[Incomplete draft of 4 lectures for Qinghua University, Oct.2004]
[Comments welcome]
0. Preliminaries.
My purpose in these four lectures is to discuss the study of language in the context of the recent upsurge of interest of cognitive neuroscience. I will draw from many fields in addition to linguistics, including anthropology, genetics, neuroscience, and computer science. One of the founding fathers of modern genetics, Th.Dobzhansky, observed that ‘nothing in biology makes sense except in the light of evolution.’ I think the same can be said for the study of language. The parallels between biology and linguistics were observed by Charles Darwin 150 years ago. Although language is largely a product of culture invented by the human mind, it was possible only because a number of biological mechanisms evolved earlier and somehow converged into a new function. Evolutionary linguistics is an area whose time has come; it is attracting attention from an increasing number of researchers from a variety of disciplines. So we may expect the area will advance very rapidly in the coming years. Here I can only offer some background information and preliminary thoughts on a very large topic.
I plan to divide my remarks as follows. In the first lecture, I will begin with an informal account of how language relates to communication in general, comparing it to some forms of animal communication. I will give a very brief and selective account of the history of linguistics, with a slant toward this scholarship in China. This will be followed by a discussion of the basic anatomy and neurology that language requires, specifically the vocal tract and the brain.
The evolution of language may be discussed at three different time scales. Microhistory deals with language acquisition as well language changes which can be observed in current speech communities. Mesohistory deals with language change in the order of centuries and millennia, the scale at which the traditional sound laws operate. The second lecture will consider how language evolves at these two time scales. We compare modern languages to infer what their ancestors were like many millennia ago. Whereas biologists compare bones and DNA, linguists compare words and sentences toward the same end of recovering the past. Linguistic comparisons are made much more difficult by the abundance of horizontal transmission. I will give a synopsis of what is know of the ancestry of English and of Chinese, among the languages of the world.
The third lecture will be concerned with language macrohistory - evolution at its grandest time scale, since its origin. Little is known from ancient times of 100 millennia ago. Nonetheless, anthropologists and geneticists provide recent data that is helpful for understanding the emergence of language. We need to consider evolutionary linguistics in ways which are constrained by their data. In recent years, computational methods have been used to simulate the emergence of various aspects of language by modelling multi-agent interaction.
The fourth lecture will recapitulate the major points that have been made. I will briefly offer my views on diverse matters related to language, such as the relation between spoken language and written language, teaching and learning of languages, the preservation of endangered languages, etc. Mostly, I hope to discuss questions from the audience that have accumulated from earlier talks.
1.1 Communication and language.
Communication may be defined as the deliberate transmission of information by an organism. According to this definition, all animals communicate, in order to protect its resources, attract mates, form alliances, and so on. The communication methods of some animals have been studied extensively. For example, the honey bee informs its mates the location of a food source by dancing on the surface of the hive. The dance usually contains three types of information: the angular direction of the food, its distance from the hive, and its relative abundance.
It is interesting to find that honey bees in different parts of the world vary in the details of the dance, much like dialectal variation in human language. Bees of the XX species in Italy do the round dance when then food is up to XX meters from the hive, but switch to the wagging dance when the food is farther away. Bees of the XX species in XX, on the other hand switch at XX meters. Unlike human dialects, however, such variations in bee dance must be due to genetic mutation rather than social learning since the communication is completely transmitted in the DNA.
Another example that has attracted attention is the call system of the vervet monkey. These African monkeys emit different calls according to whether the predator is an eagle in the sky, a leopard on the ground, or a snake in the grass. Such calls would elicit very different escape behaviors from their mates. That the signals are auditory has been confirmed by investigators, who are able to elicit the same behaviors by playing tape recordings of the calls.
Animal communication may signal by any combination of modalities - auditory, visual, by smell, and/or by touch. Wilson has estimated the number of communicative signals used by a wide range of animals, which typically range between 3 or 4 dozens. This number is dramatically less than the quantity of words used in human languages. Miller and Gildea estimate that an American high school student knows tens of thousands of words, in addition to numerous phrases, idioms, and other prefabricated linguistic patterns.
This remarkably rich lexicon is in part made possible by our agile speech mechanisms and the compositional design we use. Each language has dozens of segments in the form of consonants and vowels which it combines to construct the words. While the segments are very few in number, they combine multiplicatively to construct numerous words. Furthermore, the segments are short in duration, averaging several hundred milliseconds each; this allows the transmission of many words within a short span of time.
