Papers in Historical Phonology · Phonotactics plays a rather minor role in historical phonology. Most diachronic work deals with segmental change, or with (higher-level) prosodic

Papers in Historical Phonology

http://journals.ed.ac.uk/pihph ISSN 2399-6714

Volume 4, 83–135 (2019) DOI: 10.2218/pihph.4.2019.4192

LicensedunderaCreative

Commons4.0International

License

Phonotactics,prophylaxis,acquisitionismandchange:*RIME-xxŋandash-tensinginthehistoryofEnglish

PATRICKHONEYBONEUniversityofEdinburgh

Abstract

Thisarticle revisits,extendsand interrogates thepositionadvocated inHoneybone (2019)— that phonotactic constraints are psychologicallyreal phonological entities (namely: constraints on output-like forms),whichhaveadiachronyoftheirown,andwhichcanalsointerferewithdiachronic segmental change by inhibiting otherwise regularinnovations. I focus in the latter part of the article on the role of onephonotacticconstraint in thehistoryofEnglish: *RIME-xxŋ. Iargue thatweneedtoinvestigatetheemergenceofsuchconstraintsinthehistoryof languagesand I showhowthisparticularconstraint,once innovated(whichoccursthroughconstraintscattering),canbeunderstoodtohaveinhibited thepatterning ofash-tensing in certain varieties ofAmericanEnglish(andalsothat itmaynowhavebeen lost insomevarieties).Todo this, I adopt a phonologicalmodelwhich combines aspects of rule-basedphonologyandaspectsofconstraint-basedphonology,andwhichisfirmlyrootedinthevariationthatexistswhenchangesareinnovated.Finally,Ievaluatetheextenttowhichthetypeofphonotactically-drivenprocess-inhibition that I propose here involves prophylaxis inphonological change (I show that it doesn’t), and I consider theinteraction of these ideas with the proposal that all change occurs inlanguageacquisition(‘acquisitionism’).

1 IntroductionPhonotactics plays a rather minor role in historical phonology. Mostdiachronic work deals with segmental change, or with (higher-level)prosodic phenomena, such as stress or tone. This article, in contrast,aimsemphaticallytobeacontributiontodiachronicphonotactics;thatis, to the part of historical phonology which deals (i) with theinnovationofnewphonotactics,(ii)withthelossorchangeofexistingphonotactics, and (iii) with other ways in which phonotactics mightinteractwithphonologicalchange.Therehasbeensomeseriousworkindiachronic phonotactics, such as Lutz (1988, 1991), Dziubalska-

CORE Metadata, citation and similar papers at core.ac.uk

Provided by Papers in Historical Phonology

https://core.ac.uk/display/275887328?utm_source=pdf&utm_medium=banner&utm_campaign=pdf-decoration-v1

PatrickHoneybone 84

Kołaczyk (2005) and the work of Ritt and others at Vienna (e.g.Baumann et al. 2016) which also led to a Special Issue of FoliaLinguistica Historica (number 50:2) on diachronic phonotactics.Additionally, general volumes on historical phonology and/orphonological history do sometimes contain some consideration ofrelevantmatters (as in Hogg 1992 andMinkova 2014). It is not thatthere is no work in diachronic phonotactics, but it is surprising howlittle phonotactic issues are considered when dealing with cases ofphonologicalchange.

Fordiachronicphonotactics tobeworthwhile,weneed tobesurethat phonotactic entities are real phonological phenomena, and weneedawayofintegratingthemintophonologicalderivationsandintoamodel which allows for their interaction with other aspects ofphonologyinphonologicalchange.Idiscussthatinsection2.Section3thenfocusesontheroleofoneparticularphonotacticconstraintinthehistory of English: *RIME-xxŋ. I consider how it entered the language,how it has interacted with the innovation of other phonologicalphenomena,andalsohowitseemstobebeinglostincertainvarietiesofthelanguage.Section4considerssomefundamental implicationsofthe positions on diachronic phonotactics that are proposed earlier inthearticle,broadeningthescopeof thepiecetoaddressmoregeneralquestions in historical phonology, such as whether change involvesprophylaxisorrepair.Section5concludes.

This article is a companion piece to Honeybone (2019),1 where Iconsider some fundamental issues in phonotactics and in phonotacticchange.Idonotrepeatheremostofthepointsmadethere,althoughIsummarise(insection2)themodelthatIdevelopedthere,inordertoshow how it aims to understand the ways in which phonotacticconstraints can interactwith other aspects of phonological diachrony(specifically, to account for the way in which already-existingphonotactics can affect changes which are otherwise general in theirphonological environment,butwhichhavebeen inhibited inoneveryspecificphonologicalenvironment).IextendheretheideasconsideredinHoneybone(2019),Iconsidersomeoftheimplicationsofthem,andIshowhowtheyalsoapplytothehistoryofthephonotactic*RIME-xxŋ.

2 PhonotacticsinphonologyandinphonologicalchangeTostartatthestart,itisobviousthattherearesequencesofsegmentswhichdonotoccurinthephonologicallexiconofspecificlanguages.Forexample, [baɪŋ] is not the surface representation of any word in the 1Thismeansthat,inmymind,thatarticleisHoneybone(2019a)andthecurrentpieceisHoneybone(2019b).

85 Phonotactics,prophylaxis,acquisitionismandchange

variety of English that I speak2— it is a ‘gap’ in the lexicon. It is notobvious,however,thatspeakersrealisethattherearesuchgapsinthelexiconand that they conceiveof themas a genuinepropertyof theirlanguage. In other words, a key question for phonology is: are suchsequences absent from languages because they are forbidden by thephonological grammar? If so, then this must be due to some type ofphonological entity which forms part of a speaker’s phonologicalknowledge.Andifso,awholehostoffurtherquestionsarise:whatkindof thing are the entities involved (rules, constraints, statisticalgeneralisations...)? over what kind of structures do they generalise(words, morphemes, syllables, feet...)? are they all the same kind ofthing?aretheythesameordifferentfromtheentitiesthatareinvolvedin other aspects of phonology (alternations, non-contrastivedistributions, prosodic structure...)? are they categorical or gradient?are theyallalwaysequally important in thephonologyofa language?whatkindofconsciousaccessdonativespeakerstohavetothem?andwhatisthepreciseinventoryoftheentitiesinvolvedforanyindividuallanguage?

In Honeybone (2019) I set out a model which answers many ofthesequestionsbyadoptingtheposition(proposedanddefendedinaconsiderableamountofphonologicalwork,butnotacceptedbyall)thatphonotactics are psychologically real constraints on surface forms. InthispaperImodifythatpositionslightly(asnotallconstraintsworkonstrictlysurface forms),but thespiritof thepositiondevelopedhere isverymuch the same as that set out in the earlier article (in fact, thepositionthatIendupadoptingherewasalreadysuggestedthere).Onecrucialpointofthemodelisthatitrecognisesthetraditionalcategoricaldistinctionbetweendifferentkindsofgapsinlanguages’lexicons.Somegaps are assumed to exist in languages because the grammar forbidsthem,whileotherscouldeasilybefilled.InHoneybone(2019),Icalltheformer ‘S-gaps’, recognising that they have been called a number ofthingsintheliterature:‘systemic’,‘systematic’or‘structuralgaps’.Icallthelatter‘A-gaps’,astheyareoftendescribedas‘accidentalgaps’.

Someworkon the typeof topicsdiscussedheredissents fromtheidea that the A-gap/S-gap distinction is a valid one (Algeo 1978 andBauer2015arecautious, forexample). Insection2.1, Iconsidersomeobservationsthatmakemostsenseifweassumethatitisindeedarealdistinction, and that S-gaps really are enforced by the grammar, apositionwhichIassumeintherestofthisarticle.Insection2.2,Isetout

2It,orsomethingverysimilar,mayoccurincertainothervarietiesofEnglish,whichdiffercruciallyinthisonephonotactic.Thatwillbepartofthepointofsection3.4.ThevarietythatIspeakisageneralisednorthernEnglishdialectofBritishEnglish.

PatrickHoneybone 86

how I assume phonotactics are implemented in phonology, and insection2.3,Iconsidertheinteractionofphonotacticsanddiachrony.

2.1 DoS-gapsreallyexist?IfS-gapsareenforcedbythegrammarandA-gapsarenot,wepredictthat theyshouldbehavedifferently.Forexample,A-gapsshouldeasilybe fillable (by the adoption of loanwords, for instance), but S-gapsshouldnot(andtheyshouldforceloanwordstobeadapted).Loanwordadoption and adaptation are in fact more complex than this simpledichotomysupposes—forexample,it ispossibleforEnglishspeakerstoproduceformslikecarteblanchewitha[ʁ]andanasalvowelwhichisnotfollowedbyanasalcoda,whichisnotexpectedtobepossibleinmostvarietiesofEnglish,butthisinvolvesakindofcode-switching,asanattempttoconsciouslypreservetheoriginalFrenchphonologyoftheform, importingaspectsofFrench intoEnglishutterances(perhapsaspart of a phonological ‘periphery’ which is not allowed to affect thephonological ‘core’); also it is surely the case thatwhenspeakersof alanguage are overwhelmed by large numbers of loanwords from aprestigious superstrate language (such that there is pressure topreserve thephonologyof theborrowed forms) that this could breaktheir currentphonotactic grammar (penetrating into thephonologicalcoreofalanguage)andleadingtophonotacticchange.

Issues surrounding loanword adaptation deserve seriousconsiderationintheirownright,toadegreethatcannotbegivenhere(see,forexample,Uffmann2015),butasIdiscussinsomemoredetailin Honeybone 2019 (and also, a little, below), the behaviour ofloanwordscanindeedbeimportantindeterminingthestatusofatypeof gap in a language: is it forbidden by phonotactics or not? Somephonological work has argued carefully that certain quite noticeablegapscannonethelessbeshowntobeA-gaps,ratherthanS-gaps,andtobehave as A-gapswould be expected to behave in terms of loanwordadoption (see, for example, Iverson & Salmons 2005 on the English‘tensevowelplusfinal[ʃ]’gap).Thiscanonlybedoneifthedistinctionisarealone,suchthatitallowsustoexpectthatthedifferenttypesofgapwillexhibitdifferenttypesofphonologicalbehaviour.

A further prediction of an A-gap/S-gap distinction which is oftenassumedisthatwemightexpectthatspeakerswillhaveintuitionsthatforms which violate S-gaps are phonologically ‘wrong’. A substantialstrand of research has considered the extent to which speakers’judgements about ‘wordlikeness’ or ‘wellformedness’ of segmentalsequencesreflectthegapsthatcanbefoundinthephonologicallexicon.Theresultsofthisresearchare,however,complexandconflicting,and


it is widely recognised that understanding such judgements requiresmoreofusthansimplydistinguishingbetweenS-gapsandA-gaps.Thisneed not vitiate the distinction between the two, however.Wordlikeness judgements are (like all reflections on linguisticintuitions)notgrammaticalityjudgements(becausegrammaticalityisaproperty of a linguistic system which is not directly accessibly tospeakers), but are rather acceptability judgementswhichare likely tobe influenced (as well as by some reflection of our knowledge ofphonological structure), by our knowledge of other things that wemightbeawareof insomesense,suchas lexical frequencyandlexicalneighbourhooddensity.Thestudyofacceptabilityjudgementsinsyntaxhasdeveloped toa sophisticated level in interpretinghow informantsreacttojudginglinguisticdataandhasshownthattheinterpretationofsuch results canbe complex (see, for example, Schütze to appear andSprousetoappear),soweshouldnotexpectphonologicalwordlikenessjudgements to simply reflect a categorical distinction between S-gapsandA-gaps.

Some of those who have worked on related issues argue thatintuitionsaboutwhichgapsareA-likeandwhichareS-likearesimplydue to statistical generalisations by speakers over the items in theirlexicon. If this is true, there is no role for categorical phonologicalstatements.3 However, much evidence indicates that “a simplemodelbased purely on statistical properties of the linguistic data isinadequate, justasonebasedpurelyonphoneticbiaseswouldbe”,asAlbright (inpreparation)argues.4Gorman(2013,88)adds that “therearemanystaticphonotacticconstraintswhicharestatisticallyreliable 3Therearethusclaimsintheliteraturethatjudgementsofphonotacticwell-formednessare entirely gradient, whichwould bring the categorical S-gap/A-gap distinction intoquestion.However, asGorman2013 shows, there are reasons to be sceptical of suchresults,especiallyoftheconclusionthattheymightbeevidenceagainsttheexistenceofmultiple specific synchronically-active phonological generalisations. For example, asGorman (2013, 16) emphasises, Chomsky & Halle (1968, 416-417) already pointtowards amodelwhich aims to “define the ‘degree of admissibility’ of each potentiallexicalmatrix”beyonda simple two-way (grammatical vsungrammatical)distinction,ontheassumptionthat,while[bnɪk]isanimpossiblesurfacerepresentationinEnglish,[bznk] istakentobe“evenlessEnglish-likethan[bnɪk].” Incomparingtwosequenceslikethis,whicharebothabsentfromEnglish,thepointiseasytoacceptbecause[bznk]violatesmultiplephonologicalrequirementsofEnglish(onsetsequencing,presenceofavowel in a stressed syllable),whereas [bnɪk] only violates one (onset sequencing). Inaddition to this accumulation of violations by whole strings of segments, it is notimpossiblethatsomephonologically-enforcedgapsaremoreseriousthanothers.4 Albright is arguing in favour of there being a role for universal phonotacticknowledge (which does not derive from the statistical tendencies present in thephonological lexicon of a language) in understanding speakers’ judgements aboutboth attested and unattested phonological forms in a language. A similar line ofargumentationhasbeensetoutpersuasivelybyBerent(2013,andelsewhere).

PatrickHoneybone 88

but synchronically inert”— that is, the statistical generalisations thatspeakersshouldmakeovertheirlexicon(becausetheycanbeshowntoexistifweanalysethelexiconofalanguagecomputationally)donotinfact replicate the kinds of differences that speakers actually make interms of the kinds of phonological behaviour predicted above for A-gapsandS-gaps,sowestillrequireadistinctionbetweenthetwo.

It strikesme that understanding phonology (like other aspects oflanguage)involvesadelicatejourneytoworkouthowseveralkindsofthing influence both what comes out of our mouths and whatjudgementswereportwhenquestionedonintuitions—manyofthesethings are grammar-external, and may involve usage-basedgeneralisation,butsomehaveallthehallmarksofgrammar-internality.Shatzman&Kager (2007) andLentz&Kager (2015) are examples ofcareful work probing wellformedness judgements which shows thatboth categorical phonotactics andprobabilistic knowledge play a roleandthatthetwoareofdifferentnatures. If this isright(andIassumethat it is),weneedthenotionoftheS-gap,andofpsychologicallyrealphonotactics toenforce them. In thisarticle, I adopta robustpositionthatphonotacticconstraintsare indeedphonologicallyreal,andgoonto investigate the possibilities that this gives us for understandingphonologicalchange.

It isworthnothing thatotherworkwhichassumes fundamentallycompatible ideas to those adopted here has not always talked of‘phonotactics’—theideasexistedbeforethewordwascoined,anditisnotalwaysusedtodiscussthemnow.Bloomfield(1933), forexample,includes somedetailed considerationof relevantmatters in a chapteron“phoneticstructure”(sothisisacasewhere‘phonetic’reallymeans‘phonetic-and-phonological’), and this kind of terminology was alsoused elsewhere in earlier writings, as in Kruisinga’s (1943) detailedvolumeonthesubject,whichiscalledThePhoneticStructureofEnglishWords.Hockett(1955,92)talksabout“thedistributionalclassificationof consonants” and describes “the phonologic system: a stock ofphonemes (or phonologic units) and the arrangements inwhich theyoccur relative to each other” (1955, 14). The extent to which suchearlierwork is describingwhat occurs (in terms of gaps) rather thanwhat can possibly occur (that is, assuming that they are S-gaps) issometimesmootbuttheideathatphonotacticsforbidsthingsisbynomeansexclusive togenerativematerial (where it is,however, robust).For example, Abercrombie (1967), writing the ‘British phonetictradition’, talks about how “[s]uch structural regularities in thephonologyofalanguageproduceinitsspeakers,deep-rootedhabitsofspeech which are difficult to change. This is shown by the way newwords introduced into a language — slang, trade names, borrowing


fromforeignlanguages—conformtotheexistingstructuralpatterns.”This is not far from saying that, if the structural regularities of alanguageavoidaspecificsequence,these‘habits’forbidsuchsequencesinloanwords.

TheOxfordEnglishDictionary (OED) has attestations of the actualwordphonotactics from1956onwards, and refers toHill (1958)whoattributestheinventionofthetermtoRobertStockwell’suseofitinalecturein1954.5Thetermphonotacticswasnotusedinearlycoreworkonphonologyfromthegenerativeparadigm(suchasHalle1959,1962,andChomsky&Halle1965),butthisdoesnotmeanthatrelevantideaswerenot important in it.Forexample,Chomsky&Halle (1965)madefamousthedistinctionbetween[blɪk]asanon-occurringbutadmissibleform and [bnɪk] as a non-occurring but “inadmissible” form,6 andHarms (1968, 85), in a textbook introducing standard generativephonology written right at the point of initial pre-eminence of themodel, talks of “sequential constraint rules”which “describemany ofthe samephenomena traditionally treatedbyphonotactic statements”(thusacknowledgingthenotionsinvolved,butdistancinghimselffromtheterm).

