Underspecification Theory Research Articles

Morpheme Structure Constraints Solve Three Puzzles for Theories of Blocking in Nonderived Environments

In nonderived environment blocking (NDEB), a phonological process applies across morpheme boundaries or morpheme-internally when fed by another phonological process but is otherwise blocked. I present a theory of NDEB that attributes blocking to an interaction between morpheme structure constraints (which constrain possible underlying representations in the lexicon) and the usual phonological mapping from underlying to surface forms. The theory has some unusual aspects that make it conceptually suspicious, but I will argue that it receives empirical support. Using several case studies, I discuss three puzzles for theories of NDEB previously proposed in the literature, including the Strict Cycle Condition (Mascarό 1976), Kiparsky’s (1993) theory of underspecification, Sequential Faithfulness (Burzio 2000), Colored Containment (Van Oostendorp 2007), and Optimal Interleaving with Candidate Chains (Wolf 2008). I show that none of those theories can deal with all three puzzles and that the proposed theory with morpheme structure constraints succeeds. This result supports a dual-component architecture of phonology (as in Chomsky and Halle 1968) over architectures that eliminate language-specific morpheme structure constraints (i.e., the principle of Richness of the Base in Optimality Theory).

Emphasis Spread in Qassimi Arabic Within the Underspecification Theory

In Qassimi Arabic, the emphatic segments /tˤ, sˤ, ðˤ, rˤ/ spread their emphasis feature to the neighboring vowels turning the front feature of the vowels /ɪ,æ,æː/ into back. However, the front vowels /iː,eː/ are not affected by this process and consequently maintain their front feature in the adjacency of the emphatic segments. In this work, we provide a theoretical analysis of Emphasis Spread in the dialect of Qassimi Arabic within the underspecification theory. It has been concluded that the vowels /ɪ,æ,æː/ are underlying underspecified for the back feature, whereas the vowels /iː,eː/ are underlyingly specified for the back feature. The emphatic segments spread their secondary feature [Dorso-Pharyngeal] to the adjacent underspecified vowels and thus make them back. However, they fail to spread their secondary feature to the vowels /iː,eː/ because they have an underlying back feature. Therefore, unlike many other phonological theories, the underspecification theory can provide a more straightforward and precise analysis of Emphasis Spread. As a result, that would allow us to account for the various effects of emphatics examined in the different Arabic dialects, on the one hand, or, more broadly, any similar assimilation process in other languages, on the other.

Neurobiological evidence for voicing underspecification in English

In long-term memory, the phoneme units that make up words are coded for the distinctive features and feature values that are necessary to distinguish between words in the mental lexicon. Underspecification theory says that the phonemes that have unmarked feature values are even more abstract in that the feature is omitted from the representation altogether. This makes phoneme representations in words more sparse than the fully specified phonetic representations of the same words. Eulitz and Lahiri (2004) demonstrated that this theory predicts certain asymmetries in the Mismatch Negativity (MMN) response to phoneme contrasts. We expand on this research by demonstrating underspecification-driven asymmetry in the brain response to laryngeal feature contrasts in English (i.e. what makes /d/ and /t/ different). We add a new test by showing that the asymmetry disappears if the MMN paradigm is modified to encourage the formation of phonetic memory traces instead of phonemic memory traces. This result adds further neurobiological evidence that long-term phonological representations are more sparsely represented than phonetic representations.

On Gothic in the computer age

The Gothic language spoken some 1500 years ago in the area west and north of the Black Sea remains the darling of the older Germanic languages, as well as the preferred Germanic representative in Comparative Grammar. In the computer age researchers have readily accessible digitized Gothic texts, grammars, glossaries, bibliographies, as well as tagged corpora studies. In demonstrating the family resemblance of the inflectional morphology of the Gothic demonstrative pronoun, the strong adjective, and the third person pronoun, this paper makes use of the concept of inheritance networks, developed in computer linguistics, and of underspecification theory.

