What proves that all these tongues are related to each other and distinct from other language stocks? One obvious clue is shared vocabulary, as illustrated by the table of vocabulary on page 226 and thousands of other examples. A second clue is similar word endings (so-called inflectional endings) used to form verb conjugations and noun declensions. These endings are illustrated by part of the conjugation of 'to be' below. It becomes easier to recognize such similarities when you realize that word roots and endings shared between related languages are generally not shared identically. Instead, a particular sound in one language is often replaced by another sound in the other language. Familiar examples are the frequent equivalence of English 'th' and German 'd' (English 'thing' equals German 'ding, 'thank' equals 'danke'), or of English V and Spanish 'es' (English 'school' equals 'escuela, 'stupid' equals 'estupido'). Those resemblances among the Indo-European languages are detailed, but much grosser features of sounds and word formation set Indo-European languages apart from other language families. For example, my atrocious French accent embarrasses me as soon as I open my mouth to ask, 'Ow est le metro' But my difficulties with French are nothing compared with my total inability to produce the click sounds of some southern African languages, or to produce the eight gradations of vowel pitch in the Lakes Plain languages of the New Guinea lowlands. Naturally, my Lakes Plain friends loved teaching me bird names that differed only in pitch from words for excrement, then watching me ask the next villager I met for more information about that 'bird'.

INDO-EUROPEAN LANGUAGES

English (I) am (he) is

Gothic im ist

Latin sum est

Greek eimi esti

Sanskrit asmi asti

Old Church Slavonic jesmi jesti

NON-INDO-EUROPEAN LANGUAGES

Finnish Oien on

Ore miyuwe miye

As distinctive to Indo-European as its sounds is its word formation. Indo-European nouns and verbs have various endings that we memorize assiduously when we learn a new language. (How many of you ex-scholars of Latin can still chant amo, amas, amat, amamus, amatis, amanf?) Each such ending conveys several types of information. For example, the 'o' of 'amo' specifies first person singular present active: the lover is I, not my rival; one of me, not two of me; I am giving, not receiving, love; and I am giving it now, not yesterday. Heaven help the serenading lover who gets even a single one of those details wrong! But other languages, like Turkish, use a separate syllable or phoneme for each such type of information, while still other languages, like Vietnamese, virtually dispense with such variations of word form.

Given all these resemblances among Indo-European languages, how could the differences among them have arisen? A clue is that any language whose written documents span many centuries can be seen to change with time. For example, modern English-speakers find eighteenth-century English quaint but completely understandable; we can read Shakespeare (1564–1616), though we need notes to explain many of his words; but Old English texts, such as the poem Beowulf (circa 700–750 AD), are virtually a foreign language to us. A good example of how English has changed over the last 1,000 years is provided by the Twenty-Third Psalm:

As speakers of one original language spread into different areas with limited contact, the independent changes of words and pronunciation in each area inevitably lead to different dialects, such as those that have arisen in different parts of the US in the few centuries since permanent English settlement began in 1607. With the passing of more centuries, dialects diverge to the point where their speakers can no longer understand each other and they now rank as distinct languages. One of the best documented examples of this process is the development of the Romance languages after the break-up of Latin around 500 AD. Surviving written texts from the Eighth Century onwards show us how the languages of France, Italy, Spain, Portugal, and Rumania gradually diverged from Latin—and from each other.

The derivation of the modern Romance languages from Latin thus illustrates how groups of related languages develop from a shared ancestral tongue. Even if we had no surviving Latin texts, we could still reconstruct much of the Latin mother tongue by comparing traits in its daughter languages today. In the same way, one can reconstruct a family tree of all the Indo-European language branches, based partly on ancient texts and partly on inferences. Hence language evolution proceeds by descent and divergence, just as Darwin demonstrated for biological evolution. In their languages as well as their skeletons, modern Englishmen and Australians, who began to diverge with the colonization of Australia in 1788, are much more similar to each other than either are to the Chinese, from whom they diverged tens of thousands of years ago. Given time, the languages within any part of the world will keep on diverging, held back only by contacts between adjacent peoples. An example of the result is New Guinea, which had never been unified politically before European colonization, and where nearly one thousand mutually unintelligible languages—including dozens with no known relation to each other or to any other language in the world—are now spoken in an area the size of Texas. Thus, wherever you find the same language or related languages spoken over a wide area, you know that the clock of language evolution must have been restarted recently. That is, one language must have recently spread, eliminated other languages, and then started to differentiate all over again. Such a process accounts for the close similarities among southern Africa's Bantu languages, and among Austronesian languages of Southeast Asia and the Pacific.