Linguistics Consonant Epenthesis
by
Rebecca Morley
  • LAST MODIFIED: 27 April 2017
  • DOI: 10.1093/obo/9780199772810-0158

Introduction

Use of the term epenthesis implies an input-output mapping relationship in which the output contains more segmental material than the input. Other terms that are often used synonymously with epenthesis include “insertion,” “intrusion,” and “linking,” although the latter two may also be used to refer only to certain specific kinds of epenthesis. Epenthesis may occur in a variety of environments: intervocalically, interconsonantally, word or syllable initially, and word or syllable finally. The term epenthesis may also be used to refer to the addition of segmental material to satisfy a morphological template, or minimal word length requirement. Theoretically, epenthesis may occur as the result of a phonological, morphological, or phonetic rule. Within generative theory, epenthesis is “triggered” or “conditioned” by the presence of specific environments. Such environments may consist of sequences that are disallowed or dispreferred within the language (*XY), and that are prevented from surfacing by the operation of epenthesis (/XY/ surfaces as [XBY]). Agreement does not always hold regarding whether a given set of surface forms results from a process of epenthesis or from a process of deletion in complementary environments (/XBZ/ surfaces as [XZ]). Competing claims also are made about whether the set of universally possible epenthetic segments is restricted in any way; whether those restrictions might derive from historical, perceptual, articulatory, or grammatical forces; and which segments comprise the (potentially) restricted set.

Theoretical Overview

To describe epenthesis within a given language, it is necessary to adopt a particular theoretical framework; most fundamentally: is epenthesis taken to be a rewrite rule, as exemplified in Chomsky and Halle 1968, or the result of the interaction of a set of ranked constraints, as proposed in Prince and Smolensky 2004? In the notation of a rule-based system: 0 → B /X__Y, where 0 represents the absence of segmental material, and B is the epenthetic segment. Within this framework epenthesis can occur in any environment and involve any segment. Furthermore, a rule of epenthesis may be ordered with respect to other rules in any sequence whatsoever. This means that epenthesized segments may actually fail to surface—if a later rule deletes that segment. The pattern may also be rendered opaque if the original triggering environment is altered by the action of subsequent rules (counter-bleeding); or if the relevant environment surfaces only later, failing to trigger epenthesis (counter-feeding). However, in a theoretical framework lacking derivations, such as optimality theory, it is possible to refer only to surface-true epenthesis. In what follows only apparent cases of surface-true epenthesis will be discussed; this is partially for practical reasons—the burden of proof is higher for cases of “covert” epenthesis—and partially because optimality theory provides a more restrictive prediction about the contexts in which epenthesis can occur, and which segments can epenthesize. Specifically, optimality theory requires that epenthesis occur only if it would result in a surface form that is better than the input. In the terms of the theory, this output would be structurally “less marked” or “more harmonic”: a property that is determined by the language-specific ranking of a set of universal constraints. The epenthesized segment itself is also predictable by the same mechanism.

  • Chomsky, Noam, and Morris Halle. 1968. The sound pattern of English. New York: Harper & Row.

    Save Citation »Export Citation »E-mail Citation »

    A groundbreaking work in phonological theory, focusing specifically on English, but elaborating a framework of serially ordered rules applying to underlying forms and operating on a set of universally available articulation-based sub-segmental features.

    Find this resource:

    • Prince, Alan, and Paul Smolensky. 2004. Optimality theory: Constraint interaction in generative grammar. Malden, MA: Blackwell.

      DOI: 10.1002/9780470759400Save Citation »Export Citation »E-mail Citation »

      The original and foundational work on optimality theory. A full development of the theory of markedness sketched out in Chomsky and Halle 1968. Available as a manuscript since 1993; published in book form in 2004.

      Find this resource:

      Typological Studies

      The description of epenthesis up to this point has been completely general. However, this article will be restricted to discussion of consonant epenthesis exclusively going forward. The following sources provide cross-linguistic data on consonant epenthesis, each couched within a particular theoretical framework. Most of these sources focus on consonant epenthesis pre-, post-, and/or inter- vocalically. Vaux 2001 argues for a rule-based approach to consonant epenthesis based on what the author takes to be the arbitrary nature of the epenthetic consonant cross-linguistically; in other words, the failure of consonant epenthesis to conform to general principles of markedness (see Segmental Markedness). Vaux 2001 takes hypercorrection, or rule inversion, to be a productive source of epenthesis patterns, thus providing an explanation for the fact that the segments that are inserted are often the same segments that delete (see Hypercorrection/Rule Inversion). De Lacy 2006 is a general theoretical work on markedness in optimality theory. The author takes markedness reduction to be the driving force in consonant epenthesis, especially in the selection of the epenthetic consonant. The strong claim is made that only certain segments are possible as “default” epenthetic segments, and only certain of these in onset position (nasals may appear only in coda position). Lombardi 2002 also argues for a restricted set of possible epenthetic segments based on markedness reduction but differs in its interpretation of the data. Zygis 2010, like Vaux 2001, proposes a typology of consonant insertion without reference to markedness. The scope of what the author terms epenthesis is broader than the other works in this section, and she divides her cases into three different types: grammatical, phonetic, and prosodic. No justification of the divisions or the analyses is offered. Morley 2015 provides a reanalysis of languages cited in de Lacy 2006 and Vaux 2001, as well as other sources. The author concludes that the evidence for epenthesis in the majority of cases is insufficient to rule out alternative analyses, such as deletion, or suppletive allomorphy (see Alternative Analyses). The conclusion is that default consonant epenthesis may be much rarer than is usually assumed, and that there is no support for a coronal preference in epenthetic segment.

      • de Lacy, Paul. 2006. Markedness: Reduction and preservation in phonology. Cambridge, UK: Cambridge Univ. Press.

        Save Citation »Export Citation »E-mail Citation »

        Consonant epenthesis is covered in Section 3 of the book. An in-depth analysis of Mabalay Atayal and Axininca Campa are provided as examples of epenthesis. An in-depth analysis of Buriat is provided as an example of a pattern that has been incorrectly analyzed as epenthesis.

        Find this resource:

        • Lombardi, Linda. 2002. Coronal epenthesis and markedness. Phonology 19.2: 219–251.

