Frequency of violation and constraint-based phonological learning

Abstract

This paper provides two arguments that constraint-based grammars should not be learned by directly mirroring the frequency of constraint violation and satisfaction in the target words of a language. The first argument comes from a class of stages attested in phonological development, called Intermediate Faith (IF) stages, in which children produce marked structures only in privileged positions. Two such stages are presented and analyzed, from the literature on English and French L1 acquisition, and their learning consequences are examined. The second argument concerns the degree of restrictiveness that a learner's end-state grammar encodes, using two hypothetical interactions between learner's assumptions about hidden structure and developing constraint rankings that can trick a learner into adopting a superset grammar. These two arguments are used to support an approach called Error-Selective Learning (ESL), in which errors are learned and stored gradually, in a way that relies on violation frequency, but rankings themselves are learned in a non-gradual way (relying on the algorithms of Prince and Tesar 2004;Hayes 2004). It is also shown that violation frequencies can still cause problems regardless of a learner's method of grammatical evaluation -either ranked constraints as in Optimality Theory, or weighted constraints as in Harmonic Grammar.

Learnability Shapes Typology: The Case of the Midpoint Pathology

“…This can be compared to a batch or elementary ranking condition (ERC) learner, which can make decisions about how to adjust its grammar based on generalizations extrapolated from the entirety of data presented to it (e.g. Hayes 2004, Prince & Tesar 2004, Tessier 2009, Brasoveanu & Prince 2011.…”

Trajectories of faithfulness in child-specific phonology

“…Second, its individual stages must allow for the production of multiple outputs for the same input, mimicking the curves graphed in w2 -recalling our broad goal of producing this variation not in the grammar but in the learning mechanism, or rather in the way the learner uses the grammar. Our proposal to meet both of these goals is a version of Error-selective learning (ESL; Tessier 2007Tessier , 2009) which uses an intermediate buffer called the CACHE. The Cache creates incremental learning by restricting the speed and scope of the errors that are allowed to trigger the re-ranking algorithm.…”

“…We are indebted to Peter Graff and Corey Telfer for their advice on statistical methods. optimality-theoretic learning models in Hayes (2004), Prince & Tesar (2004) and Tessier (2007Tessier ( , 2009). Both S-shaped and U-shaped learning will be shown to emerge from the interaction of an OT grammar, a constraint-demotion algorithm that favours ranking markedness constraints above faithfulness, a stored Cache of previously produced forms and a 'dual route ' mechanism for producing variable outputs over the course of acquisition.…”

“…13 This description of ESL differs from the description given inTessier (2009) in at least two key ways: there, the Cache only stored errors, not all forms produced, and it was emptied after each cycle of re-ranking, while the Cache here changes more subtly.…”

Input frequency and the acquisition of syllable structure in Polish

“…Similarly, Stoel-Gammon (1998) found that frequency predicts the overall shapes of children's early words but that certain patterns, such as word-initial /b/, were overrepresented in the children's early words as compared to the input. A final group of studies advocates explanations crucially incorporating both syllable structure universals and frequency (Boersma & Levelt 2000;Edwards & Beckman 2008; A c c e p t e d M a n u s c r i p t 11 2010Levelt, Schiller & Levelt 2000;Roark & Demuth 2000;Tessier 2009). For example, Edwards and Beckman (2008) show that the inherent phonetic difficulty of certain sound classes, such as affricates, can be modulated by language-particular frequency.…”