Aarhus University Seal

Jan Rijkhoff

PhD (University of Amsterdam 1992), !!Emeritus

Below are my recent views on functional categories (pp. 343-348 from: Rijkhoff, Jan. 2016. Crosslinguistic categories in morphosyntactic typology: Problems and prospects. Linguistic Typology 20-2, 333-363):


3.2. Members of functional categories are the products of a speech act

Since morphosyntactic units such as clauses, phrases, words and free or bound morphemes can be characterized in terms of formal, semantic and functional properties, they can simultaneously belong to a formal, a semantic and a functional category.[1] Functional categorization is not directly concerned with the formal or semantic properties of a linguistic unit, but rather with the actual job of a form or construction in the process of verbal communication. Membership of an interpersonal functional category is determined by the way a linguistic unit is used in actual discourse (hence the alternative label ‘discourse unit’). The functional approach to categorization has a long tradition, going back to the Prague School of Linguistics (if not earlier, cf. Weil 1844: 19), whose members were “seeking to understand what jobs the various components were doing […]” (Sampson 1980: 104). A considerable amount of literature has be devoted to pragmatic or functional processes, forces, factors and principles, but so far this has not resulted in a more or less comprehensive classification of the interpersonal (‘communicative’, ‘discourse’) functions that are served by the various kinds of linguistic units in languages across the globe. Hence a tentative list of functional categories is proposed below.

   Functional categories are here defined as categories whose members are the product of an interpersonal speech act, i.e. a speech act that first and foremost concerns the role of a linguistic unit in the interaction between speaker and addressee from a communicative or discourse perspective. Speech acts are commonly divided into locutionary acts (the act of saying something), illocutionary acts (the act of providing an utterance with a communicative intention, e.g. an assertion, a question or a command), propositional acts (the acts of referring, predicating, and, as we shall see below, modifying), and perlocutionary acts (the actual effect of the speech act).[2] For our purposes we only need to concern ourselves with illocutionary acts and propositional acts, but we need two additional kinds of speech acts to account for functional categories across languages: thetical acts and pragmatic acts. Thus there are four main types of interpersonal functional categories, which I will call illocutions, theticals, propositionals and pragmaticals for ease of reference. Each of these functional categories is briefly introduced below.


3.2.1. Functional categories I: linguistic units that are the products of an illocutionary act (‘illocutions’)

Illocutionary acts, introduced in Austin 1962 as part of his speech act theory, result in linguistic units which have a specific communicative function usually called ‘illocutionary force or value’, which is captured by labels such as Declarative, Interrogative or Imperative.

    Each type of illocution comes with its own set of grammatical properties or ‘mini-grammar’, involving syntactic, morphological and/or prosodic features. For example, Nama Hottentot has dedicated particles to mark different kinds of illocutionary acts, whereas speakers of Greenlandic Eskimo use special affixes for this purpose:

  -- note: not included here: examples from Greenlandic Eskimo (Inuit) --

Notice that illocutions do not necessarily take the form of a complete clause, as in the case of the normal answer to a question such as ‘Where is she?’, where a simple ‘In her office’ suffices as a complete, declarative response. This implies that the formal expression of the illocution should be determined at an early stage (e.g. a complete clause or just the focal part, as in the example above) to make sure that the cross-linguistic category will contain comparable forms or structures (more on this below). For a recent overview of the grammatical properties of illocutions, I refer to König & Siemund 2007.


3.2.2. Functional categories II: linguistic units that are the products of a thetical act (‘theticals’)

Theticals cover more or less the same linguistic expressions as Parentheticals (Kaltenböck et al. 2011: 855-856) or what Dik (1997 II: 379-407) called Extra-Clausal Constituents in his Theory of Functional Grammar (now called Subsidiary Discourse Acts in its successor Functional Discourse Grammar; Hengeveld & Mackenzie 2008: 47, 53-56). A recent discussion of theticals is provided by Kaltenböck et al. (2011: 857), who list the following set of characteristic grammatical properties: (a) they are syntactically independent, (b) they are set off prosodically from the rest of the clause, (c) their meaning is “non-restrictive”, (d) they tend to be positionally mobile, and (e) their internal structure is built on principles of sentence grammar but can be elliptic (see also Heine et al. 2014).

