Abstract. Self-repair is a common phenomenon in communication. The present thesis discusses self-repair proposed by Levelt, based on the Relevance Theory. By analyzing self-repair, the research presents the purpose and strategy of self-repair and systematically analyses three phases of self-repair in order to display how Relevance Theory performs in the process of self-repair and thus conduct the pragmatic functions of it. Therefore it is a useful tool to help communicators grasp the conversational strategies and improve their communicative abilities.
Key words: self-repair； relevance theory； pragmatic functions
Self-repair refers to a speaker spontaneously corrects the errors or modifies the inappropriate information in his own speech. It is a common phenomenon in spoken language that the study of it can help us know better the nature of spoken language. Studies abroad dwell on conversation analysis, psycholinguistics, sociolinguistics, computational linguistics, etc.. They have gained fruitful results which arouse the interests of researchers at home in recent years. They present their points of view in many different perspectives. However there are few researches mentioned self-repair in relevance-theoretic perspective.
Relevance Theory aims "to identify underlying mechanisms, rooted in human psychology, which explain how humans communicate with one another" (Sperber&Wilson, 1986: 32). It can be regarded as a general principle to interpret many linguistic phenomena. It is based on the definition of relevance and two general principles. Relevance is defined in terms of contextual effect and processing effort. On the basis of this definition, two general principles are proposed: the Cognitive Principle that human cognition tends to be geared to the maximization of relevance, and the Communicative Principle that every act of inferential communication communicates a presumption of its own optimal relevance. Relevance Theory claims to provide such a theory on which we base to account for how utterances are interpreted. In the present thesis, the author attempt to study self-repair in relevance-theoretic approach, for the sake of enriching the interpretability of Relevance Theory and discussing the pragmatic functions of self-repair in daily communication based on Relevance Theory.
II.Overview of Self-repair
Schegloff et al. (1977) define self-repair from the perspective of conversation analysis. They referred to the repair as a variety of ways of handling troubles that arise in the process of speaking, understanding, and communicating in an interactional setting. They also point out that if the trouble source, the initiation and the repair are within the same turn, then the repair is self-initiated self-accomplished repair.
What’s more, Schegloff et al. (1977) distinguished the term "correct (errors)" and "repair". "Correct" refers to replace a slip of the tongue or an error with a correct form while what needs to be repaired is more than that. For example, repair occurs as the speaker is doing a word search in some instance.
Another statement that "self-repair" refers that the speaker corrects his own errors in his speech was given by Van Hest (1996)
From the psycholinguistic perspective, Kormos (1999) argues that self-repair comes about when the speaker detects that the output has been erroneous or inappropriate, halts the speech flow, and finally executes a correction.
Postma (2000) states that self-repair is the correction of errors without external prompting, frequently within a short span of time from the moment of error occurrence. Self-repair implies the existence of specialized control devices or ‘monitors’ which verify the correctness of ongoing motor activity and response output.
Rieger (2003) demonstrated that the repair behavior includes "error correction, the search of a word and the use of hesitation pauses, lexical, quasi-lexical, or non-lexical pause fillers, immediate lexical changes, false starts, and instantaneous repetitions".
In conclusion, these statements from different angles present different researchers purposes. They help people understand self-repair in a broad sense.Schgloff et al’s contribution makes a clear distinction between "correct" and "repair" for the first time but their definition is too general and inconvenient to operate. Riegers’s definition seems even bigger including almost all the techniques for gaining time in spoken language but he only puts emphasis on repetition,which seems narrow the strength of explaination. Potsma's and van Hest’s definitions are simple and superficial. Self-repair is not simply a behavior of correcting errors. Kormos’s definition together with Postma’s and Hest’s are all from the perspective of psycholinguistic, it gave an inspiration to other perspective studies of self-repair.
Comparably speaking, self-repair in the present thesis means that when the speech errors occur, speakers will correct the errors or modify the inappropriate expressions simultaneously.
III. Relevance Theory
Relevance Theory is proposed by Sperber and Wilson in their work Relevance: Communication and Cognition (1986/1995). It focuses on two issues: communication and cognition. According to them, human communication is an ostensive-inferential process with informative and communicative intention. Utterances automatically create expectation of relevance which is in relation to the context, individual and phenomenon. They propose two principles of relevance: cognitive principle and communicative principle.
3.1 Ostensive-Inferential Communication
There are two models of communication in Sperber and Wilson’s opinion: the code model and the inferential model .the former is more traditional, in which communication is seen as the encoding and decoding of messages. The latter, the name they apply to the approach initiated by Grice, in contrast, is of recent origin, which views communication as essentially the recognition of the speaker’s intention by the audience.
