Preface
I. Types of languages
II. Inflectional languages
III. Agglutinative languages
IV. Isolating languages
V. The question of how to classify Esperanto
VI. Various language planes
VII. Where does Esperanto fit?
VIII. Conclusions
 

PREFACE

       The national and local languages of the world are firmly linked to specific peoples and places. But the International Language Esperanto is identified with no particular nation or geographical area.

       Nevertheless, scholars, particularly those who do not speak Esperanto, occasionally express doubts about this principle, which dates from the time of Zamenhof himself, who first created the language, "If Zamenhof really knew only European languages," they ask, "how could he avoid building European grammatical and semantic principles into his language?" Even if a scientist feels obliged to caution that there are no purely European or purely Asiatic languages, and that semantic relations between words are primarily the result of the way a language is used by speakers and not of a priori definition, this perfectly serious question still merits a considered answer.

       If in fact the language created by Zamenhof were shown to be linked exclusively to Europe, its claims to linguistic neutrality would obviously be compromised. One could then truthfully assert that, although it is easier than other languages now used in international relations and therefore deserves serious consideration as an international language, nevertheless it is not a neutral medium of communication among cultures.

       But the scientific test of this question lies not in the historical limits of Zamenhof's knowledge, or in superficial characteristics of the language, but in the actual experiences of learning the language in various parts of the world and in the fundamental structure of Esperanto itself. With respect to the first problem, there exists a general knowledge about the actual learnability of Esperanto, not only in contrast to other languages, but also in comparisons among students from various parts of the world and various language groups. But there are very few truly scientific and objective studies. Such studies are urgently needed.

       The second question, as to whether Esperanto is a European language in any but the most superficial ways, is a matter which has long interested scholars, but has only recently received serious attention by people with a thorough knowledge both of Esperanto and of comparative linguistics. The present study, by one of our most distinguished linguists, Claude Piron (who feels at home not only in his native French, but also, for example, in Chinese), is a pioneer work in a field still barely explored. It is not, and is not intended to be, a polished study. It is an informal address presented in Geneva, on May 15, 1976, at a weekend meeting of Esperanto speakers. Fortunately we can include the study in the present series as a first step in the identification of the full internationality of Esperanto. We hope that it may also stimulate other competent researchers to enter this important field.

       Humphrey Tonkin
       University of Pennsylvania
       Philadelphia, February 1981

I. TYPES OF LANGUAGES

       Do the expressions "European language" and "Asiatic language" mean anything? In fact they do not. In Asia many languages are spoken (including Persian, Bengali, and Sinhalese) which, structurally and historically, belong to the same family as the majority of the European languages, the so-called "Indo-European" family. And in Europe millions of people speak languages (such as Turkish, Finnish, Hungarian, and Maltese) that belong to categories much more widely represented outside Europe. Generally and traditionally, languages are divided into three categories:

a) inflectional languages, such as the Indo-European and Semitic languages,
b) agglutinative languages, such as Hungarian and Turkish, and
c) isolating languages, such as Chinese and Vietnamese.

       Two criteria are traditionally used to classify a language into one of these categories: first, the way in which the language in question expresses grammatical relationships and, second, the processes by which its word elements are transformed or grouped to acquire new significance or a new role in a sentence. According to a more structuralist formula, one can say that the criterion is the proportion of morphemes whose form may vary. (1)

       A morpheme is defined as the smallest unit with linguistic significance. For example, the French word reverrai, 'I shall see', contains three morphemes: re signals repetition, ver the idea of seeing, and rai is an amalgam at the same time expressing the senses of 'person who speaks', 'singular' and 'future time'.

       The various morphemes of a language have different frequencies, of course. Of the three cited, re is more common in French than ver (and other forms of ver, such as vis, voi, voy, vu, and so on), since we find it in all sorts of words with the same sense of' repeat',' start again',' return' and the like. And rai is commoner than re: in fact it can occur with any French verb with the three senses just mentioned for it.

       The most frequent morphemes are those that signal grammatical functions. Their meaning content is very impoverished. (If I were to say, - 's has -ed me -ly! practically no information would be transmitted to you.) We call these grammatical morphemes. What are called semantemes (or lexemes) are the less common morphemes, whose meaning content is much richer. (The line "Take spade work garden," even if not entirely clear, nevertheless transmits a considerable amount of information. We would never say such a sentence, but we use very similar structures in classified ad sections of our magazines and newspapers every day.) It is among these semantemes that we usually class the "affixes''. Affixes, according to the definition generally used in linguistics (but not really valid for the morphemes which Esperanto speakers call afiksoj), are useful for derivations and cannot be used alone. Re in the French word reverrai is an affix: it can be attached to many roots to form derivatives, but always appears linked to another semanteme and never stands alone. We can now define the three traditional categories of language this way:

       1) If there is variation in the forms of all the kinds of morphemes, including semantemes other than affixes, the language is inflectional. For example,

       2) If only the grammatical morphemes change form, and among the semantemes, only the affixes, then the language is agglutinative. For example, Hungarian:

       3) If none of the morphemes vary in form, the language is isolating. For example, Chinese:

       Comparing the pairs brief I brevity and unjust/injustice is a good way to throw the three types of variation into relief:

       In French, an inflectional language, variation occurs both at the level of the root - bref changes to brièv - and at the level of the suffix. Thus one finds for example the suffix -té in one case (brièveté) and -ice in another (in-justice). In Hungarian, an agglutinative language, only the affixes vary: the radicals rövid and igazságtalan do not change; the suffix is the same, but it appears in two forms: -ség and -ság. In Chinese, an isolating language, no variation occurs: the radicals jiănluè and fēizhèngyì do not change, and the suffix -xìng is used in both cases and without change.