          DOI: 10.1017/S0952675702004323Save Citation »Export Citation »E-mail Citation »

          An argument for pharyngeal as the universally least marked place of articulation, and of glottal segments, specified as pharyngeal, as the least marked segments. Coronal epenthesis is predicted to occur only in restricted environments. Provides varying amounts of data for twenty-five possible cases of epenthesis.

          Find this resource:

          • Morley, Rebecca L. 2015. Deletion or epenthesis? On the falsifiability of phonological universals. Lingua 154:1–26.

            DOI: 10.1016/j.lingua.2014.11.002Save Citation »Export Citation »E-mail Citation »

            Consonant epenthesis is used largely to motivate a discussion of ambiguity in phonological analysis. It is argued that a formal diagnostic is needed for deciding whether alternation data constitute a process of epenthesis or a process of deletion. Analysis of fifty-three languages; all data are provided in an appendix.

            Find this resource:

            • Vaux, Bert. 5 January 2001. Consonant epenthesis and hypercorrection. Talk presented at the Linguistics Society of America Annual Meeting, Washington, DC.

              Save Citation »Export Citation »E-mail Citation »

              Argues that many epenthetic segments are essentially arbitrary in nature because they result from historical changes and reanalysis. Thus they cannot be captured by universal markedness constraints. Contains a sizable listing of diverse epenthetic segments and the languages in which they are said to occur, although few actual language data are provided.

              Find this resource:

              • Zygis, Marzena. 2010. Typology of consonantal insertions. Zentrum für Allgemeine Sprachwissenschaft Papers in Linguistics 52:111–140.

                Save Citation »Export Citation »E-mail Citation »

                A brief survey of seventy languages (many of which are discussed in more depth in other works cited in this article). The author claims that coronal segments are the most likely in grammatical insertion, and coronals and glottals in prosodic insertion, but offers no explanation for why this should be the case.

                Find this resource:

                Intrusive Stops

                Intrusive stops are a specific kind of epenthesis that occurs inter-consonantally, and in which the epenthetic stop shares certain features with the flanking segments. This phenomenon was well known to philologists, as well as to phoneticians by at least the late 19th century. Grandgent 1896 provides a word list with rates of stop insertion across dialects of American English, attributing the phenomenon to “a lack of simultaneousness in the action of the different organs.” This is remarkably similar to the current view, as espoused in Anderson 1976, Wetzels 1985, and Clements 1987, among others; namely, that intrusive stops derive from the timing, or mis-timing, relationship between adjacent articulatory gestures. This might involve anticipatory movements, such as lowering of the velum in preparation for production of a non-nasal segment; or preservatory movements, such as a lag in opening the mouth after closure. A number of different consonant clusters can undergo this process. Wetzels 1985 proposes a division into two types of consonant intrusions, those into liquid-final clusters and those into clusters that do not end in liquids. The former is claimed to be phonological (i.e., segmental) insertion and, the latter, phonetic insertion. Clements 1987 amends Wetzel’s proposal, and makes a new division into Type A: occurring in clusters of nasals or laterals followed by non-labial voiceless fricatives, as in the pronunciation of the English word “sense” as “sents”; and Type B: occurring in sonorant-final clusters. However, Recasens 2011 reports cases of intrusive stops in additional cluster types containing sibilant fricatives in first or second position. The author also makes no distinction between different types of clusters. Ohala 1997 provides a unified account in terms of the transition between different types of oral cavity constriction for the two consonants. A period of overlap is predicted to lead to a brief stop-like interval at the release of the first consonant, resulting in what the author calls “emergent” stops. Recasens 2011 follows the framework of Ohala 1997, providing a misperception account in which the amount of intraoral pressure, among other aerodynamic cues, affects the probability that listeners will attribute the cues to the presence of an underlying stop segment.

                • Anderson, Stephen R. 1976. Nasal consonants and the internal structure of segments. Language 52.2: 326–344.

                  DOI: 10.2307/412563Save Citation »Export Citation »E-mail Citation »

                  Argues that certain nasal segments must be analyzed as internally structured, namely pre- and post-stopped nasals. This structure is conceptualized as internal timing relationships involving the lowering of the velum. Such structures present a problem for linear phonology in which insertion can occur only between, not within, segments.

                  Find this resource:

                  • Clements, George N. 1987. Phonological feature representation and the description of intrusive stops. In CLS23: Papers from the 23rd annual regional meeting of the Chicago Linguistic Society; Part 1: The general session. Edited by Barbara Need, Eric Schiller, and Anna Bosch, 29–50. Chicago: Chicago Linguistic Society.

                    Save Citation »Export Citation »E-mail Citation »

                    Building directly on the work of Wetzels 1985, the author proposes the following reanalysis: that it is the features on the oral cavity node that spread to the adjacent supralaryngeal tier node, specifically [-cont], which, when combined with the existing [-son] and [-nas], produces the intrusive stop. This analysis accounts for additional consonant cluster types that undergo insertion.

                    Find this resource:

                    • Grandgent, Charles H. 1896. Warmpth. Publications of the Modern Language Association of America. Edited by the Modern Language Association of America, 63–75. New York: Modern Language Association of America.

                      Save Citation »Export Citation »E-mail Citation »

                      Provides early statistics on rates of stop insertion and deletion in nasal+fricative sequences across different dialects of American English. A total of 140 responses were collected by correspondence using written word lists. Speakers were instructed to indicate the variant that corresponded to their own pronunciation in “unstudied” speech.

                      Find this resource:

                      • Ohala, John J. 1997. Emergent stops. In Proceedings of the 4th Seoul international conference on linguistics (SICOL), 11–15 August 1997. 84–91. Seoul: Linguistic Society of Korea.

                        Save Citation »Export Citation »E-mail Citation »

                        Intrusive stops are analyzed as the result of a difference in the openness of the oral cavity between two sequential consonants. Anticipation of the change, resulting in both closures being active at the same time (both “valves” closed), produces the percept of a stop at the point of release of C1’s closure.

                        Find this resource:

                        • Recasens, Daniel. 2011. Articulatory constraints on stop insertion and elision in consonant clusters. Linguistics 49.5: 1137–1162.

