Skip Navigation



All times are EDT (UTC−04:00).

Short program (full version below):

FRIDAY, September 17
12:50 – 1 PM  Opening Remarks
1 – 2 PM #01   Invited talk: Hannah Sande
2 – 2:30 PM #02   Juliet Stanton & Sam Zukoff
2:30 – 3 PM #03   Edward J. Rubin & Aaron Kaplan
3 – 4 PM  Break
4 – 4:30 PM #04   Giuseppina Silvestri
4:30 – 5 PM #05   Sedigheh Moradi, Lori Repetti & Mark Aronoff
5 – 6 PM  Break
6 – 7 PM #06    Invited talk: Jongho Jun
7 – 7:30 PM #07   Ji Yea Kim
7:30 – 8 PM #08   John Mansfield, Rosey Billington & Hywel Stoakes
  SATURDAY, September 18
10 – 11 AM #09    Invited talk: Martine Grice
11 – 11:30 AM  #10   Veronica Miatto
11:30 AM – 12 PM #11   Silke Hamann & Veronica Miatto
12 – 1 PM  Break
1 – 1:50 PM  Blitz talks:
#12   Luc Baronian & Nicolas Royer-Artuso
#13   Rachel Christensen

#14   Hassan Bokhari
#15   Olivia Mignone
#16   Stacie Chadwick
2 – 2:30 PM #17   Brett C. Nelson
2:30 – 3 PM #18   Jennifer Bellik
3 PM – 4 PM #19    Invited talk: Nancy Hall
  SUNDAY, September 19
9 – 9:40 AM  Blitz talks:
#20   Renato Oniga & Alessandro Re
#21   Brian D. Joseph & Angela Ralli
#22   Graziano Savà
#23   Michael Ramsammy
10 – 10:30 AM #24   Christian Uffmann
10:30 – 11 AM #25   Martin Krämer
11 AM – 12 PM  Break
12 – 12:30 PM #26   Kate Mooney
12:30 – 1 PM #27   Andrija Petrovic
1 – 2 PM #28    Invited talk: Rebecca Morley
2 – 2:10 PM   Closing Remarks

Full program:

FRIDAY, September 17
12:50 – 1 PM

Opening Remarks (Mark Aronoff, Andrija Petrovic, Ji Yea Kim)

1 – 2 PM

Chair: Andrija Petrovic

#01    Invited talk

Epenthesis or deletion? CVCV~CCV alternations in Kru languages

Hannah Sande, UC Berkeley

Kru languages, spoken in Liberia and Côte d'Ivoire, show an alternation between CVCV and CCV. The alternation is variable, where the same word can appear as CVCV or CCV in a single environment, and lexically specific, where only a subset of morphemes alternate in this way. In this talk I describe the V~Ø alternation in a set of Kru languages, showing that whether this pattern is best categorized as an instance of epenthesis or deletion differs by language. Additionally, I provide an analysis that relies on Gradient Symbolic Representations (Smolensky & Goldrick 2016) and Maximum Entropy Harmonic Grammar (Goldwater & Johnson 2003) to model the morpheme-specific, variable V~Ø alternation across Kru languages. The same analytical tools can capture both the deletion and epenthesis patterns, and extend to nearby Mande languages which show a similar alternation.

2 – 2:30 PM

Chair: Michelle Mayro

# 02    Emergence of the unmarked in Scottish Gaelic copy epenthesis

Juliet Stanton, New York University; Sam Zukoff, Leipzig University

We document an emergence of the unmarked effect in Scottish Gaelic copy epenthesis. As shown by Clements (1986), Scottish Gaelic copy vowels obligatorily agree in [±back] with a preceding consonant, leading to mismatches in segmental quality between a copy vowel and its host. However, an analysis of the lexicon shows that while [αback] vowels tend to precede [αback] consonants, the effect is only gradient (i.e., there are many cases of disagreement). On this basis, we argue that copy vowels, unlike lexical vowels, are not faithful to an underlying representation, and thus can freely change to reduce markedness. We analyze this pattern in a correspondence-based approach to copy epenthesis (Kitto & de Lacy 1999, Stanton & Zukoff 2018).

