VIII. Direct and Indirect Derivation

(The term derivation indicates the relation of words in various groups, i.e., verbs, nouns, adjectives, adverbs. It is not used to indicate the etymological origin of a word.)

INTERLINGUISTS GENERALLY AGREE THAT the problem of derivation is one of the most important for the future planned language. The principles and rules of the present systems differ so widely that it has so far been impossible to formulate a final solution. We cannot model these rules exclusively on the ethnic languages because they are not regular, and we cannot contrive some purely a priori solution which would prove too difficult in everyday use. The only thing we can do is to compare the fundamental principles of each system to get a clear idea of the problem itself and its possible solutions.

In chapters I-IV the rules governing derivation in each of the four systems have been summarized. A comparative examination of these proposals can now be attempted which will more fully illustrate the problems involved.

VIII - 1. Derivation in ethnic languages

The ethnic languages use both indirect (or mediate, i.e., by means of a medium), and direct (or immediate) derivation. In indirect derivation affixes are used to modify the root [change, unchangeable, changeability; resist, resistible, resistibility, etc.]. The method of indirect derivation is exact if the meaning of the affix in use is monosignificant, i.e., if it stands for one meaning only and is invariable. We do not learn our mother tongue by an analysis of these elements but rather by the use of ready-made forms, and later by analogy to forms so learnt. Learning is made more difficult for the foreigner when he discovers a number of affixes which apparently fulfil the same or very similar functions [ir-resistible, il-logical, un-natural, in-variable, de-centralize, dis-interested, mis-represent, etc.].

Far more difficult to master are the usages of direct derivation which have been shaped by convention, and which vary considerably from one language to another. A nominal or substantival root is used to derive the verb directly, i.e., without the addition of an affix or a new verbal root. Examples in English are very common, and the meaning of the verb thus derived is fixed solely by usage. Shop, to shop, meaning to visit a shop with intent to purchase, ship, to ship, meaning to transport goods by means of ships, paper, to paper, meaning to affix paper to something, for example to paper a room; iron, to iron, meaning to smooth linen, etc., with iron, or to shackle with iron according to the context used; chain, to chain, meaning to secure or confine with chains; house, to house meaning to provide one or several people with habitation, the housing problem; to table a bill in the House, but German (auf-)tafeln, or -tischen means to dish up food; to feather one's nest, to book a seat, to water a plant, etc. This practice extends equally to adjectives from verbs or vice versa [dry, dry; cool, cool; warm, warm; clean, clean; open, open], and adjectives from nouns [secret, secret; equal, equal; right, right]. We may assume that these usages are familiar to all who have an adequate knowledge of English, and for them direct derivation is a very convenient way of expression. The difficulty only arises when this kind of convention is carried into the making of a planned language which should be equally easy for all who use it. For the meaning attached to verbs derived from nominal roots varies in the ethnic languages. Professor Stör (University of Prague) has given some examples which illustrate these divergences. In German Feder is used as a verb federn with three meanings, (1) to tar and feather, (2) to adorn with feathers, (3) to be elastic, Feder being the translation of feather and spring. In French plume becomes plumer with a completely different conventional meaning, it does not mean to adorn with feathers, but to pluck feathers. The meaning of to chain happens to agree in English and German, but in French chaîner means to measure with a chain [mesurer avec la chaîne]. We can well imagine that an auxiliary language which permits the direct derivation of nominal roots may cause considerable misunderstandings among those people who use this rule according to the traditions of their own language. The science of interlinguistics is today sufficiently advanced to enable us to avoid such complications in the planned language.

Another classification may here be of interest; it has already been used to examine some systems of planned language in previous chapters. It was proposed by Schleicher under the influence of Hegel. Two elements were distinguished by him, the elements carrying meaning, roots or full words, and the other elements which serve to indicate grammatical relations, affixes and terminations. Firth (The Tongues of Men) gives three formulae to illustrate the type of language:
(1) Meanings indicated by words, relations by position. R = root or full word, a = affix (prefix and suffix).

R¹ + R² +R³ + R^4.

