Research

This project deals with how agreement with multiple arguments is expressed on a verb. A language can have a marker for subject and object agreement. With two arguments present there are multiple options for agreement: (i) both markers are expressed; (ii) one marker is expressed differently in the context of the other marker; (iii) the features of both markers are expressed as a new, unsegmentable marker: person portmanteaux agreement (PPA).

I show, based on data from four distinct language families, that PPA only occurs in a subset of a verbal paradigm, thus the markers can also be expressed as options (i) and/or (ii). Secondly, all languages have evidence for two slots for agreement.

Based on this, I argue that PPA is epiphenomenal and is actually an instance of contextual allomorphy (option ii) (CA, Bobaljik 2000, Trommer 2006). Since it is CA, it obeys the same restrictions. CA has been argued to only been possible if two elements are in the same complex head (Bobaljik 2012, Embick 2010, Moskal 2015). I show that this holds for PPA, which only occurs when both agreement markers are in the same domain. Secondly, CA obeys a directional restriction in that only higher heads influence lower heads (Bobaljik 2000), which also holds for PPA. This account allows for a verbal paradigm with allomorphs, separate markers and PPA.

V-C movement is easily detectable in Germanic languages: the order of the verb and its arguments differ in matrix and embedded clauses. In polysynthetic languages, such as Algonquian languages, most information is encoded onto the verb, which makes it hard to detect where the verb is in the clause. We argue that by looking at agreement paradigms we find evidence for V-C movement even in Algonquian languages. There are multiple types of evidence.

(i) Most Algonquian languages exhibit two distinct verbal agreement paradigms: one involving suffixes, the other involving both prefixes and suffixes (Conjunct and Independent). The environments where two paradigms occur in most Algonquian languages is strikingly similar to where V2 occurs in Germanic languages. More importantly, the outliers in both groups, Arapaho and English, pattern alike. Thus, whereas most Algonquian languages have independent morphology in the same contexts as V2 in Germanic, Arapaho has conjunct morphology in these contexts and only has independent morphology in the same environments as English. We argue that the Arapaho/Algonquian split is explained under an analysis where the alternation with agreement correlates to presence or absence of V-C movement (following Richards 2004).

(ii) The morphology of the agreement markers provides the second piece of evidence. We show that the suffixal agreement marking is the same marking in independent and conjunct agreement, but can undergo Contextual Allomorphy when the agreement is in the same domain as the higher head (cf. Bobaljik 2012), e.g. when the verb has moved. Secondly, the prefix can only attach to the verb as a clitic when the verbal host is local enough, e.g. when the verb has moved.

The Dutch verbal endings are peculiar in that they allow different affixes depending on the order of the verb and the subject. In Standard Dutch, the second person singular agreement is different when the subject follows or precedes the verb. Looking at 267 Dutch varieties (SAND 2005), we find four generalizations: (i) Third person agreement never differs; (ii) no new markers appear when the verb precedes the subject, only different markers from subject-verb order; (iii) when there is a different marker in verb-subject order, it is never syncretic with 3sg; (iv) the 3sg marker cannot be dropped in present tense, but is always dropped in past tense (1/2 markers can show up with past tense).

We account for these generalizations by assuming that verb movement and the subject can interact and only person features can be influenced. Thus, only when the pronoun carries a person feature and follows the verb, agreement on the verb can change. This means that third person agreement never has different endings in inversion order, since the only feature the pronoun might carry is number. We argue that this is possible because of the feature make up of pronouns. We follow Longobardi (2008) in that person is always higher in the pronominal structure than number features and are therefore more local to the verb. Moreover, since the pronoun is a clitic, it can be picked up by the verb when the verb moves from T to C. We follow Ackema and Neeleman (2003, 2012) in that only different markers show up and now new markers due to impoverishment. Finally, past tense can interact with the marker of 3sg, since the verb will have a suffix when there is past tense marking and the 3sg marker does not have to be spelled out.

GERMANIC

Overt impersonal pronouns such as English one and Dutch men in eight Germanic languages differ in 2 properties: (i) allowing multiple readings (Generic and Existential) and (ii) allowing case (nominative and other cases). I show based on novel data from ECM constructions, passives and unaccusatives, that it is not the syntactic position which restricts the distribution of men-type pronouns, but it is case (contra Cinque 1988, Egerland 2003). one-type pronouns can occur with multiple cases, but only allows a generic reading. All men-type pronouns can only occur with nominative case and can have multiple impersonal readings. I account for this correlation by assuming different feature make-ups for the pronouns, following Egerland (2003), Hoekstra (2010), Ackema and Neeleman (2016): one has phi features and therefore always needs to be obligatorily inclusive; men lacks this functional layer and therefore has no restriction on its readings. Moreover, I propose that since men lacks a phi-layer, it is too deficient to project a KP, and therefore can only occur with unmarked nominative case.

EAST ASIAN

Using the tests proposed for Germanic, I contrast the impersonal pronouns in this language family with East-Asian languages. Holmberg and Phimsawat (2015) have shown that apart from Generic and Existential readings, a non-human reading is also possible ('one (=plants) needs to be fertilized to bear fruit'). This then is a third impersonal reading. I argue that there are three type of deficient pronouns and only English type pronouns can have case. However, since some East-Asian languages have other case systems (Saito 2006), the impersonal pronouns are allowed in ECM-like constructions and object position, even when they are deficient.

A recent debate in the morphological literature concerns the status of derivational affixes. While some linguists (Marantz 1997, 2001; Marvin 2003) consider derivational affixes a type of functional morpheme that realizes a categorial head, others (Lowenstamm 2015; De Belder 2011) argue that derivational affixes are roots. Our proposal, which finds its empirical basis in a study of Dutch derivational affixes, takes a middle position. We argue that there are two types of derivational affixes: some derivational affixes are roots (i.e. lexical morphemes) while others are categorial heads (i.e. functional morphemes). Affixes that are roots show ‘flexible’ categorial behavior, are subject to ‘lexical’ phonological rules, and may trigger idiosyncratic meanings. Affixes that realize categorial heads, on the other hand, are categorially rigid, do not trigger ‘lexical’ phonological rules nor allow for idiosyncrasies in their interpretation.

This paper provides a novel syntactic analysis of the Dutch prefix be-. This prefix can derive new verbs by attaching to verbs (be-vindV-en ‘to be located’), nouns (be-dijkN-en ‘to dam up’) or adjectives (be-grootA-en ‘to economize’). It can also form new adpositions by combining with adjectives (be-needA-en, ‘below’) or prepositions (be-ove(r)P-(e)n ‘above’).

We propose an analysis of be- based on Aboh’s (2010) account for complex locative expressions in typologically different languages, including Gungbe, Zina Kotoko, English and Dutch. We extend this analysis to Dutch verbal complexes, and argue that be- expresses a functional category (F°) that embeds a predicate phrase containing the element it attaches to, in both adpositions and verbs. Our analysis goes against Hoekstra, Lansu and Westerduin’s (1987) small clause account of be- in verbal complexes in which be- is the head of a predicate phrase (Pr°)