From the viewpoint of the speaker, this means the words can be easily packaged into convenient syllabic and breath units. From the viewpoint of the listener with a limited buffer of short-term memory, this means that the words in a long sentence can be all stored simultaneously in the same buffer for analysis and interpretation.
The compositional design, which is a fundamental design feature of language, allows the multiplicative construction of numerous larger units out of a small number of basic units. This feature is used at other levels of language as well. At the level of phonology, the basic unit is the segment, or the phoneme (when we consider its language specific functions). At the level of grammar, the basic unit is the morpheme – the roots and the affixes which combine to make up words; and words in turn
combine to form constructions of various sizes.
We should note in this regard that linguistic units typically do not combine in a simple linear manner, like the beads on a necklace. Rather, they combine in a hierarchical manner, like the successive layers of onions, with larger units containing smaller units. The Chinese syllable, for instance, is composed of a tone and a segmental syllable. The segmental syllable in turn is composed of an initial and a final. The
final is composed of a medial, and a rime. Finally, the rime is composed of a nucleus and an ending. This hierarchical structure can be bracketed as follows:
[tone [initial [medial [nucleus ending]]]].
Or, it may be shown in greater detail with a box diagram, where the larger units can be labeled. The syllables then combine to form larger units of stress, rhythm, and prosody.
[[Diagram of Chinese syllable]]
In grammar, the hierarchical structures are much more varied and complex as more and more morphemes are added to the construction. Since the same construction can be added repeated in a recursive manner, there is no upper limit to such compositional complexity. For example, the three morphemes [[da zi]ji] 'typewriter' has a 2+1 construction, whereas [xin [wan ju]] 'new toy' has a 1+2 construction. A sequence like [xin shu jia] can be bracketed either way. With a 2+1 construction, it means a case for new books; with a 1+2 construction, it means a new bookcase. We
will return to such complexities later when we approach the question of
ambiguity in language more fully.
The rich inventory of words coupled with compositionality give the language user an extremely powerful mental instrument by means of which he can represent and manipulate the world around him. Other animals communicate because they are automatically programmed to do so, such as the dancing of the bees. Or because they are reacting emotionally to some strong stimuli, such as the predator calls of vervet monkeys. These are all instinctive reflexes built into the genes over which the communicator has little voluntary control.
Language, on the other hand, is deliberate, voluntary, and does not depend on external stimuli. It enables us to represent complex objects and events and process this information in our minds - that is, to think - completely within ourselves, as well as to communicate this information to others. These representations may relate to the immediate environment, such as picking berries off a tree. Or, these representations may relate to distant events or imagined objects, such as a hunt of long ago, or dragons and demons, and square roots of negative numbers.
Language thus is an instrument that allows us to imagine other worlds, real and hypothetical, and, if we choose, to lie about them. Perhaps more than any aspect of our behavior, this cognitive ability makes us unique and defines our species. Biologists have proposed that in the billions of years that life forms have been evolving on earth, from the earliest cells to the elaborate organisms of today, there were only seven major transitions. All of these transitions were based on what kinds of information the organism can process. Within this framework, language is the latest to evolve, and one that has by far the largest impact on earth.
1.2 Brief history of ideas on language.
Curiosity about language must have come early to our ancestors. Perhaps the best known accounts of how words originated and why languages are different are those given in the Bible. For how words originated, the Bible tells of Adam giving names to all the creatures in the Garden of Eden, including to the woman that God made from his rib. To the question why languages are different from each other as human population proliferated, the Bible offers the Tower of Babel . When humans spoke only one language, God noted, their combined resources were so powerful that they try to reach heaven. So the diasporas of peoples and diversification of languages were God's response to this human hubris.
The several centuries before the Common Era saw philosophical creativity bursting forth independently in many parts of the world, in China, India, Greece, Persia, etc. Of special interest to us here are the coincident remarks on the nature of words, recorded by Plato in Greece, and only decades later, by Xunzi in China. Both advocated the idea that words arise from conventional agreement, and that the relation between a word and what it represents is an arbitrary one.
