Lexical derivation involves change in meaning; the semantics of the lexical derivatives differs from that of their base words.
There are two types of lexical derivation that are captured as such at the tectogrammatical level:
Individual kinds of definite numerals (ordinal, sort, set and fraction numerals) are taken to be derived from the corresponding cardinal numerals and are represented by their t-lemmas.
Examples:
trojí (=three_kinds_of) is represented by tři (=three);
třetina (=(one) third) → tři;
třetí (=the_third) → tři.
See Section 6.2.4, "Definite quantificational semantic adjectives".
Relative, indefinite, interrogative, negative and totalizing pronouns and pronominal numerals and adverbs of a similar type (e.g. někdo / nikdo (=somebody/nobody), několik (=several), někdy / nikdy (=sometimes/never)) are represented by the t-lemma of the corresponding interrogative or relative pronoun, numeral or adverb.
Examples:
někdo (=somebody) is represented by kdo (=who);
nikdo (=nobody) → kdo;
nic (=nothing) → co (=what);
několik (=several) → kolik (=how_many);
všechen (=all) → co (=what);
žádný (=no) → který (=which).
See Section 6.1.4, "Indefinite pronominal semantic nouns", Section 6.2.3, "Indefinite pronominal semantic adjectives".
The semantic feature distinguishing individual types of numerals from their cardinal counterparts is encoded in the value of their numertype grammateme (see Section 5.5, "The numertype grammateme"). The semantic feature distinguishing different types of pronominal numerals and adverbs from each other is encoded in the value of their indeftype grammateme (see Section 5.6, "The indeftype grammateme").
As for type A, the form of the m-lemma follows from the combination of the used t-lemma and the value of the numertype grammateme (e.g. třetí (=the third) has the t-lemma tři (=three) and the value ord is specified in the numertype attribute). As for type B, the form of the m-lemma follows from the combination of the used t-lemma and the value of the indeftype grammateme (e.g. the m-lemma někdo (=somebody) is represented by a node with the t-lemma kdo (=who) and the indef1 value in the indeftype grammateme).
!!! Only those types of lexical derivation are represented as such at the tectogrammatical level which are grammaticalized to a sufficiently large extent. Hence, we only represent those cases of lexical derivation as such that belong to a closed class of words (these are pronouns, numerals and pronominal adverbs). So far, productive types of lexical derivation (derivation of feminine forms from masculine forms, derivation of diminutives or agent nouns) are not represented as such in PDT.