* * *

       Let us now make some observations on what we have discovered. For convenience we keep the traditional terms " inflectional"," agglutinative " and "isolating", although they are poorly chosen and derive from an insufficient analysis of the facts. For example, Chinese is generally cited as the type case of an isolating language, but in fact it contains many morphemes which cannot be used in isolation. This is true not only of many affixes, such as the nü and hùa mentioned above, but also of many other semantemes. The morpheme fù, for example, which means 'father', is never used alone in ordinary language (i.e. exept in proverbs, maxims, and poetic expressions); one says fùqīn 'father', fùmù, 'parents' fùxìzhìdù, 'patriarchal system', etc. On the other hand, Chinese constantly uses a system traditionally regarded as typically agglutinative: it adds morphemes to each other to form sometimes very long words:

       (Chinese morphemes are used singly in the old written language, the so-called wényán. But wényán was never a spoken language, and it would be wrong to confuse it with the language customarily called "Chinese".)

       To classify a language into one of the three categories, the criterion need not apply one hundred percent. A small error must be anticipated, if for no other reason than the regular variation caused by the sound system. For example, in Chinese the suffix -ĕr added to a morpheme ending in a consonant masks that terminal consonant and sometimes modifies the preceding vowel. And when it occurs with a reduplicated root, the second appearance of the root shifts to the first tone (-) if it is not already in that tone:

       Sound variation in a semanteme is relatively frequent in Japanese:

       We may count a language as belonging to the category in question if the criterion applies to at least ninety percent of the morphemes appearing in, say, ten minutes of ordinary conversation. We are now able to examine the three language categories in more detail.

II. INFLECTIONAL LANGUAGES

       In all of these languages, roots transform both in derivations and with changes of grammatical function. Comparing them with Esperanto, in which the root never varies, makes this property of the inflectional languages clear. To underline the unchangeability of the Esperanto morphemes, we separate them with a hyphen in the following list and compare them with German, English, and Russian:

       The Semitic languages are also considered inflectional .But their inflection is somewhat different: the form of the derived words changes, but the basic consonant frame remains constant. In Arabic, for example, the consonant frame KTB means "write, compose":

III. AGGLUTINATIVE LANGUAGES

       Agglutinative languages are characterized by invariant stems to which are added suffixes which cannot be used alone and whose vowels may change depending upon the vowel types found in the roots to which they are attached. Let us take the Turkish expression kīrīlmadīlarmī? by way of an example. It is composed this way:

       The word thus means 'Weren't they broken?'

       Here are some other examples which underline the law of vowel harmony in Turkish. Depending upon whether the vowel of the root is e or a, the plural is expressed by -ler or -lar, 'my' by -im or -īm, and 'to' (direction, attribute, destiny) by -e or -a.

       Two facts show that these suffixes are not words, and therefore, are en-tirely different from Esperanto (given in parentheses in the following examples). First, the Turkish vowels change:

       Second, they are not used independently. For example 'my' is -im, -īm, -urn, or -m:

       But these suffixes (-um,-im, and so on) cannot be used separately to mean 'my'. If a separate word is needed, one must say benim (literally 'of me', the personal pronoun (ben "I", sen 'you') and the suffix that translates our possessive adjective (-im 'my', -in 'your').

IV. ISOLATING LANGUAGES

       The majority of the isolating languages are spoken in Asia, but some also exist in Africa. In the Americas the so-called Creole languages also belong in this category. They are called isolating because it is imagined that in these languages words consist of elements that are self-sufficiently (isolating) usable. In fact this is not the case. Many Chinese morphemes, as we saw above, occur only in connection with others. What really distinguishes the isolating languages is the fact that the morphemes are invariant, each having a form that remains constant in all of its appearances, whether they be combinations, derivations or grammatical variations. In such a language, a person who knows how to say "I" and knows the rules of the language, automatically knows how to say 'me', 'my', 'mine', etc. All nuances, specifications, derivations which, in other types of languages, are expressed by changes of (or in) the morphemes are here translated by word order or by perfectly regular use or combination of morphemes which cannot vary. There exist, also, various processes such as reduplication - cf. Malay puteh 'white', puteh puteh 'whitish' - which it is unnecessary to treat fully here. The important point is that, whatever the change in meaning or grammar, the morpheme itself remains untouched. A grammatical function is signaled either by word order or by invariant grammatical morphemes.

       Both in Malay and in the French-derived Creole of the Caribbean, the modifying word stands after the modified one: my mother's house is translated as rumah emak saya (Malay) or as caïe manman moin (Creole):

       In Chinese, the order is the other way around:

       (Wŏ mŭqīn fāngz and wŏde mŭqīnde fāngz are grammatically correct, since the word de, which shows modification, can be omitted if the sense is clear without it. The expression wŏ mŭqīnde fāngz is to be preferred only for stylistic, rhythmic considerations.)

       In an isolating language the reciprocity in the things spoken of is necessarily reflected in a parallel reciprocity in the elements signifying them.

Creole:
li ka-aller caïe manman      ou      ka-aller caïe manman li

Chinese:
tā dào nī mŭqīnde fángz qù      ou      nī dào tā mŭqīnde fángz qù

       The contrast appears if we compare the same sentence in, for example, an Indo-European language such as French or English:

French:
il va chez votre mère         vous allez chez sa mère

English:
he goes to your mother's house         you go to his mother's house

       There is similarity, but not identity, between he and his, but no similarity at all between il and sa, in contrast to the Creole li/li, the Chinese tā, tā(de). In the same way the verbs in French and English - he goes but you go, il va but vous allez - differ in form, whereas they do not in isolating languages (ka-aller, qù).

       In Creole, as in Chinese, to show the interchange of persons one simply interchanges the relevant pronouns; the other elements remain invariant. In French and English, as in all inflectional languages, this beautiful parallelism is entirely lacking. Also in agglutinative languages, because of the system of suffixes, the parallelism does not exist. (Cf. Turkish: 'he goes' gidiyor, 'you go' gidiyorsunuz, 'your mother' ananīz, 'his mother' anasī.)

