I shall further distinguish between direct derivation - that is derivation without the use of affixes - and indirect derivation - that is derivation with the use of affixes.
Zamenhof, after selecting some 2,600 words, formulated the rule that, given
the root, the noun is formed by adding -o, the adjective by adding
-a, the adverb by adding -e, the infinitive by adding
-i, etc. It follows that grammatical terminations, or as we may
call them here, indicative endings, are interchangeable. But no rule
defines the meaning of verbs derived from nominal roots and we must assume
that they are formed in analogy to our ethnic tongues. For indirect
derivation Zamenhof followed the agglutinative pattern, adding prefixes
and suffixes as required to the root, the whole word acquiring the meaning
of the root plus the meaning of the affixes [mal-san-ul-ej/o = a
place for unhealthy person(s) =hospital]. Later René de Saussure's
rule of necessity and suffiency was added to Esperanto, saying that
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.
[bel/a = beautiful; -ec/o = abstract quality;
bel-ec/o = beauty; but according to de Saussure also bel/o
= beauty]. Further Zamenhof admitted the use of suffixes as independent
roots. A suffix so used is bound to be totally artificial and
incomprehensible at first sight [an/o = adherent, estr/o =
master].
Couturat was highly critical of the lack of precision in Zamenhof's
Esperanto. In his Étude sur la dérivation (Paris
1907) he made detailed proposals to introduce logicality into the
constructed language. A distinction is made between the verbal root which
expresses an action or state, and the nominal root which denotes the
object (living being or thing) or expresses an aspect of it (adjective).
Thus if we derive directly a noun from a verbal root it can only express
the action or state [pens/ar = to think; pens/o = thinking;
dans/ar = to dance; dans/o = dance, or dancing]. But we cannot,
in Ido, derive directly a verb from a nominal root as we do in English, and
that with great convenience to ourselves: to shop, to house, to water, to
hammer, to cycle. Couturat formulated the following three rules which
govern the verbal, nominal and adjectival roots:
(1) The noun immediately derived from the verbal root signifies the action
or state expressed by the root [pens/ar = to think; pens/o].
(2) The adjective directly derived from a nominal root describes how the
thing or being is [arjent/o = silver; arjent/a kulier/o = a
silver spoon]; inversely we may derive a noun from an adjectival root with
the meaning `the thing or being which is'; [blind/a = blind;
blind/o = a blind person].
(3) An adverb directly derived from an adjective expresses `of the manner'
[agreabl/a = agreeable; agreabl/e = agreeably, pleasantly].
In indirect derivation - Couturat argued - the meaning of every derivative
can only be the meaning of the root plus the meaning of its affix or
affixes, nothing more or less. Each affix must have one invariable
function, and the meaning of that function must be part of the meaning of
the whole word. Each element must be monosignificant. And he deduced from
that basic principle the rule of reversibilty which says that
every derivative must be reversible; that is to say, if one passes from
one word to another, one must be able to pass inversely from the second
to the first
and thus return to the basic signification of the root [regul/o =
rule; regul-oz/a = regular; ne-regul-oz/a = irregular;
ne-regul-oz-es/o = irregularity]. Again Ido followed the pattern
of agglutination.
I am borrowing here, from Schleicher, a classification which has not previously been applied to constructed languages, though I am fully aware that this division into isolating, agglutinative and flexional languages is neither sufficiently comprehensive, nor does it indicate the degrees of development in language or in the systems under discussion. It is used only as it serves as a convenient classification for our purposes.