An analysis of vowel harmony in ciNsenga and ciCewa: A comparative study

This article presents an analysis of vowel harmony that occurs in the verbal system of ciNsenga and ciCewa. In these languages the nature of suffix vowels is determined by the nature of the root: mid vowels trigger the vowel /ε/ on the suffixes, non-mid vowels, on the other hand, trigger the vowel /i/ on the suffixes. An examination of monosyllabic verbs which, by definition, have no root vowel, reveals interesting patterns between different Bantu languages: in some languages such verbs consistently select suffixes with the vowel /i/ whereas in other languages such verbs select suffixes with the vowel /ε/. Drawing on insights from Underspecification Theory (Archangeli, 1984, 1988; Archangeli & Pulleyblank, 1989, 1994) the article argues that a unified and coherent account of vowel harmony can be made for these languages by positing that the languages are parameterised with respect to the harmonic feature specified in their underlying representations and the concomitant default rule needed to derive the surface representations.

Assimilation of word-final nasals to following word-initial place of articulation in United Kingdom English

Using very large speech corpora, we can study rare but systematic pronunciation patterns in spontaneous speech. Previous studies have established that word-final alveolar consonants in English (/t/, /d/, /n/, /s/ and /z/) vary their place of articulation to match a following word-initial consonant, e.g., “ran quickly” → “ra[ŋ] quickly.” Assimilation of bilabial or velar nasals, e.g., “alar[ŋ] clock” for “alarm clock,” is unexpected according to linguistic frameworks such as underspecification theory. The existence of systematic counterexamples would challenge that theory, but these might have been previously overlooked because they are infrequent. From the c. 8-million word Audio BNC (http://www.phon.ox.ac.uk/AudioBNC) we extracted more than 4,000 tokens of relevant word pairs, to determine whether non-alveolar assimilations occur and with what distribution. Word and segment boundaries were obtained by forced alignment, and F1–F3 formant frequencies were estimated using Praat. Formant frequencies in assimilation environments were compared to non-assimilating controls (e.g., them down vs. them back/then down). We also examined patterns of variability in different contexts. We will present evidence that velar and bilabial nasals sometimes do assimilate, though less frequently than alveolars.

The mental lexicon is fully specified: Evidence from eye-tracking.

Four visual-world experiments, in which listeners heard spoken words and saw printed words, compared an optimal-perception account with the theory of phonological underspecification. This theory argues that default phonological features are not specified in the mental lexicon, leading to asymmetric lexical matching: Mismatching input (pin) activates lexical entries with underspecified coronal stops (tin), but lexical entries with specified labial stops (pin) are not activated by mismatching input (tin). The eye-tracking data failed to show such a pattern. Although words that were phonologically similar to the spoken target attracted more looks than did unrelated distractors, this effect was symmetric in Experiment 1 with minimal pairs (tin-pin) and in Experiments 2 and 3 with words with an onset overlap (peacock-teacake). Experiment 4 revealed that /t/-initial words were looked at more frequently if the spoken input mismatched only in terms of place than if it mismatched in place and voice, contrary to the assumption that /t/ is unspecified for place and voice. These results show that speech perception uses signal-driven information to the fullest, as was predicted by an optimal perception account.

Anything is nothing is something

Three main types of syntactic contexts and accordingly three types of indefinites may be distinguished with respect to negation and polarity. This paper investigates diachronic changes of indefinites from one polarity type to another. A discussion of the development of the indefinites in German illustrates the effect these changes may have on the entire indefinite system of a language and crucially provides evidence for changes towards both ‘more negative’ as well as ‘more positive’. Additional data from a wide range of languages further supports the conclusion that polarity-type changes are not unidirectional. The observed changes are analyzed in terms of lexical underspecification theory as the result of one of only two possible diachronic operations: introduction of a plus-valued feature or complete loss of a feature in the lexical entries of the respective indefinites.

