• The Base-generation Approach to the Spray/Load Alternation in Japanese

      Miura, Kaori; Kyushu Sangyo University (University of Arizona Linguistics Circle (Tucson, Arizona), 2012)
      This article presents a syntactic account of the spray/load alternation in Japanese (e.g., nuru ‘smear’, tsumeru ‘pack’, etc.,). Fukui, Miyagawa and Tenny (1985) claim that the alternation verbs show the double object structure under a single VP node; namely both the Material (i.e., something that is painted) and the Location (i.e., somewhere that is painted) participants are co-sisters of the verb. When the Material participant is affected, it will be realized as the direct object of the verb. On the other hand, when the Location is (completely) affected, it will be realized as the direct object of the verb. I build an account on their intuition that the Material and the Location elements are thematically connected with the lexical verb within the binary branching hypothesis (Kayne 1994, among others); thus, they are arguments of VP. But this structure is valid only for the ni-variant where the Material is the single sister of VP. In the structure of the de-variant, the Location is a single sister of VP but the Material element is a PP, merging above VP. Under this view, the two syntactic alternants are available because they are derived from different numeration arrays. The present analysis minimizes the burden on the syntax by eliminating the affectedness condition for determining argument distribution of spray/load verbs in Japanese.
    • The behavior of velar nasal and syllabification in Korean

      Chung, Chin Wan; Lim, Byung-jin; Indiana University (University of Arizona Linguistics Circle (Tucson, Arizona), 2000)
      Across languages including Korean, a string C1V1C2V2C3 is generally syllabified as C1 V 1 .C2 V 2C3 but not as C1 V 1 C2.V 2C3. This is largely due to the fact that languages prefer a syllable with an onset, which is explained by Ito (1986) as a Universal Core Syllable Condition or by Prince and Smolenky (1993) as an optimality-theoretic constraint 'Onset'. In Korean, a C1V1C2V2C3 string is generally syllabified as C1V1.C2V2C3. However, when the velar nasal lr.JI is C2, it is not syllabified as the onset of the second syllable but rather as the coda of the preceding syllable. The main purpose of this paper is twofold. On the one hand, we investigate the behavior of the velar nasal with respect to syllabification in syllable coda position. On the other hand, we also investigate the variation among speakers regarding strategies to avoid an onsetless syllable over a morpheme boundary in which the second morpheme begins with the glide /y/ in Korean, as well as the variation between Seoul dialect (SD) and Kyungsang dialect (KD) with respect to ways to deal with the velar nasal in the syllable coda position. We provide an analysis within Optimality Theory (Prince and Smolensky 1993), particularly the more enhanced version called Correspondence Theory (McCarthy and Prince 1995). This paper tries to shed more light on various cases of syllabification in Korean. The organization of this paper is as follows. In section 2, we present the data. In section 3, we propose optimality-theoretic constraints and their interaction and provide an analysis based on the constraints and their ranking. Finally, we sum up the paper in section 4.
    • Bibliography (Coyote Papers, Volume 16)

      University of Arizona Linguistics Circle (Tucson, Arizona), 2008
    • Bimoricity in Northern Greenlandic Eskimo

      Meador, D. (University of Arizona Linguistics Circle (Tucson, Arizona), 1992)
      Multiple strong stresses on heavy syllables within words in Northern Greenlandic Eskimo indicate the absence of word trees in the metrical grid. That final light syllables also selectively receive stress in the absence of a word tree presents a challenge to available mechanisms which attempt to account for such alternations. These include destressing (Hammond 1989; Halle and Kenstowicz 1990) and extrametricality (Halle and Kenstowicz 1990). The problems presented by these mechanisms are avoided in an analysis based on bimoricity. The analysis proposed here presents a modification of the iambic template in Hayes' (1987) typology.
    • Blocking and causatives: unexpected competition across derivations

