Zwicky (1985) has argued that an adequate realizational theory of morphology must incorporate RULES OF REFERRAL in order to account for some kinds of inflectional syncretism. In this paper I propose a formal theory of rules of referral within the broader framework of PARADIGM FUNCTION MORPHOLOGY. This theory affords a precise account of a range of rule interactions involving rules of referral; in addition, it furnishes a simple explanation for the fact that syncretisms don't always encompass whole words, for the fact that some referrals are bidirectional, and for the fact that two or more referrals may participate successively or simultaneously in the definition of a single instance of syncretism.

In this paper I present a tonal analysis of inflection, derivation, and compounding for Standard Lithuanian nominals. I argue that nominal accents in this language surface as a result of rules of high tone insertion and high tone deletion, and that accented morphemes are represented with underlying high tones linked to the appropriate moras. The predictability of circumflex accents in stem-initial and stem-final contexts for unaccented stems is handled by invoking limited use of extraprosodicity and floating high tones. In addition to improving the empirical coverage of earlier proposals, this analysis offers three central arguments for underlying tones as opposed to accentual diacritics or metrical representations. First, the general status of diacritic accents is suspect; various pitch-accent systems have been successfully reanalyzed without the use of such diacritics, and independent motivation for such diacritics outside the realm of pitch-accent systems is meager. Second, there is evidence of tonal stability under segment-deletion and demorification in Lithuanian, while other accentual alternations such as that captured by de Saussure's Law have markedly tonal characteristics. Finally, tonal representations are supported by data from a Zhemayt dialect in which words may surface with multiple accents.

We argue that young German children have the major functional sentential heads, in particular the inflectional and complementizer systems. The major empirical basis is natural production data from a 25-month-old child. We perform quantitative analyses which show that the full complement of functional categories is available to the child, and that what crucially distinguishes the child's grammar from the adult's is the use of infinitives in matrix clauses. The evidence we consider includes the child's knowledge of finiteness and verb placement, agreement, head movement, and permissible word-order variations. We examine several accounts which presuppose a degenerate grammar or which deviate from the standard analysis of German and conclude that they provide a less adequate explanation of the acquisition facts.

This formal study of morphologically induced rime-change (bianyun) phenomena in Chinese argues that rime change is effected by affixation of subsegmental morphemes and that the derived forms are subject to prosodically defined templatic constraints. The Affix Manifestation Principle is proposed to account for the tendency of affix preservation. Various processes are motivated to manifest the effect of affixation and to satisfy the templatic constraints. This study not only shows that one unitary process accounts for different cases of rime change and crossdialectal variation but also has broader theoretical implications regarding the notion of maximality under affixation, the nature of output templates, and the light-heavy distinction within moraic theory.

Though both Chinese and Japanese allow resultative compounds of the form Vcause-Vresult, Japanese resultative compounds do not show the semantic ambiguities that their Chinese counterparts have. I argue that this difference results from the interaction of three factors. First, resultative compounds are morphologically head-initial in Chinese but head-final in Japanese. Secondly, resultative compounds are also associated with an event structure whose head is always represented by Vcause of the compound. Thirdly, a universal condition requiring iconic representation of the temporal relations between two subevents makes Vcause linearly precede Vresult. The result of this work supports the claim that many differences among languages can be accounted for with universal principles plus simple parametric variations.

Since McCawley 1970, the Luganda tone system has been well known for its property of allowing at most one H[igh] to L[ow] pitch drop per word. To account for this property, the underlying system has been analyzed both in terms of underlying accents of various kinds (e.g. diacritic) and in terms of underlying tone (e.g. H vs. 0). Most accentual proposals, however, fail to account for the fact that THREE marks are necessary to characterize the high to low ‘melody’: a mark for the first H mora, a mark for the place of the H-to-L drop, and a mark for the place of the last L mora. After evaluating previous accentual and tonal analyses, we present a new approach to tone in Luganda that integrates tone and accent in the following way: (i) Accent in Luganda consists of designating certain moras as metrically strong (and hence capable of attracting tone); (ii) Tone in Luganda consists of lexical and grammatical occurrences of underlying HL contours (or ‘potential pitch drops‘). This new analysis, which continues to recognize the importance of the H to L pitch drop, provides additional evidence for the coexistence of tone and accent, which may interact in complex ways in the same language.

