving individuals.




Language is a species-specific trait that evolved in humans uniquely, and develops in the young universally. A strong biological endowment was suspected a half-century ago when it was noted that infants acquire linguistic material fairly rapidly, considering the inadequacy of ambient models and the rarity of teaching or corrective feedback. Later, it was noted that infants invent components of linguistic structure in certain cases, as when infants transform pidgins into creoles, and deaf infants normalize manual languages that have been awkwardly modeled by non-natively signing parents. By the 1990s, it was possible to argue that “the capacity for language is part of human biology, not human culture” (Pinker 1995, p. 264), and supporting evidence has continued to accumulate in the interim (Locke in press).

Considering the broad appeal of biological explanations, it is paradoxical that contemporary perspectives on language, including its evolution and development, have been heavily influenced by cultural factors, ones that operate so subtly as to escape notice. I am not thinking here about the quotidian fact that infants learn the language of their culture. What I do refer to is a cultural invention and practice––academic training in literacy and language––that has spread in the last 1% or so of linguistic history. This training makes language appear less “biological” than it truly is, with serious theoretical consequences.

My purpose here is first to bring to light some unusually clear evidence of cultural effects. It suggests that the language and linguistic practices of modern speakers are altered by formal instruction. With few exceptions, all who are able to learn spoken language and attend school get this training before they receive any systematic assessments of their linguistic ability. I then turn my attention to standardized tests of language, and to problems faced by unsuspecting theoreticians who attempt to explain the evolution, development, and nature of language based on data that were influenced by formal instruction.

Selection or instruction?

The problem begins with the fact that language takes many years to develop, from infancy to sexual maturity (Locke and Bogin 2006). The mechanisms responsible for basic linguistic systems––phonology, morphology, syntax, and the lexicon––are fired up largely by familial stimulation, and operate with some degree of efficiency by the end of infancy, at about 3 years (Locke 1993). However, in modern societies a protracted course of training in language and literacy is undertaken in mid-childhood, when school begins, with continuing instruction in juvenility and the better part of adolescence.

Before I identify the problems that these practices pose for language, it may help to consider a different trait––music. No one doubts that there is a universal disposition to engage in some form of music making, from singing and chanting to drumming and other forms of instrumental music (Harwood 1976; Merker 2000). Indeed, some have argued that this trait may have contributed to the evolution of language (Masataka 2007; Mithen 2005; Wallin et al. 2000). It is customary in many cultures for parents or institutions to offer, even mandate, musical training in the young. In cases where this proves successful, it is generally concluded that the key ingredients were some sort of native ability––the product of selection––as well as instruction, and a great deal of practice.

Normally, we think of musical ability as a “talent.” Simonton (1999) defined talent as “any innate capacity that enables an individual to display exceptionally high performance in a domain that requires special skills and training.” (p. 436). It is, in other words, a product of the combined forces of selection and instruction. As examples, Simonton offered artistic creativity, concert performance, and competitive chess. Of course, no linguists would assert that the trait of human language requires instruction, but my claim here is that language, as it has come to be defined, reflects it.

In traditional societies, there is a tendency to think of language as a form of action––as a dynamic, socially interactive behavior that co-occurs with gestures and bodily activity, is heavily situational, and therefore is free to be relatively inexplicit (Linell 2005). When asked, members of these societies seem to have difficulty thinking about the concept of “word” (Goody 1977). They also find it strange to be asked what a word “means,” possibly because words are never taken out of context (Malinowski 1923), and cannot be “looked up” in dictionaries. In modern societies, where there is considerable emphasis on literacy, and laws that require formal instruction over a period of 10 years if not longer, there is a tendency to regard language as a body of knowledge that lies outside the individual (Goody 1977; Linell 2005; Olson 1994; Peek 1981).

Language training also affects the individuals themselves, including the way their brains and minds process linguistic material. In Castro-Caldas et al. (1998), subjects who were literate or culturally illiterate were enrolled in a lexical decision task. Concurrent brain scans indicated that in literate adults there was stronger activation of the right frontal opercular-anterior insular region, left anterior cingulate, left lentiform nucleus and anterior thalamus and hypothalamus than there was in the illiterate subjects. In a separate repetition task, the illiterate subjects substituted real words for nonsense constructions 25 times more often than the literate subjects (Castro-Caldas et al. 1998; also see Ostrosky-Solís et al. 2004).

