Betty S. Phillips – Lexical Diffusion: Conceptions and Misconceptions

Lexical diffusion happens when "a phonological rule gradually extends its scope of operation to a larger and larger portion of the lexicon, until all relevant items have been transformed by the process" (Chen and Wang 1975). For example, the laxing of early Modern English /u/ has affected the word foot, but not boot or loot; root and soot are variable. Despite the apparent simplicity of Chen and Wang's definition, a number of misconceptions about lexical diffusion have nevertheless stood in the way of an answer to the fundamental question posed by Labov (1981), namely, "In the evolution of sound systems, is the basic unit of change the word or the sound?" This talk addresses some of those misconceptions.

Lenore Grenoble – Contact-induced change, language attrition and incomplete learning: Russian-Evenki contact in Siberia

In this talk I report on the findings of recent fieldwork conducted on language shift and attrition among Evenki-Russian speakers in Tura, the center of the former Evenki Autonomous Area (now part of the Krasnoyarsk Region). Evenki is a Tungusic language spoken in Siberia by approximately 5000 speakers, none of whom are monolingual. It is characterized by rich agglutinative morphology, both inflectional and derivational, ATR vowel harmony, and SOV word order. Although it has been in contact with Russian for centuries, it has been undergoing shift since the post World War II period, a shift which, if anything, has accelerated in the post-Soviet era. As the regional administrative center, Tura is home to people originating in a number of dialect regions, which has an ethnic Evenki population of approximately 1000, although the total number of speakers here is smaller. Changes are occurring in phonology and morphosyntax, including: (1) confusion of [s] and [h]; (2) changes in vowel length; (3) changes in the system of vowel harmony; (4) morphological shrinking, in both derivational and inflectional morphology, with the complete loss of some categories; and (5) a restructuring of the verbal system based on Russian models. I argue that these changes are systematic, not idiosyncratic, and can be fruitfully compared to the morphosyntactic changes found in incomplete language acquisition of heritage speakers. I further argue that he sheer number of different dialects spoken in this small, coupled with the imperfect learning of an artificial standard language which has been imposed by the administration have accelerated language shift and, moreover, have influenced the direction of change.

Hideki Kishimoto – Focusing and Subject Positions in Japanese

In this presentation, I look at several different types of subjects in Japanese, and shows where they are located in clause structure. Even though it has been very difficult to pinpoint the position of subjects owing to the SOV order of the language, it is shown that their position can be readily evaluated by examining the focusing domain of the particle mo ‘also’ attached to the verb. In particular, this test reveals that nominative and dative subjects are raised to the clause subject position identified as Spec of TP, whereas other obliquely-marked subject remain in VP-internal subject position (unless nominative-marked arguments appear in the clause). In the light of the facts of subject raising, I suggest that in Japanese, a nominative argument needs to be present in the clause in order for a tense head to attract a subject.

David Odden – Features Impinging on Tone

This talk focuses on the nature of phonological features, especially their function as definers of segment-class for rules, and the question of their physical manifestation, specifically concentrating on the representation of tone in a substance-free model of phonological computations and representations. The paper examines typologically unusual interactions betweens segmental properties and tone, some involving laryngeal features and some involving vowel height. It is shown that, following classical autosegmental ''assimilation as spreading'' formal assumptions are adopted in Unified Features Theory, identification must be made between both vowel height and voicing features on the one hand, and tone height on the other. Such identification is possible only by abstracting away from low-level phonetics and concentrating on phonological patterns

John Goldsmith – Optimization is the answer. Now, what is the question?

The goal of this presentation is to offer a brief introduction to a view of linguistics which is empiricist, and which puts a heavy emphasis on the character of language learning, without being cognitivist. It is a view that says that the goal of the linguist is to understand how language can be learned, a goal distinct from that of the psychologist, who aims to understand (in a different sense) how language is learned. A more complete abstract is available for download here.

John Bailyn – Paradoxes of Binding Theory (and their possible resolutions)

Abstract: Certain aspects of Binding Theory (BT) have remained remarkably steady despite the great changes that other components of the generative model have undergone in recent years. It has been generally agreed for some time that anaphors (reflexives and reciprocals) require a local, c-commanding antecedent (=Principle A), while pronouns do not allow a local c-commanding antecedent (=Principle B), as shown in (1-2) (attempted co-reference indicated in bold).

    1. John respects himself.
    2. *John respects him.
    1. *[Friends of John] respect himself.
    2. [Friends of John] respect him.
C-command distinguishes the cases in (1) from those in (2), and Principles A and B predict exactly the co-reference possibilities found. At the same time, recent changes in the general shape of the grammar from being primarily representational to being primarily derivational have led to some unintended consequences for BT. This talk will present basic motivation for the shift to derivational binding, whereby BT applies at every stage of a derivation, and examine the consequences for both Principles A and B, looking primarily at data from English, Japanese and Russian. Issues of this sort come to light through examining the interaction of various kinds of movement with binding effects, in an attempt to determine which stages of a derivation are subject to BT. Beginning with Principle A, where the issues are much clearer, we see that there is an apparent paradox involved in attempting to analyze binding phenomena derivationally — some data happily support a derivational analysis, but other data remain resistant to anything but a representational analysis (one that applies only at certain levels, but not during the course of the derivation). I show that the Principle A paradox has an elegant resolution, given particular assumptions about the interpretive interface and formal features, which may allow us to salvage the derivational approach even for Principle B, where the issues are murkier, and the paradox of levels harder to resolve.