Natural languages are characterized by standards of well-formedness. These internal standards are likely to be, at least in part, a product of a consensus achieved among the users of a language over time. Nevertheless, it is possible that an individual, attempting to invent symbols to communicate de novo, might generate a system of symbols that is similarly characterized by internal standards of well-formedness. In these studies, we explore this possibility by comparing (1) a conventional sign language used by a community of signers and passed down from generation to generation with (2) gestures invented by a deaf child over a period of years and (3) gestures invented by nonsigning hearing individuals on the spot. Thus, we compare communication in the manual modality created over three different timespans—historical, ontogenetic, and microgenetic—focusing on the extent to which the gestures become codified and adhere to internal standards in each of these timespans. Our findings suggest that an individual can introduce standards of well-formedness into a self-generated gesture system, but that gradual development over a period of time is necessary for such standards to be constructed.

In this paper we propose six implicationally related cognitive statuses relevant for explicating the use of referring expressions in natural language discourse. These statuses are the conventional meanings signalled by determiners and pronouns, and interaction of the statuses with Grice's Maxim of Quantity accounts for the actual distribution and interpretation of forms when necessary conditions for the use of more than one form are met. This proposal is supported by an empirical study of the distribution of referring expressions in naturally occurring discourse in five languages—English, Japanese, Mandarin Chinese, Russian, and Spanish.

Jakob Wackernagel is well known for the observation that clitic elements generally appeared in second position in their clause in early Indo-European languages. It is less well known that he also proposed this as the basis of the ‘Verb Second’ phenomenon. Finite Verbs in main clauses were apparently accentless in Indo-European, and he suggested that they were therefore treated as clitics and located in second position. This paper examines the possibility that there is indeed a fundamental relationship between Verb Second and clitic placement.

I first argue that accounts of Verb Second do not extend correctly to the placement of clitics: if there is something common to these phenomena, it must involve generalizing clitic behavior to inflected Verbs, rather than vice versa. The range of clitics in the languages of the world is examined, and a set of parameters for the rules that locate them is formulated. I argue that these rules form a class with the rules of morphology that introduce affixes (and other phonological changes) into words. A unitary generalization is suggested: rules that provide formal markers of what Sapir called ‘relational’ or ‘derivational’ concepts realize these categories at a prominent anchor point (the beginning, the end, or the structural head of the material being marked). This class of rules also includes processes that transfer morphosyntactic features to an anchor position in the phrase, where they are reflected formally. I then suggest that existing accounts of why Verbs should move to their surface position in Verb Second languages lack independent justification; furthermore, they incorrectly miss the generalization that the Verb in such cases is always in second position. Icelandic constructions are discussed that suggest this position is not structurally uniform. Treating Verb Second as the result of a rule which says ‘realize the inflectional features of a clause immediately after its initial element’ unifies several cases and also expresses the connection between Verb Second and other second-position phenomena. I conclude that Wackernagel was right in proposing a unification of Verb Second and cliticization, though the generalization involved is not based on properties of word accent.

This article explores the nature of morphological categories by examining two of the most ubiquitous denominal adjective types, the possessionat (bearded) and similitudinal (friendly) adjectives. Sometimes these adjectives are morphologically distinct, sometimes not; compare icy road and icy hand. This suggests that they are related yet not related. Language provides two means of generating these adjectives: TRANSPOSITION, which alters the lexical category of the base but not its grammatical features, and FUNCTIONAL DERIVATION, which alters the latter but not the former. This article shows how both adjectives may be derived simultaneously by both means. In order to understand fully the semantic effect of transposition on the base, it is necessary to clarify the distinction between the morphological and semantic categories involved and to demonstrate the impossibility of reducing either set of categories to the other.