The standard terminology of early generative work talks of‘Morpheme StructureRules’. These include ‘sequence structure rules’,which enforce what I have been calling S-gaps. Stanley (1967), aninfluentialarticleinthissphere,explainsthat“sequencestructurerules,thoughincludedinthegrammartocharacterizeredundancy,provideasaby-productacharacterizationofthenotion‘possiblemorpheme’;thisobviates theneed foraseparatesetof statements tocharacterize thisnotion”. Stanley (1967)was influential in themovewithin generativephonology away from the use of rules to describe phonotacticgeneralisations towards the use of constraints (which he called‘Morpheme Structure Conditions’, which include ‘sequence structureconditions’). This is an important development in terms of themodelthatIadoptinsection2.2.Stanleywritesthatwecan“interpretthesetofsequencestructurerulesasastatementofconstraintsonsystematicphonemic sequences”. He explicitly advocated the “need for negativeconditions” — that is, constraints which rule out sequences. Stanleyalso argues for other types of constraints, but it is these ‘negative’ 5Stockwellhasconfirmedhishand in thecreationof the term(Lutz1988,231).HewasthinkinginacontextinwhichHockett(e.g.,1947)hadbeenusingtacticstotalkof“the study of the relation and arrangement of linguistic units, esp. the study of thearrangementofmorphemes”(OED).AlthoughHocketthimselfabandonedthetermin1955(16-17),thisusagewaslikelyinfluentialonStockwell’screation.6 At this point Halle and Chomsky assumed a simple two-way distinction betweenwhat I am calling A-gaps and S-gaps. Chapter 9 of Chomsky & Halle (1968)complicatedthings(asitoftendoes)asdiscussedinfootnote3.

PatrickHoneybone 90

constraintswhich forbid the sequenceswhich form S-gaps, that havetakenrootasawayof“separatingtheaccidentalsequentialconstraintsfromthesystematicones”(1967,401)inphonologicaltheory.7

The absence of the term phonotactics (but not the ideas that Iassociate with it here) is also notable in later generative work. Forexample, Hammond’s (1999) volume (of which about half deals withwhatisherecalledphonotactics,fromanOTperspective)doesnotusethe term — rather, the volume is described as focusing on “surfacedistributional regularities” and on “the allowable configurations ofconsonants[and]vowels”.Thereareseveral traditionsofusagewhichhavekeptthetermalivesincethe1950s,however,asinScholes(1966),Sommerstein (1974) and Singh (1987), some of which are broadlygenerative and someofwhich arenot. More recently there are signsthat ithasbecomeausual term ingenerativework, too (e.g.,Albright2006, Hayes &Wilson 2008, Gorman 2013).Whateverwe call it, thetopic in focus here (whatwe could call ‘phonotactic knowledge’) is aclearfocusforphonology.

2.2 ThestructureandstatusofphonotacticsGiventhepositiononS-gapsjustdiscussed,Iassumethatphonotacticsis8 fundamentally about understanding where the segments of alanguagecanoccur(andthus, inpart,howtheycombineinsequences—especiallyinlanguageswhichallowconsonantclusters,likeEnglish).Oncethebasicphonologicalstructuresofalanguageanditssegmentalinventoryareunderstood,thephonotacticquestionis:doallsegmentsoccur in all environments where a language allows segments inprinciple?Iassumehere(asinHoneybone2019,whichisthesourceofall thediscussion in this section) that the relevant ‘basicphonologicalstructures’ofalanguageareprovidedbytheslotsinsyllablestructurethatthelanguageallows,andIassumeaclassicbasicsyllablestructureofthetypegivenin(1),where‘O’,R’,‘N’and‘C’standfor‘onset’,‘rime’,‘nucleus’and‘coda’.9

7Someworkinsistsonusingpositiveconstraints(oneoftheothertypesthatStanleyconsiders)—forexampleTaylor’s (2002,250-251)usage-basedapproach, inwhichthe “grammar of a language comprises only ‘positive statements’ about what doesoccur,thereisnoneedfor‘negativestatements’ofwhatdoesnotoccur”—butthisisaminorityposition,notevenheldbyallusage-basedlinguists.8 A specific grammatically-enforced gap in a language is sometimes described as ‘aphonotactic’ (shorthand for ‘aphonotacticgeneralisation’), sowecanaskbothwhatphonotacticsisandwhatthephonotacticsareinanylanguage.9 I follow the convention of using the spelling ‘rime’ for the syllabic constituent, todifferentiateitfromthepoeticnotionof‘rhyme’,whichisnotthesamething.


(1) σ O R N C

This assumption that syllable structure is relevant in describingphonotactics isnotuniversallyagreed—it ispossible to focuson thesimpleoccurrenceofstringsofsegments(asforexampleSteriade1999hasadvocated),orondescribingthesegmentalstructureofwordsorofmorphemes (as is implied in the term ‘Morpheme Structure Rule’,whichcroppedupabove).However,theslotsprovidedbybasicsyllablestructure have shown themselves to be so useful in describingphonotactic generalisations insightfully that most phonotactic workuses themwithout comment, or ends up reinventing them by talkingabout‘word/morphemeinitials’insteadof‘onsets’or‘word/morphemefinals’insteadofcodasorrimes.

As mentioned above, I see phonotactics as being implementedthrough ‘static’ constraints on phonological forms. The position that Ifollowhere(developedinmoredetailinHoneybone2019)isthattheseconstraintsapplytosurface-likeforms,atthe‘end’ofthephonologicalderivation, with earlier aspects of phonology (those dealing with‘dynamic’phonologicalprocesses)modelledas rules.Thismixedrule-and-constraint model follows work in Optimality Theory in placingimportance on surface-oriented constraints, but it also follows rule-based diachronic work in allowing the straightforward depiction of(new)processesastheadditionofarule.10Themodelisnotnew—itishighlyreminiscentofclassicrule-basedworkwhichtakesphonotacticsseriously, such as Sommerstein (1974), a piece which proposesessentiallythemodelofthegrammarthatIfollowhere(andwhichwasdescribed by Goldsmith (1993, 9) as “very prescient”). The interest

10 Or as other classically-recognised types of change to the rule component of thegrammar(followingsuchworkasKiparsky1968andKing1969),suchasrulelossorrulereordering(andrecognisingthat‘rulereordering’mightreallybethe‘rising’ofarule throughthegrammar,as in the lifecycleofphonologicalprocesses).Myoverallapproach echoes Salmons (to appear) in the rejection of OT as “ill-suited toapplicationtosoundchange”initsmostcommonform,whichItaketobetheadditionofarule(but italsorecognisesthatthefocusinOTonconstraints is important,andindeed argues that we need to consider the emergence and loss of the effect ofconstraints in diachrony, as in historical work on OT). Salmons writes further that“synchronic phonology may well look monostratal but any reasonable historicalrecordof complex soundchanges showsobviouspatternsof layeringof change, theaccumulationof stepwisechanges”—this requiresaderivationalmodelof the typethatIadopt.ThesamepositionisadoptedinRinge&Eka(2013).

PatrickHoneybone 92

fromadiachronicperspectiveisthatthiskindofmodelmeansweneedto consider both the innovation of new rules (and change in the rulecomponent) and the innovation of new phonotactic constraints (andchange in the phonotactic component). This may seem a complexmodel,butitstrikesmeasexactlywhatweneedinordertoaccountforeverything that we might want to account for in phonology, and inphonologicalchange.IgiveanumberofderivationsalongtheselinesinHoneybone(2019),andIworkthoughoneexampleinsection2.3.

As background to that, we can recognise that one example of apsychologically real (language-specific) phonotactic constraint inPresent-DayEnglishisthefactthatsequenceslike[pn,bn,fn,θn,tn,dn,kn,ɡn]areabsentinonsets.Aphonologically-informedobserver(and,wemightexpectalso,alearner)caneasilyseethatthebasicstructuralfactsofEnglishallowtwo-membercomplexonsetswith[p,b,f,θ,t,d,k,ɡ]asthefirstmemberandacoronalsonorantasthesecond(asinbrief,through, climb, gloom etc.), and that nasals can occur in onsets inprinciple (as in all nasal-initial words); it is also the case that nouniversal constraint (suchas theSonoritySequencingPrinciple) rulesoutcombinationsinvolvingobstruentsandnasals,sothismustbeafactaboutthephonologyofEnglish.ThisisthekindofS-gapthatneedstobe enforced by a phonotactic, and I represent such phonotactics asexemplified in (2). In this case, ‘T’ stands for ‘all and only the non-sibilantobstruents’,11and‘N’standsfor‘allandonlythenasals’.

(2) *ONSET-TN

asequenceofanon-sibilantobstruentfollowedbyanasalisforbiddeninanonset

Evidencefortherealityof thisconstraintcomesfromthefactthat

when words are borrowed into English from other languages whichallowsuchclusters, theyareadapted,eitherbydeletingtheobstruentorbyepenthesisingaschwabetweenthetwoconsonants.Forexample,the OED includes knackwurst,with a first attestation of the word inEnglish from 1939). This word is borrowed from German where thefirst syllable is [knak], but theOED’s transcriptionof the first syllable 11 In and of itself, this constraint does not rule out onset sequences involving theEnglish affricates, suchas [tʃn], but I assume that they are accounted forby amoregeneral constrainton thecombinabilityofaffricates,given that theydonotoccur inany complex onsets. The behaviour of [s] in onset phonotactics needs some sort ofspecial treatment,as iswellknown,given the initial strings inwords likestring andskill,which violate sonority-sequencing if they are viewed as onsets. I do not adoptanyparticularanalysisofthishere(manyhavebeenproposed),butItakeforgrantedthat some sibilant-specific phonology accounts for this, and that ‘non-sibilantobstruent’isanaturalclass.


for the word in English is [nak]. The case of gnu is instructive: thepreciseetymologyofthisborrowingintoEnglishiscomplex—theOEDgives‘Khoekhoe’and‘San’anditsorigin,butotherauthoritiesconsiderother languages tohavealsobeen involved (aspartof a transmissionchain), includingDutch. It isclear fromthespelling,however, that theword in thedonor languages isassumedtohavehadacomplexonsetinvolving[n],anditisalsoclearthatthewordisnotpronouncedwithacomplex onset involving [n] in English. The ‘classical’ EnglishpronunciationaccordingtotheOED is[njuː]or[nuː],buttheOEDalsorecognisesthe‘jocular’(butinfactcommonUK)pronunciation[ɡənuː],whichoriginates froma comic song (Flanders&Swann1957),wherethehumour inpartderives fromthe fact that the initial ‹g› isactuallypronouncedas[ɡ]),andit isnotableforourpurposesthatthe[ɡ]canonlybepronouncedwithafollowing[n]ifavowelintervenesbetweenthetwo(and,ofcourse,thevowelthatisusedforthisisschwa,whichistheminimalEnglishvowel).

Other phonotactics considered in Honeybone (2019), which willlikelybefamiliartothosewhoknowthephonologyofEnglish,include*CODA-h‘[h]cannotoccurincodas’andOCP-ONSET(CORONAL)‘asequenceoftwoanteriorcoronalsegments(suchas[tl])isforbiddeninanonset’(again,ignoring[s]).Therearemanymore.Iassumethatgapsneedtobe plausible and systematic in order to be S-gaps and henceimplemented in thephonologicalgrammarasphonotactic constraints,and that they should not replicate the basic structural facts of alanguage (whichmay be language universal, as inwhat is possible inbasic syllable structure, as given in (1), or may be accounted for bysome form of parameterised principle, such as whether a languageallowscomplexonsetsornot).Theprecisedefinitionof thesenotionsarenotnecessarilysimple,but I thinktheymakesense. It isplausible,forexample,thatEnglishwouldallow*ONSET-TN-violatingclusterssuchas[pn]and[dn]becauseotherlanguagesallowthemandEnglishallowssimilarthings,butnonethelessEnglishsystematicallyexcludesthem,onthebasisofthenaturalclassesof(non-sibilant)obstruentsandnasals.12Itmight seem to be cheating to use capital letters to refer to a set ofsegmentsintherepresentationofphonotacticslikein(2),asIcouldinprincipledefinethesetinvolvedinanywaythatIchoose,butthisisnottheintention—thecapitallettersareusedtorepresentnaturalclasses.I assume therefore that learners recognise phonological gaps (whichinvolve plausible systematic gaps) in their lexicon as S-gaps, and thattheyassumethattheyareduetophonotacticconstraints.

12 See also Honeybone (2016) for a further consideration of ‘plausibility’ inphonologicalgeneralisations.

PatrickHoneybone 94

I discuss somemore of what it takes for a gap to be an S-gap insection 3.1. I turn now to consider some aspects of the diachrony ofphonotactics. If we take such S-gap-enforcing constraints to be asphonologicallyrealassegmentsandsyllablesandrules,thenweneedaserious branch of diachronic phonotactics which investigates theirinnovation and loss, as well as change in phonotactics (as I go on toshow in the next section), and also otherways inwhichphonotacticsmightinterveneindiachrony(whichisthefocusofsection3and4).

2.3 PhonotacticsindiachronyPhonotactic entities of the type argued for in section 2.2 have adiachrony, just likeotherphonological entities (suchas segmentsandfoot structure). Part of thepoint of this article is thatwe should takethis seriously— interesting results canemerge ifwedoso. It is clearthat phonotactics can be innovated into (or ‘become active’ in) alanguage, and that they can be lost over time. It is also the case,however, that phonotactics already in a language can be observed tochange.Forexample,13theconstraintaspreciselyformulatedin(2)hasnotalwaysbeenpartofthephonologyofEnglish.Thefulldiachronyofthe case involves more interesting developments than simply theadditionofaphonotactic,however.

Firstly, we need to separate out stop-nasal and fricative-nasalclusters.While there is reason tobelieve that the currentphonotacticaffects thewholeclassof(non-sibilant)obstruents,asassumedin(2),the history of fricative-nasal onset clusters requires some separatethought.Initial[sn]and[sm]havebeenpossibleinEnglishforaslongaswecanknow,sothishasnotchanged:forexample,wordslikesnowand smear descend through transmission from forms which arerobustly reconstructed with initial [sn] and [sm] in Proto-Indo-European. In early stages of English, however, other fricative-nasalinitialswerepossible,too:theonsetsequence[fn]isfirmlyattestedinOldandMiddleEnglish.Whileithasonlyeveroccurredinafewwords—Bosworth&Toller’s (1898-1921)dictionaryofOldEnglishandtheMiddle EnglishDictionary (MED) both have 11 headwordswith initial‹fn›—thereisnodoubtthatitwasphonotacticallypossible:thesetwosources include, for example, words like fnesan ‘to sneeze’ and fnæd‘border, fringe’. The full history of the phonotactics that are relevanthere deserves serious attention, and is not fully understood (perhapsbecausesofew[fn]-initialwordsareinvolved).Weknowthat[fn]waspossibleall thewaybacktoProto-Germanic(forwhichKroonen2009 13ThisexamplealsocomesfromHoneybone(2019),butIextendthediscussionofithere.


reconstructsfneusan~fnūsan‘tosneeze’forexample),andweknowthat[fn]isnolongerpossibleinEnglish,andthatthelossofthismayhaveled to the emergence of the full formof the constraint in (2), but theprecisemannerinwhich[fn]waslostdeservesseriousinvestigationofa sort thatwould takeusoff trackhere. I think that this is somethingthatweshouldconsiderinhistoricalphonology,andIthinkitwilllikelybeinsightfulifweseethediachronyof[fn]inthisphonotacticcontext,but because it is still somewhat unclear, I set fricative-nasal onsetclusters aside in this discussion and focus on stop-nasal onsets. Itherefore redefine ‘T’ in this section tomean ‘all and only the stops’(rather than ‘all andonly thenon-sibilantobstruents’,which iswas itmeant in section 2.2). If we do this, we have access to an intriguingunderstanding of the diachrony involved in the loss of [kn] and [ɡn]onsetsinEnglish:theconstraintforbiddingcomplexonsetswithnasalswassimplerafterthechangethanbeforeit.

