On the Human Languages and the Evidence of Their Kinship

The expression ‘Human speech’ serves tolerably well until someone supposes that it names a language. A Caldric railway porter and an Aurencal advocate possess much the same equipment for speaking, yet neither will understand the other merely by employing it more vigorously. I have witnessed the experiment at a frontier booking-office, where its principal result was the detention of my luggage. Our languages belong to a family; their speakers belong to a people. These are different propositions, established by different evidence, and considerable confusion follows whenever a scholar allows his knowledge of the second to excuse his neglect of the first.

The family relationship rests upon agreements which borrowing alone cannot readily explain. A common word for a steam regulator proves little when the regulator and its name arrived upon the same Dwarven wagon. Repeated correspondences between sounds, extending through ordinary vocabulary and surviving fragments of older grammar, deserve closer attention. An Aurencal consonant may regularly answer to a different Caldric consonant in words whose outward resemblance is slight. I require the correspondence to recur before I admit it as evidence. Professor Thrupp’s celebrated table of twenty identical words contains fourteen articles of commerce and two copying errors. His industry has at least furnished students with a convenient exercise in telling a borrowed word from an inherited one.

Western and Eastern Human separated in the eighth century before the era, according to the chronology I accept. The precision sometimes bestowed upon this event is wholly unwarranted: nobody has discovered the minutes of the separation. Its approximate position follows from comparative reconstruction considered alongside the movements of settlements and the relative ages of borrowed words. The Central division into northern and southern branches belongs much later, around the eleventh century A.E. The two dates are not equally secure, the later of them falling inside the period from which documents survive. Proto-Human itself is unattested. An ancestral form printed with a mark of reconstruction represents an argument about a lost word; quotation marks would make it a fraudulent document.

Our documentary poverty has an intelligible cause. Through the centuries of Dwarven government, a Human who wished to record a debt, petition an official or dedicate property to a temple ordinarily wrote Imperial Dwarven. Imperial Aesir and Early Common Elven served less frequently. No Human language is well attested before the third century A.E.; this tells us something about the surviving business of writing and very little about the richness of contemporary conversation. I once accepted a supposed early Human fragment which proved to be bad Imperial Dwarven. The correction came from a quarry accountant, whose acquaintance with abbreviated delivery notes exceeded mine. I have since quoted him; the original editor has preferred to quote my original error.

The four languages enjoying state patronage today are Caldric, Aurencal, Lanvareg and Sudermarkish. A coloured map can make their territories appear to exhaust the family, especially when its engraver has been instructed to omit inconvenient villages. Several further branches have disappeared; others persist in minority pockets whose speech is too often entered in official returns under the language used to question their inhabitants. I have obtained a declaration of Caldric mother speech from a woman who had understood the clerk’s question only through her grandson. I required the return to be amended. Such errors affect our account of living diversity and deprive the reconstruction of earlier Human speech of evidence which the larger languages may have lost.

Caldric presents its relations chiefly through word order and the small words placed before nouns. Its nouns bear no case endings and have little work to do beyond distinguishing the plural; its verbs distribute much of their business among accompanying verbs. These arrangements have encouraged certain grammarians to describe our language as worn out. I invite them to explain the difference between ‘will have been carrying’ and ‘would have been carried’ before pronouncing upon its incapacity. The older North Central records show successive stages in the reduction of endings. As distinctions disappeared from the ends of words, other arrangements became more dependable, and departures from those arrangements acquired greater consequence.

Its vocabulary records prolonged intercourse with Imperial Dwarven, the hold languages and Common Elven. The several sources cannot always be distinguished by inspecting a modern word, since a loan may have entered twice, at different periods and through different mouths. I have traced a workshop term and a term of parliamentary procedure to the same Dwarven expression, although neither a fitter nor a member of the Assembly welcomed the relationship. Our spelling further conceals the evidence. Printers fixed many forms while their pronunciation was still changing; subsequent generations preserved the letters with a fidelity they did not extend to the sounds. A printed vowel may therefore testify more faithfully to an ancestor’s mouth than to the reader’s.

The regional varieties of Caldric expose the vanity of calling the capital’s speech intrinsically correct. The eastern interior enjoys that reputation because its usage lies nearest Tarnis, where appointments are made and examinations marked. The inner-bay varieties, though most divergent from this official preference, preserve sounds lost elsewhere. In the arid south, recent innovations furnish excellent evidence of changes in progress. I initially mistook one southern distinction for careless articulation and instructed my informant to repeat himself more distinctly. He obliged until I heard the distinction I had been correcting out of his speech. We subsequently obtained consistent results from speakers who had never met him, which disposed of my first explanation more thoroughly than courtesy required.