There is an additional problem, however, one that extends beyond the influence of literacy instruction and experience. For this new instructed material forms a significant portion of the language that is tested. This makes a great deal of sense to the purchasers and consumers––the educators and the public who want to know how well programs of instruction are working––but the taught language coalesces with acquisitions that occur during, and under the aegis of, native language development. This potentially inflates estimates of our species natural linguistic capacity. Literacy instruction also distorts our concept of the language trait, for it gives the impression that spoken language, that is, words and the rules for using them, are actually a static form of knowledge that appears on printed pages or in the mind.

It is also the case that most standard measures of “language” are, in fact, evaluations of the ability to use linguistic knowledge to solve cognitive problems, ones that are invariably set in an asocial context.1. The relevant skills are usually classified as “metalinguistic” (see reviews in Hines 1990; Kimura 2002; Maccoby and Jacklin 1974). Individuals with metalinguistic ability are able “to deliberately reflect on and manipulate the structural features of spoken language.” (Tunmer and Cole 1991, p. 387). If untrained in literacy, normally speaking adults tend to perform poorly on these tests (Adrián et al. 1995; Adrián and Morais 1995; Gombert 1994; Morais 2001; Morais et al. 1979; Navas 2004; Read et al. 1986). They have difficulty counting the sounds in a word or even saying whether a word contains a particular sound––skills that phonically trained seven- or eight-year-olds do with little effort. By contrast, individuals who receive academic training usually perform well on these tasks. They are good, according to Tunmer and Cole (1991), at “treating the language system itself as an object of thought.” (p. 387, my italics).

This “objectification” of language has caused concern for decades. Over 80 years ago, it worried Jespersen, who noted that exposure to printed language seemed to convert the constituents of speech into a collection of “things or natural objects with an existence of their own” (Jespersen 1924, p. 17). Objectification troubled Lakoff and Johnson (1980), who observed that experience with printed language can cause words and sentences to resemble “objects that have properties in and of themselves and stand in fixed relationships to one another, independently of any person who speaks them or understands them.” (p. 204). The view of language as an object concerned Linell (2005), too, who commented that when linguists consult their intuitions on matters of linguistic practice, which they do as a matter of course, they encounter “an inventory of forms, and rules for generating forms” (p. 4). Recently Hermann (2007) called the language-as-thing metaphor a “systematically misleading expression.” He suggested that it is time for a better metaphor, one that captures the fleeting and subjective nature of speech.

Clearly, it is time to re-evaluate what is entailed by human language, in its purest (uninstructed) form. One way to look at this trait would be to examine the behavior of hunter-gatherers, or other members of traditional societies. Lee (1979) once suggested that the lifestyle of the !Kung people of southwest Africa resembles that of groups that lived more than a 1000 years earlier. Their way of living does share more properties with the environment of evolutionary adaptedness than do modern societies. But it can be difficult to conduct studies of the language spoken in such societies. Cultures that have no schools are also unlikely to have a clear sense of their own lexicon and grammar, and may not be motivated to investigate the acquisition of this material by the young. If so, the research will have to be carried out by outsiders, who are likely to be linguistically and culturally unsophisticated. These investigators may be forced to accept advice from informants of unknown reliability, and to use ad hoc procedures of evaluation. This can lead to disastrous results (e.g., Straight 1976).

It is obvious that to explain the evolved trait of language, questions must be asked about the linguistic knowledge and skills of those who receive the requisite social experience in infancy but are not taught language in childhood and juvenility. For, it is presumably this more natural command of language that evolved in the species. Fortunately there is relevant evidence in an unlikely place––an official document that unceremoniously bobbed to the surface nearly a century ago, when it received no more than passing notice from a handful of bureaucrats and pedagogues before sinking back into government files.

The focus of the report was an extremely unusual group of children. Their rarity was not associated with anything physical, although they may have been somewhat healthier than their peers. The children were not the product of exceptional rearing practices, although they enjoyed greater than usual contact with their parents and siblings. They were not reared in the wild, although much of their childhood was spent out of doors. What makes the children interesting is the fact that they rarely came face to face with anyone outside of their immediate family; and they almost never went to school.

