Distributed Optimality, Schemes and Mind Maps of German

Download Distributed Optimality and more Schemes and Mind Maps German in PDF only on Docsity!

Distributed Optimality

Dissertation

eingereicht bei der

Humanwissenschaftlichen Fakult¨at

der Universit¨at Potsdam

Jochen Trommer

Abstract

In this thesis I propose a synthesis (Distributed Optimality, DO) between Optimality Theory (OT, Prince and Smolensky, 1993) and a morphological framework in a genuine derivational tradition, namely Distributed Morphology (DM) as developed by Halle and Marantz (1993). By carrying over the apparatus of OT to DM, phenomena which are captured in DM by language-specific rules or features of lex- ical entries, are given a more principled account in the terms of ranked universal constraints. On the other hand, also the DM part makes two contributions, namely strong locality and impoverishment. The first gives rise to a simple formal interpretation of DO, while the latter is shown to be indispensable in any theoretically satisfying account of agreement morphology. The empirical basis of the work is given by the complex agreement morphology of genetically different languages. Theoretical focus is mainly on two areas: First, so-called direction marking which is shown to be preferably treated in terms of constraints on feature realization. Second, the effects of precedence constraints

Thanks to

Peter Staudacher, Gisbert Fanselow, Susann Fischer, John Peterson, Alec Marantz John Frampton, Jonathan David Bobaljik, Daniel Har- bour, Wolf Th¨ummel, Martin Haase, Jens Michaelis, Christian Wartena, Rolf Noyer, Roswita Rudtke, Eva Schlachter, Karin Donhauser, Peter Gebert, Dorothee K¨asler, Sylke Eichler, Theodor Marinis, Kai-Uwe K¨uhnberger, Christoph Gabriel, Albert Ortmann, Claus R. Rollinger, Anna and Jona Grimm.

I acknowledge the funding by the DFG in the graduate school ”Econ- omy and Complexity in Grammar”.

Since the writing of this thesis I have extended and revised substantial parts of the theory and the analyses: Trommer (2002b) develops an extended model of the architecture of the grammar. Trommer (2003a) fleshes out the account of affix order in chapter 7. Trommer (2002a) and Trommer (2003b) revise substantially the account of feature hier- archy effects in chapters 3 and 8.

Tecum principium in die virtutis tuae in splendoribus sanctorum: ex utero ante luciferum genui te.

Contents

7 A Minimalist Account of Agreement in Direction-Marking Languages 389 7.1 Menominee and the Unity of Direction Marking............. 393 7.2 Hierarchy-Based Competition.............. 399 7.2.1 Turkana...................... 400 7.2.2 Western Warlpiri................. 403 7.2.3 Dumi........................ 404 7.2.4 Menominee.................... 412 7.2.5 Other Approaches to Hierarchy-Based Competition.................... 428 7.3 Direction Marking.................... 437 7.3.1 A Minimal Representation for Direction Markers...................... 438 7.3.2 Systems with Inverse Markers only....... 439 7.3.3 A Full Direction-Marking System: Menominee.................... 450 7.3.4 Other Formal Accounts of Direction Marking...................... 464 7.4 Related Systems...................... 485 7.4.1 Arizona Tewa................... 487 7.4.2 Dumi........................ 491

Chapter 1

Introduction

What is systematic about affix order in the following data from Geor- gian?^1

(1) (^) v-xedav v-xedav-t xedav-s xedav-en

S1-see S1-see-PL see-S3s see-S3p ‘I see’ ‘we see’ ‘he sees’ ‘they see’

Distributed Morphology, like most current accounts of inflectional mor- phology, would state for each affix if it is a prefix (v-) or a suffix (-t, -d, -en). This is unsatisfying even under a purely descriptive perspective. A more parsimonious account would state that subject agreement af- (^1) See appendix C for the full present paradigm of xedav and source references. The conventions for the citation of language data that I adopt in this thesis are laid out in section 1.5.

fixes in Georgian are generally suffixes with the exception of v which is a prefix. This interaction of a general rule and an exception statement can be naturally formalized by ranked violable constraints as in (2), where (2b) can only apply if (2a) does not

