Chapter 9: Nature and Acquisition of Language

Process of Language Comprehension

There are a few different approaches to explaining the way in which we understand language. The first centers on the processes of speech perception, a second is more linguistically centered, and the third considers the level of discourse.

Speech Perception

Understanding the spoken word (a prerequisite to the written word) is essential to communication, which itself is critical to all but the most basic forms of human interaction. Without speech, communication is a crude pantomime, and interaction superficial, short-term, and largely unproductive.

We perceive speech with "amazing rapidity" and are able to perceive as many as fifty phonemes per second in a language in which we are fluent (Foulke 1969), as compared without ability to perceive non-speech sounds, which is done at a rate of 0.67 per second (Warren 1969).

The difference between the two can be considered in the course of learning a foreign language: we initially recognize it as a pattern of nonsensical sound, and later identify the individual phonemes, recognizing the combinations of sounds to form words. Before we can even begin to consider the meanings of words, we must gain the ability to perceive them as words rather than a stream of babble.

Prerequisite to identifying words in a pattern of sound is the facility of coarticulation, our understanding of the way in which phonemes overlap one another, and are shortened or lengthened in the context of speaking. This is necessary to perceive the phonemes within words as well as dividing the stream of data into syllables and words - in essence, knowing where one word ends and the next begins, even though the phonemes are distorted.

Coarticulation is a significant difference in the perception of speech different to perception any other noise: we perceive the burbling of a stream as a single sound, and do not dissect it into a sequence of syllables nor presume it to have meaning.

One set of theories equate speech perception to the perception of any sort of sound, i.e. that there are no distinct process for the perception of speech at the onset. Speech is received as any normal sound and must be recognized as speech before the linguistic processes attempt to analyze it, such that speech is simply noise until we recognize a pattern that resembles a word, at which point we attempt to analyze the noise as speech.

The author suggests that this is common on a conscious level, when we have to pause to figure out what someone just said. (EN: The example in the book is a bit sketchy, but consider meeting someone who speaks with a heavy accent - until you learn the idiosyncrasies of their speech, it takes conscious effort to determine what they have said.)

A common feature of these theories is that they all require decision-making above and beyond basic pattern-matching, particular in the instance where the phones might create multiple valid word patterns, which is often reverse-engineered from the context of the sentence. For example, the words "parody" and "parity" are so phonologically similar that we must consider the word in the context of the sentence to know which meaning was intended.

Other theories maintain that speech is treated differently to other sounds in the environment closer to the perceptual level. This is based on studies (Liberman) that demonstrate a facility for perceiving speech sounds categorically (our adeptness for recognizing and distinguishing between close phonemes such as "ba," "da," ga," and "ha") is different to the way we differentiate among non-speech sounds (we don not perceive a burbling or rattling noise to be composed of phonemes).

The McGurk effect is described, which occurs when there is disparity between the sounds a person hears and the words they are reading. Consider the frustration of a video program when the soundtrack is a fraction of a second out of synch, such that the lip-movement of a speaker doesn't correspond to what is being heard at the time, or a dubbed foreign film where the two do not correspond at all.

Various experiments with this effect note that the message that is interpreted (the words that were actually said) can be garbled when sound and visual data are asynchronous, which also supports the theory that our understanding of a message is an integration of sight and sound data.

The author notes that various cognitive psychology texts describe speech perception in different and incompatible ways - but ultimately while there is much disagreement over the way in which we come to recognize a sound as speech, the camps converge after that point.

Semantics and Syntax

The denotation of a word, the strict and limited definition recorded in a dictionary, is a poor reflection on the way that language is used in that it is historical record that omits much. The connotation of a word includes its full breadth of meaning in its present use, which is much more fluid and expansive.

It is reckoned that language is secondary to experience - we learn the words to describe experiences - and as such we adapt language to our understanding. At the same time, it can also be justly said that there are concepts with which we gain knowledge second hand, communicated to us by others through language, and in that way our ability to grasp a concept depends on our facility in interpreting language (as well as the source's facility in using language to describe.)