                          DOI: 10.1515/ling.2011.031Save Citation »Export Citation »E-mail Citation »

                          An account in line with Ohala 1997 in which intrusive stops emerge as an acoustic consequence of inter-gestural articulatory timing between adjacent consonants. Provides a review of the literature.

                          Find this resource:

                          • Wetzels, W. Leo. 1985. The historical phonology of intrusive stops: A non-linear description. Canadian Journal of Linguistics/Revue Canadienne de Linguistique 30:285–333.

                            Save Citation »Export Citation »E-mail Citation »

                            An autosegmental analysis of diachronic intrusive stops based on the proposal of segment-internal timing relations in Anderson 1976. Type 2 intrusive stops are analyzed as deriving from feature spread of [-son] across the timing tier, without the addition of any new syllable structure, and are thus realized phonetically.

                            Find this resource:

                            Skeletal Slots, Templates, and Reduplicants

                            In the metrical theory of syllabicity exemplified in Levin 1985, epenthesis is characterized as the insertion of a structural element: a skeletal slot on some tier above that of the segment. Vowel and consonant insertion can be combined in this framework by designating them both as X slot insertion, where X is a variable that can take on either a value of X’ = non-head (consonant), or X-bar = head (vowel). This is known as the skeletal tier, and it is also used in analysis of grammatical processes that require specific structural properties. Reduplication of all or part of a morpheme is one such process. The reduplicated, or copied, part of the base form may be altered by the deletion or insertion of segments in order to conform to a particular template. For example, Payne 1981 argues that the reduplicative form in Axininca Campa must always comprise a CVCV sequence. In stems that are too short to provide sufficient material, the CV sequence [ta] can be added to meet this requirement. Templates may also be part of the morphological system of a language more generally. Broselow 1984 gives an example of this in Amharic, where the gerund is specified as a CVCC template, and the infinitive as a CCVC template, to which the root must conform. Segments may be doubled, or inserted, to meet these morphological requirements.

                            • Broselow, Ellen. 1984. Default consonants in Amharic morphology. In Papers from the January 1984 MIT workshop in morphology. Edited by Margaret Spaas and Richard Sproat, 15–32. Cambridge: Massachusetts Institute of Technology.

                              Save Citation »Export Citation »E-mail Citation »

                              An argument for default [t] epenthesis in the templatic morphology of Amharic. Epenthesis occurs in certain Type A geminate verbs in the gerund and infinitive forms. In the former the epenthetic [t] occurs post-consonantally; in the latter, word finally.

                              Find this resource:

                              • Levin, Juliette. 1985. A metrical theory of syllabicity. PhD diss., Massachusetts Institute of Technology.

                                Save Citation »Export Citation »E-mail Citation »

                                An autosegmental framework with the addition of a tier of timing slots, collectively referred to as the skeleton. Phonological processes involving syllabic constituents operate on these abstract X slots, and segmental material is attached via a later phonological rule. Chapter 4.2.1 provides a brief description of epenthesis within the theory.

                                Find this resource:

                                • Payne, David L. 1981. The phonology and morphology of Axininca Campa. Summer Institute of Linguistics Publications in Linguistics 66. Dallas: Univ. of Texas at Arlington.

                                  Save Citation »Export Citation »E-mail Citation »

                                  A reference work on the grammar of Axininca Campa. Contains several transcribed narratives, with both phonological and morphological glosses. Contains data on reduplication and [t] epenthesis across morpheme boundaries, as well as morphophonological processes generally.

                                  Find this resource:

                                  Prosodic Epenthesis

                                  In this section a differentiation is made between insertion to satisfy (a potentially arbitrary) skeletal structure requirement and insertion to satisfy a universal prosodic preference. Such preferences include the proposal in Jakobson 1962, expanded in Clements and Keyser 1983, for CV (consonant-vowel; onset-nucleus) as the universally preferred (least marked) syllable structure. In Itô 1989 a theory of prosodic epenthesis is developed in which consonants can be inserted just in case a syllable lacks an onset. Consonants can also be inserted to provide a moraic coda, producing a somewhat less preferred CVC syllable structure but a more preferred foot structure: either a sequence of a heavy and light syllable (HL), a sequence of a light and heavy syllable (LH), or a single heavy syllable (H). McCarthy and Prince 1994 and Parker 1994 are analyses of coda epenthesis in terms of foot-structure requirements.

                                  • Clements, George N., and Samuel J. Keyser. 1983. CV phonology: A generative theory of the syllable. Cambridge: Massachusetts Institute of Technology.

                                    Save Citation »Export Citation »E-mail Citation »

                                    A proposal for a CV level representation, or tier, directly below the syllable level, and directly above the segmental level. This level distinguishes between syllable peaks (V) and syllable margins (C). This tier is motivated by a number of phonological processes that can be explained by reference to syllables or syllable constituents.

                                    Find this resource:

                                    • Itô, Junko. 1989. A prosodic theory of epenthesis. Natural Language & Linguistic Theory 7.2: 217–259.

                                      DOI: 10.1007/BF00138077Save Citation »Export Citation »E-mail Citation »

                                      Pre-OT analysis of epenthesis as motivated by prosodic licensing: segments must be parsed into syllables; syllables must be parsed into feet; feet must be parsed into phonological words or phrases. Language-specific requirements on the form of syllables drives epenthesis (and other processes).

                                      Find this resource:

                                      • Jakobson, Roman. 1962. Selected writings. Vol. 1, Phonological studies. The Hague: Mouton.

                                        Save Citation »Export Citation »E-mail Citation »

                                        Foundational work in phonology by one of the early proponents of a theory of markedness in language. Basis for the core syllable typology of Clements and Keyser 1983.

                                        Find this resource:

                                        • McCarthy, John J., and Alan Prince. 1994. The emergence of the unmarked: Optimality in prosodic morphology. In NELS 24: Proceedings of the North East Linguistic Society. Papers presented at the 24th annual meeting held at the Univ. of Massachusetts, Amherst, 19–21 November 1993. Edited by Mercé Gonzàlez, 333–379. Amherst: Univ. of Massachusetts.

                                          Save Citation »Export Citation »E-mail Citation »

                                          An early and foundational work within the framework of optimality theory. The focus is on “the emergence of the unmarked” (TETU) effect in reduplication cross-linguistically (see Markedness Constraints). Page 25 provides an analysis of glottal stop epenthesis in coda position in Massakareese reduplication.