By itself, this pattern of provision and privation is likely to intrigue students of human development, but the situation becomes even more interesting. For in the larger society, there was also a level of concern for these socially reared isolates that would bring about their evaluation, and a level of educational technology and linguistic knowledge that would supply the standardized instruments to do it. I speak of the children who were born on, and were destined to spend their lives on, the canal boats of early twentieth century England.

In the next section of the paper, I look at this rather quaint population, the historical records of which are in limited distribution, asking what the children can tell us about the biology and culture of human language. In the final section, I point to problems for evolutionary and developmental theory that arise when data are acquired from linguistically trained individuals who have been asked to make judgments about language, or whose knowledge is evaluated with paper and pencil tests.

The canal boat children of England

Anyone who tours England today is likely to see the canals that meander through her picturesque farms and villages. For two centuries, these canals were navigated by horse-drawn Narrow Boats.2 These boats remained the predominant means of shipping goods until the expansion of the rail system in 1840, but there were other threats besides the trains. Someone had to lead the horse, operate the locks, assist with loading and unloading, steer, and physically power the boat––when passing under a bridge––by pushing against the sides of the canal. Clearly, the captain could not do all these things. Other able bodies were required, and the obvious choices were the captain’s wife and all but his youngest children. This caused two concerns in the larger society. One was related to child labor, which was regulated by law. The second was education, which became compulsory in 1870. Still, the employment of canal boat children was not that easy to stop. Consumers needed supplies, and canal boats remained the only way to get them. Even if the captain could afford to hire a helper, family members needed to be together––captains rarely went ashore, and many had no other residence––and the cabins, though small, were usually sufficient to accommodate them. So canal boating continued, not to end, as a profession, until the middle of the twentieth century.

Life on the boats

Canal boat families were together 24 hours a day. Since the captain was also the husband and father, the structure of family life was predictably patriarchal. However, the mother typically “ran a tight ship” of her own, enforcing a strict code of morality and discipline. Observers found canal boat children to be well behaved, in general, and to be in better health than their terrestrial counterparts––whether due to the “open air,” as some supposed, or the lack of contact with other children.

It could not be said, however, that canal boat children had anything approaching a normal social life. The boats were constantly on the move, rarely stopping for more than a few days at a time to load or unload. Even when they docked, the children did not make up for lost time by plunging into village life. The local children were strangers, with preexisting social arrangements of their own, and many of the canal boat children were reticent; reports indicate that they did not readily mix with villagers, although they occasionally interacted with other canal boat children.

When they were around 6 years of age and their peers were entering school, the canal boat children remained behind, commencing their duties on the boats, and this put an entirely different cast upon their childhood. When they went to school, which was rare, teachers found them to be unusually serious. “They didn’t know how to play,” said one teacher.


But it would be a mistake to suggest that canal boat children received any sort of formal education. One authority estimated that they might have gone to class on 5% of the days that school was in session. For even when the boat was in port, the children were frequently kept on deck, waiting for departure. Few could have been “boat-schooled” either, for the parents were largely illiterate. Many had grown up on canal boats themselves, and took little interest in their children’s schooling. Some went so far as to block it. One official wrote that he had “known parents to assure the school attendance officer in Sandbach that the children were attending school in Runcorn, and then equally successfully convince the officer at Runcorn that the children were being taught at Sandbach. In this way,” the man wrote, “I have known a boy to reach the age of thirteen without having been inside a school, and to be a perfect young savage.”

The schools were reluctant to deal with these “savages.” “It is very difficult to place a child of 12 or 13 who can neither read or write ... with children of a similar age,” teachers told a Canal Boat Officer with the National Society for the Prevention of Cruelty to Children, and “it is somewhat difficult to place the same children among the infants.” Some parents complained that even when their children attended school they learned nothing, because “teachers placed them in a back row and gave them a picture book to look at,” adding that this was “all the education they gets.” As a result of all these factors, many boat children were still receiving little or no formal education as recently as 1939 (Freer 1992).

