In my dissertation, I show that a difference in structure between functional and lexical items has a restricting effect on both the morphology and the phonology. Morphologically, we observe two asymmetries: (i) in lexical nouns, number-driven root-suppletion is common whilst case-driven root-suppletion is virtually unattested; (ii) in contrast, pronouns commonly supplete for both number and case. By and large, we see the same pattern in verbs, observing a contrast between lexical verbs and auxiliaries with regard to suppletion for aspect and tense. In order to account for these asymmetries, I appeal to structural differences between lexical and functional material, combined with locality effects as proposed in Distributed Morphology (Halle & Marantz 1993). Crucially, lexical material contains a category-defining node which has a delimiting effect that causes case/tense to be insufficiently local to the root to condition its suppletion. In contrast, functional material lacks category-defining nodes and thus no delimiting effect is observed and case/tense are free to condition suppletion of the functional base. Phonologically, we see a correlation between the presence of a category-defining node and the absence of dominant prefixes in vowel harmony and lexical accent, whilst in the absence of a category-defining node dominant prefixes are attested. Thus, I argue for universal limitations on suppletion patterns and dominant prefixes, which crucially derive from a difference in morpho-syntactic structure between lexical and functional material.
Download: Domains on the border: Between Morphology and Phonology.


Alliterative Concord

(joint work with Caroline Féry)

Alliterative concord refers to concord that seems to be, at least partially, determined by similar segmental structures (Corbett 1991; see also Dobrin 1995, Sauvageot 1967, Dimitriadis 1997, Sande 2016, i.a.). This phenomenon has been put forward as a counter-example to strict modularity between morpho-syntax and phonology (Dimitriadis 1997, i.a.), since in models where the morpho-syntax precedes phonology, phonology should not be able to influence morpho-syntactic concord. Putting morpho-semantic concord aside, we focus on the morpho-phonological side of concord, and propose a classification of nominal concord according to two dimensions: (i) extent of morphological pre-specification of a morpheme, and (ii) extent of phonological copying from the stem. Crucially, we argue that some cases of alliterative concord are phonological in nature, and that a purely morpho-syntactic treatment of concord misses phonological generalisations.

Dominant Prefixes

In certain vowel harmony systems, either roots and/or suffixes can be ‘dominant’, i.e. donating the harmonic value to an unspecified vowel. However, it has been claimed that there are no languages with dominant prefixes (Baković 2000, 2001; Hall & Hall 1980); i.e. prefixes never donate their harmonic value to a following morpheme. However, when we look at some data from Tunen and other languages, we find that there are actually dominant prefixes. Crucially, though, these involve functional material rather than lexical items. As opposed to functional items, lexical items contain roots. I propose that the left edge of prosodic words aligns with roots, and that outer material cannot cross a prosodic boundary, effectively ruling out dominant prefixes in lexical material. In contrast, since functional material does not contain roots, no prosodic boundary is created and dominant prefixes are possible. We can extend this to lexical stress assignment, where we again observe the absence of dominant prefixes.

Labial Harmony in Altaic languages

When we look at labial harmony in three branches of Altaic languages (Turkic, Tungusic and Mongolic), we see that in Turkic languages there is a lot of variation in the circumstances under which labial harmony occurs. In stark contrast to this diversity, in Tungusic and Mongolic languages labial harmony occurs only in a single configuration. The goal of my research is twofold: (i) to identify the segmental conditions that govern the occurrence of labial harmony; and (ii) to account for the discrepancy between the gamut of variation in Turkic languages and the lack thereof in Tungusic and Mongolic languages.

Paper: Labial harmony in Turkic, Tungusic and Mongolic languages: An element approach. Phonology 35(4), pp. 689-725.

A second avenue I explore is the prosodic requirements on labial harmony triggers. In canonical labial harmony systems, there are no prosodic restrictions on triggers. However, in some languages the triggers must be two successive round syllables (bigger than a syllable), whereas in Baiyinna Orochen the trigger must be a short vowel whereas a long vowel does not initiate harmony (smaller than a syllable). The intuitive idea I propose is that labial harmony triggers are sensitive to various levels in the prosodic hierarchy.

Edge Prominence

The stress structure of many languages displays a single syllable that stands out among all others (the primary accent), and a rhythmic alternation of beats and non-beats on the other syllables. However, there is a small number of languages which consistently show two successive beats within a word (i.e., a clash). In order to explain this unexpected beat, I propose that these syllables bear a diacritic marking “Edge Prominence”. Furthermore, we see the effect of this marking in another type of languages (‘dual systems’ or ‘bidirectional systems’), as well as in stress shifts in languages such as Dutch and English.


Metathesis is the phenomenon in which sound units reverse their order, as exemplified in e.g. Fur, in which /k+ba/ ‘we drink’ does not surface as *[kba] but as [kab]. These reversals can roughly be grouped into three types of metathesis: (i) perceptually driven metathesis, (ii) phonotactically driven metathesis, and (iii) morpho-phonologically driven metathesis. However, perceptual metathesis is grounded in ambiguity of the signal and as such is not metathesis in its strict sense of segment reversal. Furthermore, I argue that metathesis is best analysed as a phonological by-product of other processes rather than a stand-alone process by itself. As such, I argue in favour of a view that metathesis is only an illusionary effect and that it does not exist in its traditional meaning of segment reversal.
Download: The Illusion of Metathesis.