                                          Find this resource:

                                          • Parker, Steve. 1994. Coda epenthesis in Huariapano. International Journal of American Linguistics 60.2: 95–119.

                                            DOI: 10.1086/466224Save Citation »Export Citation »E-mail Citation »

                                            An analysis of coda epenthesis of [h] in Huariapano. The fact that [h] is never epenthesized in even-numbered syllables is explained via prosodic structure. In a system of quantity-sensitive trochees epenthesis occurs in order to repair an unfootable syllable by making it heavy, thus allowing it to comprise its own well-formed trochaic foot.

                                            Find this resource:

                                            Markedness Constraints

                                            Epenthesis in optimality theory is one of numerous possible “repairs” that can be performed on phonological inputs to render them less marked. In so doing, a violation of a faithfulness constraint must result: DEP-IO (every segment of the output has a correspondent in the input, or “don’t epenthesize”). Furthermore, for epenthesis to be the selected repair it must also be the case that DEP-IO is ranked below other faithfulness constraints. For the specific case of epenthesis versus deletion, DEP-IO must be ranked below MAX-IO (every segment of the input has a correspondent in the input, or “don’t delete”). Both constraints are taken from McCarthy and Prince 1995. The original constraints in Prince and Smolensky 2004 (cited under Theoretical Overview) were FILL and PARSE, which operate similarly, but are not equivalent. Optimality theory makes the further prediction that epenthesis should be observed in some language. That prediction is embodied in the DEP-IO constraint itself and the principle of re-rankability. The full set of possible rankings of a given set of constraints is known as a factorial typology, and all and only the set of unique grammars that appear in the factorial typology are predicted to occur. Based on certain hypotheses about the makeup of CON (the universal set of markedness constraints) a number of predictions can be made about possible epenthesis patterns. For example, if epenthesis occurs in order to satisfy the constraint ONSET (all syllables must have onsets), and ONSET is undominated by any other constraint, the result will be onsets in all syllables, and epenthesis whenever necessary to ensure this. Other outcomes involve constraint interactions, for example, a language in which epenthesis occurs only to provide onsets in derived forms; a language in which epenthesis occurs only to provide onsets in certain prosodic environments, etc. These latter two outcomes represent a phenomenon in optimality theory known as “the emergence of the unmarked” (TETU). TETU represents a major success of the theory in accounting for the apparent failure of universal principles to apply at all times. McCarthy and Prince 1994 (cited under Prosodic Epenthesis) and Alderete, et al. 1999 provide examples of TETU involving consonant epenthesis. Casali 1997 illustrates a factorial typology that includes consonant epenthesis as one outcome.

                                            • Alderete, John, Jill Beckman, Laura Benua, Amalia Gnanadesikan, John McCarthy, and Suzanne Urbanczy. 1999. Reduplication with fixed segmentism. Linguistic Inquiry 30.3: 327–364.

                                              DOI: 10.1162/002438999554101Save Citation »Export Citation »E-mail Citation »

                                              An analysis of a type of reduplication in which invariant segments appear, that is, segments that are not copied from the base form of the word. “Fixed segmentism” is divided into two types: one that is morphologically specified and the other that is phonologically driven. Contains several case studies: Yoruba, Lushootseed, Igbo, Tübatulabal, Nancowry.

                                              Find this resource:

                                              • Casali, Roderic F. 1997. Vowel elision in hiatus contexts: Which vowel goes? Language 73.3: 493–533.

                                                DOI: 10.2307/415882Save Citation »Export Citation »E-mail Citation »

                                                Optimality theoretic cross-linguistic analysis of “repairs” of vowel hiatus. A set of indexed faithfulness constraints are used to predict the observed typology of vowel deletion: MAXWI, MAXLEX, MAXMS, MAXMI. Consonant epenthesis, diphthongization, gliding, coalescence, and re-syllabification are also predicted outcomes depending on the relative ordering of constraints.

                                                Find this resource:

                                                • McCarthy, John J., and Alan Prince. 1995. Faithfulness and reduplicative identity. ScholarWorks@UMass Amherst. Linguistics Department, Univ. of Massachusetts.

                                                  Save Citation »Export Citation »E-mail Citation »

                                                  Introduces the Correspondence Theory of Faithfulness. Builds on McCarthy and Prince 1994 (cited under Prosodic Epenthesis) extending the formalization of the reduplicant-base relationship to all input-output representations. Introduces the constraints MAX: penalizing segments in the input without correspondents in the output, and DEP: penalizing segments in the output without correspondents in the input.

                                                  Find this resource:

                                                  Segmental Markedness

                                                  There are differing theoretical predictions about which segments should be epenthesized in various contexts, as well as which segments should be possible epenthetic segments at all. Kean 1975 and Paradis and Prunet 1991 argue, among other works, that coronal is the universally least marked place of articulation. Prince and Smolensky 2004 (cited under Theoretical Overview) postulates a universally fixed context-free markedness hierarchy in which coronal segments are the least marked (followed by labial, followed by velar). Proposals are also found in McCarthy and Prince 1994 (cited under Prosodic Epenthesis) and in Lombardi 2002 and de Lacy 2006 (both cited under Typological Studies) to place the glottal segments (h and glottal stop) at the top of this hierarchy, leaving coronal as the next most harmonic output (Although Rice 2007 states that the most common epenthetic segments are probably laryngeal (h or glottal stop) the author also claims that the most common epenthetic stops are coronal and voiceless). The predicted segment also depends on the epenthetic environment. In syllable onset position low-sonority segments are preferred according to the syllable theory of Clements and Keyser 1983 (cited under Prosodic Epenthesis), and thus the epenthetic consonant in this position is predicted to be [t] (or glottal stop). Low-sonority segments are also preferred for syllable codas. However, because coda epenthesis is driven by foot structure constraints, the preferred segment is one that adds weight to the syllable, and is thus likely to be high in sonority. The full set of possible epenthetic segments depends on what constraints are assumed. Lombardi 2002 and de Lacy 2006 both argue for a constraint banning glottal stop. Thus, even though both assume that glottal segments are the least marked for onset position, a highly -ranked ban on glottal stop in a particular language can lead to coronal epenthesis instead. Although de Lacy 2006 allows for the possibility of epenthetic segments varying by language, the list of possible segments is still quite restricted. The author argues, in particular, for the impossibility of velar segments. Proposals have also been made for language-specific “default” segments, such as Hume 2003. The basis for the restricted approach, however, is the apparent typological asymmetry in which coronals and glottals are prevalent, but other segments are rare or unattested (see Evidence and Analysis).