   Several subtypes of theticals can be distinguished, such as Greetings and Leave-takings, Summonses and Addresses (Dik 1997: 384-5); some other subtypes are Motivation, Concession, Orientation, and Correction (Hengeveld & Mackenzie 2008: 53). Space limitations prevent me from offering a more detailed presentation, so I will end this section with some examples taken from Kaltenböck et al. 2011 and Dik 1997:

(9)     ‘The main point — why not have a seat?  — is outlined in the middle paragraph.’

(10)  ‘Is he a friend of yours, that guy?’

(11)  ‘Anyway, what should we tell her when she comes back?’

(12)  ‘Between you and me, he failed the exam.’

(13)  ‘Mary — don’t forget  — is coming over to visit.’


3.2.3. Functional categories III: units that result from propositional acts (‘propositionals’)

Propositional acts (Searle 1969) come in three major subtypes, Acts of Predication, Acts of Reference and Acts of Modification, resulting in three types of propositionals: (i) verbal and non-verbal Predicates, (ii) Referential Phrases  (such as NPs and complement clauses) and (iii) Modifiers.[3] Searle focused on acts of predication and reference; acts of modification are discussed in Croft 1990 and Rijkhoff 2014. Like other functional categories, Predicates, Referential Phrases, and Modifiers are characterized by their own grammatical properties. Here I will focus on Modifiers, because members of this functional category will be used to illustrate the new method to set up morphosyntactic categories for cross-linguistic purposes (Section 4). Modifiers enrich or supplement core linguistic material that is used in acts of predication or reference and are typically associated with formal categories like Adjective, Adverb(ia)l and other attributive, more or less optional linguistic forms and constructions in a noun phrase or a sentence.

    Especially in the case of Modifiers, it is important to emphasize that there is usually no one-to-one relationship between Form and Function (cf. Table 2): the same linguistic form or construction can often be used in more than one modifier function, and vice versa, different linguistic forms or constructions can be used in the same modifier function. This is discussed in considerable detail in Rijkhoff 2009b and 2010, basically showing that interpersonal functions come with their own set of grammatical properties or ‘mini-grammar’.[4] Notice furthermore that the functional category Modifier includes both lexical and grammatical material. For example, both the English past tense suffix -ed and the adverb ‘yesterday’ serve as localizing modifiers: they inform the addressee where to locate an event in time (see below).

   In a layered representation of noun phrases and clauses, members of functional modifier categories are distributed over nested layers around the head constituent (typically V in the clause and N in the noun phrase), reflecting differences in semantic scope between the various subtypes of modifiers.[5] At least five modifier subcategories can be distinguished, which can be used to analyze both clauses and NPs, here ranked according to their scopal properties: Classifying modifiers have the narrowest scope (only the head constituent) and Discourse-referential modifiers have the widest scope (more detailed presentations concerning the role of modifiers in a layered analysis of NPs and clauses can be found in, for example, Rijkhoff 2008b, 2010, 2014).

1. classifying modifiers specify a subtype of the kind of entity that is denoted by the head constituent (like ‘presidential’ in ‘a presidential election’; so-called stripped nouns (Miner 1986) are good examples of free classifying modifiers in the clause);

2. qualifying modifiers specify more or less inherent properties or ‘qualities’ (e.g. ‘black carsN’, ‘walkV fast’);

3. quantifying modifiers specify quantitative properties such as the number or cardinality of the thing or event (e.g. ‘two carsN’ or ‘fallV twice’);

4. localizing or anchoring modifiers specify locative properties, thus making the referent (object or event) locatable and hence identifiable for the addressee (e.g. ‘that houseN on the corner’, ‘metV in Konstanz’);

5. discourse-referential modifiers such as (in)definite articles specify the status of a thing or event as a discourse entity; (ir)realis markers are good examples of discourse-referential modifiers in the clause (Rijkhoff & Seibt 2005).