       Like the agglutinative languages, the isolating ones use element combination quite widely. This observation is valid particularly for the so-called monosyllabic languages, in which the majority of the morphemes have but one syllable, like Chinese and Vietnamese. These languages exploit quite a bit the possibilities presented by metaphorical usage, as shown by the following examples from Chinese:

diàn electricity

By addition of xiàn "line" one creates:

       This system always manages to solve terminological problems. When elevators appeared, their function was to replace stairways, and the Chinese accordingly created the following compound:

        But what should be done when escalators appeared? Simple enough:

       (It is interesting that the expression moving stairs exists beside escalator in English as well.)

V. THE QUESTION OF HOW TO CLASSIFY ESPERANTO

       That the traditional categories are too rigid in comparison with the not very classifiable reality is evident from the fact that a given "definitional" trait can often be found in different language families or categories. For example, noun compounds are often very similar in several languages of the three above-mentioned categories, as we can see from the expression people's commissar:

German (inflectional): Volkskommissar
       (Volk = people; Kommissar = commissar)

Hungarian (agglutinative): népbiztos
       (nép = people; biztos = commissar)

Chinese (isolating): rénmín-wĕiyuán
       (rénmín = people; wĕiyuán = commissar)

       The German structure in this case is more closely related to that of languages in other categories such as Hungarian or Chinese than to languages in the same, Indo-European language family such as French:
       commissaire du peuple (literally, 'commissar of the people')
or Russian:
       narodnyj komissar (literally, 'people (adj.) commissar').

       Another example is presented by the possessive adjectives. They are in-dependent words substituting for the relevant noun in the majority of inflecting languages. Thus 'my father' is mon père in French and moj otec in Russian in just the same way as it is wŏde fùqīn in Chinese, an isolating language. But another system, which uses suffixes in a way at first glance typically characteristic of the agglutinative languages (Turkish baba 'father' but babam 'my father') turns up also in some inflectional lan-guages:

Persian (Indo-European)
       pidar = father
       pidaram = my father,

Arabic (Semitic)
       ab = father
       abi = my father.

       As a third example, let us take negation. The structure subject + negating word + verb exists in all three categories:

Russian (inflectional): ja ne ponimaju (I don't understand)
Hungarian (agglutinative): én nem értem
Chinese (isolating): wŏ bù dŏng.

On the other hand, the reverse also occurs, and languages of the same category or even the same family may have different negative structures. To the same "inflectional" group and the same "Indo-European" family also belong:

       We see that it would be wrong to base our argument on separate traits like those just mentioned when addressing the question of where Esperanto is situated among languages. The criterion defined at the beginning; namely, the proportion of the morphemes in which variation is possible, is much more precise and seems more appropriate. Nevertheless, since adepts of traditional classification will perhaps not accept it, we will also consider various other traits that are perhaps less significant but which still can help to locate Esperanto a little better in the vast spectrum of the languages of Europe and Asia.

       Let us suppose that centuries after a catastrophe has destroyed our civilization, archaeologists from a new culture little by little rediscover documents written in the languages of the present time, which had vanished till then. One of them uncovers texts in Esperanto and asks him-self how this language is situated relative to the others.

       In one of the documents he encounters the phrase Li legis multajn seriozajn librojn, 'He read many serious books'. At first glance he concludes that it is a typical inflectional language because of the grammatical agreement between the adjectives and the corresponding nouns (as in several Indo-European and Semitic languages). Looking at the stock of words, he proposes the hypothesis that Esperanto is an Indo-European language. Studying the matter further, he finds confirmation of this thesis, for example, in seeming word families like the following ones, which he notices in recovered parts of a dictionary:

or like:

which seem at first glance to be united by the concept of breaking. He concludes therefore that in Esperanto there exist formal families of words similar to those of the Romance languages. Among other things he notes the alternation of two roots (direkc/direkt and frag/frak) with the same meaning value. According to this archaeologist, Esperanto is therefore an inflectional, Indo-European Romance language.

       But let us suppose he does not limit his investigation to that. Continuing his research, he begins to realize that chance initially delivered to him an abnormal specimen (a dictionary), and that in Esperanto there are also other sorts of families of words, actually much more frequent, where each new word is formed by an invariant root and affixes of fixed form:

       This system, by which one regularly forms new words by adding affixes to invariant roots, is traditionally considered typical of agglutinative languages. Our archaeologist ascertains that this system is much more productive (i.e. comprises a much larger proportion of ordinary text) than the system of alternating roots like frak/frag. Indeed the alternating-root system proves to be very much the exception. He concludes that Esperanto belongs to the agglutinative group, which is perfectly confirmed by such verbal forms as li resanigeblis 'he was curable' or la raporto tradukendos 'the report will have to be translated', which flawlessly evoke the verbal system of Turkish.

       But, plunging yet deeper into his studies, he notices also that the "affixes" behave exactly like any other semanteme, with an ability to stand alone that is not found in the agglutinative languages. Affixes indeed are found entirely by themselves (with, of course, the final vowel): aro, ebla, iĝi, eco, and so on, and in combination with each other: ebliĝi, arigi, ebleco, aĉularo, and the like.

       Now is not the tendency of morphemes to stand alone the primary characteristic traditionally attributed to isolating languages? Esperanto in this respect shows itself more isolating than Chinese! The affixes of Turkish, Hungarian and other agglutinative languages, for their part, are true affixes, always linked to other semantemes, and not independent words, as are the inaccurately named "affixes" of Esperanto.