                                                  • Hume, Elizabeth. 2003. Language specific markedness: A case of place of articulation. Studies in Phonetics, Phonology and Morphology 9:295–310.

                                                    Save Citation »Export Citation »E-mail Citation »

                                                    An argument for language-specific markedness, especially labial as least marked place of articulation. Evidence from phonetics, frequency, patterns of assimilation, deletion, and neutralization in various languages, especially Sri Lankan Portuguese Creole and Kiribatese.

                                                    Find this resource:

                                                    • Kean, Mary-Louise. 1975. The theory of markedness in generative grammar. PhD diss., Massachusetts Institute of Technology.

                                                      Save Citation »Export Citation »E-mail Citation »

                                                      A direct extension of the proposal for marked versus unmarked feature specification outlined in chapter 9 of Chomsky and Halle 1968 (cited under Theoretical Overview). Under the markedness conventions chosen [t] has unmarked values for all the features of a [-syllabic] segment. [p] differs only in that it must be specified as [+labial].

                                                      Find this resource:

                                                      • Paradis, Carole, and Jean-Françoise Prunet. 1991. Asymmetry and visibility in consonant articulations. In The special status of coronals: Internal and external evidence. Edited by Carole Paradis and Jean-Françoise Prunet, 1–28. Phonetics and phonology 2. San Diego, CA: Academic Press.

                                                        Save Citation »Export Citation »E-mail Citation »

                                                        Compilation of evidence for the special, unmarked status of coronals: acquired earlier than other places of articulation; more prone to assimilation; one of the few places known to participate in consonant harmony; more often the result of neutralization; can be transparent to vowel harmony.

                                                        Find this resource:

                                                        • Rice, Karen. 2007. Markedness in phonology. In The Cambridge handbook of phonology. Edited by Paul de Lacy, 79–97. Cambridge, UK: Cambridge Univ. Press.

                                                          DOI: 10.1017/CBO9780511486371.005Save Citation »Export Citation »E-mail Citation »

                                                          A historical overview of the concept of markedness within phonology. There is a brief discussion of epenthesis and epenthetic segments.

                                                          Find this resource:

                                                          Minimality

                                                          It has been observed that the same consonants that epenthesize also tend to delete, and it has been proposed that perceptual minimality, rather than markedness, drives the choice of epenthetic segment. Minimal segments in the perceptual sense are usually taken to include [h] and glottal stop at the beginning or end of prosodic domains, and glides intervocalically, sharing features with one or the other of the flanking vowels. Blevins 2008 and Vaux 2001 (cited under Typological Studies) place perceptual minimality within diachronic accounts of epenthesis, while Steriade 2001 and Howe and Pulleyblank 2004 propose synchronic accounts incorporating perceptual minimality. The former implements a gradient harmonic scale of perceptibility in an optimality theoretic framework. The latter proposes a harmony-as-faithfulness account of synchronic epenthesis, whereby more salient segments should fail both to delete and to insert. Other synchronic accounts appealing to either perceptual or articulatory minimality include Uffmann 2007, which uses prominence scales to select the most prominent (minimally contrastive) segment intervocalically; and Rubach 2000, which analyzes homorganic glide insertion as vocalic feature spreading to a required output X-slot, with the vowel and glide sharing a root node (following the metrical theory of Levin 1985 (cited under Skeletal Slots, Templates, and Reduplicants). Both Broadbent 1991 and Ortmann 1998 analyze English intrusive r as spreading of vocalic features (see the Case of Intrusive R).

                                                          • Blevins, Juliette. 2008. Consonant epenthesis: Natural and unnatural histories. In Linguistic universals and language change. Edited by Jeff Good, 79–106. Oxford: Oxford Univ. Press.

                                                            DOI: 10.1093/acprof:oso/9780199298495.003.0004Save Citation »Export Citation »E-mail Citation »

                                                            Divides synchronic epenthesis patterns into Natural ones: the result of direct sound change; and Unnatural: the indirect result of several changes, or rule inversion. This division corresponds closely with the distinction between minimal and non-minimal segments.

                                                            Find this resource:

                                                            • Howe, Darin, and Douglas George Pulleyblank. 2004. Harmonic scales as faithfulness. The Canadian Journal of Linguistics/La revue canadienne de linguistique 49.1: 1–49.

                                                              DOI: 10.1353/cjl.2004.0041Save Citation »Export Citation »E-mail Citation »

                                                              Argument for perceptually-motivated faithfulness scales rather than articulation-based markedness scales. Faithfulness to more salient elements is higher; thus, they should fail to delete as well as to epenthesize. Able to account for the apparent paradox that less marked segments epenthesize, but are also prone to deletion and assimilation.

                                                              Find this resource:

                                                              • Rubach, Jerzy. 2000. Glide and glottal stop insertion in Slavic languages: A DOT analysis. Linguistic Inquiry 31.2: 271–317.

                                                                DOI: 10.1162/002438900554361Save Citation »Export Citation »E-mail Citation »

                                                                Argues that a language with both glide epenthesis and glottal stop epenthesis cannot be captured by a single consistent ranking and requires derivational levels. At one level, glides can be inserted and glottal stop banned (*[CG]; CONTIGUITY); at the other level, glottal stop can be inserted and glides banned (NO-MULTIPLE-LINK; *ROOT-NODE).

                                                                Find this resource:

                                                                • Steriade, Donca. 2001. The phonology of perceptibility effects: The P-map and its consequences for constraint organization.

                                                                  Save Citation »Export Citation »E-mail Citation »

                                                                  Uses gradient scales of relative perceptibility in evaluating violations of correspondence between input and output. A requirement for perceptual minimality of change from input to output predicts a preference for homorganic glides, h, and glottal stop as epenthetic segments differing minimally from silence.