The layered model of the clause includes two additional layers for modal and illocutionary modifiers. Van de Velde (2007, 2012) has argued that a complete layered model of the NP should also provide slots for modal and illocutionary modifiers.


3.2.4. Functional categories IV: linguistic units that result from pragmatic acts (‘pragmaticals’)

Pragmaticals are linguistic units that are marked for their information value. Topic and Focus are the two main subtypes within this functional category. Dik (1997 I: 313, 327) only counts units as pragmaticals when they receive special grammatical treatment in a language, as when it gets

- a special form or is expressed in a special syntactic position;

- a special marker signaling its pragmatic status;

- a special prosodic contour;

or when it is expressed by some other special construction type.

   For an overview of the various (sub)types of Topic and Focus, I refer to Dik (1997 I: 309-338) and Hengeveld & Mackenzie (2008: 89f.); for a discussion of the grammatical effects of information marking in general, see Foley & Van Valin 1985.


3.3. Functional categories: summing up

There are four main types of functional categories, whose members consist of linguistic units that are the product of an interpersonal act:

Speech Act

Functional Category

Main Subtypes

Illocutionary Act


Declarative, Imperative, Interrogative, …

Thetical Act


Address, Summons, Greeting, Leave-taking, afterthought, …

Propositional Act


Predicate, Referential Phrase (‘noun phrase’, complement clause, etc.), Modifier (adjective, relative clause, genitive, adverb, adverbial, prepositional phrase, article, demonstrative, numeral; also TAM affixes, number affixes, …)

Pragmatic Act


Topic, Focus

 Table 1. Speech acts, functional categories and some of their main subtypes

 Notice that these four functional categories basically cover all the linguistic material that is part of a complete utterance. Dedicated markers of interpersonal acts (i.e. special markers of the various kinds of functional categories) are regarded as part of the mini-grammar of the members of the functional categories associated with these acts. Some examples are:

- distinct sets of illocutionary particles, as in e.g. Nama Hottentot (Hagman 1973: 257);

- dedicated markers of Acts of Predication (‘predicate markers’), Reference (‘noun phrase markers’) and Modification (‘ligatures’, ‘modification markers’), as attested in Austronesian languages (Foley 1976, Adelaar & Himmelmann eds. 2005):

- specialized markers of pragmaticals (‘Topic markers’, ‘Focus markers’), as in Japanese.



[1] The most obvious exceptions to this general rule are (a) zero expressions, which have a meaning and a discourse function, but no phonological form, and (b) empty forms, i.e. forms without a meaning or a discourse function. Empty forms can come in various guises, for example: (i) linking morphemes (e.g. ‘s’ in hunt-s-man), (ii) so-called cranberry morphemes (named after the meaningless element ‘cran’ in ‘cranberry’, and (iii) retinue or servant words (cf. Haiman 2013 on meaningless, decorative morphology).

[2] The notion ‘act’ is in itself ambiguous between an activity and an abstract object. In this article, I focus on the activity interpretation, because it emphasizes the actional status of the speech act.

[3] Traditionally, only definite NPs are deemed to have referential force (Reboul 2001: 515; García and Rijkhoff 2008: 15). For this reason, Dik (1997 I: 130) distinguished between acts of constructing reference (resulting in indefinite NPs) and acts of identifying reference (definite NPs).

[4] So-called one-trick-ponies, linguistic units that are invariably used in same interpersonal function (‘one form, one function’) are an important exception to this general rule (Rijkhoff 2010b).

[5] A layered analysis for sentences was introduced in the context of Dik’s Functional Grammar in Hengeveld 1989. Subsequently Rijkhoff 1990 proposed a unified analysis for a layered representation of noun phrases and sentences (see Rijkhoff (2004, 2008a, 2014) and Rijkhoff & Seibt 2005 for further developments).