2:30 – 3 PM

Chair: Michelle Mayro

#03    Segmental and Prosodic Influences on Bolognese Epenthesis

Edward J. Rubin and Aaron Kaplan, University of Utah

Bol(ognese), the Gallo-Italic grammar of Bologna, eliminates illicit coda clusters via epenthesis. This process is noteworthy for two reasons. First, as in other closely related Romance varieties  such as Don(ceto) (Cardinaletti & Repetti 2008), prosodic structure impacts whether an epenthetic vowel appears: certain clusters that are permitted within a PWd trigger epenthesis when they straddle a PWd boundary. Second, Bol displays two epenthetic vowels. [u] appears before [v, m], as in [ˈnoːr um] ‘norm.P (cf. [ˈnoːrma] ‘norm.S'), while [e] appears elsewhere, as in [ˈtɛːv el] ‘table.P’ (cf. [ˈtɛːvla] ‘table.S’). In closely related grammars like Don, only one epenthetic vowel ([ə]) appears. We build on Cardinaletti & Repetti’s (2008) analysis of coda clusters in Donceto to account for the Bolognese facts.
3 – 4 PM


4 – 4:30 PM

Chair: Jeffrey Heinz

#04    Two case-studies of vowel insertion in Italo-Romance and the reality of the phonology-syntax interface

Giuseppina Silvestri, UCLA

Two case-studies of vowel insertion in Italo-Romance and the reality of the phonology‑syntax interface. In this contribution I discuss two case-studies concerning two different types of vocalic segment insertion (or epenthesis; Cardinaletti and Repetti 2007): propagation and insertion of schwa in word-final position. The relevant empirical evidence comes from a cluster of dialects spoken in north-western Calabria, i.e. a geo-linguistic area belonging to the so-called “Lausberg Area” (Lausberg 1939) and was collected through in-loco interviews with speakers on several fieldwork trips from 2014 to 2019. The main goal of this presentation is to discuss both the phenomena in the light of their phonological as well as (morpho-)syntactic nature.

4:30 – 5 PM

Chair: Jeffrey Heinz

#05    Conditioned epenthesis across modules

Sedigheh Moradi, Lori Repetti and Mark Aronoff, Stony Brook University

We present a wide range of phenomena from many languages to show that epenthesis cannot be solely defined phonologically, and the factors conditioning it lie along a cline from phonetics to phonology, morphology, morphosyntax and the lexicon. While epenthesis may involve the insertion of the segment whose motivation and quality are phonologically determined, there are cases in which the motivation is phonological, but the choice among various epenthetic segments is morphologically determined; morphosyntax can also be a conditioning factor. Lexically conditioned epenthesis, syllable insertion, and linking elements round out the discussion of other types of semantically vacuous insertion processes influenced by morphology and morphosyntax. Phonologically conditioned allomorphy is a common account of many of these phenomena, but this approach fails to consider the fact that epenthetic patterns fall within a larger grammatical domain in the language.
5 – 6 PM


6 – 7 PM

Chair: Ji Yea Kim

#06    Invited talk

Naturalness in learning Korean n-insertion

Jongho Jun, Seoul National University

This study addresses the questions of what factors may have a gradient effect in application of a morpho-phonological process, how they interact, and which of the gradient effects speakers are aware of, by investigating the variation patterns of Korean n-insertion. From the results of the survey on n-insertion in existing Korean words, I have found several tendencies involving a variety of factors including phonological and morphological ones and their interactions. None of these factors are absolute conditions for the occurrence of n-insertion, contrary to the previous studies on Korean n-insertion, and they have gradient effect, contributing to the overall probability of the occurrence of n-insertion. Most, but not all, of these tendencies were mirrored in the results of experiments involving novel words, suggesting that Korean speakers can learn tendencies prominent in the lexicon. I attribute the observed data vs. learning mismatch to the lack of phonological naturalness involved.