(I wish (to) go there.) These were called the Isolating or Positional languages.
(2) Meanings carried by invariable root, relations by affixes which may be monosignificant.
a + R + a + a

These were called agglutinative languages.
(3) Meaning-roots and relational elements which became fused and formed an entity with a new meaning.
p + R + s¹²³ = (prefix plus root which may be modified [drink, drank, drunk plus suffix or suffixes. The suffix or suffixes necessitate and inflexion of the final root-letter or letters [concede, conce/ss/ion]). These are called flexional languages.
The languages quoted by Firth for the first pattern were Chinese and English, for the second Turkish, Swahili, Tamil and Korean, for the third Sanskrit, Arabic, Greek, and Latin.

If this classification is applied to systems of planned languages the first pattern would describe some a priori projects, the second Esperanto and Ido, and the third - to some extent - Occidental.

We have to qualify this statement in classifying a particular language according to one of the three patterns may be found in its structure. In saying that Occidental is flexional, we must also acknowledge that many words are formed according to the agglutinative pattern, but the admission of flexional elements in contrast to the other systems makes this an outstanding characteristic.

Again for the flexional pattern we shall distinguish the so-called internal and external flexion. One example, taken from English solely to illustrate the term internal flexion [sing, sang, sung], is not used in any stem of planned language.

The term `external flexion' as used in this book describes the change of the last letter of the root itself, e.g., comprehend and comprehensibility, in which the d has been changed into s. Occidental uses external flexion for a number of words and for some exceptional cases, as we have seen in Chapter III.

In applying Schleicher's classification to the present systems of planned languages, Occidental, containing a number of flexional characteristics, would rank as the highest developed system. Jespersen, however, outlines a different theory (Language, XXI, 7, 8). He finds that in early languages the number of irregularities, exceptions, anomalies, is greater than in modern ones. "It is true that we not infrequently see new irregularities spring up where the formations were formerly regular, but these instances are very far from counter-balancing the opposite class, in which words once irregularly inflected become regular, or are given up in favour of regularly inflected words. The tendency is more and more to denote the same thing by the same means in every case, to extend the meaning, or whatever it is, that is used in a large class of words to express a certain modification of the central idea, until it is used in all other words as well." Jespersen thus traces the simpler forms as the result of development from more archaic formations. He does not propose to reverse the theory `isolating-agglutinative-flexional' into `flexional-agglutinative-isolating.' Jespersen's opinion on the development of language after an enquiry of some 400 pages based on a life-long study is briefly and clearly stated, "the evolution of language shows a progressive tendency from inseparable irregular conglomerations to freely and regularly combinable short elements." If this dictum is acceptable to linguists who will formulate the coming form of a planned language, the system is likely to be based on the structural features of a perfected Esperanto-Ido type.

VIII - 2. Derivation in planned systems

Direct derivation is said to offer greater facility in the use of a planned language but surrenders for it some precision of expression. In the following paragraph we shall examine the rules of the four main systems in employing, more or less, direct and indirect derivation.

The examples in direct derivation of nominal roots given in the previous paragraph show that ethnic languages do not always agree on the meanings of the verbs thus derived. There are, however, examples where the meaning coincides in all languages and they have formed the basis of a long discussion in interlinguistic circles which has never yet been finally settled. A few typical examples may illustrate this point:

crown, to crowncouronne, couronnerKrone, krönen
hammer, to hammermarteau, martelerHammer, hämmern
pepper, to pepperpoivre, poivrerPfeffer, pfeffern
oil, to oilhuile, huilerOel, oelen


As we have seen in Chapter I Esperanto uses direct derivation which is not limited by any definite rule. The user of Esperanto may apply these forms throughout or limit himself to cases which are also to be found in ethnic languages and on which agreement of choice exists, as in the above examples. The verbs kroni, marteli, pipri, olei, which are understood by people knowing one of the great European languages, are directly derived from their respective nouns.

Some Esperantists have used the forms jes-i and ne-i meaning respectively `to say yes' and `to say no', or `to affirm' or `answer affirmatively', and `to negate' or `to deny'. Some such forms are ambiguous even to the experienced Esperantist and it would be better to replace them with new verbal roots. The necessity to use these forms in Esperanto is explained by the lack of adequate word-material. The English-Esperanto (Edinburgh) dictionary gives affirm = certigi, besides firmdiri, jesi; literally, however, cert-igi means to make certain. In this case, as in many others, the necessity for an ampler vocabulary must be faced if precision is to be attained. afirmi would well fit into the Esperanto vocabulary in the same way as it exists in the vocabulary of Ido.

