Budai’s task was made still more difficult by the lack of any single recognisable character or any related script on which he might base a working theory. Not that he could actually conjure up memories of the various, mostly long-disused runes he had never properly studied. He had no hypothesis, nothing tangible, not a word or a name that could serve as a faint guiding light. But shouldn’t there have been one somewhere?
In determining the nature of a text the first consideration must be the number of characters used. Systems where a single word represents a concept tend to have a great many: Chinese, for example, supposedly employs 50,000 of them. Systems using syllables would, naturally, need fewer. The previously mentioned Ancient Cretan uses eighty-nine, Cypriot script forty-four, modern Japanese 140. Systems that rely on individual letters, like modern European ones, require even fewer; English has twenty-six, Russian thirty-two, French twenty-two, and so forth.
So he set about scribbling down as many characters as he could find in the book. These quickly exceeded a hundred without showing any sign of running out. Did that mean the script employed syllables? Individual groups of characters seemed too long for that. Was each character a word? He carried on working but his collection of signs seemed ever more difficult to sort into groups and examine. He started to wonder whether he had written out the same thing several times.
Having got to number 237 he lost hope of ever coming to an end and gave up. Now he tried another way, picking out characters at random and making quick improvised calculations to work out what characters occurred most or least frequently. Assuming that there tended to be fewer vowels than consonants, there should be more of the former. In Hungarian, for example, the commonest letters were e and a, followed by t, s, n, and l, the least common being x, q, and w. It would be different for different languages of course. The only problem was that if it turned out to be syllabic script each character was bound to comprise at least one consonant and one vowel and then all his efforts would be in vain.
Another question occurred to him. Did this language contain articles as Ancient Greek, Arabic, Hebrew, English, German, Italian, Spanish — and a variety of other languages — did? Because, if it did, he could use that as a starting point. Better still, it would be easier to detect such things in written text than in speech where the sound of the article might merge with the sound of the noun. The way the blue-uniformed lift girl pronounced her name on the eighteenth floor might have been a clue: it sounded longer the second time, Etyetye, Pepepe or whatever. Might there have been an article there? That e, pe or tye? And how might it be written? If he could discover that, he would have one or two letters that he might be able to read, even if only as ancillaries.
That was what he looked for now: short, identical or similar words that might be assumed to appear at the head of longer rows of words, at the beginnings of sentences or paragraphs, for example. But however he leafed through the newspaper and the newly acquired book, he found few such characters and even when he did discover some that seemed to resemble each other, they comprised five or six characters at least, which seemed to suggest that they were not articles. There was a group that might perhaps have served as such, a little word that did recur and consisted of two characters only but invariably it turned up at the end of paragraphs and at the end of certain chapters or short stories. He remembered that in Romanian, Bulgarian, Albanian and Mordvine an end-sign took the place of the article — could that be the case here? Or might there be no articles at all, as in Latin, Finnish, Chinese or the Slavonic languages? This little closing word might be the equivalent of a phrase or utterance such as the Latin dixi, or the uff, familiar from Indian novels.
If the words for yes and no exist in a language then it may be assumed that no should turn up relatively often. It is the same for how and but and and although they might be combined with other words, such as the Latin — que. He had once read that in Hungarian the most common agglutinating substantives are people, house, flat, country and so forth. But would that be the same here? And if so, how should he fish these out or sift them from the ocean of text before him? How would he be able to tell which was which?
What if he could hack his way into this jungle of language by studying the syntax? In order to do this he would have to look out for groups of characters that were similar but not identical. If, for example, the first few were similar but the latter part different, then one might assume that these were various inflexions or agglutinations.
He spent a whole day looking into this kind of thing, neatly noting down everything in columns for easier scrutiny. He found mostly those where only the first two or three elements were the same. Of course he couldn’t exclude the possibility that these were individual words, their resemblance merely coincidental, such as batter and battle, or like the English six and sister. But if they were root-words or syllables, exactly as he hoped they were, then what did the various agglutinations mean? Were they substantives, verbal inflections, formative syllables, notations, postpositions? Might they represent differences of gender as in the case of the French directeur and directrice? It might be that what he took to be root was verbal prefix, or the first term of an agglutination comprising several parts. Or indeed much else.
If, however, the last syllable matched — for he did come across some in his enquiries that did — it was possible to image that these were different words with one kind of inflection. In Hungarian, for example, there were szobában (in the room), házban (in the house), városban (in the town), or szobának (for the room), háznak (for the house) and városnak (for the town) and many others that worked in similar ways depending on how many inflections there were available in the language. But there was no way of establishing what suffix corresponded with what inflection or how they were to be sounded. It all lacked foundation. And what if it was not a case of inflection but simply a coincidental rhyme, such as paper and caper in English, and what about damper and hamper, not to mention pamper and the like? It might have been a verb that took a prefix, such as undergoing, thoroughgoing, partygoing and masses of other such, to judge by the thirty-odd languages he could more or less read. But then back to the basic problem: how to establish what was what?
It was no use: the whole thing was sterile, a series of guesses with practically nothing guaranteed. He’d not get far with vague theories, logic, hopeful chance substitutions and games of patience, or if he did get anywhere it would be at a snail’s pace. And even if he was capable of the titanic effort of performing correct calculations based on notions of statistical probability, calculations that required endless supplies of energy, even if he did somehow succeed not only in assembling an alphabet but in solving all the associated phonetic problems without utterly exhausting himself — an achievement that was nowhere in sight for he had not yet identified a single letter — that would still not mean he understood any of the language. Scholars can, for example, read the ancient Etruscan script without difficulty, but not even the finest minds employing the most up-to-date means have yet succeeded in making sense of the language, bar a few dozen words and two or three grammatical features. Furthermore, the very origins of the language, its place in the family of languages, remained mysterious and controversial — and this local epepe might be an idiomatic expression just as isolated and unrelated to other idioms and languages as did the Etruscan to the Basque or some of the African to Caucasian tongues.