       His curiosity piqued, our archaeologist decides to look for other factors that might argue in favor of structural similarity between Espe-ranto and the isolating languages. And he gathers an abundance of them. Forms like:

ĝis nun (till now)      ĝisnuna (hitherto - adjective)
mi (I)      mia (my)      mia lando (my country)      mialanda (of my country)

       These all make him think of Chinese word formation. So does the fact that in the same article - we may pretend he found the 69th volume of the journal Esperanto from 1976, or anyway page 61 of it - he finds both estrarkunsido and estrara kunsido, obviously with the same meaning, and clearly typical of the Chinese tendency to use or omit at will the sign of modification de:

       Is Esperanto then an isolating language?

VI. VARIOUS LANGUAGE PLANES

       In fact, it is not possible to classify Esperanto without distinguishing at least three planes: intrinsic, intermediate and extrinsic. To define to which plane one or another language trait belongs, we shall use the following criterion: a given trait is considered as belonging to the extrinsic plane if a change can be introduced in it without giving to the speakers, generally, the feeling that the language is altered in its essence or in its identity; it is considered as belonging to the intrinsic plane if a change in it creates the feeling that the language has been fundamentally altered.

       From this point of view, the quality of the sounds is an extrinsic trait. No one feels the language very different depending upon whether a speaker has an Italian or a Danish accent in Esperanto. In both cases, Esperanto remains Esperanto. In the same way, English remains En-glish, whether it is spoken with a British, Indian or American accent. To substitute one word for another does not call forth a sense of an important change either. We do not have the feeling that we are speaking a different language if we switch from

       My father did not want her friend to use his novel automobile

to

       My dad did not wish her pal to use his brand new car.

The given arrangement of sounds that expresses a concept we can therefore regard as an extrinsic trait.

       When we reach the level of word order, the impression that we are changing the language becomes more acute. If I say My father wanted not that her friend use the car brand new, I arouse a sense of strangeness. But nevertheless this change does not render the language completely foreign. It remains English, even though perhaps poetic or archaic. We have reached a more interior plane than that of the sound system or the roots, but we are not yet at the kernel. Syntax is somewhat closer to the center. The phrase My father he wanted not that her friend she used of the brand new car sounds more foreign than the other just presented.

       And yet we do not have the same impression that the language has been attacked in its very identity as we would encountering such phrases as I's fatherman ha-unwill she's friendman go-use he's new-new earthing or Fatherem no willis friendha usu newan caron. These sentences are no longer English, despite the fact that nearly all the roots have been preserved and that the phonetic system need not be changed to pro-nounce them. Why? Because this time we have assaulted the intrinsic plane, that of fundamental grammatical conception. The verb system, the possessive adjectives and other traits are quite different from even archaic, poetic, regional or mildly foreign English.

       Proof that this plane is more fundamental than that of the forms of words we can take from the following point: the average speaker of English feels that a phrase as My moffy did not sut her shramp to gose the insable flar, although incomprehensible — it means nothing — might nevertheless be some kind of English or of English slang (in other words, it does not attack the identity of the language), whereas the sentence presented above {fatherem no willis friendha...) strikes even those who can decipher it as belonging to another linguistic universe.

       Accordingly, we can distinguish the following language planes:

       1) The kernel or intrinsic (fundamental, essential) plane: the basic type of grammar and of derivation, i.e. the manner in which the relations between words are indicated (e.g. which determines which), the details about this or that nuance (whether a thing is singular or plural, whether it is completed or continues, etc.), and the relations between the concepts (for example between 'brother' and 'brotherly', between 'avoid' and 'unavoidable', or among 'hair', 'split' and 'hairsplitter');
       2) the intermediate plane: syntax and customary word order;
       3) the extrinsic plane: the actual forms of words and the system of sounds.

VII. WHERE DOES ESPERANTO FIT?

The Intrinsic Plane

       As far as its core is concerned, Esperanto is an isolating language. It completely fulfils the structural criterion defined above: in variance of morphemes. Variations such as direkc/direkt or frag/frak make up only an exceedingly small proportion of what is said and written in Esperanto (between 0.1% and 0.3% of the sample studied by us). Further, these are not cases of a single morpheme occuring in various shapes, as with French directeur/diriger. This is shown by the fact that every Esperanto root can give rise to an entire series of new derivatives. Thus from direktor 'director' we get direktori, 'to act as a director', direktorigi 'to appoint as director', direktorado 'the exercising of the functions of a director', which are not at all synonyms with direkti 'to direct', direktigi 'to make somebody direct something', direktado 'the act of directing" and the like, from direkt. These are therefore cases of roots which are obviously related in terms of etymology but which are, structurally speaking, distinct morphemes.

       The idea that Esperanto is an isolating language is supported by the many basic features it shares with Chinese. (Since this text is meant for laymen, the linguistic facts are couched in terminology which is familiar to Westerners. It should be borne in mind that these terms, historically anchored as they are in the Indo-European understanding of language, are not fully adequate to describe the structures of other types of languages. The use of terms like "preposition" or "adverb", for example, must not be taken to mean that Esperanto and Chinese have prepositions and adverbs the way Western languages do.)

       1) The Esperanto "affixes" are actually full-fledged words. In this respect Esperanto is somewhat more isolating than Chinese is. Many Chinese affixes take on a new meaning when used alone. For example, the Chinese suffix -jia means 'specialist' in compounds:

       But jiā means 'family, home' when used alone. Many Chinese affixes cannot be used independently at all. For example the syllable nü desig-nates human females, but requires completion to stand as a word. 'Woman' is nüren (from rén 'human being') or nür or nüz (formed with noun formatives -r or -z). The suffix -huà, like the English -ation refers to a process. It occurs in lādīnghuà, 'romanization' (from Lādīng 'Latin'), but, like -ation, it cannot stand alone.

       2) The Esperanto relation between possessive adjectives and per-sonal pronouns has an exact counterpart in Chinese:

       This is no mere surface detail or coincidence, but on the contrary follows directly from the basically isolating nature of both languages. Neither agglutinative nor inflectional languages show this feature, which would not conform to their spirit.