7 – 7:30 PM

Chair: Kalina Kostyszyn

#07    Morphophonological consonant epenthesis in Korean

Ji Yea Kim, Stony Brook University

The present study shows that [s]-epenthesis and [n]-epenthesis in Korean are influenced by both phonology and morphology (hence morphophonological epenthesis). This study also shows that both [s]-epenthesis and [n]-epenthesis occur to mark a morpheme boundary and have a similar phonological distribution to some extent, but that they have distinctive morphological distributions and do not overlap. As for the epenthetic quality, I propose that the epenthetic consonants [s] and [n] are selected by factors that are independent from the universal markedness hierarchy.

7:30 – 8 PM

Chair: Kalina Kostyszyn

#08    Predictable vowels in Anindilyakwa

John Mansfield, University of Melbourne; Rosey Billington, Australian National University; Hywel Stoakes, University of Melbourne

Anindilyakwa, an Aboriginal language of northern Australia, has a vowel system which is unusual among Aboriginal languages, and the subject of several divergent analyses. Heath (nd) proposes phonemic open vowels /a, e/, and epenthetic non-open vowels [i, ə, u]; Leeding (1989) proposes phonemic /a, ɨ/, with allophones triggered by neighbouring consonants, and four phonemic vowels have been proposed by Stokes (1981) /a, e, i, u/, and van Egmond and Baker (2020) /a, ɛ, i, ə/. Regardless of the phonological analysis, all authors note that the quality of the non-open vowels is largely predictable according to the consonantal environment, and some note extensive variation in how and whether certain vowels are produced (Heath nd). This is similar to observations for epenthetic or intrusive vowels in other languages (e.g. N. Hall 2006, Blevins & Pawley 2010). In this study we analyse the phonotactic and phonetic properties of Anindilyakwa non-open vowels, with a particular focus on the extent to which their occurrence and quality is predictable.
SATURDAY, September 18

10 – 11 AM

Chair: Lori Repetti

#09    Invited talk

Vowel insertion in Italian and Tashlhiyt Berber

Martine Grice, Universität zu Köln

In Italian, loan words with a final consonant, such as <facebook>, are often pronounced with an additional word-final schwa. The insertion of this schwa is highly variable and dependent on a number of factors (e.g.  speaker-specific preferences, metrical structure and the identity of the word-final consonant). Crucially, a considerable amount of variation is conditioned by intonation: A vowel is more likely to occur – and is acoustically more prominent – if the intonation is complex or rising than if it is falling.  Tashlhiyt, known for its long consonantal sequences also inserts vowels. Here too, insertion is not only affected by speaker-specific patterns and properties of adjacent consonants, but also by intonation. Instead of asking whether schwa is epenthetic, and should be seen as phonological unit, or instrusive, and therefore the outcome of low level implementation, I shall discuss what role it has to play in the phonological structure of the two languages.

11 – 11:30 AM

Chair: Mark Aronoff

#10    The syllabicity of word-final inserted vocalic elements in Italian

Veronica Miatto, Stony Brook University

The present study addresses the ability of epenthetic vowels to be syllable nuclei. Specifically, acoustic and perceptual evidence is used to determine whether Italian native speakers will produce word-final inserted vowels as syllabic. Consonant-final words in Italian might be adapted with a word-final epenthetic vowel, so that a word like ‘jet’ can be adapted as [dʒɛt] or as [dʒɛtːə], with the inserted epenthetic vowel preceded by lenghtening of the word-final consonant, but much variation is reported.

11:30 AM – 12 PM

Chair: Mark Aronoff

#11    Three language-specific phonological interpretations of final release bursts and vowel-like formants

Silke Hamann, University of Amsterdam; Veronica Miatto, Stony Brook University

Auditory information is interpreted language-specifically. In this talk, we present and formalise three different strategies of how a final release burst and final, short vowel-like formant structures are interpreted with the example of three languages. In American English, the former is mapped onto a final plosive, and the latter onto a final vowel, in Korean, both are mapped onto a final syllable consisting of a plosive and a vowel, and in Italian, both are mapped onto a final plosive. Our example illustrates that the relation between phonetic information and phonological categories is arbitrary and learned language-specifically. This requires a grammar model which makes a strict distinction between phonetic and phonological representations and provides the means to map one onto the other.  
12 – 1 PM