The breakdown of functional categories in Greek aphasia: Evidence from agreement, tense, and aspect

Background: Verbal inflectional errors are among the most prominent characteristics of aphasic nonfluent speech. Several studies have shown that such impairment is selective: subject–verb agreement is relatively intact while tense is severely impaired. A number of researchers view the deficit as structural and attribute errors to a breakdown of functional categories and their projections. Agrammatic individuals are thought to produce trees that are intact up to the Tense node and “pruned” from this node up. A partial preliminary report of the data was presented at the “Science of Aphasia V” conference in Potsdam in 16–21 September 2004. An earlier version of this paper was presented at the Departmental Seminars of the Department of Language and Communication Science at City University (February 2005) and appeared at the Reading Working Papers in Linguistics 6 (2005). We thank two anonymous reviewers for comments on an earlier version of this paper as well as the two reviewers of Aphasiology for their useful comments and suggestions. We also thank speech pathologists M. Diamanti, A. Xofillis, and M. Moudouris for referring the patients. Note: All co‐authors have contributed equally to this article. Aims: The present study investigates (a) the relative sensitivity of functional categories related to verbal inflection in Greek aphasia and the systematicity thereof; and (b) the relation between patterns of impairment in production and grammaticality judgements. Method & Procedures: We present results from a sentence completion and a grammaticality judgement task with seven Greek‐speaking aphasic individuals and seven control participants matched for age and education. Materials were constructed to assess three functional categories: subject–verb agreement, tense, and aspect. Eight verbs were used, balancing estimated familiarity and regularity of aspectual conjugation. Outcomes & Results: A great variability was observed among participants in overall performance but the pattern of performance was quite systematic. The results indicated that inflectional morphemes are not all impaired to the same degree in Greek aphasia. In both tasks, as a group, patients made more errors in aspect than in agreement. The group differences between tense and the other two conditions did not reach statistical significance. Moreover, a comparison of individual aphasic performance in the three functional categories indicated that in every case in which statistically significant differences were observed among the three functional categories, agreement was found to be less impaired than tense, aspect, or both. Conclusions: These findings do not support a global impairment of inflectional morphemes in aphasia but support a selective one and, in particular, a dissociation between agreement, on the one hand, and tense and/or aspect, on the other hand. Moreover, our findings do not support a hierarchical account along the lines of Friedmann and Grodzinsky (1997) but are compatible with Chomsky's (2000) Minimalist Program and with Wenzlaff and Clahsen's (2004) tense underspecification theory.

The Gothic Language: Grammar, Genetic Provenance and Typology, Readings by Irmengard Rauch (review)

MLR, 100.2, 2005 539 The Gothic Language: Grammar, Genetic Provenance and Typology, Readings. By Irmengard Rauch. (Berkeley Models of Grammar, 5) Bern, Berlin, and Brussels: Peter Lang. 2003. xxii+192 pp. ?39.40. ISBN 0-8204-3720-4. At firstsight, this book appears welcome, providing a useful collection of texts, with facsimiles of the principal source texts, the Codex Argenteus, the Skeireins, and the relevant pages of Busbecq's letters (fol. 136 recto and verso, which contains the sum of our knowledge of Crimean Gothic). Sadly, it reveals itselfto be a badly focused and ill-written amalgam of information that errs at both extremes: sophisticated schol? arship sits beside trivial and poorly expressed description, with unreliable, misspelt, and inconsistent examples. This looks like rushed editing, and a second edition could rectifyat least some of the damage. Anyone familiar with the well-worn paths of Gothic philology should welcome new approaches, because, although they often describe the 'facts' less adequately, they usually attempt to go beyond the merely descriptive to provide interpretations. The book promises the application to the familiar Gothic data ofmodern and even cur? rent linguistic theories: the laryngeal theory, the glottalic theory, underspecification theory, and optimality theory. These theories require some introduction regarding, in particular, the assumptions that underlie them, and the shortcomings in tradi? tional theory which they are intended to overcome. Unfortunately, this opportunity is missed. It might have been better to dispense with the Readings and Glossary sections, which constitute half the book, in order to do more justice to methodology. Moreover, the material is less accurate than it should be. Phonology is treated briskly,with greater emphasis on morphology. After 96 pages on grammar come a selection of texts (pp. 97-153) and a glossary to them (pp. 15583 ) that sensibly includes the proper names. The bibliography, especially the author's contributions, refutes the preface's assertion (p. xvii) that research into Gothic in the last fiftyyears has been 'moribund', but naturally, there is scant hope of discovering new primary material: even the 188th leaf of the Codex Argenteus that turned up at Speyer in 1970 contributed only one new word (p. 8). Some important research is omitted, notably that of Richard D'Alquen (Gothic AI andA U: A Possible Solution (The Hague and Paris: Mouton, 1974)), whose treatment ofthe Gothic digraphs cleverly combines chronological transmission and regional in? fluence in his interpretation of the sixth-century 'Ostrogothic' digraph spellings (ai), (au) as monophthongs, regardless of length. The original (lost) Visigothic language may have had both diphthongal and monophthongal sounds in opposition to each other in the relevant lexemes, but we do not know how they were spelt in Visigothic manuscripts. Biblical Greek, the textual 'matrix' for Gothic, must already have reduced its inventory ofsounds, as nowadays in Modern Greek, where 77, t,v,et,and 01are all pronounced monophthongally as [i:], while ai is [e] and ov is rendered [u:]. There is no reference, either,to several valuable Internet sites. While such sites undoubtedly 'decay', in the short term they provide access to material unavailable in libraries with poor holdings in Germanic philology. Historical work on the Goths is also ignored. The text has been written with such compression that it is sometimes difficult to understand, or misleading. Thus, possibly by analogy with 'Hispanic', the word 'Germanic' is applied to speakers of a Germanic tongue, e.g. 'such non-Germanics as the Finns, Balts [. . .]' (p. 1); but this implies ethnicity. 'Germanic speakers' is still preferable, whatever their ethnic origins. 'Gothic frequently is the designated prototype ofthe Germanic languages' (p. 10): but Gothic is not a reconstructed prototype; it can representa prototype because it is the oldest surviving Germanic language that is relatively well attested in continuous text. The genetic relationship of Gothic to other Germanic languages is oddly described: 'privatively it yields no consonant gemination ' (p. 11): this seems to mean that Gothic lacks consonant gemination, which 54-0 Reviews separates it fromthe other languages. Moreover, the examples cited have been altered: the adjective plaqus should appear in this form, not as *plaques (cf. Mark 13. 28). An example of Gothic proclisis is taken from 11 Corinthians 8. 18, where the original...