       3) In Esperanto, as in Chinese, the verb lacks a conjugation. Esperanto verb endings play a role analogous to the particles which colour or demarcate the time and aspect features of Chinese verbs.

       4) The two languages structure the expression of negation similarly:

       5) In Esperanto it is usually prepositions, rather than suffixes (as in agglutinative languages), that introduce complements. Chinese generally resembles Esperanto in this matter, and there are Chinese equivalents of such prepositions as al, kun, per, por, anstataŭ, etc. that are used as in Esperanto. (However, for time and place complements Chinese uses a postposition. Thus zhuōz-shàng, literally 'table-above' means 'on the table'. This is often additionally heralded by a preposition: zài zhuōz-sháng, literally 'at table-above'.)

       6) As stated earlier, Esperanto word compounding also resembles that of Chinese, although Chinese uses the device much more extensively. Here are some more examples of morpheme-compounding which show an exact parallelism between the two languages.

       No such isomorphism obtains between these and the inflectional languages. In most of the latter, many of the relevant words are missing, as we see illustrated in the irregularities in the English translations above. Those which do exist are formed irregularly, as one can see from the following:

samlaridario/ tóngguórén /compatriot
samreligiano/ tóngjiàorén /coreligionist.

       Nevertheless, between Esperanto and other isolating languages there is also a difference: the indication of grammatical function is always obligatory in Esperanto and never so in other isolating languages. Because of this difference, and despite structural similarity, the style and overall sentence pattern of Esperanto diverge greatly from those of other isolating languages. In Chinese,

wŏ       I
wŏde       my
wŏmen       we
wŏmende       our

form a derivation table even more regular than in Esperanto. But the placement of the suffixes -de and -men is optional. Mia libro (Esperanto for 'my book') corresponds to either wŏde shū or wŏ shū. Sometimes an unambiguous context makes it possible to omit even the -men ending after a pronoun which, nevertheless, continues to function as a plural: wŏmende guó 'our country' can be (and usually is) clipped down to wŏ guó 'my/our country'.

       Thus the official Chinese text of the United Nations Charter begins Wŏ liánhéguó rénmín, literally 'I United Nations people', meaning 'We, the peoples of the United Nations'.

       This possibility of leaving grammatical function unexpressed enables isolating languages to neutralize the distinctions between passive and active, transitive and intransitive forms. More examples from Chinese:

I have not yet drunk (the) wine.

(Méi is an amalgam indicating at once negation and past tense.)

(The) wine has not yet been drunk.

       In the case of wine there is no risk of confusion, but in many cases only context makes the meaning clear. The construction zhè yú bù néng chī le may mean 'This fish can no longer be eaten' or 'This fish can no longer eat'.

       (Ambiguities of this sort crop up often in all languages which do not clearly mark grammatical function, including English, which seems to be evolving towards a Chinese-like structure. Thus people have different interpretations of the name of the International Auxiliary Language Association (IALA), an organization which played an influential part in the history of international planned languages. Some interpret it as 'association for an international auxiliary language'; others [e.g. Mr. Ric Berger in Historia del Lingua International, Morges: Editiones Interlingua, 1971, p. 2] as 'international association for an auxiliary language'. Oddly enough, such ambiguities usually go unnoticed: the first interpretation seems to have a kind of strength of obviousness which prevents consideration of the other possibilities and even the realization that they might exist.)

       In general, isolating languages other than Esperanto mark tense only where the context does not indicate the time of a verb's action. In Chinese, for instance, ordinary conversation distinguishes between

       tā lái ma? is he coming?

and

       tā láile ma? has he come? did he come?

       But if there is no doubt as to when the matters spoken of came, are coming, or will come to pass, time remains unmarked:

       Consider the verbs in the following sentence:

There was a person who came and said to him: "Master, wherever you go, I want to follow you."
Jen estis homo, kiu venis kaj diris al li: 'Mastro, kien ajn vi iros, mi volas sekvi vin'.

       Compare the compulsory expression of tense (in italics) in the English and Esperanto versions. The tense indications are completely absent in the Chinese original.

       In fact, the need to use verb endings in impersonal forms, in Esperanto, reminds one of Japanese, a language which is usually regarded as agglutinative.

       In some respects Esperanto resembles the agglutinative languages. But since the crucial test for being agglutinative - variability of shape of affixes or grammatical morphemes - yields negative results, one must consider Esperanto basically non-agglutinative. Yet the excep-tional visibility of the grammatical structure of the sentence is a feature which brings Esperanto closer to agglutinative than isolating languages.

       In the Japanese sentence watakushi-wa isha-o denwa-de yobimas 'I call the doctor by telephone', the entire grammatical "skeleton" of the sentence leaps to the eye (-wa marks the subject, -o the object, -de the instrument, -imas a present-tense verb).

       (Japanese is usually classified as an agglutinative language. Although most of its grammatical morphemes are invariant, it does pass our test, since there are two "conjugations", two categories of verbs with different endings. Besides, Japanese has some irregular verbs, although not enough of them to warrant its inclusion in the class of inflectional languages. As in Esperanto, the verb in Japanese has endings which contain markers of time and mode but not person. On the other hand the Japanese verb differs from its Esperanto counterpart in several ways, above all in that it shows a dimension of politeness. Thus for example Esperanto mangas 'eat(s)' corresponds to Japanese taberu if an intimate acquaintance is spoken to, but it corresponds to Japanese tabemas if one is speaking, in a main clause, to a distant person. Further, Japanese verb endings incorporate the expression of negation: koroshita 'killed', korosanakatta 'didn't kill', compared with Esperanto mortigis and ne mortigis, respectively. As a third difference, the personal pronoun is often understood, as in a telephone conversation sequence:

       The conversation partners rely on the context of the situation to make it clear that the question is "where are you?" and that the answer is I am home.")