The idea is captured in an immortal line from Shakespeare: a rose by any other name would smell just as sweet. In a lighter vein, Humpty Dumpty told Lewis Carroll's Alice: When I use a word, just what I choose it to mean, neither more or less. The critical issue, of course, is whether people will understand your words.
More explicitly, Xunzi stated this viewpoint with three succinct phrases, which are justly famous among students of Chinese linguistics :
ming wu gu yi words have no intrinsic correctness
ming wu gu shi words have no intrinsic content
ming you gu shan words have intrinsic appropriateness
Even earlier than Xunzi and Plato, a tradition of linguistic scholarship developed in India; the best-known of this work is the grammar of Panini, ca.600 BCE, on Sanskrit. This tradition grew out a religious need to preserve accurately the pronunciation of the sacred texts of India, known as the Vedas. These scholars achieved a remarkable degree of understanding of how speech sounds are produced and how they influence each other in a stream of speech, and left behind extremely fine linguistic descriptions which are studied by scholars to this day. Their term for influence among sounds, ‘sandhi’, is now part of the standard vocabulary of phonetics.
With the later spread of Buddhism to many parts of Asia, some of the concepts of the Sanskrit grammarians had an impact on the neighboring traditions. It has been proposed, for instance, that the Chinese method of spelling one character with two other characters, the fanqie method, developed under the influence of Sanskrit writing. The Japanese way of ordering the elements in the kana syllabary is also believed to be based on the descriptive order in the Sanskrit alphabet. On a broader aspect, it has been suggested that some of the styles of poetry in China can be traced to linguistic traditions in India.
At the beginning of the Common Era, two legacies from the Han scholars are particularly noteworthy. One has to do with recording lexical variants by Yang Xiong. The actual field work actually started several centuries earlier during the latter part of the Zhou dynasty, when messengers were send to many parts of the Chinese empire to record local words. It remained for Yang Xiong [53 BCE to 19 CE], a scholar in the service of the Han court, to build upon this early tradition. The capital of the Chinese empire at that time, Chang’an, was always brimming with travelers from other regions, and Yang noted how they would say certain things. This work culminated in the important volume, the Fangyan, perhaps the earliest known study in linguistic geography in the world, and provides invaluable information on the languages and cultures of China 2000 years ago.
The other great Han scholar whose work has significantly impacted our study of language is Xu Shen [ca.100 CE]. From the first extant inscriptions on oracle bones of 1200 BCE to the time of Xu Shen, many major changes have taken place in the Chinese writing system. Xu’s great achievement was to compile a dictionary of some 9353 characters, the Shuowen Jiezi, organizing them under 540 radicals. Insofar as the written language represents a version of how one sees the world, the organization of how these characters relate to each other represents a cognitive map of the speakers. Whereas the Fangyan allows us to see language across time and space, the Shuowen Jiezi gives us the first lexicographic view of the world at Han times through the written language.
Observations on how the words of a language are constructed internally [morphology], and how they connect to each other in a sentence externally [syntax] constitute the province of grammar. In ancient Europe, the grammatical tradition of grouping words into parts of speech started by the Greeks was largely adopted by the Romans to describe Latin. An early Latin grammar by Varro [116 BCE ?27 CE] consisted of 26 books, though most of these have been lost. Works by later Latin grammarians, Donatus in the 4th century and Priscian in the 6th century, provided a tradition of describing language that remains influential today, particularly in textbooks.
From this perspective, the Chinese language is different from Greek and Latin in two important ways: [1] the words do not change their shapes nearly as much as required by the inflectional morphology of the European languages; and [2] the changes in phonetic form are typically not revealed in written Chinese as they are in alphabetic scripts. Consequently, there was no comparable tradition of grammatical studies of the Chinese language until much later.
Within the Chinese tradition, an early division of words into function words and content words was made by a Song dynasty scholar, Lu Jiuyuan [1139-1192], who noted that a function word refers to the meaning of the word, but a content word to the concrete senses it implied. This division was followed by a remarkable hypothesis put forth by a Yuan scholar, Zhou Boqi, namely, the function words of today were all content words in ancient times:
Jin zhi xuzi jie guzhi shizi .
Zhou’s hypothesis is now studied as ‘grammaticalization’ a term which was first used by the French linguist Meillet in 1912. A recent definition of this term offered by Heine and Reh [1984:15] is ‘an evolution whereby linguistic units lose in semantic complexity, pragmatic significance, syntactic freedom, and phonetic substance. Such processes are indeed ubiquitous in linguistic change, and often follow similar pathways in diverse languages.