We know that Proto-Germanic, Old English and Middle English(PGmc,OEandME)didnothaveonsetsequencesofthetype[pn,bn,tn,dn] (which are those thatwewould expect ifany complexonsets arepossiblewith nasals as the secondmember, given that [n] is coronal,which is the unmarked place of articulation— the languages, in fact,lackedallkindsofstop-nasalonsetsinvolving[p,b,t,d]).14Thiscanbeestablishedwithafairdegreeofcertaintybysearchingfulldictionariesofthe languagesforwordsbeginningwithreconstructedsequencesofthesetypesinPGmc,andlettersequencesofthetype‹pn,bn,tn,dn›inOEandME,ontheassumptionthatthemuchmorephonemicspellingofOEandMEwouldshowtheseclustersinwrittenformsiftheyexisted,andthatwewouldexpectsuchonsetstoshowupword-initiallyiftheyare at all possible in a language. Kroonen (2009), Bosworth & Toller(1898-1921) and the MED are extensive dictionaries of these threestagesofthelanguage,andnoneofthemfeatureanywordsthatcanbeinterpretedasfeaturinginitialsequencesofthosetypes.15Thisisafair

14 I focus in the discussion here on clusterswith [n] as they are phonologically themostlikely,butwhatIsayalsoholdsforonsetclusterswithothernasals.Nostophasevercombinedwith [m]or [ŋ] inanonset inEnglish,as faras Iamaware,and thismay be due to a long-lasting fundamental phonotactic that requires the secondelement of an onset to be coronal when it is a non-glide sonorant (thanks to BenMolineaux for discussion of this).Where relevant I also searched for combinationswith [m]/‹m›, and also found nothing, so I assume that this is not a potentialcomplicatingfactorinwhatfollows.15Apartfromafewscribalerrorsorvariantsofformswhichnormallyhaveavowel,suchastheMEformstnykyllerewhichislikelyanerrorfortinkler‘tinker’,andpniger,whichislikelyanerrorforwiniger‘vinegar’(withwhatlookslikea‹p›infactpossiblyalateattestationoftheletterwynn—thankstoBenMolineauxforthissuggestion).

PatrickHoneybone 96

proxy for establishing the systematic absence of [pn, bn, tn, dn]-typeonsets.

On the other hand, the sequences [kn] and [ɡn] were perfectlypossible as onsets in PGmc, OE and ME. All three dictionaries listmultiplewordsbeginningwiththesesequences(forexample,theMEDhas140headwordsbeginning‹kn›and34beginning‹gn›),asinwordssuchasknee,know,gnat,gnaw,whichevennowretain thespellingoftheseearliersequences.WecanbesurethattheyhadinitialstopsuptillMiddle English at least because alliterative poetry from that periodalliterates words with [kn] onsets with those with [k] onsets, asMinkova (2003,313) shows (for examplecneouwen ‘knees’ alliterateswith king in Lagamon’s Brut from around 1200). On the other hand,early pronouncing dictionaries such as Spence (1775) and Walker(1791) list several words with initial ‹kn› and ‹gn›, and all aretranscribedasbeing[n]-initialintheirpronunciation,sothestopshadclearlybeenlostintheseonsetsbythelate18thcentury.

Ifweconsiderthegapsinvolvedinthethreeearlyhistoricalstagesconsideredhere(PGmc,OEandME),wecanrecognisethat,while[kn]and [ɡn] arepossible, and [pn, bn, tn, dn] aremissing, other complexonsetsinvolvingallstops(like[pl,dr,kl,ɡr])areentirelypossibleatallthree stages, and all this means that the [pn, bn, tn, dn] gaps couldplausiblybefilled.Inaddition,thesetsofsegmentsinvolvedinthegapsareclasses:thesecondmemberoftheS-gapis ‘allandonlythenasals’andthefirstmemberis ‘allandonlythelabialandcoronalstops’.Thelatterclasscanbeexpressed,butanysystemoffeatureswillstruggletoexpressitsimply.Ireturntothispointbelow.

If we assume that there is an S-gap in early English, banningsequences of labial-or-coronal stops and nasals, we predict that theeffectoftheconstraintshouldbevisibleinthehistoryofEnglish.Thereis some evidence that it was. Thus, for example, the early loanwordpneumatic, borrowed from Greek and/or Latin (for which the OED’sfirst attestation is from 1624), is nasal-initial in Present-Day English,and is already nasal-initial in Spence (1775) and Walker (1791) —while English spelling often preserves the orthography of a donorlanguage,thereisnoevidencethattherewaseveraninitial[p]inthisword, thus it may well be that it was adapted at the point of itsborrowing into English, which may just have been while velar-stop-nasalonsetwerestillallowed.Thepatternofadaptingloanwordswhichviolatethiskindofphonotacticbydeletingthestopiscertainlyrobustin earlier stages of English — as another example, Baldwin (1846)transcribes the Russian river Dnieper as nasal-initial in English, too.While thesource languages for these loansallowed labialandcoronal


stop-nasal onset clusters, the initial stopwas lost in English, likely assoonasthewordswereborrowedintothelanguage.

Intermsofthedatingofthelossoftheinitialstopsin[kn]and[ɡn]onsets,thereisagreementintheliteraturethatwerelostbytheendofthe17thcentury(thereissomedisputeovertheprecisedating—seeMinkova2014,Lass1999—Lassarguesonthebasisofcommentsonpronunciation from the period that the key change for our purposesbegan in the17thcentury).Theprecise stages involved in thechangemay in fact be different in different parts of Britain, and the changeconsidered here likely involved some briefly-existing intermediatestage between stop and deletion, but it can be set out in simplesegmentalformasin(3),withthefull-stop/periodindicatingasyllableboundary. The phonotactic point is that this innovation wasaccompaniedby(orcanequallybeseenas)somethinglikethechangegiven in (4). Any phonotactic from earlier stages of English, banningonsets combining [p, b, t, d]with a nasal,would have been relativelycomplex: no matter how it is formulated with phonological features,*ONSET-{T–k,ɡ}Nor*ONSET-{p,b,t,d}Nismorecomplexthan*ONSET-TN.

(3) k,ɡ>∅/.__n

(4) *ONSET-{T–k,ɡ}N>*ONSET-TN

Thechangein(4)expressesthediachronicphonotacticsinvolvedin

the change, and itmissespartof thepoint to ignore this (and toonlyconsider (3)). Honeybone (2019) speculates that we might evenperceiveapressuretosimplyphonologicalgeneralisations,asin(4),aspartoftheexplanationforthechange.Thisisreminiscentofproductiveworkwhichhasaimedtoestablishifthereisdirectionalityinchangeintermsofhowone setof rules can change intoanother,or in termsofhowarulecanchangeifitstaysinalanguagebutchangesitsstructuraldescription or environment, as in Kiparsky (1968, 1971) and King(1969), forexample(andseeRoss2011foracautionary investigationofpossiblepatterns in rulechange).Weshouldbecautiouswithsuchspeculation— such structural pressures cannot be seen to cause anyparticularchange,butitmaybethattheyconstrainwhichchangesarepossibleinanyphonologicalstate.

In any case, if we understand the change in hand here as justdiscussed,wecanseeitasacaseofphonotacticchange,ratherthantheadditionofanewphonotactic.Indeed,wearecalledtoconsiderwhichof (3)or(4) ‘camefirst’,orwhether theyare in factseparableatall. Iforeseecriticismthatthisapproachrisksa ‘duplication’dilemmaifwecanmodelachangeashavingbothaneffectintherulecomponentand

PatrickHoneybone 98

theconstraintcomponent.Iwouldcounterthatfailingtoconsiderbothrisks missing a point, and while the approach makes things morecomplex,italsomakesthemmoreinteresting(and,Ithink,morelikelytobeclosertothetruth).Ifwetakephonotacticsseriouslyinphonologyat allwe need to consider these issues related to phonotactic change(and to other types of change in the phonotactic component of thegrammar).

ThemainpointofHoneybone(2019),however,isthatthereisevenmoretotheunderstandingofdiachronicphonotacticsthanthis:wealsoneedtorecognisethatalready-existingphonotactics ina languagecaninhibitnewly innovatedsegmental changes.By this, Imean that thereare cases of segmental change which fundamentally have astraightforwardenvironmentofapplication,butwherethesignatureofaphonotacticisvisible,explainingacomplexityinitspatterning.

OneexampleofthisthatIdealwithinHoneybone(2019)istheroleof a constraint like that in (5) in inhibiting an aspect of Late MiddleEnglish syncope, which Luick (1914–40) assumes was completed by1500 (what I call Spätmittelenglischer Schwund ‘late Middle Englishloss’,followingLuick).

(5) OCP(SIBILANCE)`

a tautosyllabic or tautomorphemic sequence of sibilantsegmentsisforbidden

The phonotactic in (5) forbids sequences of the six sibilant

segmentsofEnglish:/s,ʃ,z,ʒ,tʃ,d� ʒ/.WecanseethatitisnecessaryinPresent-Day English because,while affricatesmay not combine easilywith other segments, sequences of other fricatives are perfectlypossible(thatis,thebasicstructuralfactsofthelanguageallowthem)—thus[sf]occursasanonsetsequence,asinsphere,sphinx,sphincter,16 16 Chomsky&Halle (1968, 416) in fact assume thatwordswith initial [sf], such assphere, are ungrammatical (“inadmissible”) despite surviving perfectly well in aphonologicalderivation.Whileanintriguingidea,thiswouldbesurprisinggiventheirlongpresenceinthelanguage.Thesewords,whileallloanwords,havebeeninEnglishforaconsiderabletime(doubtlessloanedandreloanedfromlearnedsourcesseveraltimes), and we might expect them to have been adapted by now if they areproblematicphonologically:theOEDgivesarobustnumberofattestationsofallthreewords (with ‹sph›) from the middle of the 16th century, with much earlierattestations,too.Interestingly,theveryfirstattestationsofsphereandsphinxarespeltspere (from before 1300) and Spynx (1420-1422), bothwith ‹sp›, whichmay showthatEnglishhaschangedinthisregardanddidnotthenallowinitial[sf]sequencessothatthewordswereadaptedwhenfirstborrowedtohave[sp].Theearliestspellinginthe OED of sphere and sphinx with ‹sph› (indicating that [sf] then survives thereloaningprocess)arefromaround1533and1579,respectively,justafterthelateMEperiodinquestionhere.


as may also [sθ], as in sthenic and sthenia; medial tautomorphemicsequences of fricatives are possible, thus [sf] occurs in asphalt andblaspheme, [fθ] in diphtheria andophthalmology, and [sθ] inaestheticandanaesthesia;andalsofinalsequencesoffricativessuchas[θs]and[fθ] are possible, as in meths and twelfth. None of these kinds ofcombinationsoccurwhenthefricativesaresibilants:noinitialclustersoccur (unlike in Polish, for example), word-medial sibilant clustersalways indicateamorphemeboundary inEnglish,withonesibilantofeithersideofit(asinmisshapenandnewssheet),andwordfinalsibilantclustersareabsent.

There is good evidence, therefore, that the phonotactic in (5)currently regulates the combinability of sibilant segments in English.MostofthesixcurrentsibilantshavebeeninEnglishsinceatleasttheMiddleEnglishperiod(allexcept/ʒ/whichemerged inEarlyModernEnglish), and there is evidence that their combinability has beenconstrainedinat leastsomeofthewaysdescribed(5)sincethen, too.Thediachronyofthisphonotacticisaninterestingquestioninitsownright, and while I think it is the type of a question that we can andshouldpursue,Ilackthespacetodothatfullyhere.AsIhintinfootnote16,itmaywellbethatOCP(SIBILANCE)haschangedoverthecenturiesinitsprecise formulation, and in its interactionwithotherphonotactics.Depending on how gemination is analysed (if geminates involveclusters, they are relevant), the ‘tautomorphemic’ aspect of (5) mayhavechanged,too.Thus,forexample,theMEwordkissen‘tokiss’hadamedial geminate (which was underlying — in the base — unlikepresent-day cross-morpheme derived geminates), and this kind ofstructure was possible in English until the loss of gemination (thecompletionofwhichLass1992datestoaround1400).Allthatiscrucialformypurposes,infact,isthattherewasanS-gapinthelexiconduetothe inability of sibilants to cluster in final sequences — I need toassumethatthiswasactiveinLateME.Absentclearevidencethattheconstraintwasdifferentfromtheformulationin(5),however,IsimplyassumethattheLMEphonotactichasthesameformasthecurrentone.

Honeybone(2019)setsoutthewayinwhichaphonotacticlike(5)interactedwith the innovationofpartofLateMiddleEnglish syncope(Spätmittelenglischer Schwund ‘SpSchw’), which is one of theways inwhichEnglishhaslostunstressedvowels.Thisisacaseof(post-tonic)syncope,whichtargetedthelastunstressedvowelinaword—thepartconsideredhere is thatwith thespecificsegmentalconditioninggivenin (6), which seems to have been regular. As shown in the syllabicrepresentations,thisinvolvedthelossofasyllable.

PatrickHoneybone 100

(6) CV� CVz >CV� Cz σ� σ >σ� CV� CVCVz >CV� CVCz σ� σσ >σ� σ

Some example forms relevant to understanding the change are

given in(7),gathered fromHoneybone(2019), towhich Ihaveaddedacute accents to indicate stressed vowels. The first three showorthographic forms of morphologically simple originally disyllabicwords which underwent the syncope (showing how spelling changerecords aspects of the change), the remaining forms are plurals,represented in part following the conventions that Lass (1999, 142)adopts to discuss this case, using [V] to represent the “weak vowel”which syncopates when the rule applies and otherwise semi-orthographicforms,whichallowustofudgetheissuessurroundingthenature of the tonic vowels. Given that the pre-syncope form of theregular pluralmorphemewas /-Vz/ the environment for the syncopewasmet inallregularplurals.Thesecondthreeformsin(7)are fromLass (1999), giving the ancestors of cats, dogs and kisses, and theremainingformsareaaddedtoshowthatnotjustdisyllabicformswereaffected.Theleft-handcolumngivesformsastheyoccurredbeforethechange was innovated into the language, and the right-hand columngives forms thatexistedonce thesyncopehadstabilised (thesyncopecanfeedlaryngealassimilation,asshownincats).(7) Témys > Thámes álmis > álms áddis > ádze

kátVz > kátz [®kats] dógVz > dógz kísVz ... kísVz

máidenVz > máidenz héavenVz > héavenz físherVz > físherz ábbessVz ... ábbessVz

The absence of syncope in kisses and abbesses (indicated by theabsenceof ‘>’ and the same formsbeforeandafter the change) is theimportant thing to note. Itmakes sense ifwe assume (i) that the keychangewas the introductionof this caseof syncopeasaphonologicalrule, which applied to every occurrence of [-Vz]; (ii) that, like allchanges,thisnewruleofsyncopewasinitiallyvariable;and(iii)thatitwasinhibitedduetointeractionwithOCP(SIBILANCE).The‘SpSchw’ruleofsyncopecanbeunderstoodasin(8),althoughthisisreallysimplya


Thealignment in(24) isslightlywrong,asthereneedstobesomemorespacebetweenthederivations.Itshouldbeasgivenhere(pleasealsomakesurethatthisdoesnotbreakoverthepage):

(24) “Anidealisedhistoryoftheearlystagesofthepluralsofkiss,cat,dog(theweakvowel

representedas/V/)wouldbe:

kiss cat dog

EarlyMEinput kis-Vs kat-Vs dog-Vs Weakσvoicing kis-Vz kat-Vz dog-Vz WeakVdeletion —*kat-z dog-z Voicingassimilation — kat-s — ”

Thealignmentofthederivationsin(26)shouldbepreciselyasshownhere:

kisses dogs heavens

/kis+Vz//kis+Vz/ /dog+Vz/ /dog+Vz/ /hevən+Vz/ /hevən+Vz/

SpSchw kisz — dogz — hevənz —

OCP(SIB) * — — — — —

* [kisVz] [dogz][dogVz] [hevənz][hevənVz]Thealignmentofthederivationsin(27)shouldbepreciselyasshownhere:

(27) kisses dogs heavens

/kis+z/ /doɡ+z/ /hevən+z/

epenthesis kisVz — —

OCP(SIB) — — —

[kisVz] [doɡz] [hevənz]

description of the process. The rule is partially prosodically defined,targeting the finalunstressedvowel inaword,soa linearrule formatlike(8)doesnotshowthisinsightfully,butitwilldoforexemplification.

(8) V®∅/C__z#

[unstressed]The interactionbetween the rule and thephonotactic is shown in

(9), which sets out the model proposed in Honeybone (2019) tounderstand cases like this. It shows synchronic derivations for threerepresentativewords once the syncope (SpSchw) had been innovated(as a variable rule), and assuming thatmorphology has concatenatedthe base and the plural morpheme before handing the forms tophonology.Theleft-handderivationforeachwordshowswhathappenswhen the rule applies and the right-hand derivation shows whathappenswhen it does not, so twopossible derivations are present in(9)foreachword.Bothderivationsaregrammaticalforwordslikedogsandheavens,butthederivationwhentheruleappliestokiss(andotherwords which end in a sibilant) is rendered ungrammatical once itentersthephonotacticcomponent(whichIrepresentattheendoftherulecomponent,surroundedbyaboxto flag itup)because itviolatesOCP(SIBILANCE),andso itcannotsurface.Even if laryngealassimilationappliesaftersyncope(asincatsin(7))asislikely,thederivationisnotrescued as the violation of OCP(SIBILANCE) remains. The alternativederivation of kiss, without the application of SpSchw, can surfacewithout problem.Thismeans that at this point in the language, therewould have been two possible surface forms forwords likedogs andheavens. As the syncope stabilised and ceased to be variable, itunderwentruleinversion,togivethesituationthatremainsinPresent-DaysEnglish(asdiscussedfurtherinHoneybone(2019)).