                                                                  Find this resource:

                                                                  • Uffmann, Christian. 2007. Intrusive [r] and optimal epenthetic consonants. Language Sciences 29.2–3: 451–476.

                                                                    DOI: 10.1016/j.langsci.2006.12.017Save Citation »Export Citation »E-mail Citation »

                                                                    Glide epenthesis minimizes contrast between the epenthetic segment and the flanking vowels. This derives from maximizing prominence, also known as sonority. English intrusive r is analyzed as a case of prominence-maximizing epenthesis, and linking r as a case of deletion (see the Case of Intrusive R).

                                                                    Find this resource:

                                                                    Evidence and Analysis

                                                                    Disagreement exists about what constitutes sufficient evidence for epenthesis and, thus, disagreement about the typological facts. It has been argued that patterns that are insufficiently general cannot be instances of epenthesis; for example, Lombardi 2002 (cited under Typological Studies) rules out patterns that occur only across morpheme boundaries; de Lacy 2006 (cited under Typological Studies) and de Lacy and Kingston 2013 also eliminate patterns limited to a subset of morpho-syntactic environments. Additional proposals, found in either or both of Lombardi 2002 and de Lacy 2006, call for an invariant epenthetic segment, both phonetically and phonologically; for the “repair” of all marked structures (e.g., sequences of two vowels) with no exceptions; and for the “repair” to always be epenthesis (with no sub-patterns of deletion, or glide formation, for example). In cases that are deemed not to constitute epenthesis, the alternative analysis is often deletion. In fact, the surface alternations that provide evidence for epenthesis are inherently ambiguous with respect to whether the underlying process is one of insertion of B in environment Y or deletion of B in the complementary environments. Morley 2015 (cited under Typological Studies) draws attention to the fact that although no formal theoretical criteria exist for distinguishing between the two, a number of informal arguments are made in practice. Spring 1990 (p. 53) and Payne 1981 (pp. 56–57) (cited under Skeletal Slots, Templates, and Reduplicants) make similar arguments regarding the pattern in Axininca Campa, a language the authors analyze as epenthesizing both intervocalically and inter-consonantally. Kenstowicz and Kisseberth 1979 (p. 87) offers a heuristic for general situations of this sort that is also consistent with their approach. This can be boiled down to the proposal that if the deletion analysis would require too much of a coincidence (such as numerous lexical items ending in the same consonant), then the epenthesis analysis must be chosen.

                                                                    • de Lacy, Paul, and John Kingston. 2013. Synchronic explanation. Natural Language & Linguistic Theory 31.2: 287–355.

                                                                      DOI: 10.1007/s11049-013-9191-ySave Citation »Export Citation »E-mail Citation »

                                                                      Argues for the necessity of innate cognitive constraints to explaining the observed distribution of phonological systems. Partially based on the claim that epenthesis of [k] never occurs despite the fact that it is expected to arise from purely diachronic factors.

                                                                      Find this resource:

                                                                      • Kenstowicz, Michael, and Charles Kisseberth. 1979. Generative phonology. San Diego, CA: Academic Press.

                                                                        Save Citation »Export Citation »E-mail Citation »

                                                                        A phonology textbook. Provides in-depth analysis of numerous languages and more data than more recent textbooks typically contain.

                                                                        Find this resource:

                                                                        • Spring, Cari L. 1990. Implications of Axininca Campa for prosodic morphology and reduplication. PhD diss., Univ. of Arizona.

                                                                          Save Citation »Export Citation »E-mail Citation »

                                                                          Uses reduplication in Axininca Campa to argue for the relevance of the Prosodic Word to morphological processes. No affix is required for reduplication, only the operation “copy,” which refers to the base form.

                                                                          Find this resource:

                                                                          Alternative Analyses

                                                                          Cases also are found in which it has been argued that neither deletion nor epenthesis is a possible analysis. These cases may involve segments that are largely predictable from phonological context (like epenthetic segments) but are idiosyncratic in that they either fail to surface in all expected contexts, surface as different segments in a way that is unpredictable, or are associated only with specific morphemes. These idiosyncrasies are taken to require that these segments be analyzed as underlying. However, deletion is ruled out because deletion of segments in the relevant context is not a general process in the language. Alternating segments of this sort are given various labels: “latent,” “ghost,” “floating,” “stray,” “extrametrical,” or “extrasyllabic.” Such segments are generally analyzed as inherently lacking a root node on either a CV or syllable tier. They are thus prevented from surfacing unless some other form provides a node to which they can dock. French liaison is analyzed as such a pattern in Zoll 1993 and Clements and Keyser 1983 (cited under Prosodic Epenthesis); Archangeli 1991 provides a similar analysis of alternations in Yawelmani templatic morphology. Alternating segments may also be phonologically unpredictable, or they may increase, rather than reduce, markedness. In such cases, the pattern may be referred to as suppletion, or suppletive allomorphy (although see Paster 2006 for what the author calls phonologically conditioned suppletive allomorphy). This label can also imply the necessity of a morphological (i.e., lexical), rather than a phonological analysis, such as that in Hale 1973.

                                                                          • Archangeli, Diana. 1991. Syllabification and prosodic templates in Yawelmani. Natural Language & Linguistic Theory 9.2: 231–283.

                                                                            DOI: 10.1007/BF00134677Save Citation »Export Citation »E-mail Citation »

                                                                            Ghost segments are defined as lacking an underlying syllabic specification: they cannot project a prosodic node. However, they may satisfy templatic slots when other consonants are unavailable. Thus, for example, in biconsonantal roots where a templatic slot is provided, ghost segments surface in coda position of the second syllable.

                                                                            Find this resource:

                                                                            • Hale, Kenneth. 1973. Deep-surface canonical disparities in relation to analysis and change: An Australian example. Current Trends in Linguistics 11:401–458.

                                                                              Save Citation »Export Citation »E-mail Citation »

                                                                              Analyzes the synchronic alternations in the Maori passive as resulting from suffix allomorphy (a “conjugation,” or morphological, analysis). Consonants historically part of the root, but deleted in word-final position, become analyzed as part of the passive suffix. The evidence for this analysis is ongoing regularization whereby speakers select a single allomorph more frequently than others.

                                                                              Find this resource:

                                                                              • Paster, Mary E. 2006. Phonological conditions on affixation. PhD diss., Univ. of California, Berkeley.