       Esperanto is Indo-European only in its extrinsic aspects. Neverthe-less it shares one fundamental intrinsic trait with many languages of the Indo-European family: the need for the adjective and some pronouns, in the plural and in the "accusative", to agree with the words which bind them. However, in view of the complete regularity of the Esperanto system, it would be wrong to regard the plural and objective endings as inflectional. This remark is all the more valid because the relevant grammatical markers (j, n) merely attach to the word: they never take the place of another ending or induce modification of the stem (2).

       Although Esperanto shares many features with Indo-European languages, it is, then, fundamentally, non-inflectional in structure. In fact, the special character of Esperanto consists in its combination of two principles: complete autonomy and invariance of lexical and grammatical morphemes (a major trait of isolating languages) and readily perceptible grammatical analysis (which is to some extent a characteristic of agglutinative languages). The Esperanto phrases mia sonĝo 'my dream', mi sonĝas 'I am dreaming' and sonĝa mondo 'a dream world' expressly indicate the grammatical role of the concept sonĝ-, whereas-in English and French phrases, even the verb or noun function of the word dream and rêve must be guessed from the context ('I dream/my dream', je rêve/mon rêve).

       Only very seldom do Esperanto sentences contain elements whose role is not immediately apparent. One of the rare structures to harbour an occasional ambiguity is the compound word: son-ĉasisto can mean 'hunter of sounds' or 'one who hunts by means of sounds'. Such ambiguities also crop up in agglutinative languages, which excel in grammatical clarity.

The Middle Plane

       At the middle plane Esperanto is indubitably Slavic. It exhibits many Slavic characteristics:

       1) in word order and style (the normal word order of Esperanto texts tends to resemble Slavic word order):

Esperanto: mi lin vidis/mi vidis lin (I saw him)
Russian: ja ego uvidel/ja uvidel ego.

(The Western European languages assign their pronouns a definite, unalterable place.)

Esperanto: kiel vi fartas? (literally: how you do?)
Russian: kak vy poživaete?
English: how are you?
Esperanto: kion li legas? (literally: what he reads?)
Russian: čto on čitaet?
English: what is he reading?

(The Western European languages tend to position their pronouns after the verb in such sentences.)

       2) in syntax:

       a) sequence of tenses

Esperanto: mi pensis, ke pluvas
Russian: ja dumal, čto dožd' idët
English: I thought it was (literally: is) raining;

       b) obligatory reflexive

Esperanto: ŝi amas sian edzon
Russian: ona ljubit svoego muža
English: she loves her (own) husband

(in contrast to:)

Esperanto: ŝi amas ŝian edzon
Russian: ona ljubit eë muža
English: she loves her (someone else's) husband;

       c) a distinction in grammatical form between modifying and predicative complements

Esperanto: la kuracisto trovis la sanan infanon
       la kuracisto trovis la infanon sana
Russian: vrač našel zdorovogo rebenka
       vrač našel rebenka zdorovym
English: the doctor found the healthy child
       the doctor found the child healthy;

       d) use of adverbial form with infinitival or clausal subject

Esperanto: laboro estas necesa
       labori estas necese
Russian: rabota nužna
       rabotat' nužno
English: work is necessary
       to work is necessary (literally: necessarily);

       e) infinitive as prepositionless complement of noun

Esperanto: la deziro venki
Russian: želanie pobedit'
English: the desire to win;

       f) asymmetry or constraints placed on the use of prepositions followed by infinitives

       While Esperanto allows us to say antaŭ ol foriri 'before leaving', it usually avoids post ol foriri 'after leaving', preferring the forms foririnte 'having left' or post foriro 'after departure'. In Russian they say prežde čem ujti 'before leaving', but not posle čem ujti 'after leaving' preferring instead ušedši 'having left' or posle uxoda 'after departure'. Compare this with the symmetry of the English forms just quoted or, for example, with Spanish: antes de salir and despues de salir. (Note that the redundant occurrence of ol 'than' in antaŭ ol foriri, literally 'before than to leave' — it would have been just as clear to say antaŭ foriri — comes by way of literal translation from the Russian prežde čem ujti, which shows the same čem (Esperanto: ol) as the expression bol'še čem on 'bigger than he' (Esperanto pli granda ol li.) In Esperanto we say por transdoni 'in order to transmit' but not pro transdoni 'because of to transmit', and in Russian čtoby peredat' but not iz-za peredat'. In Spanish, on the other hand, there is para transmitir and por transmitir.

(Extensive discussion about whether sen 'without' plus an infinitive is admissible in Esperanto derives from this same Slavic quality. The fact that the structure in question is quite frequent in Romance languages (Spanish: sin olvidar, French: sans oublier) and in Germanic languages (German: ohne zu vergessen) has led to widespread use of this structure in Esperanto. But, because it does not occur in Slavic languages (where one expresses the idea by using 'not' plus an adverbial participle, as in Russian ne zabyvaja 'not forgetting'), it was generally alien to Zamenhof s own usage and thus was thrown out by the purists.)

       3) in various non-Western distinctions of nuance (aspects):

       4) in the obligatory distinction between transitivity and in transitivity:

       5) in many turns of phrase:

       6) in the meaning of many roots even if they are from Romance languages:

       The semantic field of plena 'full' is the same as that of Russian polnyj, and does not coincide with that of French plein, Italian pieno or Spanish lleno. Esperanto plena verkaro 'complete collection of works' corresponds to Russian polnoe sobranie socinenij. In no Romance language would the word derived from Latin plenus be used in such a case.

(Note the Slavic semantics attached to a Romance root, clearly cognate with English and French 'occasion'.)

       7) in the forms taken by loanwords:

(The French forms, matin and four Martin, without Slavic influence, would have yielded the non-existent forms *matino and *martina forno. The transmutation of French -i- to Slavic -e- can be seen in the Russian term for four Martin: marten or martenovskaja peč'. Similarly Polish transcribes the name of Chopin as Szopen.)