An example is the evolution of full verbs into modals or tense markers, which has been observed in many languages. The English word have is a full verb with the meaning of ‘possess’ as in ‘they have a book’, but it has also grammaticalized to a modal verb, as in ‘they have not gone,’ which marks the aspect of the main verb ‘go.’ So the word now has both senses in modern English. A very similar pathway has been taken by the Chinese word ‘you’. In ‘tamen you yiben shu’ it functions as a full verb of possession. When grammaticalized, as in ‘tamen meiyou qu,’ the ‘you’ functions as an aspect marker of the main verb ‘qu’.
Many such examples can be given for various frequently used words, presumably from all languages, though with differences in detail. Such linguistic processes give us valuable clues to universal patterns of thought, and eventually the brain activities which underlie them.
Some three centuries after Zhou’s remark on grammaticalization, a major step was taken in the study of language history. This advance was in connection with the pronunciation of the Shijing, a collection of poems and folksongs which date to around 1000 BCE. Whereas earlier scholars failed to understand why these poems no longer rimed in the pronunciation of their times, Chen Di [1541-1617] of the Ming dynasty arrived at the true explanation that it is the natural consequence of the fact that languages are always changing. His famous words were :
Gai shi you gujin For time can be past or present
di you nanbei space can be south or north
zi you gengge [written] characters can be changed
yin you zhuanyi sounds can be transformed or moved
yi shi suo bi zhi these are but inevitable tendencies.
In his effort to reconstruct the rimes of the Shijing, Chen paid special attention to the nature of the evidence he used, dividing it into internal evidence which were derived from the Shijing itself, and external evidence based on other data, such as the phonetic components of the riming characters. In many ways, considering his remarkable insight on language change and the manner in which he amassed and treated the evidence, Chen has been regarded as a pioneer of the scientific method, not in studying the natural world, but in studying ancient texts. These early achievements were limited by a lack of phonetic knowledge with which to interpret the sounds he was reconstructing, and by not comparing his findings with other dialects and languages. These limitations were to be overcome by the emergence of the comparative method in Indo-European studies some two centuries later. We will return to this topic in Lecture 2.
1.3 A language ready body.
Let us now return to the observation made at the outset of this lecture that all animals communicate in the sense of deliberate transmission of information. However, only our species has invented the particular system of communication that we call language. Other species have not invented anything nearly as powerful; nor can they be taught easily what we have invented, even when the DNA of our primate relatives is remarkably similar to ours. This difference must be in part due to the fact that our body evolved along a unique path to make it language-ready for having invented language phylogenetically, and for learning language ontogenetically. We will now turn briefly to the question: how is our body language-ready?
We first diverged from the other primates some 4 million years ago when we changed from walking on all fours or from knuckle walking to erect posture and bipedal walking. While there is still no consensus on why this change took place, it is clear that this reorganization of the entire body had tremendous implications for all future developments of our species. Whereas earlier the forelimbs were used primarily for support in walking and running, now the hands became increasingly sensitive and dexterous, ready for making tools of much greater complexity and functionality.
Another related consequence of erect posture is a reorganization of the head and neck. Earlier our larynx was positioned close to the soft palate, a structure which controls air passage through the nose; this is the position of the larynx for human infants as well as for other primates. As the human body matures, however, the larynx descends by several centimeters, resulting in a pharyngeal cavity [throat] which plays an important role in helping us produce a variety of speech sounds. Basically, the mouth and the throat form an acoustic tube with a highly malleable shape. Speech is produced by expelling air through this tube, producing resonances of varying frequencies. Phoneticians refer to speech as an ‘overlaid function,’ in the sense that the original functions of respiration and mastication are primary. But the same organs used in these primary functions are also used in speech production.
By far the most important structure that made us language-ready is our brain, which underwent tremendous growth in the past 2 or 3 million years. Our brain is three or four times the size of that of our nearest primate relatives. This growth is no doubt related to the rapidly increasing capabilities and tasks of the hand as well as the greater communicative load of the mouth. In this extremely complex scenario of the evolution of the brain and of language, we can be sure that no single parameter developed by itself. The various parts of the body, the communicative and cognitive equipment that the body houses, and the physical and social environments that our ancestors interacted with formed a tight triad of co-evolution, each influencing the others in crucial ways.