                                                                                Save Citation »Export Citation »E-mail Citation »

                                                                                In phonologically conditioned suppletive allomorphy (PCSA), multiple underlying morphemes are still required, but the alternations can be analyzed as phonologically conditioned, whether or not the outcome is less marked, or whether the posited phonological features ever surface. A total of 137 examples taken from sixty-seven languages are analyzed as instances of PCSA.

                                                                                Find this resource:

                                                                                • Zoll, Cheryl. 1993. Ghost segments and optimality. In The proceedings of the Twelfth West Coast Conference on Formal Linguistics. Vol. 12. Conference held at the Univ. of California, Santa Cruz, 2–4 April 1993. Edited by Erin Duncan, Donka Farkas, and Philip Spaelti, 183–199. Stanford, CA: Center for the Study of Language and Information.

                                                                                  Save Citation »Export Citation »E-mail Citation »

                                                                                  Ghost segments are analyzed like floating features and tones that require a “docking” site to surface. A syllable position, root node, and default features are all added in regular epenthesis. In ghost segment epenthesis the features are already present and only need to attach to the new syllable and root node.

                                                                                  Find this resource:

                                                                                  Specific Languages

                                                                                  A number of epenthesis patterns have been extensively analyzed by more than one author. Of particular importance is the pattern in Axininca Campa (also Ajyíninka Apurucayali), frequently taken as the paradigm case of [t] epenthesis, as in Payne 1981 (cited under Skeletal Slots, Templates, and Reduplicants), Spring 1990 (cited under Evidence and Analysis), and McCarthy and Prince 1993 as well as Lombardi 2002, de Lacy 2006, and Morley 2015 (all three cited under Typological Studies). But see Staroverov 2014 for a deletion analysis. The pattern in Buryat (also Buriat), described in Poppe 1960, represents a controversial case of [g] epenthesis. De Lacy 2006 predicts that [g] epenthesis should not be possible when less marked segments, e.g., [t] or glottal stop, are available. The author argues that the pattern in Buryat is suppletive, while Morley 2015 offers an alternative analysis. Maori is instructive in assessing criteria related to morphological restrictedness and productivity. The data provided in Bauer 1993 are given differing analyses in de Lacy 2003 and Morley 2015. The phenomenon known as “intrusive r” in varieties of British and American English has been extensively written about and is discussed separately in the Case of Intrusive R.

                                                                                  Diachronic Epenthesis

                                                                                  Blevins 2004 is a diachronically based theory of phonology. In this framework there are three routes to sound change, all involving the listener/learner: CHANGE: mishearing on the part of the listener; CHANCE: phonologically ambiguous input differentially analyzed by speaker and listener; and CHOICE: different phonetic variants chosen as underlying by speaker and listener. A phonological change emerging directly via any of these routes is labeled a Natural one. Intervocalic glide epenthesis (resulting from CHANCE) would be one such change. Non-phonetically predictable (i.e., Unnatural) epenthesis is attributed to one of two diachronic sources. A naturally arising segment may undergo a subsequent independent change in its features: e.g., an intervocalic glide “strengthening” into a voiced stop or affricate. Alternatively, historical consonant deletion may be reanalyzed as consonant insertion, as described in Blevins 2008 (cited under Minimality). The observation that the segments that insert are also the ones that tend to delete is taken as evidence that this “rule inversion” is relatively common.

                                                                                  • Blevins, Juliette. 2004. Evolutionary phonology: The emergence of sound patterns. New York: Cambridge Univ. Press.

                                                                                    DOI: 10.1017/CBO9780511486357Save Citation »Export Citation »E-mail Citation »

                                                                                    A theoretical framework borrowing heavily from the work of John Ohala on the phonetic sources of phonological change. Most, or all, of synchronic phonological patterns are argued to be better explained by their diachronic origins, obviating the need for a synchronic account based on the operation of grammatical rules or constraints.

                                                                                    Find this resource:

                                                                                    Hypercorrection/Rule Inversion

                                                                                    The author of Vennemann 1972 is typically cited as the originator of the concept of rule inversion. In the case of intervocalic consonant epenthesis, the original historic change would be the loss of consonants word-finally and/or pre-consonantally. This is presumably a natural change deriving from the decreased perceptibility of consonants in these environments relative to pre-vocalic environments. The complementarity that exists between the environments in which the consonant is retained and the environments in which it is lost presents an ambiguous scenario for the listener/learner. They may either pick the analysis that the consonant is deleted in pre-consonantal environments (the actual historical change) or pick the analysis that the consonant is inserted in the complementary pre-vocalic environment (the inverted rule). The process is not entirely as straightforward as this account makes it seem, however. Morley 2012 is a detailed analysis of the necessary preconditions for historical deletion to evolve into default epenthesis, and predicts that it should be relatively rare. Critically, what is required are a number of generalizations on the part of the learner about the proper environment for epenthesis that go beyond, and may actually contradict, their input data.

                                                                                    • Morley, Rebecca L. 2012. The emergence of epenthesis: An incremental model of grammar change. Language Dynamics and Change 2.2: 59–97.

                                                                                      Save Citation »Export Citation »E-mail Citation »

                                                                                      An axiomatic approach to the question of phonological change. Argues that for rule inversion to result in a morphologically based epenthesis pattern a number of conditions must hold. They include the requirement that consonants delete both word finally and pre-consonantally, but that only a subset of consonants delete (e.g., coronals).

                                                                                      Find this resource:

                                                                                      • Vennemann, Theo. 1972. Rule inversions. Lingua 29:209–242.

                                                                                        DOI: 10.1016/0024-3841(72)90025-3Save Citation »Export Citation »E-mail Citation »

                                                                                        Classic paper on historic rule inversion. Inversion of rule of consonant loss to rule of consonant epenthesis is stated to be the most elementary type of rule inversion. The alternation of the indefinite article in English (“a/an”) and linking r in certain dialects are given as examples of this type of rule inversion.