      Miyagawa, Shigeru; Massachusetts Institute of Technology (University of Arizona Linguistics Circle (Tucson, Arizona), 2011)
      The Japanese causative verb exhibits the effects of blocking, whereby a causative verb (V-sase) is blocked from taking on lexical meaning if there is a competing lexical causative verb (Miyagawa (1980, 1984)). Given that the causative verb is most reasonably viewed as being formed in syntax, the blocking effect leads to the conclusion that the lexical causatives also are formed in syntax, contrary to the traditional view. A similar blocking effect is observed with English causatives formed with make, and this, together with what we can observe in Japanese, suggest that blocking is best viewed as one that arises in the process of deriving the causative verb (e.g., Embick and Marantz (2008)), and not as a result of a filter on the output of the generative component (e.g., Kiparsky (2005)).
    • Burmese Sandhi-Voicing: From the Perspective of Emergent Phonology

      Ni, Tianyi; The Ohio State University (University of Arizona Linguistics Circle, 2022)
      This paper deals with sandhi-voicing in Modern Burmese from the perspective of Emergent Grammar (EG). Sandhi-voicing is only found in compounds, but not all of them. EG predicts that Burmese speakers tend to store compounds with sandhi-voicing as a combination of two morphemes, while those without sandhi-voicing as a whole.
    • Can Idioms Be Passivized?: Evidence from Online Processing

      Stone, Megan Schildmier; University of Arizona (University of Arizona Linguistics Circle (Tucson, Arizona), 2013)
      This paper presents the results of an experiment designed to access native speakers’ underlying grammatical knowledge concerning the passivizability of English Verb-Object (VO) idioms. Although it has long been noted that some VO idioms retain their idiomatic meaning in the passive while others do not (Katz & Postal 1964, et seq.), the source of this variation is unclear, and native speaker intuitions on a large number of idioms are not as clear cut as previous accounts might suggest. Taking as a starting point Folli and Harley’s (2007) hypothesis that there is a structural distinction between passivizable and nonpassivizable idioms, the current study tests one prediction of this hypothesis, namely that there should be a categorical distinction between the two types of idioms in the grammars of native speakers. The experimental results contradict this hypothesis, as evidenced by a normal distribution of response times to passive idioms. However, it is hypothesized that this online task is not appropriate to access the fine-tuned syntactico-semantic judgments underlying native speaker intuitions of idiom passivizability, due to the fact that the methodology employed here—a self-paced reading task—does not yield the expected results even for canonically passivizable and nonpassivizable idioms.
    • Case and Configurationality

      Jelinek, Eloise; University of Arizona (University of Arizona Linguistics Circle (Tucson, Arizona), 1983)
    • Case and Phi Features as Probes

      Ussery, Cherlon; Carleton College (University of Arizona Linguistics Circle (Tucson, Arizona), 2012)
      This paper uses case and agreement patterns to argue for a reformulation of Agree (Chomsky 1995, 2000, 2001). Throughout syntactic literature, various proposals that account for the assignment of case and agreement have been made. Chomsky (1991) proposes that different projections are responsible for the two types of features. Case is assigned in Spec,TP, while agreement is established in Spec,AgrP. By contrast, Agree divorces feature checking from movement (Bobaljik and Wurmbrand 2005, Wurmbrand 2006). Case and agreement are assigned under c-command via the same Agree operation. A head, T, checks the case of a DP with a matching case feature and, in turn, that DP checks the agreement features on T. The prediction, therefore, is that case and agreement should necessarily pattern together: verbs should agree with DPs that are in a case relationship with T. I provide evidence not only that case and agreement features may pattern differently, but also that individual agreement features may pattern differently. As such, I argue that features on heads – not heads themselves – are probes. While I argue that case and phi features are not an indivisible bundle, I maintain the proposal that feature-checking need not force movement to a specifier, thus eliminating the need for independent agreement projections. Additionally, I illustrate probing is not restricted to c-command. I redefine Agree so as to allow a probe-goal relation to be established either under c-command or in a spec-head configuration.
    • A Categorial Treatment of Scrambling in Japanese