We will return to the scenario of language evolution in the third talk. Suffice it for now to note that the vocalizations of other primates appear to be centered on neural structures underneath the cortex, mainly the limbic system. Our speech, on the other hand involves a significant portion of the cerebral cortex, and it is the frontal cortex that has grown the most in the brain. Earlier studies of language and brain specifically emphasized the importance of three cortical structures on the left hemisphere: Broca’s Area in the frontal lobe, Wernicke’s Area in the temporal lobe, and the Arcuate Fasciculus, a bundle of fibers which connect these two areas.
[[Geschwind diagram here]]
It is becoming increasingly clear, however, that for a behavior as complex as language, a great deal of neural circuitry is involved than just these cortical structures, even if the cortex is easiest to observe. The past two decades have seen major breakthroughs in the technologies of three dimensional brain imaging, with frequent advances in the degree of resolution in both time and space, as well as of monitoring the electrical changes of the brain. These technologies allow us to observe which parts of the brain are involved in which types of language behavior by showing us the entire brain, both the surface as well as the inner structures.
Parallel with these advances at a global level, important new knowledge is accumulating on the neurobiology of how neurons communicate with each other. It used to be thought that the billions of glia cells merely serve a supportive function for their neighboring neurons. There is now an increasing awareness that glia cells actually participate in the communicative processes as well, perhaps by altering the chemical fields in which the synapses are embedded. This will surely add a new dimension to research on the relation between brain and behavior, including of course linguistic behavior. Researchers are also gradually working out the approximate schedules of when and where synapses remain plastic as well as when and where neurons die and regenerate. It has been noted since ancient times that young children appear to have a clear advantage over adults in learning languages (and many other behaviors). Perhaps we will one day understand the neurobiological bases for this advantage. All in all, our understanding of how the brain serves language will surely deepen considerably in the coming years.
2.1 Modern languages and their relations.
We cannot say exactly how many languages there are in the world, even disregarding the numerous languages that must have been used and then became extinct. We can get an approximate count of the number of biological species in the world by some criterion such as reproductive fertility – two animals belong to the same species if they can produce fertile offspring. But we cannot get a count of the number of languages by a similar criterion – intelligibility.
Reproductive fertility [RF] in biology is a relation that is both symmetric and transitive:
[RF(x,y)] implies [RF(y,x)], and
[RF(x,y)] and [RF(y,z)] implies [RF(x,z)].
Unfortunately, intelligibility in language is neither symmetric nor transitive. There are numerous cases where speakers of language x can understand speakers of language y, but speakers of language y cannot understand language x. Similarly, the fact that x understands y and y understands z does not guarantee that x and z will understand each other. Lacking these features of symmetry and transitivity makes exact classifications and counts of languages impossible for theoretical reasons.
Counting languages is confronted with practical difficulties as well. Speech communities are often grouped together or labeled separately for sociopolitical reasons, or for reasons of their writing system. Norwegian and Swedish are very similar languages and enjoy a high degree of mutual intelligibility. However, they are considered distinct languages because they represent two political states. Hindi and Urdu are also very similar; they are considered distinct languages not only for political reasons, but for cultural/religious and orthographic reasons as well.
On the other hand, Cantonese, Shanghainese, and Taiwanese are generally considered dialects of one language, Chinese. In terms of linguistic structures and mutual intelligibility, these speech communities are much more different from each other than either Norwegian/Swedish or Hindi/Urdu. Their unity is more in the cultural and political unity these communities have shared for two millennia, and in the distinct writing system they all use. Nonetheless, many scholars prefer to speak of the Sinitic languages because of the linguistic distances separating them, rather than calling them Chinese dialects for the cultural aspects which unite them.
In spite of these difficulties in defining the boundaries of languages and giving them an exact count, linguists mostly accept that currently there are some 6000 languages spoken in the world today. This number is getting reduced at an alarming rate as language after language loses its last speakers and become extinct. It is a practical reality that parents of small ethnic minorities encourage their children to learn the language(s) of greater social or commercial value, rather than invest time in the language of their shrinking community and disappearing heritage. Forceful arguments have been advanced, often successfully, to preserve the biological diversity of organisms for their actual or potential medicinal values. But pleas to preserve linguistic and cultural diversity often fall on unsympathetic ears, even though they are a precious and irreplaceable part of the human heritage.