(9)

AcentralpointofHoneybone(2019),whichIrepeatandreinforcehereisthatwecanunderstandthechangeinvolvedintheintroductionof (8) only if we recognise that the phonotactic in (5) intervened toaffect its patterning. The ‘exceptions’ to SpSchw, such as kisses andabbessesarenotduetothepatterningofthesyncopeitself,butaredueto the fact that the rule was innovated into a grammar which had a


phonotactic which ruled out forms with a sequence of sibilants. It isthusimportantforthismodeltoworkthatOCP(SIBILANCE)wasalreadyin the phonology when the syncope was innovated. Surface forms inwhichthesyncopeappliesarepreventedfromoccurringbecause—iftheydid—theywouldviolateaphonotactic.Thishasresonanceswiththe notion of prophylaxis in diachrony: that is, a language avoidingchange in order to prevent a ‘problem’ of some sort (such as aconstraintviolation).Ireturntothispointinsection4.

Themodel adoptedherewas shown inHoneybone (2019) to alsoallow for an insightful analysis of the patterning of the innovation ofMid-Scots θ-debuccalisation, and I go on to show that it allows us tounderstand other cases of change, too, in section 3. The basic modelcombinesaspectsofrule-basedandconstraint-basedphonologyinthewaythatitmodelssynchrony,andthisisacrucialpartofthewaythatitexplains diachrony. While it is derivational, the model has certaincharacteristicsthatarereminiscentofOptimalityTheory.Theexistenceof constraints in the model, evaluating forms that are close to thesurfaceisanobviousparallelwithOT,buttheassumptionhereisthatany constraint violation is fatal for a form,unlike inOTbut typical inpre-OTworkonphonologicalconstraints.AnotherparalleltoOTisthatthe model assumes that (in cases of change) there is more than onecandidatesurfaceform,andthatsomeoftheseformscaninprincipleberuled out by the grammar. The model adopted here has only twocandidates, however (rather than the infinite number in OT), whichboth can surface, and this number of candidates is determined bysomethingelseinthemodel(thefactthatarulecaneitherapplyornotwhenitisfirstintroduced).

The conclusions of this whole section are: that there is a lot fordiachronicphonotacticstoconsider,thatwecaninvestigatethehistoryof phonotactics seriously, and that some diachronically andphonologically interesting ideas emerge if we do so. As well as theinnovation and loss of phonotactics, and change in already existingphonotactics,wecanrecognise that that language-specificphonotacticconstraints on phonological forms can inhibit otherwise regularchanges. This latter point gives us a model for the interaction ofphonotacticsandsegmentalchanges.Iretainthefocusonthisinsection3,toshowthatitcanaccountforthepatterningoffurthercasesoftheinnovationofphonologicalphenomena.

3 *RIME-xxŋinEnglishsynchronyanddiachronyThekindoflexicalgapdiscussedatthestartofsection2ofthisarticleiswellknown.Ireturntoitheretoconsideritindetail.It isclearlytrue


that[baɪŋ]isagapinatleastsomeformsofEnglish,asImentionthere—nativespeakersofsuchvarietiesagreethatthereisnosuchthingintheir lexicon, and searches of electronic phonological databases findnothing. For it to be phonologically interesting as a gap, given thediscussion in section 2, it needs to be plausible (that is, it must beplausible that we might expect to find words with the structure inquestionbecausethestructurefitsinwiththe‘basicstructuralfacts’ofthe language) and systematic. [baɪŋ] is clearly plausible for mostEnglish forms of English, for example: [b], [aɪ] and [ŋ] occur assegments in these varieties, and there are surface representations ofwordswhicharesimilar: [baɪt] isbite,showingthat[baɪ] isapossiblestartofaword;and[bɪŋ]bing isapossiblesequence—ithas, infact,beenusedinarangeofwaysinEnglish(asanewword),forexampleasthenameofachildren’scharacterinaseriesofbooksandatelevisionseries (see, for example, Dewan 2003)— showing thatmonosyllabicmorphemesthatendin[ŋ](andstartin[b])canbefreelyaddedtothelanguage.Ishowinsection3.1that[baɪŋ]isanS-gap—itisforbiddenduetoasystematicphonotactic(atleastinmostvarietiesofEnglish).Iconsider a little of its history in section 3.2, and I go on to show insection 3.3 that it has acted in an inhibitoryway in diachrony, aswesawforOCP(SIBILANCE)insection2.Insection3.4,Iconsiderhowrobusttherelevantphonotacticis—Ishowthatinmostvarietiesitisrobust,but isnotsurface-true,whichhas implications forwherephonotacticsapplyinphonology;Ialsoshow,however,thatitmightnotberobustinsomeothervarieties.

3.1 *RIME-xxŋinEnglishThere has been some detailed work on the phonotactics of Englishwhich setsouta rangeofgeneralisationsabout thephonological gapsthat exist in the language. In a remarkable early volumeon the topic,Kruisinga (1943) shows (among much else) that [baɪŋ] is a gap inEnglishforarobustandsystematicreason.Histablewhichshowsthisisreproducedin(10),takenfromKruisinga(1943,54).Thisshowsallthemonophthongsof(GeneralBritish)English(alsoknownas‘RP’)andwhethertheycanbefollowedbysinglesonorants(informswithinitialconsonants). The table uses ‘—’ where no words with a particularphonologicalshapecanbefoundandanexamplewordwheretheycanbe found (and “whenaword isbetweenparentheses, thismeans thatthetypeoccursinafewwordsonly”).Vowelsin‘CloseContact’arethe


lax/short/checked vowels, and vowels in ‘Free Contact’ are thetense/long/freevowels.17(10)

Thesystematicityofthegapinquestionisclearfromtheabsenceof

examplesinthebottomrow:thereisnowordinEnglishwhichhasan[ŋ] at the end if preceded by a diphthong or a tense vowel.18 Thisrepresents a fundamental S-gap in the phonology of English, showingthat theabsenceof [baɪŋ] isdue toaconstraint that involvesawholenatural class of vowels: those which can be described astense/long/free.Ithasoftenbeenproposedthattheconstraintinvolvedcanbestunderstoodinprosodicterms. Itmakessensetoseethisasaconstraintonpossiblerimes,expressingwhichvowelscannotcombinewith[ŋ]inarime.Itisnotcrucialhowweexpressthenaturalclassofvowels involved (only that there is one). I adopt the autosegmentalconvention of representing these vowels using two rimal x-slots,assumingsuchrepresentationsasthosegivenin(11),wheretheunitsin vertical slashes at the bottom are intended as privative featuralspecifications(followingaconventionusedinHoneybone2001,2005).Tensemonophthongshavetwox-slotslinkedtotheironemelody,anddiphthongshaveonemelodylinkedtoeachoftheirtwox-slots.

17Kruisingausessomenowoutdatedbutinterpretableconventions,forexampleU=ʊandþ=θ.Healsouses[e],forthelaxDRESSvowel—Iuse[ɛ]forthisinPresent-DayvarietiesofEnglishelsewhereinthisarticle,followingadifferentconvention.Thefirstandlastrowsofwordsin(10)containinconsequentialtypos—thefourthwordinthefirstrowshouldbe[bUl],and[kɔm]shouldbe[kɔn].18 This holds quite fundamentally of the English lexicon, and any candidate wordswhichviolateitaremarginalinsomesense(e.g.boing,oinkareclearlyonomatopoeic).


(11) iː aɪ

N N xx xx

||• ••ï ï ï

|palatality| |lowness||palatality|

These vocalic representations are essentially those of the‘Dependency/Government’approach(asitisdescribedinCarr,Durand&Ewen 2005), and are sharedwith certain other approaches, too. Inmodelslikethis,itisfurtherassumedthateachsingletonconsonanthasone x-slot. There is no substantial difference in terms of the ideascovered here between rimal x-slots and moras, which are used forequivalentpurposesinmuchotherwork—thetwonotionscanbeseenas identical for the purposes of this paper (so I sometimes use termslike‘monomoraic’and‘bimoraic’torefertostructureswithoneortwox-slots,astheyarehandytermstouse).DiphthongsandtensevowelsinthevarietiesofEnglishinfocusherearethusseenaslong/heavyatthesurface.Onthisbasis, thephonotactic involvedinthecaseinquestionherecanbeunderstoodalongthelinesof(12).19(12) *RIME-xxŋ

asequenceoftwox-slotsandŋisforbiddeninarime

This phonotactic rules out [baɪŋ] and enforces all of the S-gapsidentifiedinthebottomrowof(10).Thephonotactic is indeedclearlyspecificto[ŋ]—astheotherrowsin(10)show,thereisnosystematicgapinvolvingvowelsbeforeothersimilarrimalconsonants,suchas[l,m, n]. As well as the systematicity involved in the pre-ŋ gaps, it isimportantforourpurposesisthatthegapsidentifiedinthebottomrowin(10)mightveryplausiblynotbegapsinEnglish—thisisshownbythefactthatmostoftheothercellsinthetablearefilled.

A number of other ways have been proposed in the literature tomodel this gap, but I think they fall short. Hammond (1999), forexample,assumesthat[ŋ]hastwomoras,andthatrimesmaynothavemore than three moras (coronals can be moraless), but it seemsarbitraryandadhoctomultiplymorassofreely.Jensen(1993),amongothers, assumes that all relevant occurrences of [ŋ] are underlyingly/nɡ/,andalateruleofg-deletionmakingtheassimilationthatderives 19Itcouldequallywellberepresentedas*RIME-μμŋ.


themopaque(otheroccurrencesof [ŋ]arederivedfrom/nk/,butthe/k/doesnotdelete).Thereismuchtorecommendthisapproach,asitcan allow the phonotactic in question here to be taken care of by ageneral constraint enforcing a rimal trimoraic maximum, which alsorulesout forms like [biːlk] and [baɪmp] (assuminga special status forcoronals),butthereisevidencethatitcannotberight.Ishowinsection3.4thatthereisfirmreasontobelievethatatleastsomeoccurrencesof/ŋ/ must exist in underlying representations. We are left, thereforewith *Rime-xxŋ. I say something about how it must have emergeddiachronicallyinthenextsection.

Thetablein(10)alsoshowsothergaps,andthisishighlyrelevanttotheunderstandingofthenotionoftheS-gapdevelopedinsection2.Forexample,thegapinthetoppartofthesecondcolumnindicatesthatthereisnowordwhichendsin[ɛŋ].Thismightbetakentoshowthattherime[ɛŋ]isungrammaticalinEnglish, like[aɪŋ].However,thereiseveryreasontobelievethat[ɛŋ]isanA-gap.Otherlaxvowelsoccurinthe[__ŋ]environment,sowemightbecautiousinassumingthatitisanS-gap. The CUBE searchable dictionary (Lindsey & Szigetvári 2013,whichincludesavastrangeofwordsofEnglish,transcribedforGeneralBritish) also does not find any word which ends in that way in thenative English lexicon, but it givesginseng andnasi gorengas endingwith[ɛŋ]—thesearebothloanwordswhichhavebeeneasilyadoptedinto English without adaptation (so, in fact, for speakers which usethem commonly, [ɛŋ] is not an A-gap).20 A similar story can be toldabout the [ʊŋ]gap thatKruisingarecognises: loanwords likeSamsungandKung(Fu)showthatthisisnotanS-gap.Wewouldpredictthatthenon-xxŋgapsin(10)areA-gaps,notS-gaps,becausetheydonotfeaturenaturalclassesofsegments(unlikethexxŋgap),anditseemsthatthispredictionismet.

3.2 Wheredid*RIME-xxŋcomefrom?Ifweassumethat*RIME-xxŋisindeedpartofthephonologyofvarietiesofEnglishwhich followthepatterningdescribedbyKruisinga (1943),as discussed in the previous section, diachronic questions arise: hasEnglish always been like this? and what would it look like if thissituation changed?Thesearebigquestions, andespecially the former

20TherearealsounambiguouslyEnglishformslikestrengthandlength,whichendin[ɛŋθ].Thisshowsthat[ɛŋ]ispossibleaspartofarime.Itmayevenshowthat[ɛŋ]ispossible as a full rime, if the final [θ] is shaved off from consideration because itrealises a separate morpheme to the base (which itself ends in [ɛŋ]), but evenconsideringevidencelikethis,[ɛŋ]doesnotoccurfinallyinthenativeEnglishlexiconinfreemorphemes.


would require serious studyof the kinddiscussed in section2.3. I donot explore this in detail here. I think thatwe should engage in suchexplorationsandIhopetohaveshownthattheyarebothnecessaryandpossible,butthatwouldbeajobforawholeseparatepaper.Somepartsof the diachrony of *RIME-xxŋ are relatively clear, however, and I setthemouthere.(Idiscusssomeaspectsofthelatterquestioninsection3.4.)

A lot of serious work has been carried out on the length (or‘quantity’) of Middle English vowels (see, for example, Luick 1898,1914-40, Lass 1974, 1992, Minkova 1982, Ritt 1994, Minkova &Stockwell1996,Bermúdez-Otero1998andLahiri&Dresher1999),andwhile it is clear that there were considerable lengthenings andshorteningsinthisperiod(someofwhichwereprosodicallydriven,andsomeofwhichweredriveninpartbysegmentalenvironmentinawaywhich is relevant toourpurposes, as inHomorganicLengtheningandShortening before Consonant Clusters), it is not possible to considertheseissuesfullyhere,inpartbecausetheyarecontroversial.Allthisisrelevant to our purposes, however, to the extent that English largelysettled into a pattern of allowing long/bimoraic vowels in certainenvironments and not in others, and the constraints on this can beunderstood at least in part as constraints on rime structure. Thesituation that developed during these changeswas the pre-*RIME-xxŋstate of the language. The current situation in varieties like thatdescribed by Kruisinga (1943), such as General British, is that rimesmayhave amaximumof three x-slots (as inmilk [mılk], lamp [lamp],meek [miːk], lime [laɪm], but forbidding S-gaps like *[miːlk], *[laımp],[laɪŋk]), although extra segments may occur following such a rime ifandonly if theyare coronals (as inmind [maɪnd],monks [mʌŋks]), inwhichcase,theyareoftenanalysedasbeinginan‘appendix’—thisisallwelldescribedinworksuchasFudge(1969)andGiegerich(1992).The xxŋ gap is unusual in English as it is a casewhere the languageallowsonlytwox-slotsinarime.

Ifwesetcoronalsaside(becausetheyclearlycomplicatethepictureandarenotdirectly relevant to thedevelopmentof thevelarnasal), Iassume, therefore, that thepre-change state (before the innovationof*RIME-xxŋ)wassimilartothecurrentsituationjustdescribed,inthatitallowed(i)rimeswithashortvowel(=1x-slot,monomoraic),suchas[i, e], as long as this was foot internal and another syllable suppliedanotherx-slot(ontheassumptionthatthefootwasamoraictrochee),(ii)rimeswithalongvowelordiphthong(=2x-slots,bimoraic),suchas[iː,eː,aɪ],(iii)rimeswithashortvowelandoneconsonant(=2x-slots,bimoraic),suchas[ik,ek,ib,ep,im,em],(iv)rimeswithashortvowelandtwoconsonants(=3x-slots,trimoraic),suchas[iŋk,eŋk,iŋɡ,eŋɡ,


imb, emp], and (v) rimes with a long vowel or diphthong and oneconsonant(=3x-slots,trimoraic),suchas[iːk,eːp,iːb,eːɡ,iːm,eːm,aɪk,aɪb, aɪm].21 The rime is thus fundamentally constrained by theconstraintin(13).

(13) *RIME-xxxx asequenceoffourx-slotsisforbiddeninarime

Achangewhichiscruciallyinvolvedintheinnovationof*RIME-xxŋ

wasthe‘clustersimplification’of[ŋɡ].Thischangecanberepresentedasin(14),but,aswith(3)and(4),thatinnovationwasaccompaniedbythe a phonotactic change, as in (15). There is general agreement thatthiscaseofclustersimplificationoccurred(instressedsyllables—thestoryof -ing is different)during theEarlyModernperiod, around themid-to-late 16th century, or perhaps even later (see, for example,Dobson1968, Lass 1999). As Lass (1999) discusses (and as has beenpickedupbyGarrett&Blevins2009andBermúdez-Otero2011), thischange worked its way gradually through the morphologicalenvironments inwhich[ŋɡ]occurred(followingthepatternpredictedbythe‘lifecycleofphonologicalprocesses’,asBermúdez-Oteroshows).While thechangehasbeen lexicalised intounderlyingrepresentationsin most varieties of English (so that there is now /ŋ/, as I argue insection 3.4), Bailey (2018), among others, shows that (14) is stillsynchonically active (and variable) in certain western dialects inEngland.

(14)ɡ>∅/ŋ__. (=theintroductionofaruleofɡ®∅/ŋ__.)

(15) *RIME-xxxx>*RIME-xxŋ

Unlikethesituationaround(3)and(4), it isnotthecasethat(14)canbeseenas the ‘samething’as(15),because[ŋɡ]became[ŋ]aftershort-vowelsaswellasafter longvowelsanddiphthongs.Thismeansthatwecanperceivehereanasymmetrybetweenthechangeintherulecomponentandthechangeintheconstraintcomponent:thesegmentalrule can be perceived to have hadprecedence because it has awiderapplicability,soitseemstomakesensetoseethephonotacticchangein(15)ascominginthewakeof(14).

21 Other vowel-consonant sequences and other clusters also, of course, occurred. Ionly list a few that are relevant here, in order to demonstrate the range of thepossible. I use the symbols [i] and [e] for short vowel here because laxingmay nothaveoccurredatthispoint.