Sperber and Wilson are in favor of the inferential approach. They argue: “Communication is successful not when hearers recognize the linguistic meaning of the utterance, but when they infer the speaker’s ‘meaning’ from it”(Sperber and Wilson,1986:23). “Verbal communication is a complex form of communication. Linguistic coding and decoding is involved, but the linguistic meaning of an uttered sentence falls short of encoding what the speaker means: it merely helps the audience infer what communicator means. The output of decoding is correctly treated by the audience as a piece of evidence about the communicator’s intensions. In other words, a coding-decoding process is subservient to a Gricean inferential process” (Sperber and Wilson,1986:27)
After striking a comparison between code model and inferential model, Sperber and Wilson developed a new approach to human communication, ostensive-inferential communication.
"The communicator produces a stimulus which makes it mutually manifest to communicator and audience that the communicator intends, by means of this stimulus, to make manifest or more manifest to the audience a set of assumptions"(Sperber and Wilson, 1995: 63).
Communication is an ostensive-inferential process involves informative and communicative intention. “Inference and ostention are one and the same process, but seen from two different point of view: that of the communicator who is involved in ostension and that of the audience who is involved in inference" (Sperber and Wilson, 1995: 54). By ostension, the speaker intends to make something manifest. In Sperber and Wilson’s, “ostension” is the “behavior which makes manifest an intention to make something manifest,” to put more easily,“showing someone something is a case of ostension”. On the other hand, by inference, the hearer attempts to interpret the speaker’s intention according to the speaker’s “ostensive behavior” and contextual assumption. The realization of this ostensive-inferential process depends on the mutual manifestness of cognitive environments between communicator and audience.
3.2 Principle of Relevance
Sperber and Wilson have proposed two principles of relevance, the first termed as cognitive principle and the second communicative principle. The two principles go like follows (Sperber and Wilson, 1995: 260):
a. Human cognition tends to be geared to the maximization of relevance.
b. Every act of ostensive communication communicates a presumption of its own optimal relevance.
Relevance is not a definite notion, but a matter of degree and a comparative notion that is characterized in terms of effect and effort. Sperber and Wilson (1986: Preface) claim that the Principle of Relevance is adequate enough in its own to explain how linguistic meaning and contextual factors interact to determine utterance interpretation or verbal comprehension.
According to Relevance Theory, communication is a cognitive process between communicator and audience. During the process the audience tried his best to search the maximal relevant, in order to make the communication effective.
IV. A Relevance-Theoretic Analysis of Self-repair
4.1 Self-repair: Purposes and Strategies
Spontaneous speech is different from the written words, it needs the speaker express his idea at once, in order to communicate. This online nature of human spontaneous speech gives rise to speech errors and disfluencies typical of hesitation, pause, silence, repetition, repair, etc. it has invited the attention of many researchers as mentioned before. Here the author first pays effort on the study of purpose and strategies.
4.1.1 Word Searching and Relevance Maximizing
Making a self-repair is an action of detecting an error or inappropriateness in a conversation, and then the speaker “transfer” structural properties of the original utterance to the correction. By transferring and reusing structural properties of previous speech the speaker may at the same time gain fluency, and establish discourse coherence to the advantage of the listener. That is to say, the purpose of self-repair is to search appropriate words for effective communication and in order to achieve the maximal relevance in communication.
Self-repair, in some degree, is considered as a guide of searching relevance. When the speaker finds that his utterance diverges from his intension, he makes a self-repair in searching for the correct words for the sake of offering relevant information, and makes it possible for the hearer to decide very rapidly on how the new utterance related to the interrupt one. It constrains the choice of context and cognitive effect. So it is the insurance of successful communication.
4.1.2 Error Monitoring and Speech Planning
Levelt (1983) notes that language production is composed by conceptualizer, formulator, articulator and self-monitor. Speakers get procedural knowledge and declarative knowledge from conceptualizer, and form preverbal message, then output this message to formulator. The formulator encodes both grammar and the phonology. Next the message comes to articulator, which articulatory movements are required. During the planning of word formation speech error may occur, self-monitoring can make the speaker aware of the error or in other words, an alarm signal is sent to working memory. The speaker can then take action on the information received. This can range from quite fluent adjustments, such as in loudness and rate of speech delivery, to a complete halt-and-restart action. This restarting is not neutral with respect to the interrupted utterance, it usually reinstalls some of the listener, who has to solve a “continuation problem”.
If there is speech monitor there must be speech planning. They go through the whole process of self-repair. For speakers to select the message, they must monitor the utterance they are planning and detect errors. This is a cognitive process; the speaker will adjust their output information to clarify their intension, through the planning of speech. The planning of speech embodies on the strategies of speech editing, such as, interruption, silent pause, filled pause and prosodic, acoustic, intonational cues, etc.. These strategies denote that the speaker realized the mismatched information and searching for the appropriate one, at the same time, the speaker need time to plan speech, encode new information which is proper to their intention.
4.2Three Phases of Self-repair
According to Levelt (1983) and van Hest (1996), a self-repair typically proceeds in three phases. The first part is the original utterance (OU), which includes the trouble source or reparandum which is the item to be repaired. The trouble source can be a linguistic problem at any level, such as phonological, morphological, lexical or syntactic level. It involves the monitoring of one’s own speech and the interruption of the flow of speech when trouble is detected. The second part is labeled as editing terms which is characterized by hesitation, pausing, etc.. It is the point at which the flow of speech is interrupted for editing. Schegloff et al. call it the initiation of repair. The third phase is alteration, named as repair proper by Levelt (1983) or completion of repair by Schegloff et al.. Despite the names of the three phases are different according to different scholars in different research fields, the main concept is the same.
