Chapter Utterance Complement Clauses

by Sonia Cristofaro

1. Defining the values

This chapter investigates the form of the verb in utterance complements, i.e. complement clauses introduced by utterance predicates such as ‘say’ or ‘tell’ (Noonan 1985: 110-113). Examples (1) - (2) provide instances of utterance complements in English (throughout the chapter, utterance complements will be enclosed in square brackets in the examples):

(1)

John said [that she came].

(2)

John says [he will not do the job].

Complement clauses are usually defined on the basis of syntactic criteria such as embedding, i.e. the fact that the complement clause functions as an argument of the main predicate (see for instance Noonan 1985: 46). However, as was done for purpose, ‘when’ and reason clauses (Chapters 125, 126, 127), utterance complements will be defined here in functional, rather than syntactic, terms. A complement construction is regarded as one expressing a particular relation between events, such that one event (the one coded by the main clause, or the main event) entails that another event (the one coded by the complement clause, or the dependent event) is referred to (for a detailed discussion of this definition, see Cristofaro 2003: ch. 2, 5). For instance, in (1) and (2), John's statement must refer to the occurrence of some event (coming, doing the job). Complement clauses may be introduced by different types of predicates (complement-taking predicates: Noonan 1985), including utterance predicates.

In Cristofaro (2003: ch. 2), the theoretical implications of the proposed functional definition of complement clauses are discussed at length, and some pragmatic criteria are provided to identify complement clauses under this definition.

The proposed definition of complement clauses encompasses all of the traditional cases of utterance complements (embedded clauses, possibly introduced by a complementizer, as in (1) and (2)), plus some other cases that might not be regarded as such under traditional syntactic criteria. For instance, in a number of languages direct speech is the only means available to convey reported speech. This is the case in Kobon (Madang, Trans-New Guinea; Papua New Guinea):

(3) Kobon (Davies 1981b: 1)

rol

Dusin

laŋ

nipe

ip

hag-öp

/

hag-öm

[yad

ram

ar-ab-in

a

g-öp]

yesterday

Dusin

above

3sg

acc.1sg

say-perf.3sg

/

say-ss.3sg

1sg

house

go-pres-1sg

quot

do-perf.3sg

‘Yesterday at Dusin he said to me, “I am going home.”’ /  ‘Yesterday at Dusin he told me that he was going home.’

Other languages display constructions that are not specifically devoted to the expression of either direct or indirect speech, but can be used for both, with a shift in the reference of deictics. This can be seen from the English translations of the following Pirahã (Mura; Brazil) example:

(4) Pirahã (Everett 1986: 269)

hi

gáí-sai

[xahóápátí

ti

xi

aagá-hóág-a]

3

say-nmlz

Xahoapati

1

hunger

have-ingr-remote

‘Xahhoapati said, “I am hungry.” ’ /  ‘Xahhoapati said that I am hungry.’

A direct report differs from an indirect report in a number of structural and functional ways, and is not an argument of the utterance verb introducing it (Munro 1982). However, if direct speech (or a construction making no difference between direct and indirect speech, such as the one in (4)) is the only means available in a language to convey reported speech, it should be assumed that it may also express the conceptual situation expressed by indirect speech in other languages. If this were not the case, the language would have no means to express the relevant conceptual situation (see on this point Cristofaro 2003: ch. 2, 5). Therefore, direct speech clauses are regarded as utterance complements here, unless the language also has some indirect speech construction. This is possible because the proposed functional definition of utterance complements is independent of any particular structural feature of the relevant clauses, and only refers to the conceptual relation between the linked events.

Verb forms in utterance complements may be either balanced or deranked, and the corresponding clauses will henceforth be called balanced utterance complements and deranked utterance complements. The notions of balancing and deranking were originally introduced by Stassen (1985), and are defined in greater detail in Chapter 125. A balanced verb form is one that can occur in an independent declarative clause, e.g. an indicative verb form. The utterance complements in examples (1) - (4) above are balanced. The verb is inflected for the same categories as the forms used in independent declarative clauses (for instance, tense, aspect, and mood in (1), (2), and (4), and tense, aspect, mood and person in (3)), and these categories are expressed in the same way as in independent declarative clauses.

A deranked verb form is one that cannot be used in independent declarative clauses. A deranked verb form may lack some or all of the categorial distinctions relevant to verbs in the language (such as tense, aspect, mood or person agreement distinctions), or display special markers not used in independent declarative clauses, e.g. special tense, aspect, mood or person markers, nominalizers, case markers or adpositions. Example (5) from Maricopa (Yuman; Arizona) provides an instance of a deranked utterance complement. The verb is not inflected for tense, aspect, and mood, and bears the locative case marker –k:

(5) Maricopa (Gordon 1986: 130)

nyaa

ʼi-m

[Pam-sh

mdiily

maa-k]

ʼi-sh

I

say-real

Pam-subj

bread

eat-loc

say-perf

‘I said that Pam ate the bread.’

In example (6), from Italian, the verb is in the infinitive, and is not inflected for tense, mood, or person. In addition, it is introduced by the preposition di  ‘of’. Infinitival utterance complements are used in Italian when main and dependent clause share their subject.