      Kurahone, Akira; University of Texas at Austin (University of Arizona Linguistics Circle (Tucson, Arizona), 1981)
      In generative studies of Japanese, the term 'Scrambling' has been used to account for the intuitively obvious relationship bebween sentences like (la) and (lb). (1) a. John-ga (subj) Bill-o (obj) mi-ta. (see-past) b. Bill-o (obj) John-ga (subj) mi-ta. (see-past) 'John saw Bill.' The phenomenon has presented linguists with an interesting problem. especially in conjunction with treatments of other linguistic phenomena (e.g., Case - Marking, Reflexivization, etc.). This paper presents a categorial treatment of Scrambling in a simplex sentence. The basic framework has been taken from Montague's Universal Grammar (1970) and Proper Treatment of Quantification in Ordinary English (1973). The purpose of this paper is two -fold. It attempts to provide (i) a categorial syntax capable of directly generating scrambled variants of a canonical form, and (ii) a semantic account for the truth functional meaning equivalence among variants. While a direct generation approach is not new (e.g., Whitman (1979), Hale (1980), Farmer (1980), Chomsky (1980), Ostler (1980), etc.), there is yet no universally accepted analysis that offers a rigorous semantic account.
    • The Causative/Inchoative Alternation, and the Decomposition of Little v

      Key, Greg; University of Arizona (University of Arizona Linguistics Circle (Tucson, Arizona), 2012)
      Morphological evidence in causative/inchoative pairs in Turkish is analyzed to determine the derivational relationship between the transitive and intransitive members of the pairs. Three patterns are found: 1. The transitive member is derived from the intransitive; 2. The intransitive is derived from the transitive; 3. Both members are independently derived from a common base. For a complete explanation of the data, it is proposed that the verbalizing head little v decomposes into a verbalizer (little v proper) and a discrete ‘flavor’ morpheme (CAUSE, BECOME, etc.).
    • Centering Revitalization in Remote Documentation

      Harvey, Meg; The University of Arizona (University of Arizona Linguistics Circle, 2022)
    • Change of State Verb and Syntax of Serial Verb Constructions in Korean: An HPSG Account

      Lee, Juwon; University of Texas at Austin (University of Arizona Linguistics Circle (Tucson, Arizona), 2012)
      The simple, canonical form of SVCs has been much studied (e.g. Lee, 1992; Chung, 1993; Kim 2010 for Korean, and Aikhenvald, 2006; Dixon, 2006 for various African languages). In this paper, I investigate a more complex form of SVCs (namely resultative SVCs) which are almost ignored in the literature. Specifically, I show that (i) the causing event and result state of a Korean change-of-state verb should be separately represented in the lexical information of the verb, (ii) the resultative SVCs are really a type of SVC by comparing the core concept of SVCs (i.e. serializing subevents and so non-cancellation of V1 result state or object) with the corresponding properties of the construction in question, and (iii) SVCs generally have the constraint that result state or object should be created after the event of V1 (with more evidence from light verb SVCs). Finally, I present an analysis of the resultative SVC and light verb SVC in Head-Driven Phrase Structure Grammar (Pollard and Sag, 1994; Sag et al., 2003).
    • Child Acquisition of Navajo and Quechua Verb Complexes: Issues of Paradigm Learning