Alexaner Francis – Features, not phonemes: Implications of laboratory training studies for L2 acquisition

It is well-known that a listener's native language (L1) can profoundly influence the perception of the sounds of another language (L2). Such effects are typically ascribed to the influence of the phonological inventory of the listener's L1 on the processing of the sounds of the L2, but most theories lack a clear mechanism to explain how such influence might be exerted. In this talk I will present data from a series of studies in which listeners were trained to better identify a variety of unfamiliar sounds, especially non-native lexical tone, vowel and consonant contrasts, and artificially manipulated speech and non-speech signals. On the basis of these results, as well as others’ work on the neural basis of lexical tone perception, I will argue that the effects of L1 experience on L2 perception might best be understood in terms of the effects of deeply entrenched perceptual biases that favor the processing of particular acoustic cues and/or disfavor others, and that phonemes as such may have little or no role to play in the development or maintenance of such biases.

Isabelle Darcy – The nature and scope of phonological knowledge in a first and a second language: Insights from phonological alternations

Abstract: In this talk, I will present some work in progress seeking to understand the scope and the nature of phonological knowledge, in a first and a second language. I will focus on phonological alternations, and the way listeners compensate for systematic variation in spoken language. The word recognition system copes easily with the great variability of spoken language. A particular kind of context-dependant variation is phonological assimilation (e.g., in French voicing assimilation, robe [rob] “dress” is pronounced [rop] when followed by sombre [sombR] “dark”). Previous research proposed that listeners use abstract phonological knowledge of alternations in their L1 to compensate for assimilations (Gaskell and Marslen-Wilson, 1996, Darcy et al., in press). The studies reported in this talk explore more in depth the representation of such knowledge in a first and a second language. I first report a study examining occurrences of categorical assimilation (neutralizations) in French, the perception of voiced and unvoiced word-final obstruents in different phonological contexts. We first show the categorical nature of the alternation (Exp. 1), supported in Exp. 2 by perceptual categorization data. In Exp. 3, with 18 (other) French listeners, the interpretation of this first percept appears to be corrected in certain contexts, inducing compensation. We argue that context effects are phonological in this case, rather than auditory or phonetic. We conclude that linguistic knowledge of alternations is necessary in compensation for categorical assimilation (Darcy and Kügler 2007). Experiment 3 confirms that French strongly compensates for voicing assimilation in the viable condition. However, compensation is asymmetrical, and seems to reflect a difference in production (categorical voicing but gradient devoicing) observed in Snoeren et al. (2006). When hearing categorical assimilations (which are less common in the case of devoicing assimilation), voicing is compensated for much stronger than devoicing (78% vs. 51%, p<.0001), whose compensation is less efficient because categorical occurrences challenge the recognition system. This suggests a different representation for each kind of alternation (Darcy and Kügler 2007). I will propose different explanations of this asymmetry.

The second part of this talk will consider the nature of this phonological knowledge in a second language. Recent research provided support for the evidence that highly proficient late learners had acquired the ability to compensate perceptually for the assimilatory pattern of their L2 (Darcy et al., 2007), suggesting that they developed a comparable phonological knowledge of alternations for L2.

In Experiment 4, 21 Americans, late learners of French with long (n=8, >2 years) or short (n=13, <2 years) exposure, were tested on the same stimuli as the French listeners in Experiment 3. The preliminary results show that beginners compensate less than more experienced learners (p<.004) and French listeners (p<.009). Advanced learners are similar to the French listeners (p>.1). However, no difference between voicing and devoicing was observed in American English learners, but a main effect of “L1” (French-English, p<.003). This suggests that learners might use a different processing or build less detailed representations than French listeners (Sebastian-Gallés et al., 2006). Further research is needed to understand better this phenomenon and its representational issues.

Chilin Shih – A Phonetic Implementation Model of Mandarin Tones

The best way to understand the phonology/phonetics interface is to build a phonetic implementation model that generates surface variations from the phonological representation. This talk describes such a model using Mandarin tones as an example.

In Chinese conversational speech, sometimes the pitch contours move in the opposite direction of the expected lexical tone, yet native speakers cannot hear the "errors" and consider the production to be natural. It is argued that this phenomenon is a general reduction process that occurs in all languages and that the level of reduction can be simulated using weights. The weights represent the speaker's balancing act to meet two conflicting demands in speech communication: for ease of production, the speaker would like to minimize articulatory effort; for ease of perception, the speaker would need to maintain production accuracy. Local reduction/inversion is then explained as global optimization.

