2.1. The PDT approach to valency

This section introduces the basics of the PDT approach to valency. The rules described here mainly apply to verbs. However, they are generally applicable to other parts of speech, too. The valency of individual parts of speech is discussed in Section 2.3, "Valency of individual semantic parts of speech".

2.1.1. Criteria for distinguishing between inner participants (arguments) and free modifications (adjuncts)

Any modification can be classified as either an inner participant (argument) or free modification (adjunct) - according to the type of dependency the are in w.r.t. their governing node.

To distinguish arguments from adjuncts, the following criteria are used:

  • can the given type of modification modify a particular verb occurence more than once, or at most once?

  • can the given type of modification modify any verb, or is there a (more or less) closed class of verbs that can be modified by it?

The rule is that:

  • free modifications (adjuncts in the sequel) are such modifications that can modify any verb and, moreover, they can (in principle) modify a particular verb token more than once.

  • inner participants (arguments in the sequel) are such modifications that can modify any given verb only once (except for the case of coordination) and they only modify a more or less closed class of verbs that can be listed.

The empirical observations made up to now lead us to the assumption that there are five different types of arguments: Actor (ACT), Patient (PAT), Addressee (ADDR), Origo (ORIG) and Effect (EFF). Other types of verbal modifications are considered to be adjuncts, corresponding to temporal, locational, manner and other kinds of adverbials (for the list of the recognized functors and their definitions, see Chapter 7, Functors and subfunctors).

For every verb (meaning: in one of its possible meanings), it is possible to determine what its arguments are on the basis of the criteria above.

As for the other parts of speech (nouns, adjectives and adverbs), the situation is similar but there may be also specific issues in addition to the ones common for all parts of speech (e.g. there is a special nominal argument MAT). For details regarding the valency of nouns, adjectives and adverbs, see Section 2.3.2, "Valency of nouns",Section 2.3.3, "Valency of adjectives" andSection 2.3.4, "Valency of adverbs" respectively.

2.1.2. Criteria for distinguishing between obligatory and optional modifications

A given type of modification is either an argument, or an adjunct in all its occurences. In the relation to its governing word, any given modification is either obligatory (obligatorily present at the deep structure level), or optional (not necessarily present). The obligatory - optional distinction does not apply directly to the individual types of modifications; it rather applies to their relation to individual lexical units (their governing verbs/nouns/adjectives..).

Adjuncts, defined in Section 2.1.1, "Criteria for distinguishing between inner participants (arguments) and free modifications (adjuncts)" as such modifications that can modify any verb (word) and, furthermore, sometimes even more than once, are in some cases also obligatory. Adjuncts are potentially always present; any event proceeds in time, at a certain location, in a certain way/manner. For some verbs, adjuncts with "temporal", "locative" or "manner" meanings are, however, obligatory, necessarily present in the meaning of the verb (i.e. they are required by the semantics of the verb).

The dialogue test. For determing whether a given modification is obligatory or optional, the so called dialogue test is used.

The dialogue test helps us to determine which modifications of a given verb are obligatory and which are optional. It is used whenever a modification is not present at the surface level but when it can be hypothesized that it is in fact (semantically) obligatory. The dialogue test is based on the difference between questions asking about something that is supposed to be known to the speaker - because it follows from the meaning of the verb he/she has used, and questions about something that does not necessarilly follow from the meaning of the used verb. Answering a question about a semantically obligatory modification of a particular verb, the speaker - who has used the verb - cannot say: I don't know. Compare the following dialogues:

  • determining whether an argument is obligatory:

      • A: Když to viděl, koupil to. (=When he saw it he bought it.)

      • B: Kdo? (=Who?)

      • A: *Nevím. (=*I don't know.)

      • A: Když to viděl, koupil to. (=When he saw it he bought it.)

      • B: Komu? (=For whom?)

      • A: Nevím. (=I don't know.)

      • A: Když to viděl, koupil to. (=When he saw it he bought it.)

      • B: Od koho? (=From whom?)

      • A: Nevím. (=I don't know.)

    The verb koupit (=buy) has (according to the criteria in Section 2.1.1, "Criteria for distinguishing between inner participants (arguments) and free modifications (adjuncts)") four arguments: Actor, Patient, Addressee and Origo. With the help of the dialogue test, it can be determined which of these arguments are obligatory and which are optional. In dialogues a), the speaker cannot answer the questions Kdo? (=Who?) by saying Nevím (=I don't know). It would make no sense if speaker's answer was Nevím (=I don't know). On the other hand, the speaker does not have to know answers to the questions Komu? (=For/to whom?) and Od koho? (=From whom?) in the dialogues b) and c). These modifications are contained in the meaning of the verb, but not necessarily; they are optional.

  • determining whether an adjunct is obligatory:

      • A: Moji přátelé přijeli. (=My friends have come.)

      • B: Kam? (=Where to?)

      • A: *Nevím. (=*I don't know.)

      • A: Moji přátelé přijeli. (=My friends have come.)

      • B: Odkud? Proč? (=Where from? Why?)

      • A: Nevím. (=I don't know.)