The innovation of (14)meant that trimoraic rimes like [iŋɡ, eŋɡ]simplifiedtobimoraic[iŋ,eŋ].Nocompensatory lengtheningoccurredto preserve the x-slot/mora that was lost — quite the opposite: thevocalicformsthatthelanguageallowedinthisregardfroze.Norimesofthe type [iːŋɡ, aɪŋɡ] were possible before the change (because theywould have violated *RIME-xxxx), and they did not become possibleafterthechange,becausethephonotactic*RIME-xxŋwasinnovated.Wecould assume that, as [ŋɡ] was finally reinterpreted by learners as asingle segment, the systematic gap that this leftwas noticed by themandwasassumedtobeduetoaphonotactic,turningthexxŋgapintoanS-gap,asin(15).

Afurtherpointtonoteaboutthechangein(15)isthat,whileanewphonotacticisaddedtothelanguageinthechange,*RIME-xxxxremainsinthegrammar,too.Formslike[iŋk,eŋk,imb,emp,iːk,eːp,iːb,eːɡ,iːm,eːm,aɪk,aɪb,aɪm]arestillpossibleinthelanguageafterthechange,andare still themaximum rimes possible (ignoring the complications ofadditional coronals), and *RIME-xxxx can do the job of enforcing thismaximuminthesamewayasbefore—rimesinvolving[ŋ]couldneverreach fourx-slots,as*RIME-xxŋwouldstopthem,but*RIME-xxxxdoesnot need to know or care about this and can stay as general andsystematically-justifiedasbeforethechange.

Thechangeinvolvedhereisreminiscentof‘rulescattering,’which,asBermúdez-Otero(2015,basedonRobinson1976),explains,involvesarulerising throughthegrammar(as in the lifecycleofphonologicalprocesses), but also leaving its original form behind. This involves akindof splittingof a phonological phenomenon, both leaving thepre-changeforminthegrammarandalsoinnovatinganew(changed)form.It makes sense in the phonologisation of processes because, asBermúdez-Otero (2015, 387)writes, “the original gradient process ofphoneticimplementationremainsactiveinthegrammarevenafterthenewcategoricalruleentersthephonology”.Whileitisnotclearthatthelifecycleisinvolvedinthecaseofphonotacticchangeconsideredhere,(15) can be seen as a kind of ‘constraint scattering’, in which theoriginal constraint remains active, while a new, more restrictivephonotactichasalsoenteredthephonology.

If the origins of *RIME-xxŋ proposed here are correct, we haveencountered furtheraspectsofdiachronicphonotactics in theprocessof working to understand the changes involved: constraint scatteringcanoccurinphonologicalchange.Thisshowsagainthatthereismuchofinteresttoconsiderinthisarea.Theprecisedetailsofthepre-changestate (and their interaction with ME length changes) would be wellworth investigatingproperly inconnectionwith the issues that Ihave


discussed,22 but it is clearly the case that *RIME-xxŋ was innovatedrelativelyrecently,duringtheEarlyModernEnglishperiod,aroundthe16thcentury.

3.3 Has*RIME-xxŋinhibitedanysegmentalchange?Givenallthediscussionuptillnowinthisarticle,itwillbeobviousthatIamaimingtoanswerthequestioninthissection’stitle.Ifocushereonacasewherethereisevidencethat*RIME-xxŋhasindeedinhibitedthepatterning of one well-known segmental change, in a similar way tohowwesawOCP(SIBILANCE)inhibitanaspectoftheSpätmittelenglischerSchwundsyncopeinsection2.3.Thechangeinquestionhasanumberofnamesintheliterature.Icallitash-tensing.Itisthechangewhichhasled to what was originally a low, lax, front, short vowel becoming araised,tense,oftendiphthongal,longvowel,ashasbeenwelldescribedin American dialectology and sociolinguistics for the Middle Atlanticstates of the US (see especially Labov 1994). The change is complex,and, aswith theMiddle English quantity adjustments above, I cannot

22 The full story is complex. For example, the history of aword like thing fromOldEnglish toPresent-DayEnglish looksquite straightforward: [θiŋɡ]> [θiŋ]> [θɪŋ] (ifwe followLass1999 inassuming that the laxing/loweringofhighvowelswas late),butthereisinfactgoodevidence(fromtextsliketheOrmulum—seeRitt1994)thatitfirstunderwentHomorganicLengthening([θiŋɡ]>[θiːŋɡ]),because[ŋɡ]wasoneoftheclustersthattriggeredthechange,andwasthenshortenedagain,perhapsevenaspartof ShorteningbeforeConsonantClusters (a formof closed syllable shortening),beforethe introductionofg-deletion(andthephonotactic infocushere),sothatthefull history of thing (at least, in some varieties) is: [θiŋɡ] > [θiːŋɡ] > [θiŋɡ] >[θiŋ]/*Rime-xxŋ>[θɪŋ].Weshouldalsoconsiderthehistoryofthesyllable-final[mb]cluster inEnglish,whichalsosimplified to [m].At firstglance,wemightexpect thatthiswaspartofaunifiedchangewith theŋɡ>ŋsimplificationgiven in (14),but infact it seems that they were separate events. There is certainly no *RIME-xxmconstraint in Present-Day English, as shown by the existence of words like womb[wuːm] and climb [klaɪm]. Thismakes sense if work like Dobson (1968) is right inarguingthatmb>moccurredmuchearlierthanŋɡ>ŋ.Dobsonarguesthatmb>moccurs by 1300,whereas ŋɡ > ŋ occurs about 1600,meaning that they are entirelyunconnectedevents.Itmayevenbethatthistookwordslikewomboutoftherealmsofpossibleapplication forShorteningbeforeConsonantClustersbecausetheclusterdisappeared before the shortening occurred (if not, [mb] was ignored in theenvironmentoftheshortening),sothatapossiblescenarioforthehistoryoftheword(onceoriginal/a/hadrisento/o/)is:[womb]>[woːmb]>[woːm]>[wuːm],thatis:thechronologicalorderingofthepossibilityforapplicationofshorteningandclusterreductionaredifferentforwordswithsyllable-final[mb]andwithsyllable-final[ŋɡ],whichmeantthatshorteningdidnotapply in[mb]words,xxmoccurswhen[mb] issimplified,andno*RIME-xxmphonotacticcouldbe innovated.Thescenariosketchedoutheremaynotberight,butat least Ihope tohaveshownthat thequestionas towhy there isnophonotactic *RIME-xxm inEnglish (when there is a *RIME-xxŋ) is aninterestingquestion.


deal with all aspects of its innovation, but certain points in thediachronyof thephenomenonareclear,and it is these that I focusonhere. The input to the change has variously been described in theliteratureas‘shorta’,theBATH/TRAPvowel,and/æ/(whichiswhyitisalso called ‘ash’, as that is the name of the letter ‹æ›, and iswhy thechangecanbereferredtoas ‘ash-tensing’). Iuse ‘æ’as thesymbol fortheinput.Theoutputvariesfromvarietytovariety,andistranscribedinseveralwaysintheliterature(includingnon-transparentformslike[A],[E]or[æh]).Itisoftentranscribedasadiphthongalformlike[eə],andthatiswhatIuseasthesymbolfortheoutputofthechange.

TheultimateoriginofthechangeislinkedbyLabov(1994),basedonearlierworksuchasFerguson(1972),to‘BATH-broadening’,23whichin its full extent is a complex set of changes in its own right, with achanging environment and aspects of lexical conditioning. It began inBritaininthe17thcentury,andeventuallyturnedshort,lax,lowvowels(originally[a])intolong,tensevowelswhicharebackinmanyvarieties,giving Present-Day General British forms like bath [bɑːθ] and grass[ɡɹɑːs]. At the time that this change was taken to Philadelphia andnearby areas in North America in early settlement from Britain, thevowel in these forms was still front (likely [æː]), and Labov’sassumptionisthatitwasstillinstantiatedthrougharule(æ®æː).Theoutputofthissetofinterrelatedchangeshasthereforeclearlychanged,andnowvaries fromdialect todialect. In thevarieties that I considerhere, it is typically described as diphthongal, and so (as mentionedabove),therelevantpartofthechangeasIwillbeconsideringitcanbethoughtofas:æ>æː>eə.Whatwillbeimportantisthattheoutputislong/tense—thatis,itisassociatedtotwox-slots/moras.24

Thephonologicalenvironmentinwhichthechangeoccurs(and/orinwhichtheassociatedruleapplies)hasalsochangedandgeneralisedin different ways in different places, and it is this aspect of thephenomenon that I focus on. The patterning currently described forash-tensing is famously lexically-specific in certain dialects (such asPhiladelphia English) in, for example, the pre-d environment (see, forexample,Labov1981,Kiparsky1988),suchthat—ifweassumethatitisstillarule—ithasexceptions,althoughitisregularinvariousother

23 This name is fromWells (1982), using his key-word for the lexical set of wordswhich have undergone the change (which have a vowel which is distinct from thewordsintheTRAPset).Lass(1999)callsthechange‘LengtheningI’.24 While we can’t read phonology directly from phonetics, Sneller (2019) showsclearlythatthedurationoftensedvowelsislongerthantheiruntensedequivalents,sothereisgoodreasontobelievethatthevowelinlong.Thischangecanthisbeseenasintroducingafurtherlong/tensevoweltothesetlisted(asin‘FreeContact’)in(10),althoughvarietiesthathaveittypicallydonothavealloftheothervowelsinthatset.


sub-environments. I followLabov andKiparsky in assuming thatash-tensingcanbeseenasasynchronicrule,whichIrepresentasæ®eə,andthediachronythatIconsiderisnottheinnovationofthisrule(orthechangeinitsoutput),butthewaythatitsenvironmenthaschangedover time, since it was taken to North America. Sneller (2018, 2019)makesagoodcasethatash-tensinginclassicalPhiladelphiaEnglishisasynchronic phonological rule, with a ‘tolerable’ number of lexicalexceptionsintermsofYang’s(2005,2016)TolerancePrinciple.

Labov (1994) argues that the pattern in the distribution of thetensedoutputofash-tensingfoundinlate20th-centuryPhiladelphiais“closetothatoftheoriginalshortalengthening,andthattheNewYorkCitydistributionwasoriginallyquitesimilar”.Thisgivesusapictureinwhich therulehasgeneralised, first fromthe initial ‘BATH-broadening’environment which was brought to the area (before syllable-finalvoicelessfricativesandsomeanteriornasalclusters),25thentoapatternlike that found now in Philadelphia (which includes syllable-finalanterior single nasals), and then to a pattern found inNewYork City(NYC),applyinggraduallyinmoreenvironments,withthedistributioninNYCbeinghighlygeneralised.Ithasbecomeconventional,followingdiscussions like Labov (1981, 1994), to represent the environmentwhich triggers ash-tensing as in (16), which sets out potentiallyrelevant obstruents and sonorants of English and includes in a boxthose which trigger the rule when they directly follow the vowel, insyllable-finalposition.

(16) p t tʃ k

b d d� ʒ ɡ

m n ŋ

fθ s ʃ

vð z ʒ

l rThe inner box in (16) gives the environment for Philadelphia

(wherethedashedlinesaround/d/indicatethatthissub-environmenthasaconsiderablenumberofexceptions)and theouterboxgives the

25 Lass (1999) also includes thepre-r environment in thepatterningof this change,butitmaybethatthisisaseparatedevelopment,linkedtootherBritishpre-rchangesandperhapseventhelossofrimal-r,giventhatitdoesnotfitaspartofthepatternofash-tensinginNorthAmerica.


environment for NYC.26 This means that ash-tensing occurs in allvarietiesmentionedhereinwordssuchasham,hand,man,pass,bath.NYC has a very general pattern, taking in natural classes of stops,fricativesandnasalsapartfromthefactthatonlyanteriornasalstriggerthe process:ashdoes not tense preceding a syllable-final velar nasal.Interestingly, Harris (1989) describes a similar situation in NorthernIrish English, which has generalised the same kind of ash-tensingpatterntothatfoundintheMid-AtlanticstatesoftheUS.Hewritesthat“[i]nBelfast, the tensereflexhasamuchwiderdistribution,occurringinalltheNewYorkCitycontextsaswellasbeforevoicedfricativesandl”. This distribution includes the tensing/lengthening before syllable-finallenisstopsatallplacesofarticulation,andnasalsatthelabialandcoronalplaceofarticulation,but,asinNYC,explicitlyexcludestensingbefore[ŋ].Inthistypeofash-tensing,almostanyconsonantcantriggerthe process apart from [ŋ]. A featural characterisation of the surfacepatternforNYC(andBelfast)wouldbecomplexbecausenaturalclassesof segments are involved apart from in the nasals, where only [ŋ] isexcluded.ItisnotablethatinPhiladelphia,too,syllable-final[m]and[n]trigger the rule, but [ŋ] is excluded. Why is the pre-ŋ environmentalwaysresistanttoash-tensing?

This isthephonotacticpoint:wecanunderstandthepatterninginthe generalisation of ash-tensing, as the environment for the rulechangeddiachronically, ifwerecognisethatthephonotactic*RIME-xxŋwas involved: ash-tensing itself generalised to take in full naturalclasses — it did not care about place in the nasal series — but theexistence of *RIME-xxŋ in the constraint component of the grammarprevented pre-ŋ tensed ashes from surfacing. We can model this inexactlythesameformatasthatsetoutin(9)asawayofmodellingtheinteraction of OCP(SIBILANCE) and the Spätmittelenglischer Schwundsyncope,asIshowin(17).

(17) ban bang

/bæn/ /bæn/ /bæŋ/ /bæŋ/

ash-tensing beən — beəŋ —

*RIME-xxŋ — — * —

[beən] [bæn] * [bæŋ]

26 The full environmental patterning also requires a consideration of paradigmuniformity effects, morphological structure and a few other factors which are notrelevanthere.


The derivations in (17) are from the point when ash-tensing isgeneralising toapply to fullnatural classesof segments, including thenasals. At this point, as in the early stages of all such phonologicalchanges,theapplicationoftherulewouldhavebeenvariable,andasisthe case in (9), (17) shows synchronic derivations for representativewordsatthispointofvariation,withtheleft-handcolumnforeachwordshowingwhathappenswhentheruleappliesandtheright-handcolumnwhathappenswhenitdoesnot,sotwopossiblederivationsarepresentin (17) foreachword.Bothderivationsaregrammatical forwords likeban, which features an environment where tensed/lengthened vowelssurface inallash-tensingvarietiesdiscussedabove,but thederivationwhentheruleappliestobang(andotherwordswithsyllable-final[ŋ])is rendered ungrammatical once it enters the phonotactic component(againrepresentedattheendoftherulecomponent,surroundedbyabox for ease of recognition) because it violates *RIME-xxŋ, and so itcannotsurface.Theotherderivationofbang,withouttheapplicationofash-tensing, can surface without problem, with the underlying ashvowel.Wecanthusseethat*RIME-xxŋhasindeedinhibitedasegmentalchange, explaining what seems otherwise to be an exceptionalenvironmentalpatterning.

We can understand the patterning of the generalisation of ash-tensingonly ifwe recognise thataphonotactic intervened toaffect it.Ash-tensingdidnotgeneralisetoproducedtensevowelsinpre-ŋwordssuchasbang,gangandfang,butthisisnotduetothepatterningoftherule itself, but to the fact that the rulewas innovated intoagrammarwhichhadaphonotacticwhichruledouttense/longvowelspreceding[ŋ]. As was the case in the SpSchw syncope discussed above, thelanguageseemstohaveprophylacticallyavoidedenactingthechangeifitwouldhaveviolatedanalready-existingphonotactictodoso.Ireturntothesepointsinsection4.

3.4 Wheredoes*RIME-xxŋapply?Isitrobustinallvarieties?Wehaveseenthatthereisgoodevidencethat*RIME-xxŋconstrainsthephonologyofEnglish—bothinthecurrentpatterningofthelexiconintermsofvarieties likeGeneralBritish,asshown in(10),and fromthefactthatwecanseeitseffectsindiachrony,asin(17).Inthissection,Iconsideranumberofpiecesofdatawhichseemtoinvolveviolationsof*RIME-xxŋ. In themodel that I have adoptedhere, this is problematic,giventhatanyconstraintviolationisassumedtobefatalforaform.

Gimson(1962)isoneoftheclassicdiscussionsofEnglishphoneticsandbasicphonology,givingadetaileddescriptionofaspectsofGeneralBritish,amongotherthings.Gimsonpointsoutthat[ŋ]caninfactoccur


(syllable-finally)afterlongvowels,asinformslikethosein(18),takenfromGimson(1962,270).

(18) I’vebeengardening. [biːŋ]

She’llsooncome.[suːŋ]

Doesthismeanthat*RIME-xxŋisnotarealphonotacticafterall?Or

has the phonotactic been lost in this variety?We can answer both ofthesequestionswitha firm ‘no’,but this requiresa recognitionof thewider paradigm in which the phonological model set out above isembedded. Gimson considers this data as part of his discussion ofassimilation, and it is clear that the two cases of [ŋ] in (18) areunderlyingly /n/, and have assimilated in terms of their place ofarticulationtothefollowing/ɡ/or/k/.Thisrecognitionalonedoesnotsolve the problem, however, given that the model adopted hereassumes that phonotactic constraints are evaluated at the end ofphonology—arulelikeNasalPlaceAssimilationmustoccurearlierinthederivationthantheconstraintcomponentas ithasbeensetout inexamplesofthemodelassumeduptohere,sotheconstraint*RIME-xxŋshould rule out forms like [biːŋ] and [suːŋ] after assimilation hascreatedthem.Eitherthemodeliswrong,or*RIME-xxŋdoesnotapply.