I will present relevant data and modeling solutions. The result suggests a principled way to represent phonology and phonetics, which maintains a level of abstraction in phonology and the richness of details in phonetics.

Yvan Rose – Addressing Child Language Phonological Phenomena in Their Proper Context: Empirical and Formal Considerations

In this presentation, I will discuss a series of speech patterns produced by young children. Some of these patterns pose serious challenges, if approached from a grammatical perspective only. I will argue that a more satisfactory explanation requires to look not only at properties of the production grammar (e.g. markedness, faithfulness) but also potentially non-grammatical factors such as perception-induced erroneous lexical representations and at how physiology- and motor control-related issues may affect the production of speech sounds in young children. Building on this observation, I discuss challenges posed by the determination of the source of these phenomena. I will also discuss how the multi-factorial analyses implied by the analyses of such phenomena could be encoded in formal systems.

Elena Benedicto – What must be, and what needs not be, in obligation modality: Mayangna ghost modals

Though Mayangna has some modal elements in the form of stative verbs, it does not have a modal expressing obligation. The expression of obligation is accomplished via a combination of morpho-syntactic elements, none of which, in isolation, denotes obligation. This talk will be dealing with: (a) how the syntax of the construction works in mayangna, (b) how a null modal element may not be the best option in explaining the phenomenon, and (c) how an analysis that maintains compositionality may look like. The analysis will concern the nature of infinitivals (and the existence of person-inflected infinitivals) and the possible existence of an evidentiality system in the language. Mayangna is a member of the isolate Misumalpan family in Central America. It is spoken by the Sumu people in the RAAN region of Nicaragua. This research has been conducted under a participatory research approach in conjunction with the Mayangna Linguists Team (Mayangna Yulbarangyang Balna).

Carmel O'Shannessy – Language contact and acquisition: a new mixed language in northern Australia

A new mixed language, Light Warlpiri, has emerged in a remote community in northern Australia. It is spoken by children and young adults in the multilingual community of Lajamanu and has developed within the last 30 years. Light Warlpiri is a verb-noun mixed language, meaning that it cannot be traced to a sole parent language, and that its verbal and nominal components tend to come from different source languages. Most verbs and the verbal morphology are from Aboriginal English or Kriol (AE/Kriol), while most nominal morphology is from Lajamanu Warlpiri (the variety of Warlpiri spoken in Lajamanu community). Nouns are drawn from both types of source language. An innovative auxiliary system has developed which draws on, but is not the same as, the systems in the source languages. But the system for indicating grammatical functions draws directly on the typologically different source languages. Lajamanu Warlpiri uses case-marking in an ergative-absolutive system while AE/Kriol uses word order (SVO) in a nominative-accusative system. In Light Warlpiri these two systems meet and are in functional competition. The structure of Light Warlpiri, and code-switching patterns of older speakers in the community, provide empirical evidence that languages of this type can arise from alternational code-switching practices. The language ecology in the community is complex, and code-switching between languages is very common. The complex contact situation, in which there is also rapid change, raises the question of how much variation there is in how grammatical functions are indicated within each of the two main languages spoken, for example, between different age groups. Analyses show that adults and children who speak both Lajamanu Warlpiri and Light Warlpiri distribute ergative marking differently in each language - they use it more often in Lajamanu Warlpiri and less often in Light Warlpiri. Children produce more regular patterns of interaction between case-marking and word order than adults do, suggesting that they are active agents of language change.

Damir Ćavar – The Croatian Language Repository: Quantitative and Qualitative Resources for Linguistic Research and Language Technologies

This presentation will shed a light on the background of the language resources and technologies program "Croatian Language Repository," and its core project "Croatian Language Corpus." We will discuss its current status and future perspectives, and the various related sub- projects. The repositories role as the central data basis is to provide large amounts of high-quality linguistic data and resources of various types (e.g. textual corpora, dictionaries, spoken language recordings), from different genres and domains, covering long time periods, and diachronic and synchronic aspects of variants and dialects. It serves for lexicographical, lexicological, and grammar modeling research, including computational dialectology, language evolution and change studies, as well as the development of language technologies.

On the one hand, we will touch on the role of this type of work and research for purposes of language documentation and standardization, focusing on Croatian, and its role in micro- and macro-political developments. On the other hand, we shed a light on the underlying technological and research goals, which involve, among others:

  1. Interoperable and standardized language resources and technologies:
    1. designed and developed on the basis of current technologies for annotation and representation of language data (e.g. XML, Unicode)
    2. obeying international guidelines and standards for linguistic annotation (e.g. TEI, TBX)
    3. with the resulting maximization of interoperability on the technological and data level (e.g. XML and Unicode), as well as on the linguistic information and representation level (e.g. GOLD and Ontology-based annotations and language processing components)
  2. Enabling language technologies for theoretical, and potentially applied computational linguistics (processing tools for language analysis and automatic annotation, and high performance natural language processing with cluster technologies for quantitative processing maximization)
  3. Enabling quantitative and qualitative linguistic research in the domain of e.g. dialectology, language evolution and change, linguistic contact phenomena (qualitative, but also quantitative multivariate analyses of dependencies, variation, and covariation across linguistic levels and cross-linguistically)