(2) a. v is a prefix b. All affixes are suffixes

These constraints are ranked because the higher ranked constraint (2a) has to be obeyed in case of conflict, and violable since (2b) is then vio- lated. Crucially, it is difficult to see how an equally concise characteri- zation of the position facts could be achieved in terms of a derivational account or unviolable constraints. But we can still do better and ask if there is a deeper reason why v- is a prefix but not -t or -en. Indeed, looking at the languages of the world one finds a general tendency for number agreement to be marked on the right and for person agreement to be marked on the left edge of the word (see chapter 6). Thus (2) can be replaced by (3):

(3) a. Number agreement should be maximally rightwards b. Person agreement should be maximally leftwards

crucial tenet of DM. In 1.3, it is shown that morphological constraints have to be modeled closely corresponding to DM rules. 1.4 provides an overview of the thesis, and 1.5 contains some remarks on my citation practice for language data in this thesis.

1.1 Optimality in Inflectional Morphology

Following the logic of OT, constraints, while themselves universal, can be ranked differently in different languages. Thus we would also expect a language where pure number and pure person markers are positioned as in Georgian but where mixed markers are prefixes. This is indeed the standard case in the prefix conjugation of Semitic languages such as Amharic (Leslau, 1995:301):

(4) (^) y -s¨abr y -s¨abr-u -s¨aber nn -s¨aber

S3-break S3-break-SPl S1-break S1p-break ‘he breaks’ ‘they break’ ‘I break’ ‘we break’

While pure number marking (-u, plural) is suffixal, the 1pl affix nn -, which fuses person and number, appears as a prefix, as does pure person marking (y - , 3rd person). Finally, the account in terms of violable constraints gives us an idea why subject agreement is often split in the way of Georgian. Marking

of person and number in one affix under the proposed constraints in- evitably leads to constraint violation, because a rightward position is suboptimal w.r.t the person features and the same holds for a leftward position w.r.t. number. However, with separate marking, both number and person can be expressed in their respective optimal position. Apart from linearization, other parts of DM also benefit from an OT perspective: First, we can model cases, where the same constraint is satisfied by different means in different languages. This is the case for the blocking of certain clitic sequences in Romance by an anti- homophony constraint, which is resolved by substitution in Italian (si si → ci si) and by deletion of one clitic in Spanish (se se → se; see Gerlach, 1998; Grimshaw, 1997). Second, we find the same constraint valid to various degrees in dif- ferent languages. Thus Noyer (1992) assumes that there is a universal filter neutralizing number features in verbs that agree with two non- third person arguments. While such a filter can be seen to be at work in many languages, it is suppressed in different contexts (1pl:2 stands for 1pl subject 2nd person object, = for neutralization, , for realization of plural):

c. Alignment Constraints that demand affixes with certain features align to the right or left edge of the domain

Constraints of the first type favor realization of features but not re- dundancy. Thus for any conceivable PARSE constraints all candidates in (7) are equally well-formed.

(7) xedav-en, xedav-en-en, xedav-en-en-en,...

A typical blocking or impoverishment constraint could prohibit multiple agreement suffixes and actually block all multiple instances of -en. Moreover, multiple realization will always displace features from edges. If all morphosyntactic features are subject to some edge- oriented alignment constraint, this leads to additional constraint vio- lations. Thus, in (7), assuming the already discussed constraints that person features have to be maximally to the left and number features maximally to the right, each additional repetition of -en causes addi- tional constraint violations, depicted as “*” in (8):^2

(^2) See chapter 3 for the notation of constraints and violations.

NUM ê R L^ êPER xedav-en * xedav-en-en * ** xedav-en-en-en ** *** xedav-en-en-en-en *** ****

Since PARSE constraints never requires more than the simple real- ization of features, and all other constraint types serve to minimize realization of features, redundancy is blocked without any additional stipulation^3.

1.2 Postsyntactic Morphology

So far it looks as if we have a pure OT model, so why should we need anything like postsyntactic morphology? The answer is: While pure agreement morphology induced by wellformedness constraints often does not or only partially reflect syntactic structure, there are clear instances where it does. This is especially clear for other types of inflection such as tense and aspect marking. Let us first take a look at some evidence that we do need a type of underlying abstract phrase (^3) Such as the Non-Redundancy Constraint in Wunderlich and Fabri (1994).