Theanswer to thisconundrumisstraightforward: themodeluseduptillnowneedsarefinement,or,rather,itneedstobeunderstoodinitspropercontext.Itisacommonplaceinmanymodelsofphonologytorecognise a distinction between lexical and postlexical (or ‘phrasal’)phonology.ThisideawasdevelopedindetailinworksuchasKiparsky(1982,1985),whichproposedtheLexicalPhonologymodel,butitalsohas longer roots, and it is currently pursued in Stratal Phonology, insuchworkasKiparsky(2000),Bermúdez-Otero(2003,2015).Thekeypoint in terms of the data just discussed is that the relevantassimilation-derived cases of [ŋ], such as those in (18), have beenderived from /n/ by a post-lexical process— this is clear because itoccursacrosswords.What thismeans is that,at theendof the lexicalphonologicalstratum,beenis[biːn]andsoonis[suːn],anditisherethat*RIME-xxŋapplies—inaconstraintcomponentattheendofthelexicalphonology. Post-lexical nasal place assimilation then applies in thepost-lexical phonological stratum, and all we need to assume is that*RIME-xxŋisnotpartofthepost-lexicalconstraintcomponent.Thisallseemsfitstogther:iftheideaisrightthatthereisfirstalexicalstratumof phonology, afterwhichwords are concatenated into phrases, afterwhichthereisapost-lexicalphonology,thenthemodelproposedabove


would expect there to be a rule component in each, followed by aconstraintcomponentineach,andwewouldexpectthatthecontentsofthesecomponentswillbedifferentinthetwostrata.

Itisclearthatash-tensingisalexicalrule:itisword-bound,andthetriggeringconsonantwhichfollowsthevowelneedstobeinacoda(sopotential resyllabification across word-boundaries is irrelevant).Kiparsky (1988) argues strongly that ash-tensing is indeed a lexicalrule.We thus have evidence that bothash-tensing and *RIME-xxŋ arepart of the lexical phonology, and everything slots into place if weassume that *RIME-xxŋ applies at the end of the lexical phonology,before words are concatenated to produce phrases in the postlexicalstratum. *RIME-xxŋ is not surface-true, but it is still a part of thephonologyofthevarietythatGimsonisdescribing,anditistrueatthelevelatwhichisapplies.

Whatthismeans is thatthe locusofphonotacticconstraints isnotnecessarily ‘the end of phonology’, as assumed in section 2.2 andelsewhere above, but rather: phonotactics apply at the end of aphonologicalstratum.Thecrucialpointremainsthesame:phonotacticsapplyattheendofablockofphonology,andcanruleoutcertainformswhichhavebeenderived,sothattheycannot‘escape’fromphonology.

Like *RIME-xxŋ, there is good evidence that OCP(SIBILANCE) alsooccurs in the lexical phonology: by very definition, it does not applyacross word-boundaries (cross-lexical sequences of sibilants arepossible, and all are cross-morphemic by definition). The extent towhich there are post-lexical phonotactics is an interesting researchquestion,but the idea thatphonotacticscanapplyat severalplaces inphonologyseemstherightthingtoassume—thepatterningofsyncopein present-day English implies exactly this. Thus, there are goodarguments, from both lexical forms and speaker behaviour whenconfronted with loanwords, that OCP-ONSET(CORONAL) is a robustconstraintofEnglish,asdefinedin(19).

(19) OCP-ONSET(CORONAL) asequenceoftwoanteriorcoronalsegmentsisforbiddenin

anonsetThis isdiscussed insomedetail inHoneybone(2019),and isbased

on the facts that [tl], [dl] and [θl] are not found as onsets in English(despite the fact that similar onsets like [tr], [dr], and [θr] are well-attested, and again ignoring aspects of the behaviour of [s]), and thatwords which begin with those sequences in a donor language areadapted when borrowed into English. This seems to be a robust factaboutthelexicalphonologyofEnglish,butitisalsowellrecognisedthat


pre-tonicsyncopeofunstressedvowelscancreateinitialsequenceswith[tl],inwordslikeToledo,telegraphy,andwith[dl],inwordslikedilemma,delightful(andcanalsocreateotherclustersthatwewouldwanttoruleout from the lexicon, such as [pt] inpotato— see, for example, Algeo1975andSzigetvári2007).ItseemsrighttosaythatOCP-ONSET(CORONAL)applies in the lexical phonology, but not in the post-lexical phonology,like the other phonotactics discussed in this section, but it also seemsrighttosaythatthiskindofsyncopeisstillunderphonologicalcontrol:itisnotthecasethatabsolutelyanythinggoes.Thistypeofsyncopecannotcreateafallingsonorityslopeinanonsetandsoisnotpossibleinwordslikeretire,reduction,mature,laconic(Roca&Johnson1999,35).WecanmakesenseofthisifweassumethatOCP-ONSET(CORONAL)appliesinthelexical phonology and that pre-tonic syncope of unstressed vowelsapplies in the post-lexical phonology, which must mean that theconstraint that is responsible for ruling out forms like *[lkɒnɪk] forlaconicispartofthepost-lexicalphonology,too.

ThefullphonologicalmodelthatIassumeisthereforethatthelocusfor the application of phonotactics is at the end of the relevantphonological stratum. The discussion here ties in with theLexical/StratalPhonology-typeprincipledexplanationfortheorderingof phonological phenomena and opacity (as discussed in Bermúdez-Otero2003,forinstance),whichallowsfororderingifphenomenacanbe shown to belong to different strata but does not require or allowextrinsicordering.AlthoughGimson(1962)doesnotputitinthisway,heisdiscussingthefactthat*RIME-xxŋisopaque,andthattheorderingoftherulesandconstraintsinvolvedispredictedfromtheirproperties,aslongasweallowfor(atleast)twofundamentalstratainphonology.

In fact, putting everything together, the model developed herepredicts that *RIME-xxŋ will also prove to be opaque in Mid-AtlanticEnglish, ifweconsiderits interactionwithash-tensingandnasalplaceassimilation(NPA)of thetype justdiscussed.Mid-Atlanticash-tensingislexical,*RIME-xxŋislexicalandNPAispostlexical,thereforeinpre-gcontexts like that in (20),ban should be able to surface as [beəŋ], asmodelled in (21), which shows ash-tensing at a time that it is stillvariable.

(20) Themadpresidentmightbangardening.

Themadpresidentmightbanggardengates.Forcomparison,(20)and(21)alsoshowthederivationsforbangin

pre-gcontext,whereNPAdoesnotapply,andwhichispredictednottobe able to surface as [beəŋ]. (21) shows the two phonological stratathat are assumed in the above, separated by a broken line (and


indicatedbytheverticallabels‘LxPhonology’=‘lexicalphonology’and‘PL Phonology’ = ‘postlexical/phrasal phonology’). The output of thelexical stratum is indicated in vertical slashes (to indicate that it isneitheranunderlyingnorsurfaceform),andNPAisassumedtobenon-variable(whichmaywellbewrong—ifso,[beən]and[bæn]willalsobepossiblesurfaceformsforbaninthesentencein(20)).(21) ban bang

/bæn/ /bæn/ /bæŋ/ /bæŋ/

ash-tensing beən — beəŋ —

*RIME-xxŋ — — * —

outputofLPh |beən| |bæn| * |bæŋ|

NPA beəŋ bæŋ * —

[other constraints] — — * —

outputofPLPh [beəŋ] [bæŋ] * [bæŋ]

ThereisdataintheliteratureonEnglishphonologywhichposesa

morefundamentalchallengetotheassumptionthat*RIME-xxŋispartofthephonologyofEnglish.Forexample,Donegan&Stampe(1979,149)describe a variety of English which has a process in which /æ, ɛ, ɪ/diphthongise to [æe, ɛi, ɪi] before tautosyllabic [ʃ, ŋ, ɡ].Thisproducesformslike[bæeŋ]bang.Hayes(2009,156)describesasimilarsituation,inwhich “/æ/ isdiphthongised to [æɨ] before/ŋ/”. I assume that thevarietiesdescribedinthesetwosourcesarethesameinthisregardandthat the two transcriptions for theoutput of thediphthongisation arenotational variants. I transcribe the output as [æɪ], in line with theconventions that I ammost used to—what is important for currentpurposes isthat it isabimoraicvocalicunit:adiphthong,whichtakesuptwox-slots.GiventhatDonegan&Stampelinktheprocesstootherlaxvowelsaswellasash,Icallit‘pre-ŋdiphthongisation’(PŋD).NeitherDonegan& StampenorHayesdescribe exactlywhichdialect they arereferring to (I assume the transcriptions reflect the speech of theauthors, and Hayes27 says as much). It is clear that these are NorthAmerican varietieswhich are not the same in this regard as theMid-Atlanticdialectsdescribedinsection3.3.

27Hayes’biographyonhiswebsitesay thathegrewup in Ithaca inNewYorkstate:https://linguistics.ucla.edu/people/hayes/Personal/Index.htm.

LxPhonologyPLPhonology


Donegan & Stampe’s transcription for bang makes the point, andHayes’ datamakes it clearer— this is a situationwhere a lexical ruleproduces forms which violates *RIME-xxŋ. The forms are presented ascategoricalrepresentations,due toan invariablerule(Hayes2009,156describesitas“anallophonicrule”),andthevelarnasalsinvolvedarenotderived by assimilation. Furthermore, both Donegan & Stampe andHayes point out that the PŋD rulewhich diphthongises /æ/ is opaquebecauseoccurrencesof/n/whichareassimilatedto[ŋ]byNPAdonottrigger it. Hayes (2009, 157) gives clear and compelling data for this,someofwhichisreproducedas(22)—amendedslightlytofitwiththetranscriptionpracticethatIadoptforvowels(includingflaggingupthattensevowelsarebimoraic,usingthe[ː]diacritic),andtoremovestressmarks, but otherwise following Hayes’ transcriptional decisions(including his analysis of rhoticity). The first three rows of wordscomparephonologicallysimilarwordswitheitherunderlying/n/(inthefirstcolumn)or/ŋ/(inthesecondcolumn).Thefourthrowshowswhathappens when a word with underlying /n/ occurs in a context whichtriggersNPA:NPAcanapply,but this ispostlexical,andsooccursafterthe lexical rule of PŋD has had its chance to apply. This makes PŋDopaque,asinformslikeDanGurney,andpancake,wherenon-diphthong-ised [æ]occurs inanenvironmentwhich ispre-ŋ at thesurface.Hayescontrasts this with the form pang cake which has /æ/ before anunderlying/ŋ/,which thereforehasapre-ŋ environment in the lexicalphonology, so that PŋD can apply (Donegan & Stampe cite mankind[mæŋkaɪnd]asaformthatshowstheopacityofPŋDforsimilarreasons).(22) pan pang

/pæn/ /pæŋ/ [pæn] [pæɪŋ]

fan fang /fæn/ /fæŋ/ [fæn] [fæɪŋ]

gander anger /ɡændɚ/ /æŋɡɚ/ [ɡændɚ] [æɪŋɡɚ]

DanGurney(thenameofanAmericanracingdriver) /dænɡɝːni/ [dæŋɡɝːni]

pancake pangcake‘cakeeatentoassuagepangsofhunger’ /pænkeɪk/ /pæŋkeɪk/ [pæŋkeɪk] [pæɪŋkeɪk]


This situation is not like the case in (18) from the dialect that

Gimson(1962) isdescribing(which is likeotherBritishdialectsthat Iamawareof),orliketheMid-Atlanticdialectsdiscussedabove.DialectswithPŋDmusthave lost *RIME-xxŋ inorder for forms like [pæɪŋ]and[æɪŋɡɚ] to be possible. It is not surprising if different dialects havedifferentphonotactics,butthisrecognitionopensaninterestingavenueforinvestigationandmakespredictionsabouthowotheraspectsofPŋDdialectswillbehave(forexample,regardingloanwords).

WewouldexpectthatmostotheraspectsofthephonologyofPŋDdialectsandnon-PŋDdialectswillbethesame,however,andPŋDthusprovidesinterestingevidenceintheargumentaboutwhether/ŋ/reallyexists in English (that is, aboutwhether or not all occurrences of [ŋ]derive from sequences of /nɡ/ or /nk/, as discussed in section 3.1,following the analysis of the xxŋ gap in classical generative work,including Jensen1993). Itmustbe thecase thatwords likepanghaveunderlying /ŋ/ in order to trigger PŋD in the lexical phonology28because the relevant case ofNPA is post-lexical— thismeans that atleastsomeoccurrencesofsurface[ŋ]mustderivefromunderlying/ŋ/(while others, as in pancake, derive from underlying /n/, which hasassimilatedtoafollowingvelarthroughNPA).Ifsomecaseof[ŋ]whereno [k]or [ɡ] follows clearlyderive from/ŋ/, there is every reason toassume that all cases of [ŋ] (where no [k] or [ɡ] follows in anymorphological formofaword)areunderlying.Aspromisedinsection3.1, there is therefore firm reason to believe that at least someoccurrencesof/ŋ/mustexistinunderlyingrepresentations,andthis,inturn,canbeseenasevidenceforthenecessityofa*RIME-xxŋconstraintinEnglishandagainstrelyingonageneralconstraintenforcingarimaltrimoraicmaximumtoruleoutformslike[baɪŋ]and[biːŋ](innon-PŋDdialects).

One final issue related to the interaction of *RIME-xxŋ and ash-tensing is that the phonotactic may, in fact (despite the above),currentlybeintheprocessofbeinglostinMid-Atlanticdialectssuchasthose described in section 3.3. Mid-Atlantic ash-tensing classicallypatterns as described in section 3.3, with its patterning showing arobustsignatureof*RIME-xxŋ,butotherdialects(suchasthoseinNewEngland)havea simplerpatternofash-tensing, typicallydescribed asthe‘nasalpattern’(forexample,inLabov1994)orthe‘nasalsystem’(asinLabovetal2016).Thisisoftendescribedastensing/æ/infrontofall (andonly)syllable-finalnasals(andso isobviouslyrelevant toourconcerns: thismeans in front of /m, nand ŋ/); however, Labov et al 28 Unless we allow for the unappealing prospect of extrinsic ordering to make thederivationworkinabrute-forcemanner,withŋ-creationorderedbeforePŋD.


(2016, 277) write that “[i]n earlier characterizations of the nasalsystem,vowelsbeforevelarnasalswereoftennot included.Friedman(2014) summarized the lax tendency of velar nasals in a range ofEasternandMidlanddialects.”Thisimpliesthattheclassicnasalsystemofash-tensingisinfactconstrainedby*RIME-xxŋ.ThepointofLabovetal(2016)isthatthenasalsystemisspreadingandisnowalsofoundinMid-Atlanticareas,competingwith,andlikelyreplacingtheclassicMid-Atlantic‘split-a’systems.Labovetalcontinuethat“recentstudiesoftheexpansion of the nasal system show tensing velar nasals playing aprominent role. Becker & Wong (2009, 15) reported that youngerWhiteNewYorkers showedno significantdifference in theF1of /æ/beforevelarandothernasals.Eckert(personalcommunication)foundconsistent tensing of short-a before velar nasals among youngCalifornians.”Thisflagsupthespreadofthenasalsystem,andimplies(althoughLabovetaldonotputitlikethis)that*RIME-xxŋisbeinglostin these varieties: if pre-ŋ tensing is included as part of the new(propagating) pattern, it seems that it is accompanied by a loss of*RIME-xxŋ.Itisnotclearifthepatterningofthediphthongisationof/æ/inPŋDdialectswasaninfluenceinitslossinthisgeneralisationofthenasalsystem,butthismaybeworthconsidering.

Theoverallconclusionofthissectionisthat*RIME-xxŋisrobustinthelexicalphonologyinmostvarietiesofEnglish,includingtheBritishvarietiesdiscussed,traditionalMid-Atlanticvarieties,andthevarietieswith the traditional nasal-system of ash-tensing. There is goodevidence, however, that it has been lost in some other Americanvarieties and perhaps also that it is being lost in others. A potentialimplicationof this change is that speakersof varieties like thiswouldmore readily accept forms like [baɪŋ], [biːŋ] and [baʊŋ] as ‘wordlike’thanspeakersofothervarieties.29Indeed,theform[baɪŋ],orsomethingverysimilar,isfoundinthesedialects,asflaggedupinfootnote2.

