LINGUIST List 20.2683
Tue Aug 04 2009
Diss: Historical Ling/Phonology/Lang Acq: Mondon: 'The Nature of...'
Editor for this issue: Di Wdzenczny
<dilinguistlist.org>
1. Jean-François
Mondon,
The Nature of Homophony and its Effects on Diachrony and Synchrony
Message 1: The Nature of Homophony and its Effects on Diachrony and Synchrony
Date: 04-Aug-2009
From: Jean-François Mondon <jfmondongmail.com>
Subject: The Nature of Homophony and its Effects on Diachrony and Synchrony
E-mail this message to a friend
Institution: University of Pennsylvania
Program: Department of Linguistics
Dissertation Status: Completed
Degree Date: 2009
Author: Jean-François R. Mondon
Dissertation Title: The Nature of Homophony and its Effects on Diachrony and Synchrony
Linguistic Field(s):
Historical Linguistics
Language Acquisition
Phonology
Dissertation Director:
Don Ringe
Charles Yang
Rolf Noyer
Dissertation Abstract:
An appeal to avoidance of homophony has been made by many linguists in boththe diachronic and synchronic arenas. In diachronic work it has beensuggested that a sound change can be prevented from occurring in a givenmorphological environment if its occurrence would result in homophony withanother form; this is an example of the larger concept of 'prophylaxis.'In synchronic work homophony has mainly been invoked in analyses couched inan Optimality Theory (OT) framework. Viewed as a synchronic constraintpresent in a grammar, anti-homophony effects have played a large role insuch analyses where expected surface forms are prevented from occurring ifthey would result in intra-paradigmatic 3/4 and in some casestransparadigmatic 3/4 homophony with another form. Once one attempts toseriously model language change in an OT framework it becomes possible tounite prophylaxis with synchronic anti-homophony effects by assuming thatregular sound changes start off as variable rules which are due to two ormore constraints which are not ranked with respect to each other. If,however, an anti-homophony constraint is consistently ranked above thenon-strictly ranked constraints then the variable rules should fail toapply in those environments (be they morphological or lexical) wherehomophony would result.
Despite the apparent simplicity of this approach several criticisms can beraised which highlight the problems which accompany such a theory. Thefirst half of the dissertation is devoted to proposing that no notion ofhomophony-avoidance can play an active role in any component of thegrammar, whether phonological, morphological, syntactic, or what have you. It reassesses the cases cited in the literature and concludes that mostare not strong examples of homophony avoidance in language change. Forthose that are, however, a model of language change which does not violatethe Neogrammarian Hypothesis of the regularity of sound change is advanced.
Homophony has not only been discussed in the diachronic and theoreticallinguistic realm but it has also occasionally appeared - though ever sobriefly - in child language acquisition literature. The exact impact onfirst language acquisition which homophony in the data imposes upon a childseems to be ambiguous. Slobin theorized that 'if there are homonymousforms in an inflectional system, those forms will tend not to be theearliest inflections acquired by the child' (1973: 203). In fact, thehomonymous forms should tend to be replaced by those endings which havefewer functions (i.e. 'inflectional imperialism'). As Smoczyńska (1985:674) showed for Polish, however, 'the ending -i, which serves a number offunctions in the noun-declension system, is among the earliest endings usedby children.'
It seems as if homophonous endings are sometimes avoided while in othersituations they can even be overgeneralized or at least produced just asearly as non-homophonous ones. What explains this apparentstrengthening/retardation effect of homophony? The second part of thisdissertation develops a morphological learning algorithm which makes clearpredictions as to when homophony will be treated systematically as opposedto accidentally by the linguistic system. The algorithm is tested againstlanguage errors from Polish and Lithuanian.
|