The suffix -ad- in Esperanto stands for `duration' or `continuation' of an action but is widely used to distinguish the verbal noun from the nominal root from which the verb, in turn, was derived. We thus have the original word kron-o = the crown, kron-i = to crown, kron-ad-o = the act of crowning, coronation. Esperanto goes even further than ordinary direct derivation. It admits certain word building elements as complete words. The suffix -estr- may become a complete noun by adding the grammatical termination -o to it, and thus stands for master. This practice is quite common and gives an air of artificiality to the language which could be avoided by using the known roots of the European languages.

A very important feature of Esperanto is the acceptance of the rule of `necessity and sufficiency' proposed by R. de Saussure and accepted by the Lingva Komitato, i.e., "The formation of any constructed word is obtained by combining all the word elements (roots, affixes, and terminations) which are necessary and sufficient to evoke clearly the idea to be represented." To appreciate the importance of this decision we shall have to examine more fully the practices of indirect (or mediate) derivation in Esperanto.

Zamenhof has formulated the rule that, given the root, the noun is formed by adding -o to it, the adjective by adding -a, the adverb by adding -e, the infinitive by adding -i, etc. It follows that grammatical terminations are interchangeable, but there exists no rule to define the exact meaning of the words thus obtained. It is important that all forms should carry a definite meaning. The practice to add grammatical terminations to elements without a clarification of meaning is - as we have seen in Esperanto - extended not only to roots but also to affixes which are treated as roots.

Lébasnier distinguishes three cases, (1) that of a derivative which may have several meanings. The relation between verb and noun may, as in ethnic languages, vary considerably in meaning [brosi = to act by means of a brush; versi = to make verses; ori = to cover with a coat of gold; violini = to play on a violin]. Adjectives directly derived from nouns may also express different relationships [orfa = who is an orphan; herba = that which contains herbs; sulfuro and sulfura, but fero and feraj^a]; (2) that of one affix which may express different ideas. The suffix -ad- indicates `repetition' of paf/o in paf/ad/o, the act of providing with a crown in kron/o, kron/ad/o. The suffix -ul- = characterized by, is used in virg/ul/o but considered superfluous in vidvo, used in skeptikulo but not in katoliko, in mistikulo but not in elektito; (3) that of a similar idea which may be expressed by different affixes. The suffixes -il- and -aj^- are frequently interchangeable [baraj^o, barilo], and again justified by the rule of de Saussure we find sano and saneco, abomena and abomeninda, fiksi and fiksigi. The failure of Esperanto to apply a more precise and scientific derivation has often been put forward as a serious critcism and will be familiar to those who have occupied themselves with the problem of an auxiliary language for many years. Although it is an old argument, it will have to be considered as Esperanto is one of the systems which may be selected as a base project for the coming planned auxiliary language.

Most roots in Esperanto belong to a certain grammatical category. It may be verbal, nominal, or adjectival, according to whether the first derivative is a verb, a noun, or an adjective. Consequently we must know the root-word of each word-family in order to know to which grammatical category it belongs. We then form all other derivatives according to the rules which apply to each category. Such classification is arbitrary and leads to contradictions. We find friponi (verb) and fripono (noun), besides hipokriti and hipokritulo. The explanation is that fripon- is a nominal root from which we may directly derive the verb friponi; hipokrit- is a verbal root, and `the person who is characterized by ...' should be derived by means of the suffix -ul-, which gives us hipokritulo.

This derivation leads to further complications. From armi = to arm Esperanto derives armilo = weapon, but literally it would mean instrument with which to arm; sen/arm/ig/i = to render without arms is only correct if we accept arm- as a nominal root, which is not the case in Esperanto.