1 – 1:50 PM

Chair: John Frederick Bailyn

Blitz talks

#12    The Case for Epenthesis in Armenian: Evidence from Textsetting
Luc Baronian and Nicolas Royer-Artuso, UQAC

Some Armenian schwas can be analyzed as part of the underlying representation. However, the schwas that have attracted most attention in phonology are those that interrupt a consonant sequence otherwise unattested in Armenian and assumed to be impossible to pronounce by native speakers. While, etymologically, it can be shown that some schwas were epenthesized at some point, one may wonder how to prove their synchronic status. Our contribution has two main goals: 1) The first goal concerns method: we want to show how using Textsetting as data can help the phonologist decide if a phonological element is underlyingly represented (or not), and therefore, if a given 'phonological process' – in the present case, epenthesis – is or is not involved in the surface variation that we observe in our data; 2) The second goal is practical: we want to offer our answer to an important question in the phonology of Armenian, namely: the status of some schwas in this language.


#13    The Phonology of English Loanwords in Brazilian Portuguese: A Comparative Theoretical Account
Rachel Christensen, Stony Brook University

This project proposes the presence of a standardized phonological process of epenthesis that occurs in the adaptation of English loanwords by native speakers of Brazilian Portuguese.


#14    The Patterning of Epenthesis in Hijazi Arabic
Hassan Bokhari, Indiana University

This paper accounts for the different types and motivations of epenthesis in Hijazi Arabic, such as syllable structure-driven epenthesis (SSD epenthesis), sonority-driven epenthesis (SD epenthesis), and epenthesis triggered by the Obligatory Contour Principle (OCP). It also analyzes the default quality of the epenthetic vowel in SD and SSD epenthesis and related to the type of the prosodic unit, syllable or foot, and their prominent status. Beyond that, the quality of the nondefault SD epenthetic vowel, which is affected by the stem vowel or one of the last two consonants in the word, is analyzed in detail.


#15    Glottal stop epenthesis as a hiatus resolution process in Ainu: evidence from acoustic analysis
Olivia Mignone, The Graduate Center, CUNY

The distribution of glottal stop in Ainu is predictable and it is not a lexically contrastive segment. It is an epenthetic segment which emerges as a means of hiatus resolution; this is one among several hiatus resolution processes in the language, including homorganic glide insertion. The present study contributes to the literature on Ainu phonology by demonstrating that glottal stop epenthesis is a purely phonological phenomenon, while glide insertion is a P-structure rule sensitive to prosodic word boundaries. This is done through an acoustic analysis of VV sequences in Ainu across different prosodic contexts and is summarized in an Optimality Theoretic (OT) framework.


#16    Alternative Modeling of /ɹ/ Epenthesis in Australian English
Stacie Chadwick, Stony Brook University

We propose an alternative modeling strategy and demonstrate its function utilizing the epenthesis of /ɹ/ found in Australian English as outlined by Yuen, Cox, and Demuth (2018). The model incorporates logical principles along with Articulatory Phonology to account for both the phonological and phonetic aspects highlighted in the 2018 paper.

2 – 2:30 PM

Chair: Robert D. Hoberman

#17    Insertion of [spread glottis] at the right edge of words in Kaqchikel

Brett C. Nelson, University of Calgary

This paper analyzes phonetic and phonological effects of insertion of [spread glottis] in Kaqchikel, a Mayan language of Guatemala. Plain, unglottalized, voiceless stops are obligatorily aspirated and non-nasal sonorants are obligatorily spirantized at the ends of words, while surface forms of other sounds are unaffected when positioned at the right edge. Compared to edge-marking in related and unrelated languages, this Kaqchikel process stands out for its phonological status, despite [spread glottis] not being a contrastive feature in Kaqchikel.  