Semantically, a word has a broad range of meanings, as it is a trigger for a multitude of phenomena. Consider the word "desk" activates many connections in the mind:

• The simple dictionary definition of the term
• Every instance of a desk you have ever seen, whether in they physical environment or depicted in drawings, photographs, film, or other media.
• The abstract concept of a desk that exists in your mind - the archetypical desk you would draw without copying a specific model
• All the characteristics of these various real and imaginary desks
• All the activities that take place at a desk and the environment in which a desk might be found
• An array of tangent concepts, such as things you might find on top of or in a drawer of a desk, places you might make a desk, the materials and techniques used to build a desk, objects that are used in a desk-like way, etc.
• Other desk-like objects (a table, filing cabinet, bureau, etc.) and the way in which they are similar to and differentiated from a "proper" desk.

As such the mention of a single word brings to working memory a vast amount of information, all of which figures into our understanding of a seemingly simple term.

Also, consider the constructive nature of memory. Having a word-label for an item has several effects: it facilitates understanding and remembering, enables us to quickly name a shape, influences our immediate perception and interpretation of an environment, etc.

Aside of the meaning of a single word, there is also the influence of syntax, which considers the meaning of a term in the context of a phrase or a sentence.

In a specific sense, grammar is the study of the use or words - which attempts to define the acceptable ways in which words may be constructed into a phrase, but in a functional sense enables use to consider a context of a word to consider its meaning, including meanings in which the grammar of a phrase does not follow the conventional rules.

(EN: Of note, grammarians attempt to be prescriptive in the use of grammar in constructing communication, whereas rhetoricians use grammar forensically, to determine the meaning of what happens to have been said. What a person does in everyday life is some combination of the two, but my sense is it is more in the nature of the latter.)

"People demonstrate a remarkable knack for understanding syntactical structure." That is, they can understand the way in which words function in context even if they do not understand the meaning of the words themselves. Consider the poem "Jabberwocky" in which readers easily make sense of sentences composed mostly of nonsensical words.

Or consider the way in which the meaning of a sentence is understood even when the speaker flubs a word, or when two words in a sentence are accidentally swapped ("I put the oven in the cake"). Conversation does not halt, but continues as the listener knows what the speaker intended, even if it isn't exactly what he happened to say.

All of this suggests that syntax is a separate operation within cognition, such that we recognize the qualities of a word (whether it is a noun, verb, adverb, adjective, etc.) independently of knowing its precise meaning.

There's a bit of a tedious examination of the way that a sentence is parsed, to first discover its subject, verb, and object, then to recognize the functions of the words within each of those components.

Of particular interest: the process seems to work in both ways: the function of a word may be learned form the context in which it is used, or the word itself may be learned and its function is part of its meaning (the term is identified as a noun in its definition).

The flexibility of the linguistic system is demonstrated in our ability to decipher various structures (knowing "she at the cake" is the equivalent of "the cake was eaten by here") as well as the ability to recognize multiple functions of a single word ("spread" can function as a noun, verb, or adjective)

Language Acquisition

Generally speaking, the theories of acquisition of language seem to fall into the same harkened mode of "nature or nurture" with the customary conclusion that each plays a role - i.e., facility with language combines innate talent with experience.

Stages of Language Acquisition

It is remarked that language acquisition seems to follow largely the same sequence across all cultures and languages. Specifically, infants demonstrate the ability to identify and distinguish among phonetic sounds and over time tend to specialize in those specific sounds that are components of our native tongue. The ability to produce speech sounds seems largely to parallel the ability to recognize them.

It's also noted that infants are most attuned to their mothers' voices, followed by those of other members of the household and caregivers with whom they interact frequently. They acquire language in the same manner (pronunciation, rhythm, etc.) as they hear it spoken. It's even suggested (Bertoncini, Mehler, et. al.) that infants coordinate physical motion to the pattern of sound - in effect, if you were to videotape an infant, you could clearly recognize that their movements were like "dancing in time with the rhythm of the speech."