      Courtney, Ellen; Saville-Troike, Muriel; Department of Foreign Languages and Literatures, East Carolina University; Department of English, University of Arizona (University of Arizona Linguistics Circle (Tucson, Arizona), 2000)
      Navajo and Quechua, both morphologically rich languages, present an interesting testing ground for proposals regarding the acquisition of inflectional systems. Of particular interest for these languages is the development of the verb, which encodes not only tense, aspect, and number- /person-of- subject, but also such grammatical notions as transitivity, causation, modification and internal arguments. In fact, the complex verb forms which characterize these languages often constitute the entire VP or, indeed, the whole sentence. The languages are all the more fascinating because, typologically, the structure of the verbs in Navajo is roughly the mirror image of Quechua verbs: while complex Navajo verbs are formed by appending prefixes to the root or stem, Quechua verbs are formed entirely through suffixation. This is illustrated in the Navajo and Quechua equivalents of the English sentence, `They were feeding it to me, too': (1) a. Navajo: Shí - aɬdo' - shá - da - 'í - ø - ɬ - stood; me also for Pl obj subj CL impf stem:FEED b. Quechua: Mikhu - chi - sha - wa - rqa - n - ku - pis; EAT Caus Prog l obj Past 3 subj Pl Add. In the Navajo verb form, the disjunct prefixes (those furthest from the stem) include oblique object, adverbial, postposition, and plural, and the conjunct prefixes (those closest to the stem), the direct object, subject, and transitive classifier. The stem 'feed' occurs in final position and its form indicates imperfective aspect. By contrast, in the Quechua verb shown in (lb), the stem created by affixing the causative suffix to the root means 'cause to eat' or 'feed', and the final element, the Additive suffix, is an independent enclitic meaning 'also.' The morphemes occurring between the causative stem and the final enclitic are part of the inflectional set. In both languages, a verb must minimally consist of a root and a person-of-subject affix; that is, adult speakers do not produce bare verb roots or stems. However, in Navajo, the ordering of the prefixes in relation to the verb stem is quite rigid, whereas, in Quechua, some of the suffixes attached to the verb stem may occur in varied order. While the Quechua suffixes have unique and identifiable meanings, the rules for their combination often have no basis in semantics: they are idiosyncratic, including ordering restrictions which must be formulated as negative filters (Muysken 1986, Muysken 1988). Recent work by several researchers has yielded relevant, isolated proposals, many intended to enlighten the nativist-empiricist debate. For example, Hyams proposed the stem parameter, with two possible settings: a verbal stem is/is not a well-formed word (Hyams 1986a, Hyams 1994, Pizzuto and Caselli 1994). She further claimed that children set this parameter very early on. Assuming that Hyams' parameter is a valid constraint on verb formation, we would expect children acquiring morphologically rich languages never to produce bare verb roots or stems. In a number of studies on the acquisition of synthetic languages, the perceptual salience of particular syllables is cited as a vital cue in segmentation (Pye 1983, Aksu-Koç and Slobin 1985, Mithun 1989), with the phonologically most salient morphemes occurring at the periphery of words. We would therefore expect children to produce the inflectional morphemes first, before the derivational affixes found closer to the verb stem (Peters 1995). On the way to meaningful productivity, children acquiring morphologically rich languages may at first produce frozen chunks or amalgams of affixes, even splicing together different unanalyzed strings to form novel, sometimes ungrammatical combinations (Peters 1985 and elsewhere, Franco and Landa 1998, Rubino and Pine 1998). Finally, it has often been noted that children learning a variety of agglutinative languages may insert novel filler syllables into the affixal string, (e.g., Aksu -Koç and Slobin 1985, Saville-Troike 1996). Peters (1985 and later work) views the insertion of such "placeholder" syllables as further evidence of unanalyzed amalgams in child production. As illuminating as these studies may be, none presents a model for the acquisition of morphology as comprehensive as Pinker's (1984) proposal regarding the acquisition of inflectional systems through paradigm-learning. In the Introduction to the second edition (1996), Pinker later cites a number of studies confirming his original proposal. In developing the proposal, Pinker draws on observations of inflectional learning in a variety of languages, both his own and those of other researchers such as Slobin. This brings us to the aim of the present study: an exploration of the acquisition of verb morphology by children learning Navajo and Quechua in the context of Pinker's theory.
    • CLI and Cognitive Control in the L3 Initial State

      Brown, Megan M.; Boston University (University of Arizona Linguistics Circle (Tucson, Arizona), 2021)
    • Clitics, Scrambling and Parsing