    For the verb přijet (=come), the modification answering the question Kam? (=Where to?) is obligatory, which can be seen from the impossibility to answer the question by saying Nevím (=I don't know). The speaker used the verb přijet (=come), so it would make no sense if the answer to the question about the goal was Nevím (=I don't know). A modification of this type is implied by the meaning of the verb - the speaker knows it and left it out only because he/she was sure the hearer knew it as well. On the other hand, the speaker does not need to know answers to the questions Odkud? (=Where from?), or Proč? (=Why?) in dialogue b).

    The modification answering the question Kam? (=Where to?) is (according to the criteria in Section 2.1.1, "Criteria for distinguishing between inner participants (arguments) and free modifications (adjuncts)") an adjunct; it is, however, obligatory for the verb přijet.

2.1.3. Structure of a valency frame

By combining the criteria for distinguishing between arguments and adjuncts with the criteria for distinguishing between obligatory and optional modifications, we get the four possibilitites displayed in the following table:Table 6.1, "Structure of a valency frame".

Table 6.1. Structure of a valency frame

  Obligatory modifications Optional modifications
Arguments + +
Adjuncts + -

All arguments and those adjuncts that are obligatory for a given meaning of a given verb (noun/adjective/adverb) (cf. the pluses in the table Table 6.1, "Structure of a valency frame") are understood as valency modifications in the narrow sense and are recorded in the valency frame of the verb. Every verb has at least one valency frame - and often more, with one frame corresponding to one meaning of the verb.

As for idiomatic expressions, the valency frame of the governing verb contains, apart from its arguments and adjuncts, also the dependent parts of the idiomatic expressions in question (with the functors CPHR or DPHR; see Section 2.2.2, "Valency frames of idiomatic expressions (phrasemes) and complex predicates").

A valency frame can also be empty - for the discussion see Section 2.2, "Valency frames and the way they are recorded in the valency lexicon".

Valency frames are recorded in the valency lexicon. For a discussion concerning the valency lexicon and the way valency frames are recorded in it, see Section 2.2, "Valency frames and the way they are recorded in the valency lexicon".

2.1.4. Criteria for determining the type of argument (the principle of shifting)

When determining the type of the argument in question, two kinds of criteria are used: syntactic (when only the Actor (ACT) and Patient (PAT) are involved) and semantic (when more than two arguments are involved). For a discussion of the semantics of the individual arguments (and their definitions), see Section 2, "Argument functors"

In principle, it holds that:

  • the first argument is always the Actor, the second one is the Patient. From this, it follows that:

    • if a verb has only one argument, it is the Actor (ACT) regardless of its exact semantic relation to the verb.

    • if a verb has two arguments, they are the Actor (ACT) and the Patient (PAT).

    Determining the first and the second argument. When determining which argument is the first one (i.e. the Actor), the basic rule is that the Actor occupies the subject position, i.e. the structural nominative position. Only if one of the arguments is in the dative case and the other one in the nominative case, the semantics of the arguments comes into account. If the argument in dative refers to the Experiencer (or Agent), we consider the argument the Actor and the argument in nominative the Patient.

    Cf:

    • Kniha.PAT se mi.ACT líbila. (=I liked the book; lit. book.NOM REFL to_me.DAT seemed_nice)

      The dative argument refers to the Experiencer. It corresponds to the Actor in the relevant valency frame; the nominative argument is the Patient.

      The valency frame for the given meaning of the verb líbit se (=seem_nice/be_to_one's_taste):

      ACT(.3) PAT(.1;že[.v];.f;.c)

    • Naše výrobky.ACT se vyrovnají cizím výrobkům.PAT (=Our products are as good as the products from other countries; lit. Our products.NOM REFL keep_pace_with/are_a_match_for foreign products.DAT)

      The argument in the dative case does not refer to an Experiencer/Agent; the Actor functor is assigned to the argument in the nominative and the dative argument is the Patient.

      The valency frame for the given meaning of the verb vyrovnat se (=keep_pace_with/be_a_match_fo):

      ACT(.1) PAT(.3)

  • if a verb has more than two arguments, semantic criteria come into play. Determining whether the third (fourth, fifth) argument is the Addressee, Effect or Origo depends on the semantics of the argument in question.

The arguments are, in principle, defined (also) semantically (cf. Section 2, "Argument functors"); however, when considering the first two arguments, the syntactic criteria are decisive. As a consequence of this, the so called argument shifting takes place. The argument shifting means that:

  • if a verb has no argument in its valency frame that bears the cognitive role of an Agent (or another role typical for the first participant - Actor), its position is taken up by the Patient (i.e. what would be assigned the Patient functor under usual circumstances). Cf.:

    • Kniha.ACT vyšla. (=The book was published/came out.)

      The Patient has taken up the position of the Actor (=it has undergone shifting).

