Search results
modification of the input, contravening faithfulness. Input-output faithfulness and base-reduplicant identity, we argue, are controlled by exactly the same set of formal considerations, played out over different pairs of compared structures. In aid of this conception, we revise the implementation of faithfulness presented in Prince & Smolensky ...
Feb 8, 2024 · We highlight that the current trend towards increasing the plausibility of explanations, primarily driven by the demand for user-friendly interfaces, may come at the cost of diminishing their faithfulness. We assert that the faithfulness of explanations is critical in LLMs employed for high-stakes decision-making.
output form to its corresponding input (loosely speaking, underlying) form and require the two to be identical along some phonologically relevant dimension; for example, there are different faithfulness constraints that penalize epenthesis, deletion, and featural change. Different rankings among markedness and faithfulness constraints lead
- University of California, Santa Cruz
- 1 Background
- (6) For any structure M=(N, V, formulae j and y,
- Eo[IO(i, o)] and M, o=j
- 6 A general method for stating new correspondences
- (27) M, g= j
- 7 Beyond the outer limits
- (Nuc!<%>æ)
- Coda
- input * T
- 8 Conclusion
We develop an extensible description logic for stating the content of optimality-theoretic constraints in phonology, and specify a class of structures for inter-preting it. The aim is a transparent formalisation of OT. We show how to state a wide range of constraints, including markedness, input–output faithfulness and base–reduplicant faithfulness...
Optimality Theory (OT) constraints have developed a rich array of forms. The major families of constraint, markedness and faithfulness, each have identifiable subfamilies, diering, sometimes quite subtly, in their factual coverage, computational properties and implications for learnability. The result is a theory with intrinsic interest and impress...
M, u=p M, u=ÿj M, u=j!y M, u=j/y M, u=j*y M, u=j&y M, u=<%>j M, u=<$>j M, u= j M, u= j M, u= (j1, i‰ i‰ i‰ i‰ i‰ i‰ i‰ i‰ i‰ i‰ ... , jn) i‰ D, L, f), node u of M, and well-formed usV (p), where psProp M, u j M, u=j and M, u=y M, u=j or M, u=y M, u j or M, u=y M, u=j i‰ M, u=y Eu¢[D(u, u¢)] and M, u¢=j Eu¢[D(u¢, u)] and M, u¢=j Eu¢[L(u, u¢)] ...
In other words, in the structure M at the input node i, the formula j holds if and only if there is an output node o such that the relation IO holds between i and o, and at the output node o, j holds. Formally, correspon-dence is the same as dominance: it defines a binary relation (denotes a set of ordered pairs of nodes). We match this partiti...
Input–output faithfulness is not the only kind of OT constraint that calls upon correspondence. Base–reduplicant faithfulness is another. These constraints are defined so that satisfaction of all of them results in com-plete reduplication. An input like {/dum/, REDUP}, where REDUP is some kind of reduplicative template or abstract morpheme demandin...
i‰ Eh[GH(g, h)] and M, h=j The converse is of course also definable; whether it is needed would depend on the nature of the structures that NG and NH pick out.
We now turn to two constraint types proposed in the literature for which the above technique for extending the description language to capture new classes of OT constraint fails quite strikingly. We consider output–output correspondence and sympathy constraints. For both, the question of whether a given candidate (as defined above) satisfies their ...
* ÿ<%> ! (Coda!<%>r) ‘If u is an output syllable node, then u does not have both a Nuc-labelled daughter dominating (exactly the features of) [æ] and a Coda-labelled daughter dominating (exactly the features of) [r].’ Benua’s claim about why we get [lær] and not [lAr] is that the truncated form must be faithful to an independent output, which we he...
/ A j k î m j k m / Word Nuc Ons Nuc Nuc Coda A j A î m
‘If u is an input node, then u has a sympathy correspondent.’ At an abstract level, intracandidate sympathy structures are very much like those required for paradigm uniformity. They are complex objects, but ones that we can get a grip on in a modal language using the same view of candidates that is at the heart of the classical OT formalism.
We showed in §§3–6 that a modal logic like L is a versatile and useful description language for OT grammars. The content of a wide range of OT constraints, including markedness and a variety of faithfulness constraints, can be stated using this kind of syntax and model theory. Furthermore, adding new kinds of constraints can be relatively easy. How...
Apr 1, 2014 · Faithfulness constraints (Kager, 1999: 10) can be regarded as a prime residue of generative grammar, because they crucially involve a comparison between representations at an underlying (input) level and at an output level: if there is a difference (e.g. by deletion, insertion, or other changes), then input–output (IO) faithfulness is violated. As a theory that explicitly focuses on the ...
- Jeroen van de Weijer
- 2014
interaction between output or markedness constraints and input-output mapping or faithfulness constraints. 2 In other words, optionality is an expected consequence of violable and conflicting universal constraints and their language-particular ranking, the core assumptions of OT.
People also ask
What are faithfulness constraints?
Which faithfulness constraint is violated when input place specifications are altered?
Why does M-Fort(STR) outrank the faithfulness constraint?
When is an explanation considered faithful?
What is the difference between faithfulness and plausible explanations?
Why is faithfulness important in LLM development?
correspondence should be generalized to include other kinds of linguistic relations, such as input-output faithfulness in particular (§2). In this way, the apparatus of copying constraints is combined with faithfulness into a broadly applicable Correspondence Theory. The key notion underlying this generalization is identity.