4 Prophylaxisortherapy?Acquisitionismorlifespanchange?The key point of this paper is to extend and interrogate the positionadvocated in Honeybone (2019) — that phonotactic constraints arepsychologically real phonological entities (namely: constraints onoutput-likeforms),whichhaveadiachronyoftheirown,andwhichcanalsointerferewithdiachronicsegmentalchangebyinhibitingotherwiseregularinnovations.Ihopetohaveshownallofthisinsection3forthephonotactic*RIME-xxŋ:thereisevidencethatwecanusetounderstanditsinnovationintoEnglish,itseffectininhibitingthepatterningofash-

29IthankJoeFruehwaldfordiscussionoftheseissues.


tensing, and its loss in certain varieties. The story is different indifferent dialects, as we would expect (and I not told the history of*RIME-xxŋinalldialectsofEnglish).Inthisfinalsectionofthisarticle,IturntosomeimplicationsofamajorpointofHoneybone(2019)andofthispaper:ifitreallyisthecasethatphonotacticscaninhibitsegmentalchange in theway that I discuss, then there are resonanceswith twocentralissuesinhistoricalphonology.

One of these issues is the fact that we might query whether weshould expect that already-existing phonotactics are able to interactwithchangeswhichareotherwiseindependentofphonotactics,inlightoftheimpetusthatisoftenrecognisedinhistoricallinguisticstorejectprophylaxis in change. As mentioned above, there are resonancesbetweenthepositiononphonotactically-inhibitedchangeadoptedhereandthenotionofprophylaxisindiachrony:thetwocasesconsideredinHoneybone (2019), and the further case considered here involve alanguageavoidingenactingachangeiftodosowouldhaveviolatedanalready-existingphonotactic.Thiscouldbeseenaslanguagesavoidingchange in order to prevent a ‘problem’ of some sort (such as aconstraintviolation).

There is much resistance to the notion that languages engage inprophylaxis in change.Fertig (2013), forexample, explains that “[t]heslogan‘languagespracticetherapy,notprophylaxis’iswidelyattributedtoLightfoot(1979,123)butasHarris&Campbell(1995,28)pointout,thewordingisactuallyseveralyearsolderandtheideagoesbackmuchfurtherthanthat.”Lass(1997),foranotherexample,isscepticalofbothprophylaxis and therapy as explanatory concepts in historicallinguistics. The concepts involved in discussions of diachronicprophylaxis areoften linked to the interactionof analogy and regularchange,buttheyareapplicablewaybeyondthis.Forexample,Kiparsky(1974,190)arguesthat“languagepractises‘therapy’(Gilliéron)ratherthan prophylaxis”, as part of a rejection of the idea that phonologicalchange can be inhibited if it would create homophony (which heattributes to Martinet). Kiparsky’s reference is to Gilliéron (1915-1921),whoexplicitlyarguesthatphonologicalchangeregularlycreateshomophones, and that this is followed by languages engaging in‘therapy’, in order to remove the homophony. This is one example ofdiscussionofthenotionmuchearlierthanLightfoot,andshowsthattheidea is applicable in principle to any way of preventing a regularphonologicalchange.

Fertig (2013) goes on to consider how neogrammarians such asOsthoff (1879) and Paul (1886) rejected the idea that phonologicalchangecanbepreventedfromoccurringintheenvironmentswhereitisexpected.This isall relevant tomyconcernsherebecausewhatwe


expectitthatchangeswilloccurinenvironmentsthatcanbedefinedbysimplephonologicalprinciples:weexpectphonologicalconditioningtoinvolve natural classes of segments, not complex segmentalconditioning (as is shown to be the case in section 3); if a syncopeaffects final unstressed vowels, this is a prosodically-definedenvironment,andwedonotexpectcomplexsegmentalconditioning,aswouldbenecessarytopreventit fromoccurringbetweensibilants(asinthecaseofchangediscussedbrieflyinsection2andinmoredetailinHoneybone (2019); furthermore, if a lenition occurs in ‘strong’environments,wewouldalsoexpectthatitshouldalsoapplyin‘weak’environments,yetthecaseofMid-Scotsθ-debuccalisationdescribedininHoneybone(2019)hasexactly thispatterning.These threechangeshavetheseunexpectedconditionings—inasensetheregularityofthechangesisdisturbed—andthemodelsetoutinHoneybone(2019)anddeveloped in this paper aims to explain how the phonotactics of theinnovating language in some sense prevented the changes fromapplyinginparticularenvironments,causingthisoddpatterning.

Is this prophylaxis? Prophylaxis is indeed highly controversial inhistoricallinguistics.Weshouldbecautiousofinvokingamodelwhichadopts it — of claiming that phonotactics can inhibit a phonologicalchange from occurring in a subset of the environments where it isexpected.Achangeisnotawareofitseffects,socannotbeexpectedtobeinhibitedifitseffectsmightpresentsomeproblemforalanguage.

The framework that I propose tomodel cases of phonotactically-inhibited change, however, avoids the conceptual problems that areconnectedwith this.Central tounderstanding this is torecognise thatthe changes under consideration involve the introduction orgeneralisation of phonological rules. At the relevant level of thephonological derivation where the rules apply, they are in fact notinhibited anddo affect the relevant phonological forms. In this sense,thechangesthemselves(theruleadditionsorgeneralisations)arenotinhibited — there is no prophylaxis. Forms which violate thephonotacticsarederivedinthederivationsin(9)and(17).Thepointisthat the phonotactic constraints apply later in a derivation afterphonotactic-violating forms have been derived, and these forms areruled ungrammatical at the end of the relevant phonological stratum.The phonotactics prevent the forms that violate them from surfacing,buttherulewhichisinnovated,orthegeneralisationoftherulewhichalready exists does not know anything about this. The innovations intherulesortheirenvironmentsinthesecasesarestraightforward,andare as would be expected in line with the expectation that naturalclassesshouldbe involved inrules,or that lenitionsshouldpattern intermsofthe‘strongimpliesweak’predictionsthatformlenitiontheory.


The phonotactics that are important in the three cases consideredeither here or inHoneybone (2019)— *RIME-xxŋ, OCP(SIBILANCE) and*CODA-h—donotpreventtherulesfromapplying(thiswouldbetrueprophylaxis), although they do stop ungrammatical forms fromescapingfromthegrammar.

Theotherissuethatisraisedbythismodel—thefinalpointthatIdiscuss—isthefactthatthemodelassumesthatthephonotacticswerealreadyexistinginthepre-changestateofthelanguagesinvolvedwhentheyaffectedthesegmentalchanges.Thephonotacticsconsideredhereareclearlylanguage-specific:whilesomeotherlanguagesmayalsouseOCP(SIBILANCE), *CODA-h and *RIME-xxŋ, others do not, so the fact thattheywereactive inpre-changelateMiddleEnglishandMid-Scots,andin Mid-Atlantic varieties before ash-tensing generalised, were factsabout thematurephonologyof these varieties.Thismeans that theseare cases where pre-existing language-specific structure in aninnovatinglanguageaffectedthe(surface)patterningofchangeswhentheywereinnovated.

This issurprising forcertainmodelsofphonologicalchange. Ifweconsiderthequestionwhatisthelocusofphonologicalchange?,30oneofthe main responses is ‘in acquisition’. This is standardly assumed inmuch current generative historical linguistics — for example, vanKemenade (2007, 158)writes that “[e]ver since Lightfoot (1979), thegenerative approach to syntactic change has considered that the keymechanism of change is reanalysis, essentially the language learner’sattributionofanovelunderlinganalysistothesamesurfaceform”,andYang (2000, 9) writes that “[l]anguage change is observed when agenerationofspeakersproduceslinguisticexpressionsthatdifferfromthose of previous generations”. This position that first languageacquisition is a key locus for change has long been recognised as apossibility in historical phonology (it is found in Paul 1886 andKiparsky1965),andthecertaintythatchangeonlyoccursinacquisitionhasmoved fromgenerativehistoricalsyntax intohistoricalphonology—forexample,Hale(2003,345)writesthat“webelievethat‘change’isto be conceived of as the set of differences between the grammargenerating the primary linguistic data (PLD) usedby an acquirer andthegrammarultimatelyconstructedbythatacquirer”andReiss(2003,143) writes that “reranking or rule loss/addition cannot be definedwithinasinglegrammar,butis,atbest,adescriptionoftherelationshipbetweengrammars”,neatlyexpressingtheideathatchangeonlyoccurs‘across speakers’, during acquisition. This position — the all changeoccursinacquisition—canbecalled‘acquisitionism’(asinHoneybone 30 Honeybone & Salmons (2015) suggest that this is one of the key questions forhistoricalphonology(formulatedas‘wheredoeschangeoccur?’).


2006 and Honeybone & Salmons 2015). It is also the most obviousinterpretationofOhala’sinfluentialposition(e.g.,1981,1993)thatthekey locus of change is in the hearer, due to acoustic confusability ormisperception:itisimplausibletoassumethatanadultwouldsuddenlymakeareanalysis(thatis,amistake)onmisperceivingsomethingwhentheyhaveasteadystatewithwhichtheycancomparewhattheyhear— learners engaged in First Language Acquisition, however, lack agrammaragainstwhich to checka confusableutterance, andare thusthose who are open to the type of reanalysis that is crucial on thispicture.

Strict acquisitionism is thus strong in some areas of historicalphonology.It ischallenged,however,bythosewhobelievein ‘lifespanchange’,suchasSankoff&Blondeau(2007)andBowie&Yaeger-Dror(2015),whereatleastcertaintypesofchangearearguedtobepossiblepost-acquisition. Such work can be described as ‘anti-acquisitionist’.Aspects of anti-acquisitionism are, in fact, also present in the earliestgenerativehistoricalwork,suchasHalle(1962,64),whoassumesthat“changes in later life are restricted to theadditionor eliminationof afewrulesinthegrammar”,thusacceptingthatpost-acquisitionchangeispossible—whilealsoassuming(1962,65) that “it isnotnecessarythat the child and his parents have identical grammars”, and thatchange can thusoccur across generations, too.Halle continues (1962,64), that “a wholesale restructuring of [...] grammar is beyond thecapabilities of the average adult”, making clear that he assumes thatonly certain types of change are possible post-acquisition. We candescribethepositionseenhereas‘softanti-acquistionism’,thatis,theassumptionthatchangecanoccurboth inacquisitionand in thepost-acquisition period.We can also recognise ‘strong anti-acquisitionism’,that is, the assumption that no change happens in first languageacquisition,which Foulkes & Vihman (2015), for example, tentativelyadvocate.

The model that I propose for understanding phonotactically-inhibited change is most straightforwardly compatible with an anti-acquisitionistposition:ifallchangeoccursinfirstlanguageacquisition— across generations — then language-specific structure in theinnovating language should not be able to inhibit a change, as theinnovatingchildrendonotyethaveit.Theproposalsinsection2.3and3.3 assume that the phonotactics involved are already active in thelanguage at the point at which the changes are innovated orgeneralised. For this to be possible, the changes must be added tospeakers’ phonology after the initial stage of language acquisition


(whichfixedthephonotacticconstraints).31Itstrikesmethatsoftanti-acquisitionism is likely right: there is some lifespan change, perhapsonlyinvolvingtheadditionofaruleorsimilar‘small’changes,asHalle(1962) assumes, but more fundamental reanalyses (involvingunderlyingrepresentations,forexample)occurinacquisition.32

5 ConclusionThis paper has considered quite a lot. I hope to have shown howcomplex and interesting questions in diachronic phonotactics can be,andalsothatwecanandshouldpursuethem.Certainchangesarebestviewedasphonotacticchange—astheadditionorlossorchangeofanalready existing phonotactic — and other changes involve bothphonotactic change and segmental or prosodic change occurringtogether; still other changes may involve phonotactic change beingcaused by other types of change. And we can clearly see thatphonotacticscanplayaroleininhibitingthefullsurfacepatterningthatcomes in the wake of a segmental change, although this does notinvolveprophylaxis,narrowlydefined.Thislatterpointbecomesclearifseparate out the effect on the patterning of a change of the rulecomponent of the grammar and on the constraint component of thegrammar,andifwearecleartoconsiderwhetherlexicalorpost-lexicalphonology is involved. The model that I propose to understandphonotactically-inhibitedchangeistiedtovariabilityinphonology,andneedstoinvolvecarefulconsiderationoftheacquisitionofphonologicalentitiesbylearners—itcanbeseenasanargumentfor(atleastsoft)anti-acquistionism in historical phonology. The ideas discussed hereareintendedtobeprovocative,butIhopetheyarealsointriguing.

31 This conclusion follows only if the acquisition of the precise and detailedphonotacticsofthetypeconsideredhereproceedsatfundamentallythesamepaceasotheraspectsofphonologicalacquisition.This isanempiricalquestion(andthere isevidence that some aspects of phonotactic knowledge are acquired early, as acommentonaversionofthispaperpointsout).Analternativeconclusion(ifdetailedphonotactic learningoccursbefore the learningof rule-likeaspectsofphonology) isthat—ifacquisitionismiscorrect—thenonlyphonotacticknowledgeshouldbeabletoinhibitinnovationsinthefashiondiscussedinthisarticle.Iamcautiousaboutthis:doesearlyphonotacticlearningreallyinvolvesuchdetailedandspecificphonotacticsas those considered here, which require a knowledge of the natural classes of alanguage?32ThisisalsoDresher’s(2015,515)position,forexample:“withoutclaimingthatalllanguagechangeoriginatesinacquisition,itappearsunavoidablethatcertaintypesofchangesdo.”ItisnotablealsothatFruehwald(2017),afteradetailedinvestigationofdataforseveralchangesfromseveraldecades,findslifespanchangeinonlyaminorityofthechangesinvolved:itseemsthatlifespanchangeispossible,butisnotpartofallchanges.


CommentsinvitedPiHPhreliesonpost-publicationreviewofthepapersthatitpublishes.If you have any comments on this piece, please add them to itscommentssite.Youareencouragedtoconsultthissiteafterreadingthepaper,astheremaybecommentsfromotherreadersthere,andrepliesfromtheauthor.Thispaper’ssiteishere:

http://dx.doi.org/10.2218/pihph.4.2019.4192

AcknowledgmentsEarlyversionsofsomeoftheideasinthispaperwerepresentedattheDiachronic Phonotactics Workshop at the University of Vienna inSeptember2017,andthenreviewedforpublicationinFoliaLinguisticaHistorica,leadingtoHoneybone(2019).AversionofsomeoftheotherideaswaspresentedattheP-WorkshopattheUniversityofEdinburghin 2019. I am grateful for all the comments and ideas that I got fromthis, especially from Niki Ritt, and I am particularly grateful to BenMolineaux, who read through the paper as an editor for PiHPh, andwhose comments led to many improvements in it. I alone amresponsibleforitscontents,ofcourse,however.

AuthorcontactdetailsPatrickHoneyboneLinguisticsandEnglishLanguageUniversityofEdinburghEdinburghEH89AD,UK.

[email protected]

ReferencesAbercrombie,David.1967.Elementsofgeneralphonetics.Edinburgh:

EdinburghUniversityPress.Albright,Adam.Inpreparation.Naturalclassesarenotenough:Biased

generalization in novel onset clusters. Available here:http://web.mit.edu/albright/www/

Albright,Adam.2006.Gradientphonotacticeffects:lexical?grammatical?both?neither?TalkpresentedatthemeetingoftheLinguisticSocietyofAmerica,Albuquerque.Availablehere:


http://www.mit.edu/~albright/papers/Albright-GradientPhonotactics-LSA2006.pdf

Algeo,John.1975.SyncopeandthephonotacticsofEnglish.GeneralLinguistics15.71–78.

Algeo,John.1978.WhatConsonantClustersArePossible?Word29.206–224

Bailey, George. 2018.Variation and change in Northern English velarnasals: Production and perception. PhD thesis, University ofManchester.

Baldwin, Thomas. 1846. A Universal Pronouncing Gazetteer. Thirdedition.Philadelphia:Lindsay&Blakiston.

Bauer,Laurie.2015.Englishphonotactics.EnglishLanguageandLinguistics19.437–475.

Baumann,Andreas,NikolausRitt&ChristinaPrömer.2016.DiachronicdynamicsofMiddleEnglishphonotacticsprovideevidenceforanalogyeffectsamonglexicalandmorphonotacticconsonantclusters.PapersinHistoricalPhonology1.50–75.

Becker,Kara&AmyWong.2009.Theshort-asystemofNewYorkCityEnglish: an update. University of Pennsylvania Working Papers inLinguistics1(2).9–20.

Berent, Iris. 2013. The phonological mind. Cambridge: CambridgeUniversityPress.

Bermúdez-Otero, Ricardo. 1998. Prosodic optimization: the MiddleEnglishlengthadjustment.EnglishLanguageandLinguistics2.169–197.

Bermúdez-Otero,Ricardo.2003.Theacquisitionofphonologicalopacity.InJenniferSpenader,AndersEriksson&ÖstenDahl(eds.),VariationwithinOptimalityTheory:ProceedingsoftheStockholmWorkshopon‘VariationwithinOptimalityTheory’,25-36.Stockholm:DepartmentofLinguistics,StockholmUniversity.

Bermúdez-Otero, Ricardo. 2011. Cyclicity. In Marc van Oostendorp,Colin J. Ewen, Elizabeth Hume & Keren Rice (eds),The BlackwellCompanion to Phonology, volume 4, 2019–2048. Malden, Mass.:Wiley-Blackwell.

Bermúdez-Otero,Ricardo.2015.Amphichronicexplanationandthelifecycleofphonologicalprocesses. InPatrickHoneybone&JosephC.Salmons (eds.),TheOxfordhandbookofhistoricalphonology, 374–399.Oxford:OxfordUniversityPress.