                                                                                        Find this resource:

                                                                                        The Case of Intrusive R

                                                                                        Certain dialects of English in Britain, the United States, New Zealand, Australia, South Africa, and India have experienced, at some point in their history, loss of r-sounds in specific environments. This led to alternations between the presence and absence of the rhotic—typically represented by [r]. Some of these so-called non-rhotic dialects also introduced [r] into new contexts. These “intrusive r’s” are typically analyzed as evidence of r epenthesis, and thus a case of historic rule inversion. The phenomenon has received considerable attention for at least the past three decades, and it has been documented from at least the middle of the 19th century in England. For example, Jackson 1830 references mispronunciations of the type “daughter” as “darter.” Because the historical conditions are well known in many of the dialects, and the forms can be compared to non-alternating dialects, English intrusive r represents a valuable case study in rule inversion. The source of the pattern is the loss of r pre-consonantally and pre-pausally. The [r] that is preserved intervocalically is known as “linking r” (e.g., “bar above” versus “ba below”) and faithfully reflects the historical pattern. Intrusive r’s are those that surface in contexts in which no [r] was ever historically present (e.g., “saw him” versus “sawr a”). Johansson 1973 attributes the historic loss of [r] to a process of vowel assimilation (also termed r-vocalization in other accounts) in which features of the [r] were transferred to the preceding vowel, paving the way for the loss of the [r] itself. The reanalysis and generalization of the pattern is attributed to the phonetic similarity of [r] and the vowel schwa. Broadbent 1991 also notes the similarity between [r] and the vowels it follows. Adopting a form of metrical theory, the author analyzes intrusive [r] as the result of vocalic features attaching to an empty onset position, an account consistent with Rubach 2000 (cited under Minimality). Like Broadbent 1991, Uffmann 2007 (cited under Minimality) analyzes [r] as a glide, and explains its distribution with reference to shared vocalic features, but explicitly rejects the vowel spreading account. Uffmann 2007 argues for a synchronic epenthesis process explainable with the machinery of OT. Other analyses of intrusive r as epenthesis include McCarthy 1993, Ortmann 1998, and Vaux 2001 (cited under Typological Studies). However, the innovated r’s can also be attributed to a process of hypercorrection: speakers are aware of the fact that they “delete” r’s with respect to other, possibly more standard, dialects of English, and they make an attempt to “add them back in.” This interpretation suggests a productive process of [r] deletion, and it is consistent with the appearance of intrusive r pre-consonantally, a position that would not be predicted by a markedness-driven epenthesis rule. See Gick 1999 for a deletion analysis in the framework of articulatory phonology. All accounts agree that intrusive r is restricted, never surfacing following a high vowel.

                                                                                        • Broadbent, Judith. 1991. Linking and intrusive r in English. University College London Working Papers.

                                                                                          Save Citation »Export Citation »E-mail Citation »

                                                                                          A unified treatment of linking/intrusive r in the non-rhotic dialect of West Yorkshire. [r] is grouped with [w] and [j] as part of optional glide insertion in intervocalic environments. Glides result from the spread of underlying vocalic features into an empty onset position. Provides a review of other proposals.

                                                                                          Find this resource:

                                                                                          • Gick, Bryan. 1999. A gesture-based account of intrusive consonants in English. Phonology 16.1: 29–54.

                                                                                            DOI: 10.1017/S0952675799003693Save Citation »Export Citation »E-mail Citation »

                                                                                            Evidence from production experiments of [w], [l], and [r]—all segments analyzed as consisting of both a C-gesture and a V-gesture. Shows that reduction in magnitude and delay of C-gesture occur in final position, a process analogous with “r vocalization” (or r-dropping). Intrusive r and l both analyzed as articulatory-based deletion of underlying segments.

                                                                                            Find this resource:

                                                                                            • Jackson, George. 1830. Popular errors in English grammar, particularly in pronunciation, familiarly pointed out: For the use of those persons who want either opportunity or inclination to study this science. London: Effingham Wilson.

                                                                                              Save Citation »Export Citation »E-mail Citation »

                                                                                              A twenty-seven page manual on the correct spelling and pronunciation of a set of words illustrating what the author refers to as “popular errors.” Among these are a handful of examples of intrusive r: “dawn” as “dorn,” “caught” as cort,” “dilemma” as “dilemmer,” “daughter” as “darter,” “idea” as “idear,” etc.

                                                                                              Find this resource:

                                                                                              • Johansson, Stig. 1973. Linking and intrusive /r/ in English: A case for a more concrete phonology. Studia Linguistica 27.1–2: 53–68.

                                                                                                DOI: 10.1111/j.1467-9582.1973.tb00595.xSave Citation »Export Citation »E-mail Citation »

                                                                                                A unified treatment of linking/intrusive r in the non-rhotic dialect of British received pronunciation (RP). Presents several arguments against a synchronic deletion account. Argues for epenthesis via rule inversion as a simplification of the synchronic grammar. Provides details about the relative frequency of [r] insertion.

                                                                                                Find this resource:

                                                                                                • McCarthy, John J. 1993. A case of surface constraint violation. In Special issue: Constraint-based theories in multilinear phonology. Canadian Journal of Linguistics/Revue Canadienne de Linguistique 38.2: 169–195.

                                                                                                  Save Citation »Export Citation »E-mail Citation »

                                                                                                  An analysis of intrusive and linking r in Boston English. [r] deletion is driven by a constraint against [r] in post-nuclear position; [r] insertion is driven by a constraint against short vowels at the ends of prosodic words. The featural material of the epenthetic segment is filled in by a post-lexical rule.

                                                                                                  Find this resource:

                                                                                                  • Ortmann, Albert. 1998. Consonant epenthesis: Its distribution and phonological specification. In Phonology and morphology of the Germanic languages. Papers given at a workshop at the Philipps Universität, Marburg, Germany, August 1997. Edited by Wolfgang Kehrein and Richard Wiese, 51–76. Tübingen, Germany: Niemeyer.

                                                                                                    Save Citation »Export Citation »E-mail Citation »

                                                                                                    Proposes that segments involved in “systematic” epenthesis must be phonologically “simple” or “universally unmarked,” but adopts language-specific measure of markedness. Analysis of English intrusive r similar to Broadbent 1991 and Uffmann 2007 (cited under Minimality), but [r] is not considered a glide. Also analyzes German and Dutch.

                                                                                                    Find this resource:

                                                                                                    back to top

                                                                                                    Article

                                                                                                    Up

                                                                                                    Down