Our understanding of the biological world was considerably deepened with the theory of evolution, which provides a theoretical framework for studying how organisms are related to each other by tracing their past development. Similar ideas of evolution can be applied to the 6000 some languages as well. Earlier discussions of the relations among languages were largely empty and speculative, often mired in creationist scenarios. The first major step toward a scientific approach to this question was made by William Jones [1746-1794], an English jurist, and marked by a famous lecture he gave in Calcutta in 1786. His work is generally regarded as the starting point of modern linguistics.
In this lecture, he pointed out the remarkable similarities among several classical languages, Greek, Latin, Persian, and, most importantly, Sanskrit. In his words , these languages show
“… a stronger affinity, both in the roots of verbs and in the forms of grammar, than could possibly have been produced by accident; so strong indeed, that no philologer could examine them all three, without believing them to have sprung from some common source, which, perhaps, no longer exists.”
The common source that Jones hypothesized is Proto-Indo-European, a language spoken perhaps some 8 or 9 thousand years ago. A likely site where this protolanguage was spoken is ancient Anatolia, now part of Turkey. This is also the site where agriculture probably originated in the western world, spreading northward and westward, ultimately reaching the British Isles. Furthermore, it is likely that the spread was not only of the technology of farming, but of the farmers themselves. This was demonstrated by the striking correlation between the dates of the earliest farms with the genetic clines of the European peoples, reported in an exemplary study that ties archeology with genetics.
Building on this hypothesis, more than a century of research involving numerous scholars has resulted in a great deal of solid knowledge of the distant past of Indo-European languages. An example of this knowledge is the family tree of Indo-European languages published by the linguist August Schleicher [1821-68] in 1861. Although the tree is largely correct in the relations it depicts, it is far from complete. In early 20th century, for instance, two important groups of languages preserved in the form of inscriptions were unearthed in Xinjiang, China, and in Turkey, Tocharian and Hittite respectively.
[[Schleicher tree]]
Both discoveries are very important for historical linguistics. Tocharian is the east most relative of the Indo-European languages, but it possesses certain phonetic characteristics found primarily in western Indo-European. This caste doubt on the validity of earlier east/west dichotomy proposed for the Indo-European family of languages. Hittite is important because it confirms a class of phonemes whose reconstruction on theoretical grounds by Ferdinand de Saussure [1857-1913] pre-dated the archeological discovery of the actual inscriptions which confirmed the reconstruction. From the viewpoint of the history of science, it was a prediction of a yet unobserved entity based on reasoning from phonological patterns. There is a clear parallel here to the prediction of boron by the chemist Mendeleyev, reasoning from the patterns then available in the periodic table of the elements. Saussure’s achievement is all the more remarkable since his reconstruction was published in 1878, when its author was just twenty-one!
In many ways, linguistics was ahead of biology in the methods that were invented, and in the depth and detail of its reconstructions using these methods, explaining the historical relations of a tremendous range of data from diverse languages. By the time Charles Darwin published his Origin of Species in 1859, Schleicher was able to proclaim that he was a Darwinian before Darwin. Linguists and biologists often interacted to their mutual benefit; it is well known Darwin corresponded with various linguists, including Schleicher and Max Muller. In fact, among the recent admirers of the Indo-European scholarship and the comparative method was the eminent biologist Lewis Thomas , who had lavish praise for this research.
2.2 The comparative method and multilateral comparison.
The 6000 some languages of the world today share many resemblances, much as the organisms of the animals and plant kingdoms share resemblances. Some resemblances are more obvious – such as those between a horse and a zebra, or a tiger and a cat. Other resemblances are more hidden – such as between the human forelimbs and the wing of a bat. Similarly, it is easy to observe that in Chinese and in English the verb comes before the object in declarative sentences. It takes more scrutiny to see that the consonant /t/ in English corresponds to /ts/ in German, spelled with the letter ‘z’ in German orthography, at the beginning of many words. Some examples of this correspondence are: ‘to’ and ‘zu’, ‘two’ and ‘zwei’, ‘ten’ and ‘zehn’, etc.