(6) Italian

Dic-e

di

non

pot-er

venire.

say-pres.ind.3sg

of

neg

can-pres.inf

come-pres.inf

‘S/he says s/he cannot come.’

Another case of deranked utterance complements is represented by the use of so-called dependent moods. As was pointed out in the Chapters 125, 126 and 127, dependent moods are verb forms displaying the same categorial distinctions as the verb forms used in independent declarative clauses. However, these distinctions are realized by means of special forms not used in independent declarative clauses; thus dependent moods cannot occur in independent declarative clauses. For instance, in the following West Greenlandic (Eskimo; Greenland) example, the verb is in the so-called contemporative mood. Tense and aspect are expressed as in independent clauses, but mood and person are expressed by means of special affixes that cannot be used in independent declarative clauses:

(7) West Greenlandic (Fortescue 1984: 40)

uqaar-sinnaa-vunga

tama-asa

[uuqattaar-sima-llugit]

say-can-1sg.ind

all-3pl

try-perf-1sg.3pl.contemp

‘I can say that I have tried them all.’

As Stassen (1985: 338-339) observes, if a verb form is inflected for the same categories as the forms used in independent declarative clauses, one might want to regard it as balanced. However, since dependent moods cannot themselves be used in independent declarative clauses, they will be regarded as deranked here.

As was pointed out in the Chapters 125, 126 and 127, the distinction between balancing and deranking is not equivalent to the distinction between finiteness and non-finiteness, although the two overlap to a significant extent. For discussion of the relevant issues, see Chapter 125.

For any given language, utterance complements can be coded by balanced verb forms only, by deranked verb forms only, or by either balanced or deranked verb forms. Map 128A shows the distribution of these three types in the world's languages.

Values of Map 128A. Utterance Complement Clauses
Go to map
 ValueRepresentation
Balanced 114
Balanced/deranked 18
Deranked 11
Total: 143

2. Geographical distribution

Languages with exclusively balanced utterance complements are overwhelmingly dominant in the language sample as a whole. They are also dominant in all geographical areas, except for an area extending from Europe in the west to Siberia in the east. Although several languages with exclusively balanced utterance complements are found in Europe, this area displays a high number of languages with both deranked and balanced utterance complements, as well as a pocket of languages in Siberia with exclusively deranked utterance complements.

Apart from this area, languages with both deranked and balanced utterance complements are also found in North America, Africa and India, but they are marginal in all of these areas. Languages with exclusively deranked utterance complements can be found in North and South America, India, and Australia. They are marginal in all of these areas. However, South America displays a higher concentration of languages with exclusively deranked utterance complements than the other just-mentioned areas.

3. Theoretical issues

The most significant fact about the distribution of balanced and deranked utterance complements across the language sample is the overwhelming dominance of languages with exclusively balanced utterance complements. These languages are widespread, and in most cases dominant in all genetic families and geographical areas (with the exception of Siberia). This suggests that deranked utterance complements are disfavored cross-linguistically.

In Cristofaro (1998, 2003) it is argued that the occurrence of deranked verb forms is related to a number of semantic features of particular subordination relation types, and utterance complements do not display these features. For instance, deranked verb forms are often not marked for tense, aspect, mood, or person agreement. As a result, they tend to be used when the time reference, the aspect and mood value, and the participants of the dependent event are predetermined, and thus need not be overtly specified. However, utterance complements do not involve predetermination of the time reference or the participants of the dependent event (for instance, examples (1) and (2) show that utterance complements may have different time reference, and may or may not share a participant with the main clause; on this point see also Noonan 1985). This disfavors the occurrence of verb forms not marked for tense or person agreement. Also, the occurrence of deranked verb forms is related to semantic integration between the linked events (in the sense defined in Givón 1980, 1990: ch. 13; see the discussion in Chapter 125). However, utterance complements involve no semantic integration between the linked events – in fact, the subordination relation here is not established between two events, but between an event (the main event) and a propositional content referring to another event (see on this point Dik 1989, 1997a and 1997b; Hengeveld 1989 and 1990; Siewierska 1991).

The occurrence of deranked verb forms is also related to the ability of the dependent event to be conceptualized as a thing rather than as a process (in Langacker's sense of these terms: see Langacker 1987a, 1987b and 1991, as well as Cristofaro 2003: ch. 6, 9). Conceptualization of the dependent event as a thing rather than as a process leads to the presence of nominal properties on the verb, such as case markers or adpositions, as well as the absence of typical verbal properties such as tense, aspect and mood distinctions. All of these phenomena are characterictics of deranking. However, utterance complements do not seem to easily allow conceptualization of the dependent event as a thing, and this disfavors the occurrence of verb forms displaying nominal properties, or lacking verbal properties.

Finally, the occurrence of some deranked forms such as subjunctives and the like is related to the dependent event being presented as unrealized. However, utterance complements have no implication about whether or not the dependent event is realized. This disfavors the occurrence of verb forms devoted to the expression of irrealis, such as subjunctives.

All this may explain why deranked utterance complements are disfavored cross-linguistically. In fact, as is shown in Cristofaro (1998, 2003), an implicational hierarchy (the Subordination Deranking Hierarchy) exists such that if a language uses deranked verb forms for utterance complements than it will use deranked verb forms for all of the other complement types, as well as for most types of relative and adverbial clauses.