      Lewis, William, 1938-; Department of Linguistics, University of Arizona (University of Arizona Linguistics Circle (Tucson, Arizona), 2000)
      The purpose of this paper is to devise optimal algorithms for parsing linguistic structures that contain P2 (Wackernagel) clitics. Since many languages that have P2 clitics also allow scrambling, any algorithms for parsing P2 clitics must also contain algorithms for parsing scrambled structures. Most of the energy of this paper, however, will be focused on P2 parsing. Although many languages have P2 clitics. I have focused most of my attention on Native American languages (with some exceptions). There is one major reason for this: languages of the Americas are almost entirely ignored by the computational and parsing literature, which focuses on languages of the Indo-European language family (and almost always on English, at that). By doing so, researchers deprive themselves of data and linguistic structural diversity that can help in devising more widely applicable parsing algorithms. This is a computational paper, the intention of which is to develop parsing procedures. Little attention will be paid to a specific syntactic /morphological theory, nor will much attention be paid to the form of the output. These are concerns that can be addressed in a later stage of parser design. What is an "optimal" parsing algorithm? I shall define the optimality of a given solution by the criteria in (1) below: (1) 1) The optimal solution is one which uses devices and formalisms whose generative capacity is as low as possible on the Chomsky hierarchy. 2) The optimal solution uses as few "rules" or "devices" as possible. Obviously, it will be necessary to strike a balance between these two criteria. For this reason, the issue of optimality may be somewhat lexìbìe, depending on how much weight is given to each criterion. The most optimal solutions might require the power of context -sensitive rules, but these may be used in concert with context -free or even finite-state rules.
    • Complex Verbs and the Lexicon

      Akmajian, Adrian; Miyagawa, Shigeru (The University of Arizona., 1980)
      At the stage in the development of generative -transformational grammar when the primary emphasis of research was on the syntactic analysis of sentences, morphology was largely ignored, and the lexicon was simply viewed as an unstructured list of lexical items. However, Chomsky's Lexicalist Hypothesis brought about a renewed interest in word formation, and it is now clear that "the lexicon has a rich, internal structure. In this thesis we will apply the Lexicalist Hypothesis to Japanese, a non -Indo- European, agglutinative language. The analysis presented will be referred to as the "lexical analysis." Two major theoretical issues face the lexical analysis of Japanese. First, it attempts to provide an alternative framework to the transformational analysis that has dominated Japanese linguistics for the past fifteen years. With emphasis on sentences instead of words, the transformational analysis, in a sense, "deagglutinizes" morphologically complex verbs made up of a verb stem and one or more bound morphemes, e.g., V-sase-rare-ta-gat-ta 'V-cause-passive-want-appear- past'. A complex underlying structure is postulated for a sentence with such a verb, with each of the morphemes acting as "higher" verbs in the structure. In the transformational analysis, the fact that the morphemes combine to form a word seems almost incidental. The lexical analysis starts with the assumption that a complex verb comprises a single word that is formed in the lexicon. This forces us to look at Japanese in a different light, as a language with fairly simple phrase structure, and virtually no transformations (if any), but with a rich, highly structured lexicon. This view is closer to the original intuition that Japanese is an agglutinative language. The second theoretical issue concerns the theory of the lexicon. The Lexicalist Hypothesis has been worked out mainly for English, a nonagglutinative language. Since a primary concern of the Hypothesis is with words, it makes sense to test it using a language such as Japanese that has rich and varied word formation processes. The lexical analysis of Japanese draws from major works on the lexicon in English, but because of the highly agglutinating nature of Japanese, we find it necessary to reject, alter, and extend various aspects of lexical analysis of English. The goal of the lexical analysis is to define organizations within the Japanese lexicon. Two types of organizations the lexical analysis focuses on are (a) content and ordering of rules that apply within the domain of the lexicon, and (b) arrangement of verbs, both simple (i.e., verb stem) and complex, listed in the lexicon. Regarding (a), word formation rules akin to those proposed by Aronoff bear the responsibility of forming complex verbs in Japanese; and "redundancy rules" assign rule- governed, i.e., "regular," case arrays to the NPs in the subcategorization feature of verbs. Idiosyncratic case marking is either attributed to a particular lexical item -- verb -specific case marking --or assigned by a rule with a limited scope (Marked-Case Specification Rule). As for (b), the lexicon imposes an organization on all listed verbs by providing slots within "paradigmatic structures." Verb stems automatically receive a slot, and thus they are the most basic verbs, while complex verbs formed by word formation rules can only enter an appropriate slot if the slot is not already occupied by a more basic lexical item, usually a verb stem. If a complex verb can occupy a slot, it receives a lexical entry and becomes part of the permanent lexicon. These verbs undergo lexical processes such as semantic drift and nominalization commonly attributed to the basic verb stems. It is hoped that the lexical analysis of Japanese presented in this thesis will be a model for other agglutinative languages. As a way of illustrating this possibility, Turkish, Mitla Zapotec, and Navajo are briéfly considered in light of the lexical analysis of Japanese. While all of these languages share virtually the same components within the lexicon, a slight difference in the arrangement of the components leads to the difference in the morphological characterization among the languages.
    • The contrastive reading of Japanese -wa, and the role of information structure