The report

Educational concerns such as these eventually led to the formation of a committee, appointed by the Minister of Health in 1921 and chaired by Neville Chamberlain. Its purpose was to “inquire into the practice of living-in on Canal Boats in England and Wales and to report whether any alteration in the practice is desirable.” The committee found that canal boat people were not at all unhappy with their situation. Initially, it had been thought, “the narrowness of the quarters occupied, the total absence of any sanitary conveniences upon the boats and the enforced proximity of the sexes would have led to laxity in the standards of morality and decency.” But the committee reflected on the fact that “most of those employed on canal boats have been accustomed to it from early youth, if not from birth, and that they have been brought up in traditions handed down to them by generations of canal boat people, with the result that their conventions are different from those of shore dwellers, and that circumstances which might draw attention among the latter pass unnoticed by them.” From interviews, the committee learned that when it came to health, cleanliness, morality, feeding and clothing, the boat children were “fully equal, if not superior, to town dwellers of a similar class.” (Report of the departmental committee 1921, p. 4)

Language and cognition

The Chamberlain committee’s findings were based on interviews and testimony, but they had no way to assess the intellectual development of the children directly. This was remedied in a second study, carried out by a school inspector, Hugh Gordon, under the auspices of the Board of Education (Gordon 1923). Gordon examined the cognitive abilities of several groups of children who were suspected of under-attendance at school. These included canal boat children as well as two other under-schooled populations, one with physical handicaps, the other a highly nomadic group of gipsy children. The rate of school attendance for physically defective and gipsy children was 48 and 35%, respectively. A group of mentally “backward” children was also included for comparison purposes.

We are most interested, of course, in the canal boat sample, which included 76 four- to fourteen-year-old children. Gordon gave these children what was then an innovative test of aptitude––the Stanford Revision and Extension Tests, based on a revised version of the Binet–Simon Scale (Terman 1916). He found the canal boat children to be interested, obedient, talkative, and eager to do their best on the strange new tests.

Gordon’s report contains average scores on all of the Stanford subtests, but we will limit our attention here to performance on the oral subtests that bear some relation to language: comprehension, vocabulary, sentence construction, and verbal fluency. The comprehension tests were administered at 4, 6, 8, and 10 years of age. Instructions required examiners to ask the children three questions, e.g., What must you do when you are sleepy? Questions were to be repeated as often as necessary, with about twenty seconds allotted for each answer.

Figure 1 shows the average test scores––ratios of mental to chronological age, multiplied by 100 (100 being the average score of normally schooled children). From the steady decline of test scores in the Figure, it initially appears that the canal boat children lost comprehension ability as they aged. In actuality, canal boat children would have gained in this area, purely as a result of maturation and continued familial interaction, but they obviously did so at a slower rate than other children and juveniles. In other words, since the tests get harder at each level, Fig. 1 is actually an indirect indication of the greater improvement in comprehension by the schoolchildren who formed the norm-reference group for the Stanford test.
Fig. 1

Comprehension scores of canal boat children at 4, 6, 8, and 10 years of age

Where “culture” is allowed to operate at all, it can only do so as the culture that it is––a particular culture. In that sense, Fig. 1 is also an indictment of culture sensitive tests. Consider the amazement experienced by Gordon’s testers when, in giving the vocabulary test, the word “muzzle” was defined by canal boat children as a bit placed on the mouth of horses, not, as the test required, a strap placed on the mouths of dogs. One gets the rather clear impression that answers of this sort were scored “incorrect,” even though the definition was both appropriate and correct in light of the children’s real-world experience. The comprehension portion of the Stanford tests also asked questions that would have made little sense to the canal boat children. For example, as the tests progressed, they increasingly invited, and possibly rewarded, information that could have been acquired merely by being in school, apart from the classroom instruction itself (e.g., What’s the thing for you to do when you notice on your way to school that you are in danger of being tardy?).

Vocabulary tests were administered orally at 8, 10, and 12 years, when the children were asked to define, 20, 30, or 40 words, respectively (e.g., orange, bonfire, roar). Results indicated that at 8 years, when school would have been underway for 2 years, vocabulary scores were already depressed: the canal boat children scored just 6.5. By comparison, the gypsy and the backward children achieved scores of 48.7 and 45.2. At 10 and 12 years, the canal boat children’s vocabulary scores fell to 1.3 and zero, respectively. As with the comprehension scores, this means not that the canal boat children were losing vocabulary, but that their school-going peers were learning new words at a more rapid rate.3 This can hardly be surprising, for research now indicates that some vocabulary growth occurs purely as a result of chance encounters with new words when reading (Nagy and Anderson 1984; Nagy et al. 1985).