Catherine Rudin – Multiple Wh-Fronting, Free Relatives, Correlatives, and the [±MFS] Split

Nearly 20 years ago Rudin 1988 showed that multiple wh fronting is not a unified phenomenon: In some languages (dubbed +M(ultiply)F(illed)S(pecifier)), the wh words all undergo wh-movement to SpecCP, while in other, of the wh words front to other positions. Since that time a great deal of work on multiple wh fronting in Slavic and other languages has considerably refined the typology of multiple wh fronting. The vast majority of this work, however, has dealt only with multiple questions. Despite occasional notes that multiple wh fronting also occurs in certain relative clauses, multiple wh relatives have received little attention.

This talk is a first step toward rectifying this oversight, mostly by raising questions and suggesting avenues for further research on multiple wh relative clauses. Very broad and basic questions need to be asked: How are multiple wh relatives like or unlike multiple wh questions? How are they like or unlike other relative clauses? Are they free relatives or correlatives? (There are no headed multiple wh relatives.) What can they tell us about the structure of (non-multiple) free relatives/correlatives? How do they fit into typologies of multiple wh fronting?

A comparison of multiple relatives in several languages provides insight into many of these questions. Polish, a classical -MFS language, has only multiple correlatives, while the classical +MFS languages, Bulgarian and Romanian, have multiple free relatives as well. This suggests that multiple free relatives but not correlatives require a structure with all wh words in SpecCP, and furthermore that multiple relatives, like multiple questions, differ in wh landing sites from one language to another (and sometimes within one language). Parallelism of Superiority effects in questions and relatives across languages and constructions supports this conclusion. The existence of multiple free relatives favors a Comp Account approach to free relatives due to semantic and syntactic problems with multiple heads.

Gregory Stump – Principal parts and morphological typology

Gregory Stump
University of Kentucky
& Raphael Finkel
University of Kentucky

Like the numbers in a sudoku puzzle, a lexeme’s principal parts provide enough information—but only enough—to deduce all of the remaining forms in its paradigm. Language teachers have long recognized the practical usefulness of principal parts, but their significance extends beyond the domain of language instruction. Because principal parts are a distillation of the implicative relations that exist among the members of a lexeme’s paradigm, they afford an important (but heretofore neglected) basis for typological classification, as we show here. For present purposes, we focus purely on a language’s verbal paradigms.

We recognize three logically distinct sorts of principal-part systems that might be postulated for a given language: in a static system, all lexemes belonging to the same category have the same members of their paradigm as their principal parts; in an adaptive system, by contrast, all lexemes have the same member of their paradigm as their first principal part, but the exponence of a lexeme’s nth principal part determines the identity of its (n+1)th principal part; and in a dynamic system, the number and identity of a lexeme’s principal parts may vary freely from one inflection class to another.

We propose five cross-cutting criteria for the typological classification of principal-part systems. (For present purposes, we focus on dynamic systems.)

The application of Criterion A (How many principal parts are needed to determine a lexeme’s paradigm?) in a given language depends on whether one assumes a static, an adaptive, or a dynamic system of principal parts; in general, a static system involves the largest number of principal parts, and a dynamic system involves the smallest.

Criterion B (Are the principal parts the same for all inflection classes?) distinguishes parallel systems (in which distinct inflection classes possess parallel sets of principal parts) and skewed systems (in which the principal parts of distinct inflection classes are not fully parallel); this criterion gives a positive answer for all static systems of principal parts, but it needn’t do so for adaptive or dynamic systems.

Criterion C (How many principal parts are needed to determine a given word in a lexeme’s paradigm?) distinguishes segregated principal-part systems (in which each word in a lexeme’s paradigm is deducible from a single one of its principal parts) from integrated principal-part systems (in which at least some of a lexeme’s words must be deduced from a combination of its principal parts).

Criterion D (What is the morphosyntactic relation between a principal part and the nonprincipal parts that are deduced from it?) distinguishes morphosyntactically coherent principal-part systems (in which each principal part shares a fixed set of morphosyntactic properties with the nonprincipal parts that are deduced from it) from systems that are morphosyntactically incoherent.

Criterion E (Are corresponding words in distinct paradigms determined by the same principal parts?) distinguishes isomorphic principal-part systems (in which a lexeme’s nonprincipal parts are inferred from its principal parts in the same way from one inflection class to another) from non-isomorphic systems.