Nominal suppletion

(see also my dissertation)

The phenomenon of suppletion refers to a single lexical item being associated with two phonologically unrelated forms, the choice of form depending on the morphosyntactic context. Consider good-better-best.: the adjective root is good in isolation but be(tt) in the context of the comparative (and superlative). Though rare in absolute terms, suppletion is frequently observed across languages (Hippisley e.a. 2004). Furthermore, cross-linguistically, lexical nouns frequently supplete for number, but crucially they do not supplete in the context of case. In stark contrast, pronouns regularly display suppletion for number as well as for case. To account for this discrepant behaviour, I argue that the distinct structures of pronouns and lexical nouns interact with locality restrictions, which allows for case-driven suppletion in pronouns but prohibiting it in nouns.

Paper: Limits on Allomorphy: A Case Study in Nominal Suppletion. Linguistic Inquiry 46(2), pp. 363-376.

Containment in inclusive/exclusive pronouns

Morphological marking of inclusive and exclusive (first) person (plural) is relatively frequent (WALS). Either the inclusive or the exclusive can be morphologically marked (Harbour 2011), but, as noted by others, first person inclusive is a marked category (Noyer 2002, Siewierska 2004, Cysouw 2005, a.o.). I show a further asymmetry between morphological inclusive and exclusive marking. While we observe pronoun suppletion in the context of the inclusive in a variety of languages, we do not observe pronoun suppletion in the context of the exclusive (suppletion in the context of both the inclusive and exclusive is also attested). To account for the discrepancy between the inclusive and exclusive as a context for suppletion, I propose that there is a containment relation such that the feature set that makes up the inclusive properly contains the features that form the exclusive, following the reasoning laid out in Bobaljik (2012). I further consider the makeup of person features, and argue that the lack of ABA patterns in clusivity suggest that clusivity features are privative, rather than binary (cf. Harbour 2016).

Paper: Excluding exclusively the exclusive: Suppletion patterns in clusivity. Glossa: A Journal of General Linguistics 3(1), 130.

Adjacency in allomorphy

(joint work with P.W. Smith)

In this paper, we look at the role of adjacency in allomorphy. Adjacency restrictions have been appealed to in a number of works as a restrictor in suppletion. The idea is that some form of adjacency, either structural or linear is required between two elements in order for an allomorphic relation to be established between the two. However, as we show, arguments from lexical nouns that purport to show blocking effects (where trigger and target are split by an intervening element and the allomorphy blocked), tend to fall out naturally from considerations of accessibility, in the sense of Moskal (2015). We look at adjacency effects in personal pronouns, and show that there are empirical and conceptual reasons that lead us to reject adjacency being a restrictor on suppletion, and offer a way that the apparent blocking effects can be captured. Specifically, we propose that blocking effects arise when languages employ hyper-contextual VI rules, that is, VI rules that make reference to more than one node. In this manner two nodes can jointly serve as the context for allomorphy, and produce an effect that looks like blocking under adjacency, but is rather simply a more specific rule winning out over a less specific one.

Paper: (with P.W. Smith) Towards a theory without adjacency: Hyper-contextual VI-rules. Morphology 26(3-4), pp. 295-312.

Pronominal suppletion

(joint work with P.W. Smith, Jungmin Kang, Ting Xu and Jonathan Bobaljik)

In this project, we investigate the internal structure of pronouns by looking at their suppletion patterns (cf. Bobaljik 2012). Concentrating on suppletion patterns of free personal pronouns in the context of (i) case, and (ii) number, we investigate whether we can incorporate markedness hierarchies of case and number into a structural containment relation. With regard to case, we conclude that markedness relations are best presented by containment: more marked case features contain less marked case features, e.g. accusative case contains nominative case. With regard to number, we observe that suppletion displays an interesting markedness reversal: whilst it is usually the case that dual is marked relative to plural, for certain languages this is reversed, in the sense that plural is the marked value among the two. Nevins (2011) argues that the widespread markedness of the dual comes about through the fact that [-augmented] is the general marked value of the feature, but we show that this is subject to variation, and either value of [±augmented] can be selected as ‘marked’ in a language.

Paper: (with P.W. Smith, T. Xu, J. Kang and J. Bobaljik) Case and Number Suppletion in Pronouns. Natural Language and Linguistic Theory .

Inverse languages

Languages can make use of an invariant person hierarchy; inverse languages make use of a morphological marking to indicate a reversal of the relations of the unmarked person hierarchy. That is, a ‘direct’ relation between agents and patients follows the (unmarked) person hierarchy, but an ‘inverse’ relation between agents and patients indicates that the relation is the reverse of the (unmarked) person hierarchy. I briefly looked into the verbal pronominal agreement patterns, the case-marking of arguments, and the presence and absence of an overt inverse marker in 11 languages that make use of the inverse.