2:30 – 3 PM

Chair: Robert D. Hoberman

#18    Vowel intrusion in Turkish onset clusters, in hyper- and hypoarticulated speech

Jennifer Bellik, UC Santa Cruz

Onset clusters in Turkish loanwords have previously been described as being repaired with an optional epenthetic vowel (Clements & Sezer 1982, Yıldız 2010, Kaun 2009, Yavas 1980). I present ultrasound and acoustic production data suggesting that non-lexical vowels in onset clusters in Turkish loanwords arise from an open transition between consonants (vowel intrusion), rather than insertion of an additional vowel target (epenthesis). The intrusive vowels are shorter than underlying vowels, as well as acoustically and gesturally intermediate between front and back vowels.

3 – 4 PM

Chair: Ellen Broselow

#19    Invited talk

How many ways can an inserted vowel be weak? Beyond the epenthesis / intrusion dichotomy

Nancy Hall, California State University Long Beach

In Hall 2006, I argued for a distinction between ‘epenthetic’ vowels—full segments that are inserted to repair marked structures, typically by creating new syllables—versus ‘intrusive’ vowels, phonetic transitions between consonants that do not correspond to a phonological segment. Here I discuss case studies that challenge this dichotomy: inserted vowels that are phonetically weaker than lexical vowels, yet have too much phonetic substance to be simply transitional.
SUNDAY, September 19

9 – 9:40 AM

Chair: Grace Wivell

Blitz talks

#20    Epenthesis in Latin
Renato Oniga and Alessandro Re, Università degli Studi di Udine

This 5-minute speech intends to focus on two aspects. First of all we want to present how the ancient Latin grammarians developed the concept of epentesis in their theoretical reflection. Secondly, the grammatical reflection on this topic raises questions that, in our opinion, have not been explained in a fully satisfactory way.


#21    The Aftermath of Epenthesis: From hiatus to morpheme in Greek
Brian D. Joseph, The Ohio State University; Angela Ralli, University of Patras

Epenthesis leads to positive outcomes in phonological structure, but can it lead to other sorts of outcomes. In particular, can epenthesis outcomes go beyond phonology? We argue here that certain developments in the active imperfective past paradigm in Modern Greek offer a case in which an epenthetic consonant, pressed into service to repair a vowel hiatus caused by the analogical spread of a particular verbal ending, turned into a new morphological marker, ultimately of the form /γ/, of imperfective aspectual value. In this Greek case, we thus see the progression from epenthetic hiatus-breaker to morpheme, going beyond mere phonological effects. Moreover, the emergence of the -γ- morpheme shows a case of degrammaticalization, in that a grammatical marker developed out of a phonetic effect, thus moving up the cline of grammaticality, counter to the directionality implicit in Hopper's (1994:31) statement that "phonogenesis is an advanced stage of grammaticalization".


#22    A few notes on epenthesis in Nara (Eritrea) and Amharic (Ethiopia)
Graziano Savà, "L'Orientale" University, Naples, Italy

The paper presents some aspects of epenthesis in a minority language of Eritrea, called Nara, and in Amharic, the administrative language of Ethiopia. Even if quite different the two languages have in common that syllables must have a vocalic nucleus and cannot have more than one consonant as onset and coda. If a syllable without nucleus results as a consequence of morphological processes, a phonological process, mainly epenthesis, intervenes to avoid its unlicensed structure.


#23    Towards an analysis of vowel alternations and epenthesis in Hindi causative verbs
Michael Ramsammy, University of Edinburgh

I present an analysis of phonological alternations that occur in Hindi causative derivations. The focus is given to the suffix /-ɑ/, which as well as triggering qualitative alternations to stem vowels, also causes the emergence of an epenthetic /l/ after vowel-final bases.  The analysis is couched in Stratal Optimality Theory and assumes an underlying contrast between [+tense] and [–tense] vowels in Hindi.  I argue that vowel neutralisation in causative stems is driven by tenseness dissimilation.  Epenthesis is hiatus-breaking, and I discuss the choice of epenthetic /l/ in relation to diachronic changes in the phonological history of Hindi and Indo-Aryan.  