Of course, there are, as we have seen, other reasons why failure to attend school would depress vocabulary development, e.g., lack of exposure to a variety of people. Clearly, overhearing people talk facilitates the acquisition of language (Akhtar 2005). But it is unlikely that this could have produced such low scores on tasks requiring children to construct a sentence, or that this would account for the severity of deficits in the other areas of language.

In the sentence construction task, administered at 8 years, children were asked to make sentences out of three words that were supplied by the examiner (boy, ball, river; work, money, men; desert, rivers, lakes). The canal boat children achieved a score of 6.5 on this task, a fourth of the level achieved by backward and gypsy children. In a verbal fluency task administered at 10 years, children were asked to say as many words as possible in 3 min without repeating themselves or producing sentences. Inasmuch as the canal boat children approached zero on the vocabulary test administered at the same age, it would not be surprising if they performed poorly on this task, and their performance was, in fact, a dismal 5.2. But there may have been other (task) factors at work, for the gipsy and backward groups displayed no more fluency than the canal boat children, in spite of their considerably larger vocabularies.

Nonverbal abilities

Gordon found that up to 6 years or so, canal boat children scored between 90 and 100 on the intelligence test. After 6, when their peers would be in school, the “intelligence” of these children––that is their scores on the Stanford test––began to fall below normal, and continued to decline with increasing age. But much of this decline was due to the verbal tasks, a fact noted by a second investigator, Frances Gaw. She pursued this issue in a follow-up study, giving a 14-item nonverbal test to 34 canal boat children between the ages of 6 and 13 (Gaw 1925). Items included picture completion and assembly, pattern matching, maze tracing, and other visual tasks. A total of 27 of Gaw’s subjects had participated in Gordon’s study, enabling comparisons between verbal and nonverbal items. Although in Gordon’s study the 27 shared subjects had fallen into the “mentally deficient” category, based on a combination of verbal and nonverbal items, they were close to normal on Gaw’s performance tasks. Thus, the standardized test used by Gordon––an early version of the psychometric tests used today––was particularly sensitive to trained language.

In a discussion of his findings on canal boat children, Gordon commented on a “marked decrease in ‘intelligence’ with an increase of age,” a trend that “suggests very convincingly that the low average ‘intelligence’ of these children is not due to heredity.” (Gordon 1923, p. 44) What it was actually due to had already been forecast by Cyril Burt, who was quoted in the opening pages of the report. “To achieve distinction … in a trial so academic as the Binet–Simon tests,” Burt had written, “experience must be heavily supplemented … by the artificial aids supplied by a civilized society.” (Burt 1922, p. 183, italics mine) There was one “artificial aid” that Burt had in mind. It was school. This is precisely what Gordon’s research had shown, since he had obtained positive correlations between attendance and intelligence in children whose capacity to learn was presumptively normal.

Risks for theorists

The canal boat children increasingly fell behind school-based normative samples as they matured, not because they were given too few opportunities to converse, but because they received little if any exposure to the linguistic materials and practices that are appraised on the academically oriented intelligence tests that they received. What happens to theories of language when they are based on evaluations that reward academic training in language and literacy? The answer is that theories take on unreal properties, for such procedures give us no indication of natural––one is tempted to say ecologically valid––applications of the faculty of language. From paper and pencil tests of anagrams and antonyms we do not, and cannot, learn how often individuals speak, nor do we find out in which situations they do so, with what degree of complexity, for what purposes, and to what effect. We cannot learn from verbal fluency and sentence completion tests who knows how to converse and to make sense, nor do such measures tell us how often one interrupts or yields the floor, or how well one can generate and interpret sentences in real life situations, or makes small talk. No existing test tells us if a person can persuade, tease, joke, insult, gossip, or speak sarcastically. We do not, then, find out anything about a person’s habitual uses of speech, or ability to use speech, in everyday life.