Underspecified Interpretations in a Curry-typed Representation Language

In previous work we have developed Property Theory with Curry Typing (PTCT, an intensional first-order logic for natural language semantics. PTCT permits fine-grained specifications of meaning. It also supports polymorphic types and separation types.1 We develop an intensional number theory within PTCT order to represent proportional generalized quantifiers like 'most', and we suggest a dynamic type-theoretic approach to anaphora and ellipsis resolution. Here we extend the type system to include product types, and use these to define a permutation function that generates underspecified scope representations within PTCT. We indicate how filters can be added to encode constraints on possible scope readings. Our account offers several important advantages over other current theories of underspecification.

Polish Negation and Lexical Resource Semantics

Using negative concord phenomena in Polish, we introduce and motivate Lexical Resource Semantics (LRS), a new framework for combinatorial semantics with type-theoretic representation languages in HPSG. LRS combines techniques first investigated within theories of semantic underspecification with HPSG-specific formal devices to derive fully specified logical representations of linguistic expressions.

Similarity Avoidance and the OCP

It has long been known that verbal roots containing homorganic consonant pairs are rare in Arabic, motivating the existence of an OCP-Place constraint (Obligatory Contour Principle on place of articulation) in the phonological grammar. We explore this constraint using an on-line lexicon of Arabic roots. The strength of the constraint is quantified by the ratio of the observed number of examples of each consonant pair to the number that would be statistically expected under random combination of phonemes. We show that the strength of the effect over all pairs is a gradient function of the similarity of the consonants in the pair. A similarity metric based on natural classes is developed, which solves the formal difficulties of contrastive underspecification theory while preserving the insight that contrastiveness plays an important role in perceived similarity. This metric is applied in an explicit model of the gradient OCP constraint, which achieves a better fit to the regularities and sub-regularities of the Arabic verbal lexicon than any prior approach. Lastly, we review evidence for the psychological reality of the constraint, for its existence in related forms in other languages, and for its cognitive/phonetic foundations in the speech processing system. We argue that the total body of evidence supports a model in which phonetic and cognitive pressures incrementally affect the lexicon, and phonotactic constraints are abstractions over the lexicon of phonological forms.