      The valency frame for the given meaning of the verb vyjít (=come_out):

      ACT(.1)

  • if a verb subcategorizing for two arguments has no argument that bears the cognitive role of a Patient, another argument takes up its position (i.e. is assigned the Patient functor). The following rule applies:

    • if a verb has a potential Addressee/Origo and a potential Effect but has no Patient-like argument, then the Patient position is taken up by the Effect-like argument. The Addressee and/or Origo-like arguments do not undergo any shifting. Cf.:

      • Petr.ACT vykopal jámu.PAT (=Petr has dug a hole.)

        The Patient position is taken up by the Effect-like argument (i.e. the Effect has undergone shifting).

        The valency frame for the given meaning of the verb vykopat (=dig (up/out)):

        ACT(.1) PAT(.4)

      • Jan.ACT vyspěl z jinocha.ORIG v muže.PAT (=Jan grew up into a man - he is not a child any more; lit. Jan grew_up from adolescent into man)

        The Patient position is taken up by the Effect-like argument (i.e. the Effect has undergone shifting), the Origo-like argument has not undergone any shifting.

        The valency frame for the given meaning of the verb vyspět (=grow up):

        ACT(.1) PAT(v+4) ?ORIG(z+2)

    • if a verb has no Effect-like argument, the Patient position is taken up by the cognitive Addressee/Origo (i.e. they shift to the position of the Patient). Cf.:

      • Učitel.ACT vyvolal žáka.PAT (=The teacher asked a pupil to answer a question; lit. Teacher called_upon pupil.)

        The Patient position is taken up by the Addressee-like argument (i.e. the Addressee has undergone shifting).

        The valency frame for the given meaning of the verb vyvolat (=call upon/examine):

        ACT(.1) PAT(.4)

      • Z banálního nachlazení.PAT se vyvinulo závažné onemocnění.ACT (=A slight/banal cold developed into a serious illness; lit. From banal cold REFL developed serious illness.)

        The Patient position is taken up by the Origo-like argument (i.e. the Origo has undergone shifting).

        The valency frame for the given meaning of the verb vyvinout se (=develop):

        ACT(.1) PAT(z+2)

The argument shifting can be represented schematically as follows:

  • ACTPATEFF / ADDR / ORIG.

NB! The shifting only concerns arguments. Adjuncts do not shift to argument positions. An adjunct that is obligatory for a given verb (according to the criteria in Section 2.1.1, "Criteria for distinguishing between inner participants (arguments) and free modifications (adjuncts)") is always assigned an adjunct-ltype functor. Cf.:

  • Hučí v komíně.LOC (=lit. Whistles in chimney.)

    The verb hučet (=whistle) has no argument. The obligatory adjunct with the LOC functor does not shift (i.e. does not take up the position of the Actor).

    The valency frame for the given meaning of the verb hučet (=whistle):

    LOC(*).

  • Petr.ACT přijel do Prahy.DIR3(=lit. Petr came to Praha.)

    The verb přijet has one argument, i.e. the Actor. The obligatory adjunct with the DIR3 functor does not shift (i.e. does not take up the position of the Patient).

    The valency frame for the given meaning of the verb přijet (=come/arrive):

    ACT(.1) DIR3(*).

The argument shifting applies to the valency frames of all verbs, with the exception of complex predicates, for the relevant discussion see Section 9.3.3, "Valency frames of complex predicates".

2.1.5. Relationship between the verb meanings and valency frames

Every verb meaning is assigned a valency frame. Verbs usually have more than one meaning; each is assigned a separate valency frame. Every verb has as many valency frames as it has meanings. However, in the PDT valency lexicon, only those verbs, nouns, adjectives and adverbs - more precisely those of their meanings - are included which occured in the annotated data (for a discussion, see Section 2.2.4, "Valency lexicon").

The one meaning - one valency frame principle is violated in the cases of so called competing valency modifications, described in Section 2.3.1.5, "Valency modifications competing for the same position (while the meaning of the verb is preserved)". These are the cases when a single valency position can be taken up by more different modifications with no (or almost no) change in meaning. The potential competition arises either between an argument and adjunct or between different types of adjuncts. There are two strategies how to deal with cases of competing modifications; one of them is to constitute as many valency frames as there are competing modifications. Then, two or more valency frames correspond to a single verb meaning.

Two different meanings of a verb can have the same valency frame, i.e. identical with respect to the number and type of modifications, as well as their surface form. Compare the following examples:

  • three instances of an identical valency frame of the verb chytit (=catch):

    • ACT(.1) PAT(.4)

      několik míčů (=several balls)

    • ACT(.1) PAT(.4)

      chytili pachatele (=they caught the culprit)

    • ACT(.1) PAT(.4)

      chytili poslední vlak (=they caught the last train)

  • two instances of an identical valency frame for two different meanings of the verb chovat (=raise/breed vs. nurse/cradle):

    • ACT(.1) PAT(.4)

      chová prasata na farmě.LOC (=he/she raises pigs on his/her farm)

    • ACT(.1) PAT(.4)

      chová dítě v náruči.LOC (=he/she is cradling a child in his/her arms)

Different verb meanings are delimited in a rather intuitive way; thus, it can happen that a case understood in PDT as a verb having one meaning/one valency frame is analyzed differently by someone else (who is making a finer-grained distinction) - i.e. as having more meanings/more valency frames.