We discuss the use of these criteria in classifying a range of typologically diverse languages:

    [Trans-New Guinea; Irian Jaya]
1parallelsegregatedcoherent Yes
    [Muskogean; USA]
1skewedsegregatedcoherent No
    [Indo-European; India]
>1parallelsegregatedcoherent Yes
    [Indo-European; Italy]
>1skewedsegregatedpartially coherent No
    [Dravidian; India]
>1skewedintegratedlargely incoherent No
    [Nilo-Saharan; Sudan]
>1skewedintegratedlargely incoherent No
    [Nilo-Saharan; DR Congo]
>1skewedintegratedlargely incoherent No
Comaltepec Chinantec
    [Oto-Manguean; Mexico]
>1skewedintegratedlargely incoherent No

While past research in morphological typology has tended to focus on the structure of individual words (invoking such criteria as the average number of morphemes per word and the degree of morpheme fusion within a word), the criteria proposed here extend the focus of typological classification from the structure of individual words to that of whole paradigms and to the implicative relations that paradigms embody.

Elizabeth Grace Winkler – A gender-based analysis of tag questions in Limonese Creole

In the Caribbean province of Limon, Costa Rica, many people speak both an English-based creole language as well as Spanish bains of use in the Afro-Costa Rican community, codeswitching and borrowing from Spanish into LC are common. This study analyzes the differing use by gender of native and Spanish discourse markers in LC discourse. Discourse markers are defined as "sequentially dependent elements which bracket units of talk" (Schiffrin 1987:31) and include grammatical categories such as particles, conjunctions, adverbs, tag questions and interjections. Winkler (1999) showed that discourse markers are frequently borrowed from Spanish into LC; however, the rate of use and type of both borrowed and native discourse markers are markedly different depending on the gender of the speaker. A number of factors contribute to the differences in the way men and women use discourse markers including the role of Spanish in the community and the use of discourse markers as mitigators by women. Women borrow more Spanish discourse markers because they consider the use of prestige forms important, and in Limon, prestige is equated with Spanish.

The differing uses of discourse markers by women and men have been attributed to both gender matched with the imbalance of power that generally exists between the genders; therefore, it is difficult to attribute the variation in use solely to gender. Cameron questions whether markers are used as symbols of 'power, deference, or concern for others — or are all these things' (1992:19).

This presentation will concentrate only on one aspect of this broad study: the use of tag questions (You finished the report, didn't you?). According to Cameron, "tags have been analyzed as a marker of subordination, as a marker of person-centered female values, and … as markers of authority and power in discourse" (1992:16).

Tag questions have traditionally been evaluated from two positions: from the perspective of women's weaknesses, and from women's strengths. Lakoff indicated that because of their lack of authority "women construct their propositions in a surface interrogative form, inviting someone else to confirm their validity".

Holmes takes a different approach to the classification of tags - she focuses on women's tendencies to be supportive of other participants in conversations. Thus, she believes that rather than expressing the speaker's tentativeness and lack of authority, these hearer-oriented tags should be seen as attending to the needs and the 'face' of the hearer

In Limon, women have economic and social power that differs from the middle class white women who have generally been the subjects of study for tag question research. This provides a unique opportunity to evaluate tag question use in an unusual community in which the traditional power relationship between women and men, although not completely displaced, has been greatly changed over the last few decades.

Robert Botne – Tense and cognitive space: On the organization of tense/aspect systems in Bantu and beyond

Bantu languages are well-known for their complex tense systems encoding multiple degrees of remoteness, typically construed as regular progressive intervals as one moves linearly along a time line away from the speech event. Such analyses are founded upon the concept of the traditional time line as a completely satisfactory representational model of tense relations. In this presentation, I propose a significantly different, more complex, and cognitively richer, approach to understanding tense systems. The model treats linguistic time as organized around two conceptualizations of time—moving-ego vs moving-event—in conjunction with deictically separate cognitive worlds, or domains: a contemporal world including the "present" versus non-contemporal worlds dissociated from the present. Although focused on Bantu, the model applies as well to languages as diverse as French and Upper Chinookan.

Richard Janda – “If You Go, You Can’t Come Back” Revisited: On Relexicalizations and Other Upgradings of Grammaticalized Forms

The explicit analysis of metaphor surely belongs to (the semantic subpart of) linguistic science. But at least one active area of current linguistic research — grammaticalization studies — must be viewed as unconsciously relying on unanalyzed metaphors whose implications can be highly misleading. E.g., grammaticalization is often described as a unified set of parallel erosion-processes whereby dirt-clod-like morphemes gradually are imperceptibly worn away in the course of continuously rolling down various one-way slopes leading from lexical to grammatical status. This analogy may work for lumps of earth (though not for snowballs), but the unidirectionality that it implies for the evolution of linguistic units turns out to be contradicted by a surprisingly large number of instances in which temporally later instantiations of a morpheme are more lexical, and not less lexical, than their earlier counterparts.

One of the most striking diachronic-linguistic phenomena involving a change from (lexical to) grammatical back to lexical involves euphemisms for human body-parts. Indeed, grammaticalizationists have literally been sitting on a bulky conjunction of such counterexamples. Now, it has been convincingly argued that prepositions meaning ‘behind’ often arise from words meaning, e.g., ‘buttocks’. Yet it is also the case that prepositional ‘behind’ often comes to have a related — and originally very grammatical — adjective meaning ‘posterior, hind’, and that the latter form is used absolutely, via either nominalization or use with a zero noun, as a more delicate substitute for ‘buttocks’. But, as frequently happens with euphemisms, expressions that mean ‘behind [noun]’ or ‘hind (part(s))’ can themselves acquire the highly lexical meaning ‘buttocks’. Hence the gluteus-maximally non-grammatical meaning of English behind or posterior, French (and English-loaned) derrière, Italian didietro, Spanish trasero, and German Hintere(r)/Hintern.