      Deguchi, Masanori; Western Washington University (University of Arizona Linguistics Circle (Tucson, Arizona), 2010)
      In this study, I investigate the distribution of the contrastive reading associated with the so-called Japanese topic marker –wa. The main goal is two-fold. First, I examine two previous approaches, which I call the “predicate-based approach,” and the “argument-based approach” respectively, and demonstrate that they are not sufficient to capture some empirical data. Second, based on the observation that wa-phrases in all-focus and subject-focus sentences induce the contrastive reading, I argue and demonstrate that the contrastive reading arises when wa-phrases are part of focus.
    • Copala Trique Tone and Universal Tone Features

      Hollenbach, Barbara E. (University of Arizona Linguistics Circle (Tucson, Arizona), 1984)
      In motivating a set of distinctive features, at least three criteria must be taken into consideration. The first one is comprehensiveness: the features should provide enough distinctions to cover every opposition found in natural languages. The second one is phonetic reality: each feature should have some articulatory or perceptual basis (though many oppositions will, of course, need to be defined in relative, rather than absolute, terms). The third criterion is phonological function: the features should permit phonological (and perhaps morphological) processes to be stated in a simple and insightful way. Let us consider these three criteria as they apply to tone. Even though comprehensiveness is hardly an issue in segmental phonology, in tonal phonology it is, because there seems to exist a reluctance on the part of some linguists to admit the existence of complex tonal systems, and a tendency to look for abstract analyses that assign surface tone contrasts to some other parameter in underlying structure. For example, both Gruber (1964) and Yip (1980) provided feature specifications that permit only four contrasting tone levels in underlying structure even though a number of different languages with five contrasting levels oftone had been reported in the literature by the time they wrote. At the present time, there are at least nine such languages. Tone systems with four contrasting levels are even more common than those with five levels, and yet some linguists, such as Woo (1969) and Halle and Stevens (1971), provide features that permit only three levels of tone. One purpose of this paper is to document the existence of a tenth language with a five-level tone system, Copala Trique. A second purpose is to claim, on the basis of evidence from Copala Trique and from the published material on the languages cited in footnote one, that any proposed universal feature set for tone that fails to provide a unique specification for each of five levels is inadequate on the basis of the comprehensiveness criterion. At first glance, it may seem that the second criterion, phonetic reality, should present no problems in the area of tone, because tone has a well defined acoustic correlate, fundamental frequency. On closer examination, however, a number of problems arise, although I give them very little attention in this study. I include virtually nothing, for example, about the interaction between fundamental frequency and other phonetic parameters. I turn now to the third criterion, phonological function, and to the relation between the second and third criteria. These two sometimes conflict because different languages may impose differing phonological organization on very similar phonetic material. One important problem that involves both phonetic reality And phonological function is the issue of contour, or gliding, tones. Should they be treated as indivisible units in at least some languages, or should they invariably be decomposed into sequences of level tones? Because of space limitations, I do not discuss this issue in the present paper, but simply assume that gliding tones should always be decomposed. It is therefore necessary for phonological theory to provide features only for level tones. For a detailed defense of this position, the reader is referred to S. Anderson (1978, pp.146 -61) and to Yip (1980, pp.10- 30). A second problem that involves both the second and third criteria concerns binary versus scalar features. From