The rules of Ido do not admit of direct derivation of verbs from nominal roots as these roots do not express either an action or a state. To express the meaning to crown, to hammer, to pepper, to oil, Ido uses a number of precise and invariable affixes, kron/iz/ar = to equip or supply with a crown, martel/ag/ar = to act with a hammer, pipr/iz/ar, ole/iz/ar. For Ido the rule of reversibility has become part of its grammar. "Every derivative must be reversible; that is to say, if one passes from one word to another of the same family in virtue of a certain rule, one must be able to pass inversely from the second to the first in virtue of a rule which is exactly the inverse of the preceding" (Couturat). In Ido we distinguish between verbal root and nominal root, the former expressing an action or a state [labor-, drink-, parol-] while the latter denotes an object (living being or thing), or expresses an aspect of it (as adjectives). Verbal roots produce the verb by adding the appropriate grammatical ending [labor/ar, drink/ar, parol/ar] and a noun directly derived from them can only mean the action, `working', `drinking', `speaking'. Nominal roots can produce, by direct derivation, only names of beings or things or descriptions: `man', `house', `beautiful', `blind' [Ido homo, domo, bela, blinda]. If we depart from a verbal root and wish to derive the noun [labor/ar, labor/o = to work, the work] we do not, in Ido, require an affix, as the verbal noun so derived does not contain any idea which is not already expressed by the verb. It follows that Ido cannot admit the direct the direct derivation of verbs from nouns which do not express an action or state.

The derivation in Ido is direct when it consists in changing grammatical terminations; infinitive in -ar, noun singular in -o, adjective in -a, adverb in -e. The verbal root parol- may be changed into parolar = to speak, parolo = speaking, parola = oral, parole = orally.

The derivation is indirect when it requires affixes. These affixes, as well as the roots, are invariable and modify but do not change the meaning. Affixes are not used in simple imitation of the ethnic languages [varm/a = warm, varm/et/a = lukewarm (little warm), varm/eg/a = hot].

For the direct (immediate) derivation Ido observes three rules which regulate the relationship between verbal, nominal, and adjectival roots. (1) The noun immediately derived from the verb signifies the state or action expressed by the verb [labor/ar = to work, labor/o = the work]. (2) The adjective directly derived from a noun describes how the thing or being is [orfan/o = orphan, orfan/a infanto = orphan child; arjento = silver, arjenta kuliero = silver spoon]. Inversely we may derive a noun from an adjective with the meaning `the thing or being which is' [richa = rich, richo = a rich person; blinda = blind, blindo = a blind person; dezerta = deserted, dezerto = a desert]. (3) An adverb directly derived from an adjective expresses `of the manner' indicated by the adjective [bela = beautiful, bele = beautifully; agreabla = agreeable, pleasant, agreable = agreeably, pleasantly]. The adverb directly derived from a noun forms the complement expressing the circumstances of place, time, quantity or quality [automobile vehar = to ride by car; necese ajornar = to adjourn of necessity; telefone parolar = to talk by telephone].

We cannot, in Ido, derive directly a verb from the adjective or a verb from a noun as there is no relationship which could be logically fixed. They would contradict the rule of reversibility (see Chapter II).


Occidental does not use direct derivation as freely as Esperanto. The derivation of verbs from nominal roots is restricted to four cases, namely, when the meaning of the verb so derived indicates (1) to provide with, (2) to use as an instrument, (3) to secrete, and (4) to act as. Any cases which conform to these four categories may be directly derived. Thus Occidental forms coron, coronar, pipre, piprar, martel, martelar. S. Auerbach (Pri nonmediati derivatione in li international lingues) observes that, as the infinitive ending is -r only, no explanation is given for the addition of -a- in these cases. The verbal noun is formed by the affix -ation, coron/ation, martel/ation.

In Chapter III we have already seen the departure of Occidental from lines to which the former systems have, more or less, conformed. The affixes used do, in some cases, cover a variety of meanings and examples given need not be repeated here. The model for their use could be found in most cases in the Romanic languages. Occidental does not attempt to introduce a simple and regulated system of derivation. The result was a far more natural aspect of the language and greater flexibility at the cost of precision of meaning of the derivatives so obtained.


Novial does not use the vowels -o, -a, -e, as do Esperanto and Ido, to indicate respectively the noun in singular, the adjective and the adverb. These three vowels in Novial have two distinct functions (1) to indicate the sexes [-o masculine, -a feminine, -e neuter] and (2) -e the noun or name, -a the verbal infinitive, -o the verbal noun expressing the action or state.

Direct derivation with these distinctive grammatical terminations, is used for the following three cases, (1) to use an instrument [hamre = the hammer, tu hamra = to hammer, hamro = hammering; telefone = the telephone, tu telephona = to telephone, telefono = telephoning; bicikle = the bicycle, tu bicikla = to cycle, biciklo = cycling], (2) to secrete [sange = the blood, tu sanga = to bleed, sango = bleeding; sudore = the sweat, tu sudora = to sweat, sudoro = sweating], and (3) to describe meteorological phenomena [nive = the snow, tu niva = to snow, nivo = snowing; pluve = to rain, tu pluva = to rain, pluvo = raining].