Aurencal, along the east coast below the Vasker Mountains, offers a different distribution of grammatical labour. Nouns have no case, while articles and adjectives follow two grammatical genders. Its articles descend from older words meaning ‘that’, a development whose intermediate uses can be followed in the written record. The verbs retain extensive inflection and numerous irregularities. Some apparent irregularities become intelligible when older sound changes are recovered; others remain inadequately explained. I object equally to the schoolmaster who treats every difficulty as a pupil’s moral failure and the enthusiast who promises that historical study will abolish every difficulty. The history is often recoverable, though the pupil must still learn the verb.

The loss of most final unstressed vowels has given modern Aurencal its conspicuous consonantal endings and obscured relationships once easier to hear. Fortunately, its ancestor is preserved from the third century A.E. onwards and continues as a learned written language in eastern courts and law schools. Continuity of use is itself a trap: a legal secretary composing in the ancient language may reproduce a construction absent from ordinary speech for centuries, while another inadvertently introduces a modern habit beneath antique endings. At Aurenca I disputed a lecturer’s claim that his students spoke as their remote ancestors had spoken. His evidence consisted of their examination compositions. When I requested conversation, he complained that I was changing the subject.

Lanvareg, the Western Human language of the republic of Lanvarr, places its verb before its subject. Relations between nouns are expressed through preceding particles which themselves inflect, an arrangement liable to distress visitors educated to expect grammatical information upon the noun. The distress is temporary; the arrangement is perfectly serviceable. More arresting is the alteration of a word’s first consonant according to what precedes it. A collector who records each noun only as an isolated answer will miss a substantial part of the language. My first Lanvareg assistant brought me an admirable list of citation forms. We spent the following fortnight discovering what happened to them in sentences, during which his list became admirably untidy.

Modern Common Elven also changes initial consonants, and Lanvarr’s sole land connection passes through territory held by Elven tribes. A historical connection is therefore plausible; it remains unproved. Similar arrangements can arise independently, and contact may extend a pattern which already existed in a more limited form. We need to know which preceding words caused the earliest recoverable Lanvareg changes, how the changes developed, and whether their chronology permits transmission from Elven. The oldest Western material leaves serious gaps. I have argued against both a purely Elven explanation and the patriotic insistence upon wholly independent invention. Each party has accused me of assisting the other, a judgement which gives me some confidence that they have heard the objections.

Sudermarkish, spoken in the southern thanedom of Sudermark, is Caldric’s nearest living relative. Its case endings and genders preserve distinctions our language has lost; the verb occupies the end of a dependent clause, and long compounds allow a writer to accumulate qualifications before permitting the reader to reach the governing idea. These features have been recruited into an argument that the people possess an innate taste for obedience. I have heard the same compounds employed in remarkably inventive abuse of a commanding officer. Grammar supplies means of expression to speakers with many purposes. A dictatorship’s preference for a construction establishes neither its origin nor the sentiments of everyone who uses it.

Several Sudermarkish consonants are articulated far back in the throat and require care from a Caldric observer, whose first impulse is often to hear two unfamiliar sounds as one. During my southern survey I could distinguish a pair when others spoke them long before I could reliably produce either. My notes improved when I admitted this disability. The thanedom’s office language presents another obstacle: it combines prescribed forms into a written usage spoken by no district. Local varieties must be compared directly. An officer offered to regularise my transcripts before publication, believing that he would thereby improve their accuracy. I explained that his proposed assistance would remove much of the evidence for which I had travelled.

Our neighbours’ arrangements provide useful comparisons, provided we resist turning them into universal laws. Every Dwarf learns Imperial Dwarven for intercourse between holds, while the standing of the hold forms as dialects or separate languages remains disputed. Common Elven is a compromise between tribes within a family which never split. Human political boundaries have produced a different appearance: four states support four named languages, though Caldric and Sudermarkish remain close relatives and substantial diversity exists within each frontier. Mutual understanding admits degrees and changes with exposure. The clerk who readily reads a neighbouring language may fail to understand a herder speaking his own. Neither experience, taken alone, settles where a language ends.

Railways and telegraphs now increase both intercourse and the pressure towards uniform usage. Telegraph tariffs encourage abbreviation; railway employment carries local expressions along routes which cut across older districts. Schools give children access to forms useful far beyond their homes, and I support that provision. I also oppose the punishment of children for supplying evidence their teachers have never troubled to collect. A competent survey records domestic conversation alongside official prose, asks speakers to explain differences, and preserves the circumstances of each utterance. When I next revise this account, some conclusions will almost certainly require alteration. I should prefer the necessary evidence to survive long enough to inconvenience me.