The differential substitution of English /θ ð/ in French

Summary One of the most interesting problems in second-language (L2) phonology is how to account for differential substitution. This is the phenomenon by which speakers who lack a certain segment (sequence) in their first language (L1) may adopt alternative language-specific replacement strategies in the L2 they are attempting to acquire. It has recently been claimed by Weinberger (1997) that the reason why, for example, Japanese learners of English systematically replace English /θ ð/ by /s z/ while their Russian counterparts always substitute /t d/ is that fricatives are unspecified for the feature [continuant] in Japanese while in Russian, the stops constitute the default obstruents. What is argued here is that this analysis in terms of Underspecification Theory cannot possibly work in the case of European and Canadian French which evince an equally systematic differential substitution of /θ ð/ to /s z/ and /t d/ respectively even though they have an identical system of underlying obstuents. It is also suggested that a perception-based approach to the thorny problem of differential substitution would appear to be a much more promising avenue of research.

A Course in Phonology (review)

Reviewed by: A course in phonology by Iggy Roca, Wyn Johnson Jeffrey Steele A course in phonology. By Iggy Roca and Wyn Johnson. Oxford: Blackwell, 1999. Pp. xxi, 725. Written in a clear and engaging style, Iggy Roca and Wyn Johnson’s book presents the essentials of nonlinear phonology, including an introduction to Optimality Theory. While examples are taken from English for the most part, some phenomena are exemplified with data from other languages (e.g. Turkish vowel harmony, Japanese pitch accent). The book is organized into three principal sections. Part 1, ‘Phonetics and phonology’, consists of a general introduction to articulatory phonetics and segmental phonology. Chs. 1–4 deal with the articulatory description of consonants. Using assimilation as a starting point, these chapters present the concepts of rules, derivations, distinctive features, natural classes, and autosegmental representation. In Chs. 5–7, R&J shift their focus to vowels. Beginning with a discussion of the vowel space, the learner is shown how the same types of distinctive features and autosegmental representations used for consonants can be applied to vowels. An examination of English vowels and dialectal variation constitutes the core of Ch. 7. Part 1 concludes with an introduction to the timing tier, including its role in the representation of complex segments and its potential as a mechanism for explaining the Great Vowel Shift. The discussion of the timing tier provides a good segue to Part 2, which examines suprasegmental phonology. Chs. 9 and 10 investigate syllable theory; the [End Page 589] concepts of sonority and constituency are presented and illustrated with examples from English. Cross-linguistic syllable typology is also discussed. Chs. 11 and 12 examine metrical phonology using grid theory. Ch. 13 integrates material encountered in the four preceding chapters through a discussion of syllable weight and its role in stress systems. Ch. 14 concludes Part 2 with an investigation of tonal phonology. Intonation, tonal and pitch accent languages, and the concepts of autosegmental association and floating tones are examined. Part 3, ‘Advanced theory’, investigates other topics in nonlinear phonology. Ch. 15 focuses on cyclic vs. noncyclic rules, primarily through a discussion of metrical grid construction in English. Ch. 16 investigates the concept of domain and its role in rule application; lexical phonology is introduced. In Ch. 17, R&J further investigate the lexicon through a discussion of lexical representation, markedness, and underspecification theory. Whereas Ch. 18 provides more study of rules and derivation, Ch. 19 introduces the concept of constraints and constraint interaction in Optimality Theory. Ch. 20 concludes with a summary. This book has many strengths. First, it presents the core of generative phonological theory using readily accessible examples. Second, it is clear and well-organized: The chapters follow logically, and summaries and margin notes highlight important concepts. Finally, much effort is made to guide the novice phonologist through the processes of hypothesis construction and evaluation. The book also has some weaknesses. In general, there is a lack of examples drawn from languages other than English. Second, some of the problem sets could be better constructed. Third, the chapter on Optimality Theory could be better integrated into the book as a whole. Finally, the ‘interactive’ grey boxes, intended to facilitate the reader’s understanding, are excessive in number and interrupt the flow of the text. However, all in all, this is indeed a very sound textbook. Jeffrey Steele McGill University Copyright © 2002 Linguistic Society of America

