2. LITERATURE REVIEW
2.1. Optimality Theory
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
CHAPTER 2
LITERATURE REVIEW
This chapter contains three sections. Section 1 reviews the Optimality Theory (OT).
Section 2 provides information from previous studies on reduplication in Mandarin Chinese and on Mandarin onomatopoeia. Section 3 discusses reduplication under OT and the constraints used.
2.1. Optimality Theory
2.1.1. Basic concept of OTOptimality Theory, which was first introduced by Prince and Smolensky (Prince &
Smolensky 1993/2004; McCarthy & Prince 1993a, 1993b, 1994, 1995, 1999), has aroused great interest in the study of phonology. OT is a development of Generative Grammar, a theory sharing its focus on the formal description and quest for universal principles on the basis of empirical research of linguistic typology and language acquisition. However, OT radically differs from the generative model and has the central idea that surface forms of language reflect resolutions of conflicts between constraints. The five basic tenets of OT are given below.
(5) Principles of Optimality Theory
a. UniversalityUG provides a set Con of constraints that are universal and universally present in all grammars.
b. Violability
Constraints are violable; but violation is minimal.
c. Ranking
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
The constraints of Con are ranked on a language-particular basis; the notion of minimal violation is defined in terms of this ranking. A grammar is a ranking of the constraints set.
d. Inclusiveness
The constraints hierarchy evaluates a set of candidates analyzes that are admitted by very general considerations of structural well-formedness.
e. Parallelism
Best-satisfaction of the constraint hierarchy is computed over the whole
hierarchy and the whole candidate set. There is no serial derivation. (Other views are consistent with a-d, as well)
Universality is essential to the emergence of the unmarked (TETU), one of the main points in this thesis. Since structural constraints are universal and present in every grammar, even those that are obviously and commonly violated in a given language are predicted to be available to do their work under appropriate conditions. It needs to be noticed that Universality is hopeless without Violability and Ranking, in the face of the diversity of inter-linguistic variation seen in linguistic systems.
These principles must figure in a particular conception of how grammar is organized.
Universal grammar must minimally provide the following:
Con. The set of constraints out of which grammars are constructed.
Gen. A function defining for each possible input i, the range of candidate linguistic analyzes available to i.
Eval. A function that comparatively evaluate sets of forms with respect to a given constraint Γ, a ranking of Con.
The following schema sketches the way input-output pairing is accomplished using these notions. Suppose we have a grammarΓm, the mth ranking of Con, and an input
in
i— a lexical entry, if we are looking at word phonology.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
(6) Schema for an OT grammar
Gen (ini) = {cand1, cand2, …}Eval (Γm , {cand1, cand2, …} ) → (candk) This grammar pairs input ini with output candk.
The function Gen emits a set of candidate analyzes consistent with a given input. Gen consists of a very broad principle of linguistic forms, essentially limited to those that define the representational primitives and their most basic modes of combinations.
Eval deals with a system of ranked constraintsΓ: a formal construction on Con that yields the grammar of an individual language. It rates the members of the candidate set in terms of their relative harmony, or degree of success with respect to the language‘s ranking Γ of constraints. It imposes an order on the various candidates, and a maximally harmonic candidate is optimal. Such a candidate best-satisfies or minimally violates the grammar‘s constraint ranking. It is output associated by the grammar with the specific input ini. The various non-optimal candidates have no grammatical status; no direct inferences about plausible patterns of variation or historical changes can be drawn from their ordering. The constraint ranking for the Mandarin core lexicon is different from that of Mandarin onomatopoeia so that they have various phonological structures.
2.1.2. Constraint family in OT
Before coming to the discussion of actual constraints and their ranking in a specific situation, let us first find out in a general way about the five major forces embodied by constraints.
a. Faithfulness Constraint. It requires that outputs preserve the properties of their basic forms, requiring some kind of similarity between output and its input.
b. Markedness Constraint. It is a general denominator for grammatical factors that exert pressure toward unmarked types of structure. The markedness constraint
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
requires that the output forms meet some criterion of structural well-formedeness.
c. Generalized Alignment Constraint. It was developed in the analysis of relation between syntactic constituents and prosodic phrases (McCarthy & Prince 1993b;
Kager 1999). Generalized Alignment Constraint serves the match between the edge of various linguistic constituents in prosodic category and grammatical category.
d. Anchoring Constraint (McCarthy & Prince 1993a, 1995a, b). It is member of the Faithfulness constraint family and be taken as positional faithfulness constraint.
The anchoring constraint serves to ensure that a certain edge of the reduplicant will correspond to a certain edge of the base.
e. Adjacency Constraint. Kitto & de Lacy (1999) proposed the BE-A
DJACENCY
constraint and tried to account for the tendency of copied epenthetic segments to be near the segment form which they have copied their features. A
DJACENCY
BR
requires the same of reduplicant segments. Thus, there potentially is a family of A
DJACENCY
constraints which are restricted to an output relation only, as any ADJACENCY
requirement in the input-output (IO) domain would be nonsensical.2.1.3 Cophonology
The cophonology approach, developed in Orgun (1996) and Anttila (1997), and much subsequent work, holds that within a single language there can be co-existing distinct phonological systems, indexed to such components of language as register, lexical class, morphological category, and most conspicuously in the context of this thesis, individual morphological construction. The cophonology approach‘s diversity is captured by associating morphological constructions or lexical classes with different phonological grammars, i.e., constraint rankings. All constraints within a given cophonology are fully general (e.g., M
AX
, the ban on deletion, or *COMP
-ONSET
, the‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
ban on complex onset); and morphonological differentiation of phonological patterns results from different ranking of the constraints across cophonologies.
In Mandarin onomatopoeia, several morphological constructions can occur and the same input form can generate various output forms. This phenomenon can be explained by a cophonology approach since there are many phonological grammar/constraint rankings in the generation of Mandarin onomatopoeia.
2.1.4 The emergence of the unmarked
Under the OT rubric of the emergence of the unmarked (McCarthy and Prince, 1994a), which provides a way to allow only unmarked structures in a domain like reduplicant while permitting the corresponding marked structure to occur elsewhere in the
language. The idea is that non-copying of a base segment, with substitution of some fixed, default segment, decreases phonological markedness.
If a given markedness constraint M crucially dominates an appropriate faithfulness constraint F (and no constraint dominating M somehow vitiates its force), then no M-offending structure will appear in a surface form, even at the expense of
imperfectly reproducing some underlying forms. If we rank constraints in the other way and faithfulness takes precedence, so that the M-offending structure can be found in surface forms. Differences in ranking give differences in activity of markedness constraints, so it is possible to say that every constraint is present in the grammar of every language, though if a constraint is crucially dominated, its activity may be limited or non-existent. The limited but nonetheless visible activity of dominated markedness constraints is essential to the theory of fixed segmentism (Alderete et al.
1999).