The challenge for the linguist as well as for the biologist is to make sense of these innumerable shared resemblances at different levels of comparison. The biologist can turn to fossils for phylogeny and for dating, even to very great time-depths. Languages do not leave fossils, of course, and written texts unfortunately do not go back more than seven or eight thousand years. Given a particular shared resemblance, how do we explain it? Is it due to sheer chance, or is it due to a similar internal response to a similar external stimulus, or is it due to imitation of some sort, or is it due to common inheritance.
For convenience, we may refer to these four types of explanations as [1] chance, [2] convergence, [3] borrowing , and [4] inheritance. Confirming which of these explanations applies to each resemblance is informative on human language in general, and, more particularly, on the development of the languages and the peoples who speak them. In fact, before the advent of large-scale genetic analysis, our knowledge of prehistory and ancient migrations was primarily based on language classification.
Biologists studying higher organisms typically have little difficulty distinguishing the process of imitation from that of common inheritance. Genetic materials in mammals, for instance, and the components of behavior which they help determine are transmitted only vertically from parents to offspring. Biologists studying plant life, on the other hand, have much greater difficulty distinguishing these two processes, to sort out horizontal transmission from the vertical; cross fertilization takes place much more commonly in the plant kingdom. Yet a true phylogeny must be constructed from just common inheritance.
Language in this respect is much more like plants, where hybridization is commonplace. Whereas early Indo-Europeanists, such as Antoine Meillet, believed that changes induced by language contact can only be peripheral, a school of thought championed most forcefully by Hugo Schuchardt did not accept such boundaries. Indeed, studies of language contact over the past century have shown that anything in language can be imitated, though with different probabilities.
Although the likeliest features to be borrowed are words, prolonged and intensive contact between languages can result in the borrowing of any features. The most difficult aspect in linguistic investigations is to sort out the borrowed features from the inherited features, to distinguish horizontal transmission from vertical transmission. Only the inherited, vertically transmitted features can inform us of the true phylogeny of a language.
To a large extent, to determine how informative a feature is depends on probabilistic reasoning. To take the example given earlier, the fact that both Chinese and English have the verb in the middle position in declarative sentences is not at all informative. The three major elements of a sentence, Subject, Verb, Object, can only be arranged in 3! or six possible syntactic orders; it is likely that thousands of other languages have the order SVO as well. Furthermore, a language may change its syntactic order, as it can change any of its features. The Chinese language probably had the order SOV several thousand years ago, as suggested by the fact that the closest relatives to Chinese, the Tibeto-Burman languages, all have retained this order.
Contrasted with syntactic order with only 6 possibilities, the pairing of sound with meaning to make a word is virtually infinite with possibilities, and hence is highly informative. If two languages share a resemblance in a word, in both its sound and meaning, it is much more probable that the explanation lies in the history of these two languages. The two languages may have descended from the same ancestral language, and both have preserved this word. The word may have changed to different forms over time, as in the numeral /er/ in Putonghua and /ni/ in Shanghainese, both meaning ‘two’, or as in the noun /agua/ in Italian and /o/ in French, both meaning ‘water’.
Or, the two languages may have no demonstrable genetic relationship , but the word was borrowed from one language into the other. The Cantonese of Hong Kong, for instance, is rife with words borrowed from English, having been a British colony for a over a century. Typically, the borrowed word is adapted to the sound system of the host language. The adaptation becomes quickly conventionalized in this case when the borrowed word gets written with Chinese characters, each with its own indigenous lexical tone. Thus the final consonant in a noun like ‘ball’ is truncated because Cantonese has no syllable-final /l/, i.e., /bo/, and nouns like ‘tips’ and ‘store’ are borrowed as disyllables because Cantonese has no consonant clusters, i.e., /tip-si/ and /si-do/.
Some words are more likely to be borrowed for various cultural reasons, such as ‘ball’, ‘tips’, and ‘store’ above. Other words are less likely to be borrowed and stable for the same reasons, such as ‘two’ and ‘water’ above. The former are sometimes referred to as ‘cultural words’ while the latter as ‘basic words’. However, it is clear that there is actually a continuum of ‘borrowability’ among words, and that this continuum probably varies according to cultural region.