10 – 10:30 AM

Chair: Yang Liu

#24    Epenthesis as a Matter of Faith

Christian Uffmann, HHU Düsseldorf

This talk argues that in an Optimality-Theoretic framework epenthesis of default segment ts should not be understood as epenthesis of a maximally unmarked segment but instead as epenthesis of a maximally unspecified segment, to satisfy feature faithfulness constraints. I will show why a markedness-based account is problematic in several ways and show how a faithfulness-based account avoids these problems but instead provides the basis for a unified theory of epenthesis that takes both default segmentism and spreading-based epenthesis into account.  

10:30 – 11 AM

Chair: Yang Liu

#25    Degenerate syllables and excrescent vowels. Evidence from Yuman

Martin Krämer, University of Tromsø - The Arctic University of Norway

Excrescent vowels in two Yuman languages and the phonotactic restrictions for their occurrence show that vowels that fulfill some criteria of excrescent vowels are not always a phonetic reflex of articulatory timing, but rather signal the presence of an additional, albeit degenerate, syllable. They are inserted in degenerate syllables that lack a mora, which renders them inaccessible for higher level prosodic computation.
11 AM – 12 PM


12 – 12:30 PM

Chair: Scott Nelson

#26    Not all consonant epenthesis is low-sonority

Kate Mooney, New York University

In this paper, I examine consonant epenthesis patterns from two dialects of Uab Meto (Austronesian, West Timor). In the first dialect, Amarasi, the epenthetic consonant bears a fixed quality as /g/. In the second dialect, the epenthetic consonant alternates based on the quality of the preceding vowel (/b, l, ɟ͡ʝ/). Each of these consonants were previously unattested as epenthetic segments (cf. de Lacy 2006: 81). Based on these data, I argue that consonant epenthesis can be decomposed into two steps: slot insertion, followed by insertion or spreading of features (Uffmann 2006, a.o.). Languages with default epenthesis have feature-insertion for the second step (Dep-F, Krämer 1998, Archangeli 2000, a.o.), whereas those with vowel-conditioned epenthesis have spreading (Uffmann 2006, Staroverov 2014). I then go on to show how spreading and feature-insertion dialects have different behaviors with (spreading-based) metathesis: vowel-to-consonant spreading bleeds metathesis, but feature insertion does not.  

12:30 – 1 PM

Chair: Scott Nelson

#27    Modeling morphological epenthesis

Andrija Petrovic, Stony Brook University

Various consonant insertion processes in different languages, even when they are phonologically motivated, are not phonologically optimizing, and are heavily influenced by morphological factors. This work focuses on characterizing such processes as predictable, rather than artifacts of listed stem allomorphy, and the analysis includes such processes in a wider interpretation of morphological processes as logical transductions on strings. In terms of computational power, such a morphological system is equivalent to a collection of regular relations. I show that limiting morphology to string transductions, and placing such a module between syntax and phonology, correctly predicts that morphological processes at the interfaces should exist, and be modelled by regular relations. Morphological epenthesis (insertion of semantically vacuous material) is predictable from the phonological shape of morphological constituents and the morphosyntactic properties of the lexeme, and can also occur as a realization of a specific syntactic structure.

1 – 2 PM

Chair: Veronica Miatto

#28    Invited talk

Why epenthesis can help us answer the question of where phonology comes from

Rebecca Morley, The Ohio State University

In this talk I present a framework for linking synchronic phonological representations with their diachronic sources via explicit models of the learner and the properties of the learner’s input. The significance of harmony-increasing consonant epenthesis (e.g., as a “repair” for an onset-less syllable) lies in its status as an unequivocally phonological phenomenon. I review 3 linked papers that deal with the status of synchronic epenthesis typologically, the set of sufficient diachronic conditions for synchronic epenthesis, and the learnability of synchronic epenthesis in both impoverished, and non-impoverished conditions. Conclusions drawn from this set of papers form the foundation of a theory in which many of the distinctions between phonetic and phonological representations are collapsed. I will talk briefly about more recent work, in which I have built on this framework to model contrasts based on nasalization, vowel length, and laryngeal features.
2 – 2:10 PM

Closing Remarks (Lori Repetti, Veronica Miatto)