A fourth category of verbs may be derived in the same fashion, this group being described as cases in which no doubt about the meaning of the verb so derived is possible. This may clearly cover a very wide field. S. Auerbach gives two examples for this case [parte = the part, tu parta = to part, parto = parting; honore = the honour, tu honora = to honour, honoro = honouring]. Where the verb clearly indicates `to provide with ...' Novial uses the suffix -is [Ido -iz-], [krone = the crown, tu kron/is/a = to crown, kron/is/o = crowning; pipre = the pepper, tu pipr/is/a = to pepper, pipr/is/o = peppering]. Verbs in -a cannot be derived directly from names of men or animals as -a is used to indicate the female sex. By adding `tu' to the infinitive of the verb Jespersen has added a further distinction to differentiate between noun and verb.

The solution which Jespersen has devised for the cases of direct derivation in Novial were made possible by the suppression of definite grammatical terminations for the various classes of words. Novial uses a great many affixes with precise meanings, which are in most cases identical with those of Ido.

VIII - 3. Commentary

The problem of derivation cannot be said to have been finally solved by any of the existing systems. At the time of the delegation, Couturat's solution, based on the logical conception of the relationships of word families, seemed to some the ideal solution. It offered precision for the words so derived and apparently made Ido a more perfect instrument for use in science than any other system known before. But criticism was soon heard, being based on the difficulty of using correctly certain affixes, particularly -if-, -ig-, and -iz-. They were said to be difficult to master by the average student. Direct derivation in cases where these affixes should logically be used were easier for everyday use. Any departure from the logical conception of reversibility would destroy the logical structure of Ido and could therefore not be considered by Idists. Janis Roze has collected some interesting proposals in his Report on the Decisions of the Interim Academy of Ido (Riga, 1937). It was proposed to reduce the three affixes to their common denominator (-i-) for ordinary speech without necessarily suppressing the affixes - which had proved useful - for scientific terms. Thus -i- would stand for a variety of meanings and indicate that a more precise suffix had been replaced for the sake of ease. A number of proposals by S. Quarfood, Fr. Honoré, de Belie, and others, have been submitted for the sense for which -i- should stand. Profet/o, profet/i/ar = prophet, to prophesy; filozof/o, filozof/i/ar = philosopher, to philosophize; instead of martel/agar = to act with a hammer, martel/i/ar = to hammer, which would be the nearest approach to direct derivation without abandoning the principle of reversibility, analogically fusiliar, kroniar, etc. in the sense to act with an instrument. Alan Kelso has suggested deriving verbs directly from adjectives with the aid of -i, red/i/ar = to make red, net/i/ar = to make clean, instead of red/ig/ar and net/ig/ar. No decisions have been taken on these proposals, but they may well be kept in mind for the question of derivation is not as yet finally settled for the planned language.

Another criticism has been raised against the derivation in Ido. Dans- is a verbal root in Ido and the noun directly derived [dans/o] expresses the act or state. In ethnic languages `dance' may express either `the action of dancing' or `the dance' as an abstract designation. For common usage Ido has used the suffix -ad- to express the continued action, and the termination -o to describe the act or state, but this is not a purely logical solution. Other words belong to the same class, parol/ar = to speak, parol/o = speaking or speech, pens/ar = to think, pens/o = thinking or thought, koncept/ar = to conceive, koncept/o = conceiving or conception. The naturalists say that as perfect logicality in language cannot be achieved, the principle itself should be abandoned. It is questionable whether it is better to retain a few imperfections for which a better solution has yet to be found, or to reject the principle with which so many perfect forms can be obtained.

These problems are only briefly mentioned here to show that a profound comparative study will be required before the planned language can be fixed in its ultimate shape. If a linguistic commission decides that the planned language shall be formulated on autonomistic principles, i.e., be precise and independent of the irregularities of ethnic tongues, the principles of Ido may be developed in the sense of the discussions and proposals sketched above. If, however, a more naturalistic solution is attempted, the principles offered by Occidental may be acceptable, i.e., a development in close analogy to the Romanic languages, less precise but independent of the complications of a logically developed and coherent system of word derivation.

Back to Contents Page

This page is hosted by Geocities.
James Chandler 1997.