A Constraint-based Account For Mandarin Speech Errors

A speech error refers to the disruption in the production of speech through a conscious or unconscious linguistic deviation from the apparently intended form of an utterance.According to the instant psychological statement of the speaker when speech errors occur,they can be divided into conscious speech errors and unconscious speech errors.According to the linguistic categories involved in the speech errors,they can be divided into lexical speech errors,syntactic speech errors,semantic speech errors,phonological speech errors and so on.In this research,the study focuses on the unconscious phonological and phonetic speech errors.Previous speech error studies focus on the classification of different types of speech errors and speech error corpus building.In the traditional studies,speech errors are classified by the linear sequential distortion,that is,by the linear deviation of the target units from the proper form to the error form.In this way,the typology of speech errors is built and all speech errors are generally classified into anticipation,preservation,exchange,addition,deletion and so on.This typology based on linear alternation has its linguistic meaning,which helps to evidence different linguistic units and structures and to illustrate the psychological process of language production.However,due to its descriptive nature,this method fails to demonstrate the mechanism behind any type of the speech error.It just restates the form of speech errors in a general way,instead of giving any scientific illustration of the speech error mechanism.In order to overcome the descriptive limitation of the previous study,to meet the pursuit of linguistic descriptive accuracy,and to better answer the above questions,this thesis adopts the constraint-based approach as the basic methodology.Many modern linguistic theories regard constraints as the basic mechanism in human linguistic system,which holds that there is the constraint-based approach between the input form and the output form of human languages.This input-to-output correspondent relationship can also be projected to the output-to-output counterpart.Therefore,the correspondence and variations between different outputs can also be demonstrated by the ranking of faithfulness and markedness constraints.The proper form and the deviated form of a speech error are two types of outputs.Accordingly,there are the correspondence and discrepancy between the two,which can be illustrated by a set of ranked constraints.In this way,the typology of any type of speech errors can be built by a certain ranking of constraints.For example,the mechanism behind the onset anticipatory speech error is the mismatch of the right onset unit in the proper form to the left onset slot in the deviated form.As a result,it seems that the right onset unit anticipates and replaces the left onset unit in the deviated form.This mechanism of mismatch can be represented by the constraint ranking : ALIGN ONSET(P,R,D,L)>> LINEARITY–PD.This constraint ranking can be projected to the more general one: ALIGN(P,R,D,L)>> LINEARITY–PD,which represents all segment-based anticipatory speech errors.In other words,it stands for the typology of all segment-based anticipatory speech errors.