The Semantics of Lexical Underspecification

In this paper, I examine the mechanisms responsible for enabling underspecified lexical forms to acquire a determinate and unique interpretation in the composition of the sentence. In particular, I study the behavior of several classes of lexical items exhibiting degrees of systematic or logical polysemy. These words present an interesting challenge to conventional treatments of polysemy and full specification in natural language semantics. Extending the analysis of these classes presented in Pustejovsky (1994,1995), I elaborate the formal mechanisms responisible for capturing the syntactic and semantic behavior of nouns such as lecture, prize, book, and lunch, and verbs such as read, rent, and break. I argue that members of these classes must be represented as complex types, as they do not allow for simple conjunctive typing; furthermore, these nouns and verbs display a peculiar behavior regarding the quantificational force of the type which is absent from other nominal and verbal classes. I conclude by discussing the consequences of this analysis for a theory of semantic underspecification.

Long-distance place assimilation with an interacting error pattern in phonological acquisition

Two commonly occurring and independent error patterns in children's early speech are examined to determine how and to what extent they might interact. One error pattern replaces velar consonants with coronals, and the other replaces a coronal with a consonant that agrees in place of articulation with some other consonant elsewhere in the word. A range of interactions is observed within and across children with regard to whether the product of one error pattern can serve as the target of the other. The different interactions motivate different claims about the nature and substance of children's underlying representations, which in some cases may differ from those of the ambient system. An extension to underspecification theory is advanced which allows underlying representations to be radically underspecified and in certain cases also to be specified for a default feature.

Consonant Harmony as a Compensatory Mechanism in Fluent Aphasic Speech

This study addresses how fluent aphasics construct complete phonological representations, given the premise that their phonological speech errors result from faulty information about stored lexical representations. We explored whether consonant harmony, a common rule-governed process of feature copying, operates as a compensatory device for completing phonological representations in fluent aphasia. This was examined in a corpus of phonemic paraphasias (n = 543) produced by 8 fluent aphasics during picture naming. Consonant substitutions due to a single feature change (n = 143) were analyzed for the properties of consonant harmony predicted by the phonological principles embodied in a Universal Markedness version of Underspecification Theory (e.g., Chomsky and Halle, 1968). Results indicated that harmony constrained the feature substitution errors involving the feature class of voice (e.g., calendar→ /gæl▪viŋ/), but not place of articulation (e.g., igloo→/ıdlu/); substitutions due to an error in manner were rare. These findings were used to argue that for English-speaking fluent aphasics a consonant harmony rule for the feature voice is incorporated into a compensatory output mechanism that is used to complete faulty lexical-phonological representations.

The representation of the place of nasal stops in word recognition

Predictable aspects of the representations used in the perception of a word may not be necessary for recognition of the word to be successful. The present study investigates whether relative predictability of place in nasal stops affects recognition of words. Subjects were given an auditory lexical decision task in which each target time was preceded by an auditory prime. Target words ended in /mp/ (BUMP), /nt/ (BUNT), /m/ (BUM), or /n/ (BUN). Primes ended in /m/ (UM), /nl/ (UN), or /k/ (UK). Place of the nasal stops is predictable in /mp/ and /nt/ words given the place of the following obstruent. Some current theories of phonological underspecification also claim that place in /n/ words is predictable, assuming that coronal is universally the default place. It was found that /mp/, /nt/, and /n/ words were primed equally well by /m/ and /n/ primes, implying that the specifics of place were not relevant in these words. By contrast, /m/ words were primed significantly better by /m/ primes relative to /n/ primes, implying that the representation of place in /m/ words is relevant for recognition. [Work supported by NIDCD.]

Underspecification in texistepec popoluca phonology

Abstract The theory of underspecification (Archangeli 1984, Pulleyblank 1986, 1988, Ito and Mester 1989, Mohanan 1991, papers in Yearbook of Phonology 5), received its primary motivation from the analysis of phoneme - especially vowel - inventories. Additional motivation is needed from the workings of phonological rules (as well as typology). It is my purpose in this article to treat some aspects of Texistepec Popoluca phonology that testify to the necessity of some version of underspecification theory in addition to a non-linear approach to phonology (Goldsmith 1976, Clements and Keyser 1983).