(Note that the alternation between /s/ and /š/ is identical in Esperanto and Russian, though both have borrowed both words from languages where the alternation does not occur in this way.)

       8) in the writing system:

       Accent-marked consonants occur in Czech, Slovak, Croatian and Slovenian. Esperanto has ĉ, ĝ, ĥ, ĵ, and ŝ. The invariant pronunciation of c, even in front of a, o, and u, occurs in no Western language, but does occur in Romanized Slavic languages. The traditional way to abbreviate in Esperanto, with a hyphen, follows the Russian model, unused in the West:

The Extrinsic Plane

       As far as the origin of its root words is concerned, Esperanto is mostly Romance and Germanic, with a predominance of the Romance (specifically French) element. In the Germanic element one notices a prevalence of the German contribution. The word-initial clusters /šp, št, šm/, for instance, occur only in German and Yiddish.

       The Esperanto sound system approximates that of the Romance languages, specifically Italian, but with some Eastern European features. The latter include the complete series of palato-alveolars /č, š, dž, ž/ and perhaps also the frequency of the sequences /oj, aj/, although in this case one might postulate concomitant influence of Yiddish and of the traditional pronunciation of Ancient Greek. Stress follows the Polish model.

       It would be interesting to test the following "law" in detail: «Except in those cases where only the two basic principles of invariance of autonomous roots and readily available grammatical analysis apply, if a linguistic feature is shared by Germanic and Slavic languages, Esperanto has it; if a linguistic feature is shared by Romance and Slavic languages, Esperanto has it; if no two of the three groups share a way to solve a particular problem, Esperanto follows (a) Slavic languages if the matter pertains to the middle-plane (syntax, style, idiom) or (b) either Germanic or Romance languages if the matter pertains to the extrinsic plane (phonetics, word shapes).» The word 'law' is obviously too strong. Rather it would be more accurate to say that, when the conditions it indicates are not met, tensions show up in the language. We have already cited the case of sen + infinitive. Other examples can be found. The current forms jarcento 'year-hundred' and jarmilo 'year-thousand' for 'century' and 'millenium', for example, follow the Germanic model, not the Slavic and Latin models which underlay the older forms centjaro and miljaro.

       Another example of strain is the passive participles. When one means 'the contract was signed at 10 o'clock', should one say la kontrakto estis subskribita je la 10a or estis subskribata? Subskribata seems to mean 'being signed'. (The most logical form would probably be the German/Dutch form igis subskribita 'became signed'; At a few seconds before ten, the contract is being signed; at a few seconds after, it has certainly already been signed; at the second when the last pen leaves the paper, it is transformed from the state of 'being signed' to that of 'having been signed', and thus it 'becomes signed'. Hence if one speaks about it later, one ought logically to say, 'at ten it became having-been-signed': ĝi iĝis subskribita. But the usage habits of Slavic, Romance and English speakers are perhaps too strong for them to accept such a form.) The -ita form seems to be winning now, although only after facing serious resistance. In the usage of the first speakers of Esperanto, who mostly lived in Eastern Europe, the -ita form perfectly corresponded to the Slavic past passive participle of the perfective aspect {subskribita = Russian podpisannyj) and the -ata form to the present passive participle of the imperfective aspect {subskribata = Russian podpisyvaemyj). In that system, endings indicate more than time; they intertwine notions of time and aspect: -ata stresses the fact that the action takes place over time, without regard to a definite end point, if any, while -ita underscores the reaching of a definite end point.

       What strikes Slavs as obvious in this is inscrutable to the Germanic speaker. German er ging 'he went/he was going', like Esperanto li iris, translates Russian on šël, French il allait, Spanish andaba (action regarded as repeated or continuing) as well as it does Russian on pošël, French il alla, Spanish anduvo (a precise, one-shot, definite action). Consequently, Germanic-language Esperanto-speakers fail to find in the phrase estis subskribata the feeling of extendedness in time, in duration, which it conveys to the Slavs.

       As for Romance speakers, they find this shade of meaning less foreign than the Germanic speakers do, since they have it in their conjugation; but in their languages it never affects participles, so that it is hard for them simply to follow the system which comes naturally to the Slavs. As a result, passive participles now constitute a point of tension in the language, and the usage is not too coherent here. One often notices Westerners using -ita participles in situations where the action is clearly a repeated one.

       Let us now turn to some examples of the "law" in effect, a) Examples of features shared by Germanic and Slavic languages:
       - In the sound system, the use of /kv/ where Romance languages have /kw/ or /k/ (a feature characterizing most but not all Germanic languages).
       - Distinction between 'her' and 'his' (Esperanto ŝia and lia), unlike Romance languages. (French son livre, Spanish su libro, can both mean either 'his book' or 'her book'.)
       - Habit of placing the attribute before its head. In a Romance language one would never speak of a terrible, for me intolerable situation (Esperanto: terura, por mi neelportebla situacio); the adjectives would come after the noun. (Interestingly enough, in Zamenhof s usage the attributive adjective generally follows the Polish and not the Russian model. For international language, for ex-ample, Zamenhof tends to say lingvo internacia, corresponding to Polish jezyk międzynarodowy, rather than internacia lingvo, corres-ponding to Russian meždunarodnyj jazyk.)

       b) Examples of features shared by Romance and Slavic languages:

       - In the sound system, the voiceless consonants are unaspirated: /p, t, k/ are pronounced as in Polish and Italian, not as in English, German, and the Scandinavian languages.
       - The Esperanto prefix mal-, used for the derivation of antonyms, was probably selected in preference to Latin in- and dis- or to Germanic un-(or on-) because, although it helps form many derivatives in Romance languages, where it specifically means 'bad', it also occurs as a Slavic prefix with the sense of 'little':

       The use of the Esperanto negative prefix ne-, we note in passing, is also Slavic:

       - Negative form of verb. The Esperanto structure (ne + verb) follows the model of all Slavic languages and of all Romance languages except French. It does not occur anywhere in the Germanic languages.