If, on the other hand, individuals are merely good talkers––merely entertaining or persuasive or eloquent––they will be unable to pass the tests, or even to sit for them. For this, speakers must also be able to read and write, and to access their tacit language knowledge. Selection must be augmented by instruction. Unfortunately, the data produced by psychometric tests have not been seen as irrelevant, or too irrelevant to model. As a consequence, theories of language may be theories about something else. If the scores on these tests were correlated with real-life measures, the harm might not be too serious, but there are few if any correlations. In fact, highly skilled speakers––those who excel in verbal jokes and games, and in the cut and thrust of rhetorical contests––occasionally appear to do worse than others on academic exercises (Dandy 1991).4


If we are to explain the evolution of language, there must be a role for natural or sexual selection. Throughout the animal kingdom, there is a strong disposition for males to display and females to choose. If this pattern also applied to ancestral humans, and left behind perceptible traces, we should expect to find evidence that males use speech to display, and that females choose males based on the quality of these displays. The expectation, on this logic, is that males would be superior in the linguistic skills that are displayed. However, there have been many reports of female superiority in language (see reviews in Halpern 1986; Maccoby 1966; Maccoby and Jacklin 1974). In fact, these differences are generally slight. Hyde and Linn (1988) called them “negligible.” But even if the differences had been substantial, it would be difficult to interpret such findings, for most of the tests were administered in writing. Written tests produce scores that are largely irrelevant to spoken language skills, but they also lead to faulty conclusions. Females tend to score higher than males in the area of reading, writing, and spelling (Allred 1999; Horne 2007; Lindsay and Desforges 1999). Thus, even if males excelled in spoken language––the evolved trait––the mode of testing could well produce the appearance of female superiority.

This can, and arguably has, caused serious theoretical problems. Recently, two scholars sought support for a sexual selection hypothesis for language, but they had read literature reviews that seemed to suggest that males perform worse than females on a variety of verbal tests. This had little effect on one of the scholars, Miller (2000), who went ahead anyway, asserting that males are verbally dominant, based on the fact that they write more books, give more lectures, and tend to dominate in mixed-sex interactions. There is, Miller said, a “greater male motivation to produce public verbal displays” (p. 376). Anthropological and other literatures contain a great deal of evidence to support Miller’s assumption. Males do evince a greater disposition to engage in public displays of speech and language (Locke 2001; Locke and Bogin 2006), and may be more likely to succeed in such rituals (Friedley and Manchester 1985, 1987; Manchester and Friedley 2003; Stepp and Gardner 2001).

The other scholar who read a review of the test-based studies, Fitch (2004), was perceptibly discouraged. “If language originated in sexually selected displays,” he wrote, “we would ... expect human males to have more highly developed linguistic capacities than females. In fact, just the opposite is the case: all available data suggests, where they differ, female linguistic abilities exceed those of males.”5 Fitch moved on to propose an account that made no reference to sexual selection. The point is not that his alternative proposal was of no interest, because it is (Locke 2006; Locke and Bogin 2006). Rather, it’s that Fitch might have come up with something different if the available reviews were based on more ecologically valid populations and modes of measurement.


For at least 75 years, sporadic reports have suggested that girls acquire language faster than boys in early infancy (Berglund et al. 2005; Fenson et al. 1994; Huttenlocher et al. 1991; Lutchmaya et al. 2002; McCarthy 1930; Morisset et al. 1995). It has never been easy to interpret these reports since the advantages were slight, may have been associated with sex differences in rate of overall development, and usually disappeared a few years later (e.g., Baron-Cohen et al. 2004).

The language tests that are given around the time of school enrollment typically treat language as a “container,” asking if the standard areas of language––phonology, morphology, syntax, and vocabulary––are filled to the age-appropriate level. Scores on these tests tend to peak at about 6 years (Shriberg et al. 1999; Tomblin et al. 1997), and show no strong sex effects one way or the other. But we should wonder what kind of language measure allows young people to peak at 6 years. Surely no test could do so if it included items relating to pragmatics and performance (Nippold 1998). These include joking, arguing, teasing, gossiping, chatting up, putting down, and a variety of other verbal-social skills, some as ordinary as conversing. These kinds of verbal actions are difficult if not impossible to test (Adams 2002; Tomblin et al. 1996), and lacking firm connections to academic success, teachers may see little reason why they should care if adolescents do poorly in these areas (Reed and Spicer 2003). However, as I have argued elsewhere, such performance-oriented applications of language would have been likely targets for selection (Locke and Bogin 2006) much as they figure into intra-sexual competition and mate selection today (Bale et al. 2006; Bressler and Balshine 2006; Bressler et al. 2006; Cunningham 1989; Lundy et al. 1998).

