Morphologically-conditioned tonotactics in multilevel Maximum Entropy grammar

Shih, Stephanie S.; Inkelas, Sharon

doi:10.3765/amp.v3i0.3659

Cited by 19 publications

(7 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…I call this model HIERARCHICAL MAXENT. Hierarchical regression is used widely to capture statistical generalizations together with idiosyncrasies in variable datasets: linguists in particular have employed random intercepts to measure by-word/lexical class idiosyncrasy (Fruehwald 2012, Zuraw & Hayes 2017, Smith & Moore-Cantwell 2017, inter alia); Shih & Inkelas (2016) and Shih (2018) even adopt the hierarchical model as a theory of learning and competence for their data.…”

Section: Hierarchical Regression Backgroundmentioning

confidence: 99%

Learning a Frequency-Matching Grammar together with Lexical Idiosyncrasy: MaxEnt versus Hierarchical Regression

Zymet

2019

AMP

View full text Add to dashboard Cite

Experimental research has uncovered language learners’ ability to frequency-match to statistical generalizations across the lexicon, while also acquiring the idiosyncratic behavior of individual attested words. How can we model the learning of a frequency-matching grammar together with lexical idiosyncrasy? A recent approach based in the single-level regression model Maximum Entropy Harmonic Grammar makes use of general constraints that putatively capture statistical generalizations across the lexicon, as well as lexical constraints governing the behavior of individual words. I argue on the basis of learning simulations that the approach fails to learn statistical generalizations across the lexicon, running into what I call the GRAMMAR-LEXICON BALANCING PROBLEM: lexical constraints are so powerful that the learner comes to acquire the behavior of each attested form using only these constraints, at which point the general constraint is rendered superfluous and ineffective. I argue that MaxEnt be replaced with the HIERARCHICAL REGRESSION MODEL: multiple layers of regression structure, corresponding to different levels of a hierarchy of generalizations. Hierarchical regression is shown to surmount the grammar-lexicon balancing problem—learning a frequency-matching grammar together with lexical idiosyncrasy—by encoding general constraints as fixed effects and lexical constraints as a random effect. The model is applied to variable Slovenian palatalization, with promising results.

show abstract

Section: Hierarchical Regression Backgroundmentioning

confidence: 99%

Learning a Frequency-Matching Grammar together with Lexical Idiosyncrasy: MaxEnt versus Hierarchical Regression

Zymet

2019

AMP

View full text Add to dashboard Cite

show abstract

“…I will employ a scaling system in which scaling involves the addition of a penalty to the basic weight w of a constraint (Hsu & Jesney 2016, Shih & Inkelas 2016). This amount is determined by a constraint-specific scaling factor s , whose value is multiplied by a numerical value d , which is determined by the corresponding point on the hierarchy of the constraint violation.…”

Section: Scalar Constraint Analysis Of Restrictions On ṽXmentioning

confidence: 99%

Exceptional prosodification effects revisited in Gradient Harmonic Grammar

Hsu

2019

Phonology

View full text Add to dashboard Cite

This paper presents an analysis of exceptional prosodification effects, in which exceptional lexical items appear to follow a regular pattern that is found in a different prosodic context. These patterns have been analysed as cases of prosodic prespecification, where morphemes select a non-default prosodic representation. I argue that prespecification approaches should be reconsidered, and that such patterns are predicted without morpheme-specific prosody in Gradient Harmonic Grammar, a weighted constraint system with gradiently active symbols. Exceptional prosodification effects result from the interaction of two influences on constraint penalties: (i) scaling of constraint violations by prosodic context and (ii) contrastive activity values in underlying forms. This interaction is illustrated with the distribution of French nasal vowels and linking [n]. This approach reduces the amount of structure posited for URs, and provides new arguments for a more uniform syntax–prosody mapping.

show abstract

“…Phonological exceptions are a subset of lexical class-conditioned phonological phenomena, but differ from similar phenomena (e.g. morphosyntactic classes (Smith, 2016); ideophones (Dingemanse, 2012;Shih & Inkelas, 2016), inter alia) in terms of arbitrariness. Exceptions are only distinguishable from non-exceptions by virtue of their idiosyncratic behavior, as opposed to membership in a lexical class that is distinguishable on other grounds (e.g.…”

Section: What Is An Exception?mentioning

confidence: 99%

The disambiguating effects of phonological exceptions in grammar

Hout

2019

AMP

View full text Add to dashboard Cite

This paper explores the analytical and typological consequences of lexically-indexed constraints in Optimality Theory; namely, that the introduction of an indexed constraint by an exception can force the determination of otherwise unspecified rankings in the grammar. This “disambiguating” effect makes two important predictions. First, the typology of exceptional blocking is more complex than has been traditionally assumed: in addition to typical non-undergoing (=faithfulness-obeying) exceptions, exceptions that block a regular repair but undergo an alternative instead (referred to here as “walljumping” exceptions) are also predicted to exist. Second, the forced determination effect means that the existence of an exception determines the possibility or impossibility of other exceptional and regular patterns in the grammar. Using two exceptions to the hiatus resolution conspiracy in Mushunguli (Somali Chizigula, ISO [xma]) as a test case, I show that these predictions hold true, and demonstrate that indexed constraints allow for a unified analysis of hiatus resolution in the language. This result builds upon and strengthens prior observations made by Ito & Mester (1995) regarding the behavior of stratal faithfulness constraints, and supports a view of exceptions as clarifying and reifying agents within the grammar.

show abstract

Morphologically-conditioned tonotactics in multilevel Maximum Entropy grammar

Cited by 19 publications

References 22 publications

Learning a Frequency-Matching Grammar together with Lexical Idiosyncrasy: MaxEnt versus Hierarchical Regression

Learning a Frequency-Matching Grammar together with Lexical Idiosyncrasy: MaxEnt versus Hierarchical Regression

Exceptional prosodification effects revisited in Gradient Harmonic Grammar

The disambiguating effects of phonological exceptions in grammar

Contact Info

Product

Resources

About