In many language families, pronouns for the first and second persons are particularly stable, presumably because they are fundamental for most communications. Thus in the Indo-European family of languages, the 1st p pronoun typically contains an /m/ and the 2nd p pronoun contains a /t/. English exemplifies this retention, with the /m/ in ‘me, my’, the /th/ in ‘thou, thy’ derives from an earlier /t/ from a regular sound change to be discussed below. In the Sino-Tibetan family of languages, the 1st pp typically contains an /ng/ and the 2nd pp has an /n/. Cantonese has largely preserved the ancient forms with /ngo/ and /nei/. Similarly for Amerind, a family of probably greater time-depth than either Indo-European or Sino-Tibetan, Joseph Greenberg has identified /n/ for the 1st pp and /m/ for the 2nd pp.
Another group of words which are relatively stable are the low numerals. This remark should be supplemented by two observations. It has been recently reported that some languages in Amazonia do not have numerals above one and two. Also, a language may retain its indigenous numerals while additionally borrowing another set. We may see an instance of this in the table constructed by Colin Renfrew for the first ten numerals in six different languages. The first five are all Indo-European languages, related to each other genetically. The 6th is Japanese, which is generally considered an Altaic language.
[[Renfrew table here]]
The Japanese words in the table are all indigenous numerals. At the same time, however, Japanese has borrowed a full set of numerals from Chinese, whose pronunciations are demonstrably of Chinese origin. In fact, if we look at the numerals ichi ‘1’, roku ‘6’, shichi ‘7’, and hachi ‘8’ we can see that the process of adaptation here is the same as we noted earlier in Cantonese ‘tip-si’. Since the Cantonese sound system does not allow the syllable to end in an /s/, the vowel /i/ was added to make a second syllable. To understand the adaptation here, we need to know that the Chinese numerals for ‘1’, ‘7’ and ‘8’ actually ended in a /t/ in earlier times, which is lost in Putonghua but preserved in many dialects, including Cantonese. Since the Japanese sound system did not allow the syllable to end in a /t/, a front vowel /i/ was added to support the /t/, creating a second syllable. This added vowel caused the borrowed /t/ to change to a /ch/, in a type of sound change called ‘palatalization’, which is widely attested in the languages of the world. The Chinese numeral ended in a /k/, which is also lost in Putonghua. In the adaptation, a back vowel /u/ was added to support the /k/, which is a back consonant.
Returning now to the Renfrew table, the forms are given in various types of spelling, which are less suitable for technical discussions than phonetic transcriptions. Furthermore, the five Indo-European languages are all of different time depths. English is written in its modern form, whereas Gothic, another Germanic language which has become extinct, dates to the 4th century CE. Classical Greek, Latin and Sanskrit all date still further back.
Nonetheless, some interesting patterns emerge when we examine these forms, as shown in the table below, which suggest that these languages, except for Japanese, are members of the same family, namely Indo-European. For instance, for the numerals 6 and 7, all the languages begin with /s/ while Greek has /h/. Thus a 6-sided figure is called a hexagon, one of many English words of Greek origin. Similarly, the ‘hemi-‘ in hemisphere is of Greek origin, corresponding to the ‘semi’ in semi-annual, each meaning ‘one-half’.
More interestingly, for the numeral 3, the Germanic languages all begin with /th/ while the other languages begin with /t/. On the other hand, when the Germanic languages have /t/, the other languages begin with /d/.
Integer English Gothic Latin Greek Sanskrit
two, ten t- t- d- d- d-
three th- th- t- t- t-
eight, ten ‘gh’ -h- -k- -k- -s-
six, seven s- s- s- h- s-
Some Indo-European Correspondences in Integers
bh > b b > p p > f
dh > d d > t t > th
gh > g g > k k > h
1 ashes belly bird bite all earth
2 bark blood claw burn big I
3 cloud bone dog come black name
4 fire breast feather die cold night
5 leaf ear fish drink dry not
6 man egg horn eat fat one
7 moon eye louse fly full road
8 mountain foot tail give good
9 person hair hear green that
10 rain hand kill long this
11 root head know many thou
12 sand heart lie new two
13 seed knee say red we
14 smoke liver see round what
15 star meat sit small who
16 stone mouth sleep warm
17 sun neck stand white
18 tree nose swim yellow
19 water skin walk
20 woman tongue
21 tooth 100
salt year
wind
7 9 5 3 2 9 35
发表于 2004-12-4 10:30:00