VIII. CONCLUSIONS

       The problem of where to place Esperanto in the vast gamut of human languages is not easily solved. We have approached it here essentially from the point of view of the intrinsic structure. This has yielded the con-clusion that Esperanto is basically an isolating language.

       With respect to the origin of lexical material, we would have to classify it between the Romance and Germanic families, with predomi-nance of the Romance element. The two most closely connected languages would then be English and Romansh.

       A criterion focusing on style and syntax would accentuate the Slavic quality of the International Language. But we have also noted that Esperanto borders on the agglutinative type in many ways.

       The problem turns out to be especially complex because of two factors. On the one hand Zamenhof probably wanted to construct a highly coherent system, and the invariance of the morphemes most likely owes more to this intention than to any wish to follow a Creole or Chinese model. However, we cannot exclude the hypothesis that he might have been affected by an awareness that in the typical intercultural situation, when two people know just a few basic elements of a common language and try to get across to each other, they end up spontaneously transforming this poorly known language into a sort of isolating language.

       Since Zamenhof himself had a command only of inflectional languages, their influence, though contrary to the basic principles chosen for the linguistic instrument which he forged, dominated his way of writing and speaking the newly built language, and the model offered to the public was subject to internal strains right from the start.

       On the other hand, the isolating structure of Esperanto and its extreme regularity were sharply criticized by sophisticated people in Western Europe, leading Zamenhof more and more to mix the initial language with elements more consonant with the major Western structures; hence the existence of doublets like redaktisto/redaktoro 'editor', redaktejo/redakcio 'editorial office', etc. This tendency seems to contradict his early ideas, judging from the remark made in the fifteenth rule of the Fundamento (Fundamentals) to the effect that for the so-called "international words" one should follow a policy of taking only the root and then constructing the derivatives according to the autonomous rules of Esperanto itself.

       Be that as it may, another factor made its presence felt: the "substra-tum". The community which adopted Esperanto speaks, for the most part, inflectional languages, and consequently is unfamiliar with or dislikes the latent potential of isolating languages, with the result that it tends to solve problems of expression (particularity of terminology) along lines which may be viewed as antithetical to the basic spirit of the language. Zamenhof s vocabulary, especially in the texts of the early years, is much more "Chinese" than that of most later writers. Thus Zamenhof used to say ununombro 'singular' (an exact counterpart to Chinese dānshù from dān 'single' and shù 'number') where later grammarians introduced the term singularo.

       Given a heterogeneous substratum, the language has been pulled and stretched by divergent tendencies. The vocabulary shows a tension between, on the one hand, a "naturalistic" tendency, which shows up in many places in the Plena Ilustrita Vortaro (Complete Illustrated Dictionary)-a tendency to borrow profusely from Greek and Latin, more or less respecting their spelling systems (leading to words like relegacii and ekshibicio)—and on the other hand an "Esperantist simplicity", i.e. an inclination to use short roots (like rilegi and ekzibo) and to exploit Esperanto's derivational and compounding possibilities instead of intro-ducing neologisms.

       The grammar shows another tension: that between "conservatism" (for example in a refusal to use the form gis kiam 'until when' for 'until' or the form sen .../ 'without ...ing') and "boldness" i.«. a wish to exploit as completely as possible the latent possibilities of the language, whatever the usage in Zamenhof s day may have been. Examples of the latter include introduction of participles ending in -unta and -uta and abbreviation (on the part of Lanti, among others) of the traditional forms such as junulino or malsanulino to junino or malsanino. Many other examples readily come to mind.

       An individual may, of course, be conservative on one point and bold on another. At first sight it would appear that the majority of the Esperanto-speaking public leans towards the "conservative" end, while authors, especially poets, prefer to be "bold".

       It is to a large extent in these tensions that Esperanto's qualities as a living language are rooted. From a structuralist perspective, it is fascinating to observe the evolution of this extraordinary phenomenon. We find here a structure created by a single person but eluding his control and obeying laws whose existence he, the author, was unaware of. We find it turning into the locus of a remarkable dialectic in the hands of an international community constituting a true diaspora. And no authority, even if prestigiously termed an Academy, can ever freeze the conflicting tendencies toward assimilation and conformity which force the linguistic structure to adjust to the community that uses it, and in turn force the community to adjust to a linguistic structure whose laws are stronger than the community itself.

NOTES

       1. The reader with a high level of linguistic competence will rightly criticize the use of only one criterion. But he should remember that this text is directed to laymen. It would not be possible in so short a compass to treat the very complex question of criteria for linguistic typology. For example, many take the prime criterion of isolation to be the fact that in an isolating language the majority of the words are monomorphemic. But if we applied this criterion, Chinese would cease to belong among the isolating languages and would become agglutinative. That would be an interesting and defensible thesis, although presumably surprising for many. But since linguists generally continue to class Chinese among isolating languages, we limit ourselves to a single, if fundamental, characteristic, enabling us to retain the traditional divisions of languages. The typological considerations presented in the present work should be regarded more as a device for clarifying the position of Esperanto in relation to other languages than as a new way of approaching the problem of typology. We are well aware that our criterion would raise difficult problems if one were to apply it, for example, to situating the Bantu languages. For the same goal of simplicity and the same limitations of time, we have not considered the so-called "polysynthetic" type, into which some American Indian languages, among others, may be classified.

       2. In this regard Esperanto differs significantly from the Esperanto-derived project Ido. The transition in Esperanto from infano 'child' to infanoj 'children' is not-inflectional; it is an additive process: infan-o-j (child-noun-plural) exactly identical to the Chinese hái-z-men. The Ido plural is formed not by addition but by substitution of -i for -o: infanto 'child', infanti 'children'. Structurally, this is quite a different matter.

Esperanto Documents, number 22A (1981)