Bloomfield,Leonard.1933.Language.NewYork:Holt.Bosworth, Joseph & Thomas Toller. 1898-1921. An Anglo-Saxon

Dictionary. With supplement. Oxford: Clarendon Press.Electronicversion:http://bosworthtoller.com/


Bowie,David&MalcahYaeger-Dror.2015.Phonologicalchangeinrealtime. In Patrick Honeybone & Joseph Salmons (eds.)The Oxfordhandbook of historical phonology, 603-618. Oxford: OxfordUniversityPress.

Carr, Philip, Jacques Durand & Colin Ewen (eds). 2005. Headhood,Elements, Specification and Contrastivity. Amsterdam: JohnBenjamins.

Chomsky,Noam&MorrisHalle.1965.Somecontroversialquestionsinphonologicaltheory.JournalofLinguistics1(2).97–138.

Chomsky,Noam&MorrisHalle.1968.ThesoundpatternofEnglish.NewYork:HarperandRow.

Dewan,Ted.2003.Bing:getdressed.Oxford:DavidFicklingBooks.Dobson, E.J. 1968. English pronunciation 1500-1700. Second edition.

Twovolumes.Oxford:Clarendon.Donegan, Patricia & David Stampe. 1979. The study of natural

phonology. In Daniel Dinnsen (ed.), Current Approaches toPhonological Theory, 126–173. Bloomington: Indiana UniversityPress

Dresher,B.Elan.2015.Rule-basedgenerativehistoricalphonology. InPatrickHoneybone&JosephSalmons(eds.)TheOxfordhandbookofthehistoricalphonology.Oxford:OxfordUniversityPress.

Dziubalska-Kołaczyk,Katarzyna.2005.PhonotacticsofconsonantclustersinthehistoryofEnglish.InBertacca,Antonio(ed.),HistoricallinguisticstudiesofspokenEnglish,15-34.Pisa:PLUS–PisanaUniversityPress.

Ferguson,Charles.1972.‘Shorta’inPhiladelphiaEnglish.InM.EstellieSmith(ed.),StudiesinlinguisticsinhonorofGeorgeL.Trager,259-74.TheHague:Mouton.

Fertig, David. 2013. Analogy and morphological change. Edinburgh:EdinburghUniversityPress.

Flanders,Michael&DonaldSwann.1957.Agnu.London:Parlophone.Foulkes, Paul&MarilynVihman. 2015. First language acquisition and

phonological change. In Patrick Honeybone & Joseph Salmons(eds.) The Oxford handbook of historical phonology, 289–312.Oxford:OxfordUniversityPress.

Fruehwald, Josef. 2017. Generations, lifespans, and the zeitgeist.LanguageVariationandChange29(1).1–27

Fudge,Erik.1969.Syllables.JournalofLinguistics5.253–286.Garrett,Andrew&JulietteBlevins.2009.Analogicalmorphophonology.

InKristinHanson&Sharon Inkelas (eds),Thenatureof theword:essays in honor of Paul Kiparsky, 527–45. Cambridge, Mass.: MITPress.


Giegerich,Heinz.1992.Englishphonology:an introduction.Cambridge:CambridgeUniversityPress.

Gilliéron, Jules. 1915–1921.Pathologie et thérapeutique verbales. Fourvolumes.Variouspublishers,includingParis:Champion.

Gimson, A.C. 1962. An introduction to the pronunciation of English.London:EdwardArnold.

Goldsmith,John(ed.).1993.Phonologicaltheory:theessentialreadings.Malden&Oxford:Blackwell.

Gorman,Lyle.2013.Generativephonotactics.PhDdissertation,UniversityofPennsylvania.

Hale,Mark.2003.Neogrammariansoundchange.InBrianJoseph&RichardJanda(eds.),Thehandbookofhistoricallinguistics,343–368.Malden:Blackwell.

Halle,Morris.1959.ThesoundpatternofRussian.TheHague:Moulton.Halle,Morris.1962.Phonologyingenerativegrammar.Word18.54–72.Hammond,Michael.1999.ThephonologyofEnglish:aprosodic

Optimality-Theoreticapproach.Oxford:OxfordUniversityPress.Harms,Robert.1968.Introductiontophonologicaltheory.Englewood

Cliffs:Prentice-Hall.Harris,John.1989.Towardsalexicalanalysisofsoundchangein

progress.JournalofLinguistics25.35–56.Harris,Alice&LyleCampbell.1995.Historicalsyntaxincross-linguistic

perspective.Cambridge:CambridgeUniversityPress.Hayes, Bruce. 2009. Introductory phonology. Malden& Oxford:Wiley-

Blackwell.Hayes,Bruce&ColinWilson.2008.Amaximumentropymodelof

phonotacticsandphonotacticlearning.LinguisticInquiry39.379–440.

Hill,Archibald.1958.Introductiontolinguisticstructures:fromsoundtosentenceinEnglish.NewYork:Harcourt,Brace.

Hockett,Charles.1947.Problemsofmorphemicanalysis.Language23.321.

Hockett,Charles.1958. Acourseinmodernlinguistics.NewYork:Macmillan.

Hogg, Richard. 1992. A grammar of Old English, vol. 1: Phonology.Oxford:Blackwell.

Honeybone, Patrick. 2001. Lenition inhibition in Liverpool English.EnglishLanguageandLinguistics5.213–249.

Honeybone, Patrick. 2005. Sharing makes us stronger: processinhibitionandsegmentalstructure. InPhilipCarr, JacquesDurand& Colin Ewen (eds.), Headhood, elements, specification andcontrastivity:phonologicalpapersinhonourofJohnAnderson,167–192.Amsterdam:JohnBenjamins.


Honeybone,Patrick.2006.Reviewof:D.ErcHolt(ed.)2003.OptimalityTheoryandLanguageChange.JournalofLinguistics42,726-731.

Honeybone,Patrick.2016.Arethereimpossiblechanges?θ>fbutf≯θ.PapersinHistoricalPhonology1.316–358.

Honeybone, Patrick. 2019. Can phonotactic constraints inhibitsegmental change? Arguments from lenition and syncope.FoliaLinguisticaHistorica40.9–36.

Honeybone,Patrick&JosephSalmons.2015.Introduction:keyquestionsforhistoricalphonology.InPatrickHoneybone&JosephSalmons(eds.),TheOxfordhandbookofhistoricalphonology,3–10.Oxford:OxfordUniversityPress.

Iverson,Gregory&JosephSalmons.2005.Fillingthegap:Englishtensevowelplusfinal/š/.JournalofEnglishLinguistics33.207–221.

Jensen,John.1993.Englishphonology.Amsterdam:JohnBenjamins.King, Robert D. 1969. Historical linguistics and generative grammar.

EnglewoodCliffs,NJ:Prentice-Hall.Kiparsky,Paul.1965.Phonologicalchange.PhDthesis,MIT.Kiparsky, Paul. 1968. Linguistic universals and linguistic change. In

Bach,Emmon&RobertHarms(eds.),Universalsinlinguistictheory,171–202.NewYork:Holt,RinehartandWinston.

Kiparsky,Paul.1971.Historicallinguistics.InWilliamO.Dingwall(ed.),A survey of linguistic science, 576-649. College Park: University ofMarylandLinguisticsProgram.

Kiparsky,Paul.1974.Ontheevaluationmeasure.InAnthonyBruck,RobertFox&MichaelLaGaly(eds.),Proceedingsoftheparasessiononnaturalphonology,tenthregionalmeeting,ChicagoLinguisticSociety,328-37.Chicago:ChicagoLinguisticSociety.

Kiparsky, Paul. 1982. Lexical morphology and phonology. In In-SeokYang (ed.), Linguistics in the morning calm: selected papers fromSICOL–1981,vol.1,3–91.Seoul:Hanshin.

Kiparsky, Paul. 1985. Some consequences of Lexical Phonology.PhonologyYearbook2.85–138.

Kiparsky, Paul. 1988. Phonological change. In Frederick Newmeyer(ed.), Linguistics: the Cambridge survey, volume 1, 363–415.Cambridge:CambridgeUniversityPress.

Kiparsky, Paul. 2000.Opacity and cyclicity. InNancyA. Ritter (ed.),Areview of Optimality Theory. Special issue of Linguistic Review17.351–67.

Kroonen,Guus.2013.EtymologicaldictionaryofProto-Germanic. Leiden:Brill.Availablehere:http://dictionaries.brillonline.com/proto-germanic

Kruisinga,Etsko.1943.ThephoneticstructureofEnglishwords.Bern:A.Francke.


Labov, William. 1981. Resolving the Neogrammarian controversy.Language57.267–308.

Labov,William1994.Principlesoflinguisticchange,vol.1:internalfactors.Oxford:Blackwell.

Labov,William,SabriyaFisher,DunaGylfadottír,AnitaHenderson&BetsySneller.2016.CompetingsystemsinPhiladelphiaphonology.LanguageVariationandChange28.273–305.

Lahiri,Aditi&B.ElanDresher.1999.OpensyllablelengtheninginWestGermanic.Language75.678–719.

Lass, Roger. 1974. Linguistic orthogenesis? Scots vowel quantity andthe English length conspiracy. In John Anderson & Charles Jones(eds.), Historical linguistics: proceedings of the First InternationalConferenceonHistoricalLinguistics,Edinburgh,2-7September1973,volume2,311-52.Amsterdam:North-Holland.

Lass,Roger.1992.Phonologyandmorphology. InNormanBlake(ed.),The Cambridge history of the English language, volume 2: 1066-1476,23–155.Cambridge:CambridgeUniversityPress.

Lass, Roger. 1997. Historical linguistics and language change.Cambridge:CambridgeUniversityPress.

Lass,Roger.1999.Phonologyandmorphology.InRogerLass(ed.),CambridgeHistoryoftheEnglishLanguage,volume3:1476–1776.Cambridge:CambridgeUniversityPress.

Lentz,Tomas&RenéKager.2015.Categoricalphonotacticknowledgefilterssecondlanguageinput,butprobabilisticphonotacticknowledgecanstillbeacquired.LanguageandSpeech58.387–413.

Lightfoot,David.1979.Principlesofdiachronicsyntax.Cambridge:CambridgeUniversityPress.

Lindsey, Geoff & Péter Szigetvári. 2013. Current British EnglishSearchableTranscriptions.http://cube.elte.hu/.

Luick,Karl.1898.BeitragezurenglischenGrammatik,III:DieQuantitatsveranderungenimLaufederenglischenSprachentwicklung.Anglia20.335–62.

Luick,Karl.1914-1940.HistorischeGrammatikderenglischenSprache.Stuttgart:BernhardTauchnitz.Reprinted1964.Oxford:Blackwell.

Lutz,Angelika.1988.OnthehistoricalphonotacticsofEnglish.InDieterKastovsky & Gero Bauer (eds), Luick revisited:papers read at theLuick-Symposium at Schloss Liechtenstein, 15-18.9.1985, 221-239.Tübingen:GunterNarr.

Lutz, Angelika. 1991. Phonotaktisch gesteuerte Konsonanten-veränderungen in der Geschichte des Englischen. Tübingen: MaxNiemeyerVerlag.

MED=MiddleEnglishDictionary.Originalversion1952–2001.Updatedonline version, revised 2018. Michigan: University of Michigan


Library. Available here: https://quod.lib.umich.edu/m/middle-english-dictionary/dictionary

Minkova,Donka.1982.Theenvironmentforopensyllable lengtheninginMiddleEnglish.FoliaLinguisticaHistorica3(2).29–58.

Minkova, Donka. 2003.Alliteration and sound change in early English.Cambridge:CambridgeUniversityPress.

Minkova, Donka. 2014. A historical phonology of English. Edinburgh:UniversityofEdinburghPress.

Minkova, Donka & Robert Stockwell. 1996. The origins of long-shortallomorphy in English. In Jacek Fisiak & Marcin Krygier (eds),Advances in English historical linguistics 1996, 211–39. Berlin:MoutondeGruyter.

Ohala, John.1981.The listenerasa sourceof soundchange. InCarrieMasek,RobertaHendrick&MaryFrancesMiller(eds.),Papersfromthe parasession on language and behavior, 178–203. Chicago:ChicagoLinguisticSociety.

Ohala, John. 1993. The phonetics of sound change. In Charles Jones(ed.), Historical linguistics: problems and perspectives, 237–278.London:Longman.

Osthoff,Hermann.1879.DasphysiologischeundpsychologischeMomentindersprachlichenFormenbildung.Berlin:Habel.

OED=OxfordEnglishDictionary.Originalversion1884–1928.Supplements1933-1986.OnlineversionconsultedinDecember2019.Oxford:OxfordUniversityPress.Availablehere:http://www.oed.com/

Paul,Hermann. 1886.Prinzipien der Sprachgeschichte. Second edition.Halle:MaxNiemeyer.

Reiss,Charles.2003.Languagechangewithoutconstraintreranking.InD. Eric Holt(ed.), Optimality Theory and Language Change, 143–168.Dordrecht:Kluwer.

Ringe, Donald & Joseph Eska. 2013. Historical linguistics: toward atwenty-first century reintegration. Cambridge: CambridgeUniversityPress.

Ritt, Nikolaus. 1994. Quantity adjustment: vowel lengthening andshortening in Early Middle English. Cambridge: CambridgeUniversityPress.

Robinson,Orrin.1976.A‘scattered’ruleinSwissGerman.Language52.148–62.

Roca,Iggy&WynJohnson.1999.AWorkbookinPhonology.Oxford:Blackwell.

Ross,Erica.2011.Diachronicdevelopmentsinthescopeofphonologicalgeneralisations.PhDthesis,UniversityofCambridge.


Salmons, Joseph. Inpreparation.SoundChange. Edinburgh:EdinburghUniversityPress.

Sankoff,Gillian&HélènBlondeau.2007.Languagechangeacrossthelifespan:/r/inMontrealFrench.Language83.560–88.

Scholes,Robert.1966.Phonotacticgrammaticality.TheHague:Mouton.Schütze,Carson.Toappear.Acceptabilityratingscannotbetakenat

facevalue.InSamuelSchindler,AnnaDrozdzowicz&KarenBrøcker(eds.),Linguisticintuitions.Oxford:OxfordUniversityPress.Availableat:https://ling.auf.net/lingbuzz/004862

Shatzman,Keren&RenéKager.2007.Aroleforphonotacticconstraintsinspeechperception.ICPhCXVI,Saarbrücken,1409–1412.

Singh, Rajendra. 1987. Well-formedness conditions and phonologicaltheory.InWolfgangDressleretal.(ed.),Phonologica1984,273-285.Cambridge,CambridgeUniversityPress.

Sneller, Betsy. 2018. Mechanisms of Phonological Change. PhDdissertation,UniversityofPennsylvania.

Sneller,Betsy.2019.Allophonicsystemsasavariablewithinindividualspeakers. In David Lightoot (ed.),Variable properties in language:their nature and acquisition. Washington, DC: GeorgetownUniversityPress.

Sommerstein,Alan.1974.Onphonotacticallymotivatedrules.JournalofLinguistics10.71–94.

Spence,Thomas.1775.ThegrandrepositoryoftheEnglishLanguage.NewcastleUponTyne:printedbyT.Saint,fortheauthor,andsoldbyhimathisschoolontheKeyside,andbyallthebooksellersintownandcountry.

Sprouse,Jon.inpress.Acceptabilityjudgmentmethods.In:JonSprouse(ed.),TheOxfordHandbookofExperimentalSyntax.OxfordUniversityPress.

Stanley,Richard.1967.Redundancyrulesinphonology.Language43.393–436.

Steriade,Donca.1999.Alternativestothesyllabicinterpretationofconsonantalphonotactics.InOsamuFujimura,BrianJoseph&BohumilPalek(eds.),Proceedingsofthe1998LinguisticsandPhoneticsConference,205-242.Prague:KarolinumPress.

Szigetvári,Péter.2007.BranchingonsetsandsyncopeinEnglish.LanguageSciences29.408–25.

Taylor,John.2002.Cognitivegrammar.Oxford:OxfordUniversityPressUffmann,Christian.2015.Loanwordadaptation.InPatrickHoneybone

&JosephSalmons(eds.),TheOxfordHandbookofHistoricalPhonology,644-665.Oxford:OxfordUniversityPress.

van Kemenade, Ans. 2007. Formal syntax and language change:developmentsandoutlook.Diachronica24.155–169


Walker, John. 1791.A critical pronouncing dictionary and expositor ofthe English Language. London: sold by G. G. J. and J. Robinson,PaternosterRow;andT.Cadell,intheStrand.

Wells, John C. 1982.Accents of English. 3 vols. Cambridge: CambridgeUniversityPress.

William Labov, Sabriya Fisher, Duna Gylfadottir, Anita Henderson&Betsy Sneller. 2016. Competing systems in Philadelphiaphonology.LanguageVariationandChange28(3).273–305.

Yang,Charles.2005.Onproductivity.Linguisticvariationyearbook5(1).265–302.

Yang, Charles. 2016. The price of productivity. Cambridge, Mass.: MITpress.

Documents

Papers in Historical Phonology · Phonotactics plays a rather minor role in historical phonology. Most diachronic work deals with segmental change, or with (higher-level) prosodic