Indeed, German offers an even more extreme instance of lexicalization via the tainting of a euphemism. The Old High German preposition/adverb for ‘after’ was the highly grammaticalized form aftar, from which had been formed the (likewise originally grammatical) adjective aftero ‘following, hind’, whose nominalized form was used to mean ‘behind, buttocks’. In early Modern High German, though, the adjective and preposition/adverb after o; of the formerly euphemous but by-then dysphemous cognate noun After ‘anus’, which has survived.

The misleading dirt-clod metaphor for grammaticalization recognizes not only erosive semantic bleaching, but also phonetic erosion and morphosyntactic downgrading (categorical demotion, as well as cliticization often followed by univerbation). It is thus unsurprising that diligent inquiry has unearthed numerous cases in which the phonological forms of grammaticalized words have been expanded, or their morphosyntactic status has been upgraded via decliticization and even deaffixation. Here, hypercorrection often appears to play a role, in that speakers seem willing to counteract reductive tendencies in order to pursue prestige or avoid stigma. E.g., a number of grammaticalized English words like against, amidst, and whilst have gained an excrescent (a.k.a. parasitic) final -t that presumably reflects a hypercorrect attempt to undo t/d-deletion (cf. respective Middle English againes, amiddes, and whiles). Similarly, late Middle English texts show an upsurge in the use of possessive constructions like Gwenayfer his love ‘Guinevere’s love’, which suggest that speakers of the time were tempted by the accidental homophony of the determiner (h)is and the moribund genitive-suffix -es to reanalyze the latter as a reduced form of the former.

These and similar examples show that grammaticalization is by no means a unified set of irresistible erosions carried out via the gravity-induced downhill rolling of morphemic clods. It may be an uphill battle to reverse this trend, but, if you go, you actually can sometimes come back.

Bob Port – ‘High-D Phonology’: A new approach to linguistics but without phonetic segments

The claim will be defended that phonetic segments, like [p] and [n] and [i] which are so intuitively real to us, actually play no role in real-time processing of language. I am aware of no experimental result whatever that offers clear support for the widespread view that words are mentally ‘represented as’ (or spelled from) a short list of abstract, speaker- and context-independent tokens arrayed in serial order. I will review a number of kinds of data that are incompatible with the traditional view. It is only our conscious awareness of speech that relies on phoneme-like segments. The reason it does is because of the letter-based education shared by all linguists (and most others in the west, of course). What then are the phonological units that linguists observe in various languages? — ie, the list of possible syllable-types, the segment inventory and the distinctive features, etc? They are generalizations across the speech of some community over a historical time interval. They exist as descriptive units in a database of speech, and are experientially salient for those with literacy education, but they play almost no role in speech perception or in other real-time language processing.

Michael Adams – Watch What You Say: Television, New Media, and Language Change

Generally, sociolinguists have argued that mass media have no discernible effect on language change, and they were probably right until recently. Changes in mass media during the last couple of decades and currently, however, especially changes in relations among media, which are increasingly “networked” and construct what Henry Jenkins calls “convergence culture,” suggest that we should question assumptions about the relative importance of various linguistic “levels”; about what constitutes a legitimate site of language use, development, and change; and about the nature of media and its relationship to language.

Scott Kiesling – Style as stance: Can stance be the primary explanation for patterns of sociolinguistic variation?

In this talk I will explore the relationship between stance and sociolinguistic style. I argue that stance is a precursor, or primitive, to style in sociolinguistic variation: sociolinguistic variants are initially associated with interactional stances and these stances become reified in a speech community over time and repeated use. In fact, I want to test the more extreme hypothesis that stance-taking is where indexicality in variation begins; stance is, in Silverstein's (2003) terms, where the 'baptismal essentializations' of indexicality occur, and is the original first- (or, possibly, zero-) order indexicality.

Using data from my own work on variation and stance-taking, I evaluate how far I can take the proposition that stance is the basis on which all speech differentiation is based. In other words, any choice of linguistic form made by a speaker is based ultimately on the interpersonal or epistemic stance they wish to take with their various interlocutors. I will argue that these stance indexicalities become 'short circuited,' so that ways of speaking become associated with situations and speaking roles in which certain stances are customarily taken.

The overall goal of this thought experiment to to understand how much of the sociolinguistic literature on style variation is attributable to stance. My ultimate goal is to connect the everyday use of language variation in discourse to the ways that it patterns on larger social scales, and to test the hypothesis that this connection can be made through the concept of stance.