In general, language production follows a sequence:

1. Responsive to voice but no attempt to speak (birth to six months)
2. Cooing, which is limited to vowel sounds (prior to 6 months)
3. Babbling, which includes consonants. (6 months to 18 months)
4. One-word utterances within the limits of their capabilities to pronounce both vowels and consonants (18 months to 3 years)
5. Two-word utterances and "telegraphic speech" (noun-verb pairings) (3 to 4 years)
6. Basic adult sentence structure with continuing vocabulary acquisition (age 4 to 5)

Linguists have paid particular interest to cooing, and note that there is no distinguishable difference in the cooing of infants in various cultures and that even deaf infants make the same cooing noises. It's further suggested that infants show the ability to distinguish among consonant sounds, even though they are incapable of making them.

It is added (Wekrer) that infants can make phonetic distinctions that adults have lost. For example, Japanese infants can distinguish between an 'l' sound and an 'r' sound even though the former is not an element of the Japanese language - however, this distinction is lost by one year of age. The same has been evident in American infants when tested using certain consonant sounds in the Hindi language that are absent in English.

At the babbling stage, deaf infants no longer vocalize and the sounds produced by hearing infants, both vowels and consonants, align with those of the language of their household. Thus, while cooing is universal, babble is language-specific.

The development of one-word utterances marks the recognition that certain combinations of sounds have a specific meaning, and uttering them causes predictable changes in the environment. In essence, the utterances are "holophrapses" ion that a single word such as "juice" is the functional equivalent of stating "I'd like some juice" or "Give me that juice."

By 18 months, the typical human vocabulary is between three and 100 words (Siegler) and overextension errors become common, such as using he word "dada" to describe any adult male or "doggie" can be used for any four-legged animal.

There is some speculation as to whether overextension errors represent an inability to distinguish between specific things in terms of their features (they use "juice" for milk because they do not recognize the two to be different), or a rudimentary form of categorization because their functions are not recognized (to an infant, either juice or milk serve the same purpose). Neither can be proven to the exclusion of the other, and it seems likely both mechanisms may be influential.

Gradually between 1.5 and 2.5 years, children begin combining single words into two- to three-word utterances, representing the formation of syntax. This begins with simple noun-verb parings and later appends adjectives to nouns, either directly or reflexively.

In terms of vocabulary, it grows rapidly in early stages, going from 100 words at 18 months to 300 words by age two, to about 1,000 at age three.

By age four children acquire the foundations of adult syntax; by age five they can produce "quite complex and uncommon" sentences; and by age ten their use of language is fundamentally the same as that of adults.

Nature and Nurture

The development of language, particularly in the universal behavior in the cooing stage and the differentiation in the babbling stage, suggests that there are innate tendencies that support learned behaviors.

In particular, it is theorized that every human being has a "language-acquisition device" that is biologically innate - a core set of capabilities that can be leveraged, at least some of which atrophy quickly in the first few years of life.

The author provides a litany of reasons that support a predisposition toward language in humans that is pronounced when compared to other animal species:

• Our auditory perception is rather dull in most regards, but highly attuned to speech
• Children acquire language far more rapidly than any other ability
• It is very easy for a child to learn its first language, and far more difficult for an adult to learn other languages
• The ability to understand syntax without understanding words facilitates learning language

There is a brief mention of metacognition, the ability to think about learning and test perception against a theoretical frame. This facilitates rapid assimilation of information. In language, we learn the qualities of language itself, syntax and structure, as a container for the vocabulary we will later acquire.

In particular, the metacognitive aspect of language learning makes it easier to learn a second language that has the same sounds and follows the same rules as the primary language. For example, English and French share much of the same structure and syntax, so it is far easier for an English-Speaker to learn that language that it is to learn Russian or Chinese, whose very structure and rules are different.

Back to development, it is observed with some amazement that children in all linguistic environments acquire language by the same processes and at the same rate,, though other factors (such as the amount of language they happen to hear in their environment) can speed of slow the process.