a strictly phonetic point of view, fundamental frequency is a single, potentially multivalued, parameter. This fact can be captured simply and naturally by the use of a single scalar tone feature; such a feature is capable of handling any number of tone levels. At least one linguist, Stahlke (1977), has argued for this position. From a phonological point of view, however, binary features for consonants and vowels have proven so useful in expressing underlying oppositions and in writing rules that it seems desirable to employ them for tone as well. To my knowledge, all linguists except Stahlke who have proposed feature sets for tone have assumed that tone features should be binary, and I consider only binary features in the remainder of this study. It is clear that the choice of binary features to be used for partitioning a single phonetic parameter into three or more values cannot be made solely on a phonetic basis. There are various ways of juggling two features in order to characterize systems with three or four levels of tone, and there are even more ways of juggling three features in order to characterize systems with five levels of tone. In order to choose among these alternatives on a principled basis, it is essential to consider phonological function. Phonological processes differ significantly from language to language. In order to capture different kinds of processes in an equally insightful way, therefore, some latitude must be permitted in the way that features are selected and assigned. Most linguists who have proposed feature sets for complex tone systems, however, provide only a single choice of features, and a single way of assigning those features to different levels of tone. Even though such linguists often state that their feature set permits an insightful statement of phonological processes, their claims are usually based on a very small sample of languages. A third purpose of this paper is to claim that none of the feature sets for tone proposed to date permits enough latitude in either feature choice or feature assignment to capture the range of significant relationships among tone levels found in natural languages. Al l of these proposals are therefore inadequate as universal feature sets. To support this claim, I show precisely how each of them fails to provide an insightful description of Copala Trique tone. A fourth purpose of the paper is to present a set of three features that succeeds in capturing the significant relationships among the five tone levels of Copala Trique. The fifth purpose of this paper is to propose a new set of universal tone features that is flexible enough to accommodate the relationships among levels found in all tone systems described to date. I believe that phonological theory must provide five different underlying features, of which a language may select as many as three. I also propose a number of constraints on feature combinations.
    • A Correspondence Theory of Morpheme Order

      de Lacy, Paul; University of Massachusetts, Amherst (University of Arizona Linguistics Circle (Tucson, Arizona), 2000)
      The aim of this paper is to explain how the grammar distinguishes prefixes from suffixes. More generally, a theory that accounts for the variation in direction of attachment in both affixes and bound roots is presented, set within Optimality Theory. The core of the proposal is that direction of attachment is a property of morphemes; specifically, direction of attachment is indicated in the phonological string of an morpheme by an empty position. More formally, I propose that phonological strings can be partial functions from positions to phonological features; 'empty positions' are just those positions which do not map onto phonological features. The formalism behind this proposal is presented in section 2. The empirical implications of this approach are examined in section 3. Two phenomena are shown to follow straightforwardly from the present approach: (I) the implicational relationship between prefixes and suffixes (if a language has prefixes it also has suffixes, but not vice-versa - Hawkins & Gilligan 1988), and (2) the Affix Ordering generalization (that class I affixes must appear closer to the root than class II affixes - Siegel 1974). The typology of morpheme types produced by this theory is also discussed.