Chris Beckwith – Three Fruit Banana : Phoronyms and the Pseudopartitive Construction

Although there is a vast typological literature on sortal unit classifiers (as in Mandarin sân zhî mâo three CLF[animals] cat ‘three cats’), and a small but important formal literature on the pseudopartitive construction in which classifiers occur, functional typologists have been unable to define what a classifier is, or explain what it does, while formalists continue to be stumped by the pseudopartitive. The latter have much preferred to study the ‘true’ partitive (as in a piece of the chair or a cup of the tea). The ‘pseudo’-partitive (as in a piece of furniture or a cup of tea) has largely been ignored since its discussion by Jackendoff in his X͞ syntax (1977). I will argue that the perplexity about the pseudopartitive has arisen primarily because English is unusual, possibly unique, in having developed morphological marking for it, which I have dubbed the pseudopartitive case. I will also show that sortal unit classifiers, group classifiers (as in a pack of wolves), mensural classifiers or ‘measure-words’ (as in a pound of sugar or a barrel of beer), and repeaters (as in Japanese hako hito-hako box one-box ‘a box’) are all simply subtypes of the phoronym, the pseudopartitive function term, a previously unnoted grammatical category. Since the mensural classifier, a type of phoronym, occurs in the pseudopartitive construction and is thought to be found in all languages, the phoronym and pseudopartitive appear to be ‘linguistic universals’. Finally, it has not previously been noted that Russian, a ‘gender language’, has classifiers, or that in Thai—a stereotypical ‘classifier language’, with the type of morphophonology that is thought to go along with classifiers—extended phoronym constructions take obligatory, overt concordial agreement, exactly as in Russian and other ‘gender languages’. These and other points suggest the necessity of a radical reanalysis not only of classifier typology but also of the phonology, morphology, and syntax of the putative ‘isolating’ languages of East and Southeast Asia in general. I will illustrate the lecture with examples from relevant Eurasian languages.

Caroline Féry – The localization of given and new objects in space: grammar and cognition

In the talk, I will report two identical experiments, one in German (joint work with Robin Hörnig), and one in French (joint work with Serge Pahaut) in which participants described spatial altered layouts, where new and given objects were located in different ways in relation to each other. Two aspects of natural speech are investigated in depth: first the linguistic correlates of new and given, and second the grammatical devices used to denote the spatial organization of new and given, or moved objects relatively to each other. Intonation, word order and definiteness play a role in distinguishing the new from the given elements. As far as the spatial relationship is concerned, different strategies are conflicting: first the word order, which predicts that the locatum, i.e. the object to be located, in general the new object, comes before the relatum, the object relatively to which the locatum is located, in general the given object. And second, the given-new hypothesis, which predicts that a new object has to follow a given one. I will show how the two languages solve the conflicts, and will draw generalizations on word order and intonation.

Jennifer Cole – Emergent feature structures: Harmony systems in exemplar models of phonology

In exemplar models of phonology (Bybee and Hopper 2001, Pierrehumbert 2003) phonotactic constraints are modeled as emergent from patterns of high activation between units that co-occur with statistical regularity, or as patterns of low activation or inhibition between units that co-occur less frequently or not at all (e.g., Dell 1990). Exemplar models posit no a priori formal or representational properties to the phonological units or sound patterns that emerge from the statistical regularities of speech, in stark contrast to analyses in the generative phonology tradition, including Optimality Theory, where sound patterns are determined by well-formedness constraints on phonological structures. This paper focuses on the analysis of long-distance assimilation, i.e., harmony systems, comparing the predictions of generative analyses based on constraints on representation, with an exemplar model account. Representational approaches model assimilation with constraints that favor long-distance feature structures. The question addressed here is whether and how the feature structures of harmony systems can be modeled as emergent structure. I argue that an exemplar account can model the co-occurrence patterns of harmony systems in the transitional probabilities between segments that share the harmony feature (Newport & Aslin 2004), without invoking feature structure, but that the domain properties of harmony feature structures (as in Cole & Kisseberth 1995) emerge due to the associations between phonological units (the harmonizing segments) and the morphological units that delimit harmony domains. This association grounds the sound pattern in the lexicon, and provides a “convergence of regularities” which facilitate learning (Frisch 2007). Evidence from phonological accounts of assimilation, together with acoustic evidence and behavioral evidence from our recent experimental studies of speech production are presented in support of the exemplar model of harmony.

Brian D. Joseph – Paradigmatic and Inflectional Change: Some Theoretical Lessons

Several paradigm-based morphological changes in Greek and Latin are presented, including a persistent and recurring type of change involving the reshaping of inflectional endings based on other endings, that provide evidence first for the existence of paradigms and then for their internal structure. These considerations lead to an exploration of the relevance of various theoretical constructs and notions (e.g. rules of referral, OO-correspondence relations, constraints on syncretism, and directionality in grammaticalization), and to (hopefully, informed) speculation on what speakers really know about morphology and what they really do with it.