Studies of the deaf (Meier) suggest there are critical periods in the development of language: consider that children who acquired sign language before the age of six are more fluent than those who learned it after age twelve - regardless of the number of years of experience they have at signing, failure to acquire the language before a certain age results in less facility and a more superficial understanding of its distinctive syntax.

(EN: This strongly suggests that the process of learning a language as a child and that of learning it as an adult may be entirely different. That is, once our linguistic wiring is soldered into place, we use process other than the innate learning capacity to acquire language.)

There have also been studies of linguistically isolated children, which is quite rare but when it occurs it is found that they do not as readily acquire sophisticated language structures when the linguistic skill is cultivated later in life. This also supports the notion of critical periods in development.

Biologically, it is found that humans possess certain physiological structures for which no other purpose has been identified except the production of speech, and that this biology is universally determinant of the kinds of sounds humans can produce. That is to say we are born with the ability to speak phonemes of every human language, but tend to specialize in a single set that corresponds to a specific language.

According to the hypothesis testing view, children acquire language by forming tentative hypotheses regarding language and test them in their environment: they make an utterance and observe reactions of their caregivers, alter the sound to determine whether the reaction is different, or make a different sound to determine whether there is any reaction at all. While this seems highly mechanical, the behavior can be easily observed. It's further reckoned that children who are acquiring language do not pay attention to all its aspects, but seek to discover its mist salient aspects.

While there has been no tenable theory that language is entirely natural rather than learned, certain aspects of language do seem to exhibit a correlation to genetics.

There is also some consideration of the behavior patterns evident in the learning process: imitation, modeling, and conditioning.

In imitation children merely mimic the language patterns of others. This alone is not sufficient for acquiring language but does often lead to the isolation of specific phonemes as they loosely follow what they hear. If imitation were the primary means by which language was gained, children would speak in complete sentences of words they did not yet understand, but this is not the case.

The pattern of language acquisition follows one of modeling - imitating word sounds at first, but later learning words and assembling two-word utterances that are not imitative of the way in which language is used by adults in their environment.

This is also observed in the phenomenon of overregulation: when children say "mouses" instead of "mice" and "runned" instead of "ran" they are not imitating adult speech, but applying the rules of grammar and syntax as they have come to understand them independently of what was heard.

Some mention is made of child-directed speech, which is so widespread a practice as to seem a natural proclivity in adults, who raise the pitch of their voice, enunciate more carefully, and simplify their sentence structure when speaking to infants. The intent of the speaker is clearly to be more understood to the infant, and it has been found that infants give better attention to child-directed speech.

It's further noted that adults universally adopt certain patterns of speech under different circumstances - using rising intonations to gain attention, falling intonations to give comfort, and rapid speech to warn against prohibited behavior. The speech alterations by adults also decrease as the child's language develops.

The conditioning behavior is relatively straightforward - by gesturing at a thing and speaking its name, they are rewarded by the parent for having spoken (either by being handed the thing or by verbal assurance). The reward becomes a conditioning device that assures them they have used the word correctly, and withholding reward indicates they have gotten it wrong.

Conditioning can be interrupted by a parent who withholds reward when a word is spoken correctly or granted when it is spoken incorrectly, and alter in development the parent generally makes an intentional effort to correct the child's grammar and pronunciation.

However ,conditioning alone does not explain language, as children constantly use novel utterances for which they have never previously been rewarded, and apply the words and language structures they already know to novel situations for which they have not received reinforcement. As such, conditioning enhances and tunes language, but is not solely responsible for its development.

Beyond the First Years

Much of the study of language development in children focuses on development until age four. Many of the most basic skills have been mastered by that age, but there is still significant development afterward. Vocabulary increases and sentence structure becomes more sophisticated, but at a far slower pace than during the early years of language development.

It is theorized that language is built by a process of comprehension monitoring: we recognize that we do not understand a sentence or a word and ponder it, adding what we have learned to our own linguistic abilities. (EN: This seems a generalization, and people have varying levels of curiosity. Distressingly many simply ignore information they do not understand and do not bother to learn it.)

