The surface form of a valency modification is the form in which the given modification is represented on the analytical level (i.e. on the level (lower than the tectogrammatical one) where all words contained in the surface form of a sentence are present). The surface form specification contains the following information:

Sometimes, it is necessary to specify the lemma of a preposition, for example.

In PDT, in contrast to the original system of valency frame representation (known from the literature on valency), an enhanced encoding system is used, which enables a uniform treatment of simple cases (like capturing the case requirements independently on the part-of-speech membership of the modification and other properties), as well as more complex cases (like idiomatic expressions; see Section 2.2.2, "Valency frames of idiomatic expressions (phrasemes) and complex predicates").

Dependency specification. To indicate the dependency, square brackets ( [ ] ) are used; sister nodes are separated by a comma ( , ). The notation is, then:

Specification of the part-of-speech membership and morphemic properties. The requirements as to the part of speech and morphemics of individual nodes are encoded in an abbreviated form (using one symbol for each class), introduced after a period or colon (the difference between a period and colon - as a means of separating the lemma and the morphological information - is discussed in Section 5, "The notation of valency frames and its semantics"), in the following order: part of speech, gender, number, case, degree. 4 means that the governing verb requires a modification in the accusative case, P6 refers to 'locative plural'. If a surface-level category is not specified, it means that the given valency modification may get any value of the category.

Examples of surface-form encoding:

In some cases, surface-form specifications include also the information regarding the analytical forms (lemmas) of dependent (analytical) nodes, which are part of the surface form of a modification in question; these are prepositions, subordinating conjunctions and also dependent parts of idiomatic expressions (see Section 2.2.2, "Valency frames of idiomatic expressions (phrasemes) and complex predicates"). For example, the requirement that a modification have a form of a subordinate clause with the conjunction že is encoded like this: že[.v].

For the sake of simplicity, when specifying what kind of prepositional phrase is required by a given verb, an abbreviated form is used (for the list, see Section 5.5, "Abbreviated forms of realization records"). For example, na+4 is short for: na-1[.4].

Cf. other cases of surface-form specifications:

Surface-form specification contains all surface-form variants of a given modification found in the analyzed data, stylistic variants included. What is not included in the information on the surface form of a modification are the changes in form that result from productive processes (syntactic transformations, e.g. passivization, reciprocity); for a detailed discussion see Section 2.2.3, "Productive changes in the surface form (not specified in the valency frames)".

A surface form of an obligatory adjunct is usually not specified, which means that all usual forms can be used. This is indicated by the star symbol ( * ), which is used instead of the explicit specification of a surface form. With arguments, the surface forms are always specified.

Idiomatic expressions (see Section 8, "Idioms (phrasemes)") and complex predicates (see Section 9.3, "Complex predicates") represent more complex cases; their dependent parts are included in the valency frames of the relevant head verbs, among other valency modifications (their functor is CPHR or DPHR).

Valency frames of idiomatic expressions. Specifying the surface forms may get rather complicated with idiomatic expression. When specifying the surface form of the dependent part of an idiomatic expression, it is necessary to capture the following facts: how many parts (words) the dependent part has, what are their morphological categories and often also the precise lexical content of these parts. There is a convention adopted for representing these requirements.

Examples:

Valency frames of complex predicates. All complex predicates (with the CPHR functor) that have the same verb in their verbal part and the nominal part of which may be formed by various synonyms and antonyms are assigned the same valency frame. The surface form of the nominal part of a complex predicate is specified as follows: a list of possible synonyms and antonyms in curly brackets ( { } ) is followed by the list of possible morphemic forms. The list of the synonyms and antonyms (their lemmas) ends with three dots, which indicates that the list is not exhaustive; it only contains the cases collected so far. The rule of argument shifting does not apply here (see Section 2.1.4, "Criteria for determining the type of argument (the principle of shifting)").

Valency frames of complex predicates are discussed in more detail in Section 9.3.3, "Valency frames of complex predicates".

Surface-form specifications contain all variants found in the analyzed data, with certain exceptions, though. These exceptions are cases when the change (shift) in form is caused by a productive process.

The cases when a surface-form variant is not recorded in the valency lexicon include:

Transformational rules may be applied to the original valency frame - this is a guarantee (or rather, a way of testing) that the verb was assigned a correct valency frame.

Valency frames (assigned to individual meanings of words) are recorded in the valency lexicon. The valency lexicon contains valency frames of semantic verbs, nouns, adjectives and adverbs. Individual valency frames are clustered on the basis of what t-lemma they are related to (for a discussion of t-lemmas see Chapter 4, Tectogrammatical lemma (t-lemma)).

The valency lexicon does not contain t-lemma substitutes (#Colon, #EmpVerb etc.) and t-lemmas of those nodes present at the surface level that are expressed by pronouns (it means that the valency lexicon does not contain t-lemmas like the following: který, jaký (=which, what) etc.). For a discussion of pronouns standing in place of lexical units with subcategorization requirements, see Section 2.4.3.4, "Pronouns in place of words with valency".

The valency lexicon was being constituted during the annotation; therefore, only those verbs, nouns, adjectives and adverbs - i.e. those of their meanings - are included which occured in the analyzed data. For example, if a verb has two different valency frames in the lexicon, it means that these two meanings of the verb were found in the analyzed data; however, the given verb may have other meanings (i.e. other valency frames), too.

!!! The current version of the valency lexicon contains:

The valency lexicon only contains the t-lemmas of those nodes that have the value of the nodetype attribute specified as complex. The t-lemmas of traditional verbs, nouns, adjectives and adverbs the nodetype attribute of which has a value other than complex (according to the rules in Chapter 3, Node types) are not included in the valency lexicon (even if they have argument modifiers).