4.3 Pragmatic Functions
Self-repair performs a variety of pragmatic functions such as ambiguity avoiding, turn holding, face saving, contemplating, and delaying. Sometimes making a self-repair may perform more than one function, here we only focus on the main function of a self-repair.
4.3.1 Ambiguity Avoiding
Making a self-repair is a process of detecting errors and replanning speech, that is, making the expression clearly without ambiguity. Speaker will try and interpret his own speech in the context of what was previously said by himself or by another person. He may thus become aware of ambiguity, vagueness, indeterminacy of reference. He makes a self-repair immediately in order to avoid ambiguity, and express relevant information based on the context assumption. Listeners are intended to see what the speaker is saying as relevant to the current situation. They are to ask themselves, “Why is the speaker saying this now?” and construe the mutually recognizable reasons as implicatures.
4.3.2 Turn Holding
Self-repair is a mean for the speaker to hold the turn. Speakers always make a silent pause or produce some kind of signals, such as uh, um, well, look, etc, which is called filled pause or filler, to show that he doesn’t fininsh his turn as Maclay&Osgood (1959:41) presented their point of view：
Let us assume that the speaker is motivated to keep control of the conversational “ball” until he was achieved some sense of completion… therefore, if he pauses long enough to receive the cue of his own silence, he will produce some kind of signal which says, in effect, “I’m still in control-don’t interrupt me.”
A related proposal is that fillers are elements “whereby the speaker, momentarily unable or unwilling to produce the required word or phrase, gives audible evidence that he is engaged in speech-productive labor” (Goffman, 1981:293). Previous researches is a powerful evidence to show that self-repair performs as a marker of turn holding.
4.3.3 Face Saving
Error speech is a face-threatening act to the speaker, so implementing self-repair helps face-resuming. People try to express their intensions correctly and fluently in order to maintain face and pursue equality in society. Sometimes self-repair reveals people’s uncertainty in communication.
Spontaneous speech is different from the written discourse, speakers need time to contemplate and compose their language before proceeding. Speakers attempt to express clearly to the hearers that what they are saying is indeed relevant, so contemplation is performed to seek memory for a proper word and plan the continuous speech. Under the circumstance, editing terms show that speaker's own ability to produce the optimally relevant response is compromised in some way, and that the hearer needs to take that into account when interpreting the response.
When speaker make a self-repair, he may choose delaying time on purpose, in order to search words, compose utterance. For speaker to select the message “I am now initiating what I expect to be a minor, or major, delay”, they must monitor the utterance they are formulating and detect an imminent delay. There is much evidence that speakers scan internal representations-a type of inner speech-for material they have formulated but not yet produced (Levelt, 1983).
The relevant-theoretic study of self-repair enriches and develops the study of this phenomenon in natural language. Based on Relevance Theory, making a self-repair ensures the hearer's correct interpretation of utterance with the least effort. The study has made further efforts to broad relevance-based view on the pragmatic functions of self-repair and it is a powerful evidence to display the interpretability of the relevance theory. Meanwhile, the thesis can be considered as a blank filling in the studies of self-repair has been done so far. It can also fill the blanks of studies on conversation analysis. These findings can also induce the further studies on Chinese and other languages. Moreover, the appropriate use of self-repair can also develop and improve the efficiency of communication.
Goffman,E. Radio talk. In E.Goffman(Ed.), Forms of talkPhiladelphia, PA: University of Pennsylvania Press,1981:197-327.
Kormos , J. Monitoring and self-repair in L2 [J]. Language Learning 49, 1999, (2): 303-342
Levelt, W. J. M.. Monitoring and self-repair in speech [J].Cognition, 1983,(14):41-104.
 Postma, A.. Detection of errors during speech production: a review of speech monitoring models [J].Cognition, 2000,(77):97-131.
 Rieger, C. L. Repetition as Self-repair Strategies in English and German Conversations [J].Journal of Pragmatics, 2003, (35): 47-69.
 Schegloff, E. A., G. Jefferson, &H. Sacks. The preference for self-correction in the organization of repair in conversation[J].Language, 1977,(53):361-382.
 Schegloff, E.A. When others initiate repair [J]. Applied Linguistics, 2000, 21(2): 205-243.
Sperber, D& Wilson, D. Relevance: Communication and Cognition (1st edition) [M].Cambridge: Harvard University Press, 1986.
 Sperber, D& Wilson, D. Relevance: Communication and Cognition (2nd edition) [M] .Cambridge: Harvard University Press, 1995.
VanHest, E.Self-repair in L1and L2 Production [M].Tilburg, The Netherlands: Tilburg University Press,1996.