Parenthesis and the various types of parenthetical expressions have attracted constant but never intense attention. This monograph is one of those very few specifically dedicated to parenthetical verbs or, as I prefer to call them, reduced parenthetical clauses (RPCs) in three Romance languages, French, Italian, and Spanish. Typical French RPCs are, e.g., je crois, disons, je dirais, je veux dire, je pense, je sais pas, and je trouve.
The study draws extensively on the methodology of contemporary corpus linguistics and is based on the major corpora of spoken French, Italian, and Spanish comprising several millions of words that document spoken language over the last 30 years. The selection of RPCs from the corpora is based on formal, i.e., semasiologic properties.
The RPCs are described from several perspectives or at several analytic levels: sociolinguistic, pragmatic, semantic, syntactic, and prosodic. The pragmatic, the most dominant part of the analysis, owes much to the theory first sketched out by Urmson (1952) and Hare (1970), and recently further developed in Caffi's (1999, 2001) mitigation framework.
Its major findings consist in a clear definition of RPCs, in a list of the relevant expressions in the three languages, in the understanding of their pragmatic function and of the factors influencing their use, in a coherent classification of RPCs, and in the description of their syntactic and prosodic properties. Other findings are the fact that RPCs are not always restricted to statements but may also occur in questions and that belief verbs are not as frequent among RPCs as commonly assumed. Although the book is about French, Italian, and Spanish parentheticals, its findings are relevant for other languages such as English and German.
In general, the frequency of RPCs is high in Italian, lower in French and lowest in Spanish. However, this is due mainly to the persistent use of diciamo in the Italian texts. If we exclude Fr. disons, It. diciamo, and Sp. digamos from the group of mitigating RPCs, there is no difference between French and Italian. Whereas in the French and Italian data mitigating RPCs achieve high percentages in dialogic texts face to face with regulated turn-taking produced in the institutional-public domain and low percentages in monologic texts produced in all kinds of domain, no clear tendency can be observed in the Spanish data.
Based on deictic orientation and sentence type (declarative, imperative, interrogative), I distinguish clauses alleviating speaker responsibility from those removing and those dividing speaker responsibility. All these clauses reduce communicative responsibility and they are at the center of this study. Initially I also took into account RPCs with phatic function, i.e., I isolated a group of addressee-centered imperative and interrogative clauses which do not reduce speaker responsibility but invite the addressee to co-operate and control the effectiveness of communication. Based on the perspective of the logical structures of utterances and on the functions of RPCs, I identified four main classes of RPCs that reduce communicative responsibility:
1. clauses mitigating the phrastic (propositional content);
2. clauses indicating the tropic (illocution) and mitigating the phrastic or the neustic (speaker commitment);
3. clauses directly mitigating the neustic;
4. clauses indirectly mitigating the neustic.
Several clauses mitigate more than one utterance component. That is, they may alternate between two components (typically phrastic or neustic) or operate on two at the same time (e.g., tropic and phrastic, tropic and neustic). Furthermore, I identified a class of clauses reporting discourse. Since, ideally, these clauses take communicative responsibility completely away from the speaker and speaker commitment is not at stake, they constitute an extreme form of responsibility reduction or lie even outside it.
The data show that parentheticals may be inserted in virtually any sentence position, even within noun phrases or prepositional phrases. Sentence-initial parentheticals are more common in Italian and French than in Spanish. I defend the view that in sentence-initial position and in the absence of a complementizer RPCs do not govern the following clause. In a section dedicated to questions of scope, I maintain that the scope of an RPC is largely predetermined by its pragmatic function and the structure and meaning of the host. Three major scopes, phrase-limited, clause-limited, and sentence-limited, can be distinguished. RPCs mitigating the phrastic generally limit their scope to the phrase they are in, most other RPCs are constrained to the surrounding clause, only rarely does an RPC have a complex sentence in its scope.
Previous studies as well as the corpus data suggest that in terms of prosody, a difference should be made between parenthetical expressions that interrupt the utterance's intonational contour and those that do not. The expressions of the second type are usually but not always RPCs. Their prosodic properties correspond partly to those of the expressions of the first type. Yet, there are at least two very important differences: They are not introduced by a sudden break of the intonational contour and they are not separated from their intonation units by pauses. Hence, with respect to intonation, in most cases RPCs are not really parentheticals, in the sense of being separate and isolated from the host sentence.