Rusty Barrett – Language revitalization and language change in K'ichean

In response to the ethnic violence of the civil war in Guatemala in the 1980’s, the Maya movement emerged as a cultural revitalization movement in which language is the central focus of cultural activism Issues of language ideology and the relationship between Spanish and Mayan languages have been central concerns of language revitalization efforts. This talk examines language revitalization in two K'ichean languages, Kaqchikel and Sipakapense. These two languages are spoken in very distinct sociocultural contexts and have quite different language ecologies (Mufwene 2001). Kaqchikel is spoken by 400,000 Maya in central Guatemala and has a number of language planners involved in standardization. In contrast, there has been virtually no work on standardization in Sipakapense, a language spoken by about 10,000 speakers in an isolated region of western Guatemala. This talk describes the ideological basis for prescriptive proposals made by Kaqchikel language planners. These proposals are compared to patterns of language variation in Sipakapense, based on quantitative analyses of grammatical variation and code-switching patterns in the speech of three generations of Sipakapense-Spanish bilinguals. Five grammatical variables were chosen as potential indicators of convergence with Spanish: SVO word order, overt object pronouns, antipassive/inverse constructions, the use of the Mayan resumptive pronoun, and the use of Spanish discourse markers. The frequency of antipassive/inverse constructions and the use of Spanish discourse markers show no significant changes across generations. The remaining grammatical variables all show signs of emerging hyperdifferentiation between Spanish and Maya. The results suggest that younger speakers are hyperdifferentiating by avoiding traditional Sipakapense constructions that coincidentally overlap with Spanish. Although Sipakapense has had less Spanish influence compared to Kaqchikel, younger speakers of Sipakapense show a preference for grammatical forms similar to those proposed by Kaqchikel prescriptivists. The similarity suggests that the language ideology of the cultural revitalization movement is driving language change even in communities where standardization efforts have been minimal.

Detmar Meurers – Intelligent Computer-Assisted Language Learning – An Opportunity for Interdisciplinary Research

Complementing the general focus on communication and culture in foreign language teaching, a growing body of research since the 90s has established that awareness of language categories, forms and rules is important for an adult learner to successfully acquire a foreign language. But given the limited amount of time an instructor can spend with a student and the communicative focus of instruction, there are few opportunities in class for fostering linguistic awareness through focus on form and providing individual feedback on errors.

Computer-Assisted Language Learning (CALL) systems could be one way to address this problem, but traditional CALL systems lack the ability to analyze language and provide diagnostic feedback or input enhancement on that basis. Natural Language Processing can, in principle, be used to automatically obtain such linguistic analyses. However, such "Intelligent" CALL (ICALL) research has largely focused on computational issues disconnected from the needs and objectives of foreign language teaching and the insights from second language acquisition (SLA) research. As a result, there are virtually no ICALL systems used in foreign language teaching today - which is all the more regrettable given that such controlled environments could in principle provide large data sets for SLA research.

Putting a positive spin on the situation, we interpret it as a concrete opportunity for an interdisciplinary approach combining linguistic modeling, learner modeling, and activity modeling with computational linguistic algorithms and representations capable of integrating the information from these models. The talk will discuss some of our efforts in the OSU ICALL group to proceed along this path, in particular my joint work with Luiz Amaral on TAGARELA, a web-based electronic workbook accompanying the instruction of Portuguese, and the WERTi prototype exemplifying on-the-fly generation of language awareness materials based on web pages selected by the learner.

Diane Brentari, Caroline Gonzalez and Amanda Seidl, Purdue University – Perception of sign language prosodic cues by signers and non-signers

In these studies, we focused on the prosodic cues that mark the edge of an Intonational Phrase [IP] in American Sign Language [ASL]. Signers and non-signers were tested on their ability to use visual-gestural prosodic cues to segment IP units in ASL. The purpose of the work is to determine whether nonsigners are sensitive to the prosodic cues that mark the edge of an IP of a sign language [in this case, ASL], and if so, do they weight cues in a similar way to the signers' cue weighting.

The stimuli consisted of ASL passages that contained identical pairs of signs in 2 conditions —in one condition an IP break was present between the two signs, in the other there was no IP break. Subjects were from 3 groups: adult proficient ASL signers, adult nonsigners, and a group of 9-month old hearing infants from hearing families. We first analyzed the stimuli to determine what prosodic cues were present; 5 robust cues were measured—length, pause, hold, blink, and drop-hands. The adults watched videotaped passages and responded to a questionnaire about where 'sentence breaks' were; the infant study used the Preferential looking procedure; all tasks included a familiarization and a test phase.

All groups were sensitive to ASL IP boundaries at a greater than chance level. The reduced number of items used with the infants allowed us to determine only that they are sensitive to ASL prosodic cues; of the 24 9-month-old infants who participated, 18 looked longer at the familiarized passage that contained an IP than the one without the IP-boundary [p=.049]. In the adult groups we were able to determine that nonsigners and signers overlapped on some of the cues that were salient [e.g., pause], but there were also important differences [e. g., 'hold' was more salient to signers, while 'drop-hands' was more salient to nonsigners]. The implications for these results will be discussed.