A commonly made, but rarely defended, assumption is that phonetic reduction processes apply to hyperarticulated phonetic targets. Results from experiments reported in this paper support this assumption. In various experimental conditions, listeners adjusted the input parameters of a speech synthesizer until the vowels it produced sounded like the vowels found in a set of example words. A preliminary study indicated that the method of adjustment is a feasible tool for studying vowel systems. Interestingly, listeners in the study chose vowels that were systematically different from those measured in productions of the set of example words: high vowels were higher, low vowels were lower, front vowels were farther front, and back vowels were farther back. We hypothesized that this extreme vowel space corresponds to phonetic targets that are hyperarticulated: HYPERSPACE. This hypothesis was tested in the two main experiments. The first experiment controlled for possible effects of instructions and phonetic training on the listeners' choices. In the second experiment, we improved the naturalness and distinctiveness of the synthetic vowels. The results indicate that the extreme vowels chosen by the listeners were consistent with those produced in hyperarticulated speech; moreover, the hyperspace effect is robust across experimental conditions. These results validate the hypothesis that phonetic targets are hyperarticulated, and are consistent with a two-stage model of phonetic implementation: at the first stage distinctive features are mapped to hyperarticulated phonetic targets, and at the second stage these phonetic targets are reduced.

This paper argues for a feature called Sonorant Voice (SV), which replaces the traditional feature [sonorant]. It is underlyingly present in all sonorants and in obstruents that function as sonorants within a system. This feature differs from [sonorant] in several ways: (a) it defines the type of voicing present in sonorants, (b) it can be present in obstruents, and (c) it bears a special relationship to the feature Nasal in that a nasal consonant is the unmarked realization of a consonant with the feature SV. Two major kinds of evidence are given for SV. First, alternations between voiced stops and nasals in some languages suggest that the two are allophones (e.g. Southern Barasano, Slave) or form a natural class (Rotokas), a relationship that is captured through the feature SV. Second, cases in which voicing and a nasal surface in the same environment (e.g. Navajo, Irish) can be accounted for by the use of SV.

Since Brown 1973, the study of the acquisition of the items generally referred to as ‘grammatical morphemes’ has tended to focus on inferring the linguistic and cognitive factors governing the ORDER in which these items reach some designated level of acquisition. What is less often addressed is the PROCESS of acquisition of these common inflections and closed-class functors: how it unfolds, why children learning the same language take such different paths, why some morphemes are harder to learn than others, and why the path to adult performance is so uneven. We adopt a microgenetic approach to these questions, using longitudinal data from two children learning English in extremely different ways. One produces protomorphemic filler syllables from which true grammatical morphemes develop smoothly; the other has no filler syllables and displays a more discontinuous developmental pattern, marked by false starts. The production of particular grammatical morphemes as filler syllables or other precursor forms is influenced by phonological salience, position in the utterance, frequency in the input, and morpheme homonymy. Turning to the growing body of crosslinguistic data on protomorphemes, we then propose eight general attributes of morphological systems which, we predict, will promote or inhibit the emergence of filler syllables during development. Finally, we argue that the pattern of smooth development via protomorphemes and their apparent adult-language analogues requires models/theories of morphology that allow for a state of ‘partial knowledge’ of a morpheme within each level of representation.

‘Geminate inalterability’ is the well-known phenomenon whereby a rule affecting singleton segments fails to alter comparable geminates. Past accounts have attributed inalterability to geminate-specific constraints on rule application. Two new generalizations point in a radically different direction. First, rules respecting inalterability are purely structure-filling (never structure-changing). Second, inalterability occurs also among singletons. We develop a new theory attributing inalterability to prespecification: geminate (and singleton) inalterability follows solely from the Elsewhere Condition. This explains not only the convergence of regular geminates and irregular singletons, but also the asymmetry among rule types sensitive to inalterability. Geminate-specific constraints on rule application are neither necessary nor sufficient to account for inalterability.