When one looks at qualitative evidence––ways of using language that are typically acquired outside of academic contexts––one finds that males, from an early age, are more competitive verbally, using more commands, boasts, and threats than females, who are more likely to use indirect forms of language, to be more cooperative, and more responsive to others (e.g., Leaper, 1991; Sachs 1987; for reviews, see Maccoby 1990). Some of these differences are already evident in the second or third year––well before school––when males speak more assertively than girls, and use less affiliative speech (Leaper and Smith 2004). Such variations have been attributed to the fact that boys and girls differentially accommodate to the speech of men and women, who speak differently (e.g., Tannen 1999). But this requires a (cultural) account of gendered speech at the adult level, and psycholinguistic evidence that divergent patterns of exposure (and imitation) are responsible for the children’s own differences.

I suggest another approach. It starts with the documented tendency of males to perform and to compete verbally, as early as infancy and early childhood (Gossen 1976; Wyatt 1995, 1999), with increases in juvenility, when joking and riddling become more prominent (Dundes et al. 1970; McGhee 1979; Shultz and Horibe 1974). In adolescence, males are more openly competitive than ever, but verbal aggression in females tends to occur covertly, through the agency of friends (Björkqvist et al. 1992a, b). Precisely why these differences occur remains to be established, but those who invoke cultural explanations should take note: there are sex differences in aggression in the other primates (Wrangham and Peterson 1996), and there are congruent findings from human endocrinology. Abnormally high levels of prenatal androgen are associated with increased masculine play behavior in girls (Berenbaum and Hines 1992), and there are indirect links, in men, between testosterone and verbal competition (Dabbs et al. 1998).

Concluding remarks

Evolutionary theories require a reasonably clear sense of the traits that evolved. If it is important to know about the environment of evolutionary adaptedness, it is equally important, by analogy, to entertain sensible impressions of the environment of linguistic evolution. This is not the prevalent environment in modern societies. In this sense, the lesson of the canal boat children is in the questions they raise about methodology, from population selection to the means of observation and data acquisition on which we are willing to base evolutionary claims. Many of the findings that have been reported in the last half-century, and are reported still, distract us from the less easily quantified or “measured” behaviors that are characteristic of traditional societies, and of people in modern societies who use speech naturally in social contexts. But it is these behaviors that biological theories must explain. The need to address interactions between linguistic evolution and development has become more important than ever (Locke in press), but the “evo-devo” framework requires behaviors that are in suitable form. These cannot have been influenced significantly, and to an unknown degree, by “the artificial aids supplied by a civilized society.”


Many neuropsychologists, as well as educational psychologists, use “language” interchangeably with “verbal behavior,” where the latter was elicited on an anagram or verbal fluency task, both of which require knowledge of letters and word spellings (e.g., Halpern 1986; Hyde and Linn 1988; Maccoby and Jacklin 1974).


The material in this section was supplied by Bowen (2001), Gordon (1923), Hanson (1978), and the Report of the departmental committee (1921).


This is not to say that the boat children were exposed to the same concepts as their more freely ranging peers, who may have been more likely to know the meaning of some words (e.g., bonfire) purely through contact with the relevant actions and objects.


Much the same trend has been observed in children who develop cognitive traits in the context of their everyday lives. In Recife, a city on the northeast coast of Brazil, street-vending children perform mental calculations quickly and accurately while working, but do poorly on decontextualized “story” problems (Carraher et al. 1985). On the shores of Lake Victoria in Kenya, there are children with extensive knowledge of herbal medicines who fail to display comparable cognitive skills on standard measures of academic achievement (Sternberg et al. 2001).


The source, Henton (1992), rests her statements about sexual dimorphism on an article by staff writers that had appeared 11 years earlier in a popular magazine (Newsweek, May 18, 1981). In that article, a discursive treatise on “Just How the Sexes Differ,” there is a suggestion at one point that women “seem better at languages.” Nothing else was said about language, or how this female superiority had been uncovered or documented. Paradoxically, Henton also claimed, in a different part of her review: “Males are expected to speak more in public, to speak with authority and to do so at younger ages than females are.” (p. 45) This, of course, is precisely what one would expect if males’ ability to display verbal skills had been selected.


Copyright information

© Springer Science+Business Media B.V. 2007