It's also noted in studies of children ages 8 to 11 (Markman) that contradictory information often goes unnoticed. In sentences as short as four paragraphs, a blatant contradiction is presented - almost half of children do not notice the contradiction at all, even when told in advance that they would be asked to identify the problem with the story.

(EN: The author does not go on to indicate when the ability to detect contradiction evolves, and given that NCES studies suggest that the average reading level is grade 8, such that half are below that mark, chances are some people never do develop the ability to comprehend more than a sentence or two. But in fairness, this may relate to attentiveness rather than cognitive ability - a choice not to pay attention rather than the inability to comprehend. Functionally, it comes to the same thing, but it would be difficult to separate for purposes of experimentation and measurement.)

Animal Language

Some cognitive psychologists have turned to animal experimentation in order to understand language, which seems an unsound approach especially given that language is a distinctly human behavior, but the author lists a number of reasons that such studies may provide valid insight:

1. Animals have simpler cognitive structures and are easier to model and comprehend. Whereas human beings consciously alter their behavior, it is presumed that animals merely react and their behavior is more consistent and easier to attribute to specific stimuli.
2. The ethical constraints when dealing with animal subjects are far looser than those pertaining to human ones, largely because it is presumed they have no memory and quickly recover from experimental conditions, whereas humans retain more long-term effects.
3. Animals can be kept in controlled environments for long periods of time, whereas human beings can be brought into a lab for a short period of time, and the external influences upon them are unknown and uncontrollable.
4. Animals provide a glimpse into more primitive states of consciousness and therefore are presumed to provide a glimpse into development in early stages of human development. One cannot test on a Neanderthal man, but one can test on an ape, which is presumed to have the same level of cognitive ability.
5. Measurements of animals, when compared to humans, provides a clearer perspective on which abilities and behaviors are distinctly human. Particularly in the examination of "nature versus nurture" the behavior of animals is presumed to fall almost entirely in the "nature" category and can be assumed to be a biological rather than learned factor in humans.

Naturally, the findings of animal experimentation cannot be bootstrapped to humanity, and there are many assumptions and presumptions inherent in any animal study that may be entirely incorrect. Ultimately, it must be conceded that animal studies do not yield much useful information for understanding human cognition, but neither can it be stated that they are entirely useless.

In terms of animal studies, it is reckoned that primates offer the most promising insights into nonhuman language. (EN: glancing ahead, I see no mention of studies of language in aquatic mammals, such as dolphins, and much research has been done in this area, though it is difficult in that their methods of communication are not entirely verbal but instead include visual and tactile elements that are difficult for researchers even to observe. Hence the focus on primates is likely understandable, and likely to remain the primary subject of study.)

While tamed primates have shown some proclivity to communicate using sign language in captivity, this behavior is not evident in the wild, in which their linguistic capabilities do not show much more sophistication than canines or birds - a limited range of vocalizations that communicate basic concepts such as "beware of danger" or "stay away from me." Such behavior does not satisfy the criteria for communication.

Insofar as tamed primates are concerned, it has been found that chimps, in particular, demonstrate linguistic skills beyond the stages that a human toddler can reach. In one example (Premack) a female chimp acquired a vocabulary of over 100 words of various parts of speech and showed rudimentary communication skills in appropriately assembling sentences that were independent and contextually relevant (e.g., to sign "I am hungry" to communicate a desire for food, without being asked "are you hungry?")

There is some dispute over whether this qualifies as communication, or whether it is simply imitative behavior (a dog will beg unprompted when it is hungry, even in the absence of food, correlating the behavior to the result rather than to the meaning). Moreover, the signing patterns of primates do not demonstrate knowledge of syntax, in that their patterns will be randomized - as likely to sign "give me food" or "food give me" or "food me give" without developing a preference for a given sequence. Even in the telegraphic stage of speech development, infants develop a preference.

As such, while language-like capabilities have been observed in animals, no subject has sufficiently or consistently demonstrated sufficient fluency to support the thesis that they are capable of language - they may use sounds and gestures to communicate, but in a very primitive manner.