Where? Dwinelle Hall 1229
When? In Spring 2017, every Thursday from 12:30-2pm, and as announced
What? *dhworom is a semi-monthly talk series showcasing diachronic work in linguistics, including genetic historical linguistics, sociolinguistics, and work on language contact. In Spring 2017, *dhworom will be supplemented by an informal reading group on phylogenetics, to meet in all weeks without invited talks. Please find our schedule and reading list on this page.
Who? *dhworom is organized by Erin Donnelly and Tyler Lau. We welcome linguists and philologists from all departments and institutions.
Past and Current Topics:
- Spring 2017: Phylogenetics
- Fall 2016: Sound Change
- March 9: Luc Baronian (Stanford University, Linguistics)
- February 17: Lev Michael (UC Berkeley, Linguistics)
Different choices, different trees: Comparing approaches to lexical homology in computational phylogenetics (a Tupí-Guaraní case study)
One of the major developments in recent years in historical linguistics has been the application of computational phylogenetic tools, originally developed to study biological evolution, to language evolution. Much important early work in this paradigm was carried out by non-linguists, and as linguists are engaging with these new methods, linguistically-informed criticism of, and reworking of, previous approaches is emerging (e.g., Change et al. 2015). In this talk I use a testbed of lexical data from Tupí-Guaraní (TG) languages to address a core conceptual issue in linguistic phylogenetics: the understanding of cognacy — or ’homology’ in phylogenetic parlance — underlying the application of these methods. I show that the understanding of homology used in previous phylogenetic studies diverges from how historical linguists typically define cognacy, so that such studies are not based on cognate sets, strictly speaking, but rather ‘root-meaning’ sets that do not consider lexical items to be homologous if their meanings differ. I demonstrate that the two understandings of homology yield noticeably different classifications of the TG family, which suggests that abandonment of the classical linguistic notion of cognacy may have significant consequences for previous linguistic phylogenetic results. I also demonstrate that ‘binary recoding’, a procedure commonly carried in previous studies based on root-meaning sets, has the effect of inflating the resolution of classifications based on these sets (Chousou-Polydouri et al. 2016). Situating these results in a broader discussion of phylogenetic methods, I argue that they remain promising tools for historical linguistics, but that their ability to produce robust results depends on linguists engaging substantively with the linguistic entailments of their implementation.
- November 17: Matt Faytak (UC Berkeley, Linguistics)
Encroachment without crisis: a case study of Wu Chinese vowel fricativization
Vocalic chain shifts have often been divided into “push” and “pull” or “drag” chains. It has been argued that only pull chains, whereby a leading change creates a “vacuum” in acoustic space that a category adjacent in acoustic space will tend to fill, actually exist (King, 1969; Samuels, 2006). This argument typically has two parts (1–2):
(1) The chronology required for push chains is not attested: that is, one category adjacent to another in acoustic space ”encroaches” upon another, which subsequently distances itself from the first.
(2) The very mechanism of "encroachment” leading to a push chain is undesirably teleological, since it suggests an impending merger due to overlap of two or more categories presents a "crisis" that a language is able to detect and react to.
Some push chains have been rigorously confirmed (Langstrof, 2006), reducing the power of (1). The teleology problem as in (2) persists, however, and I focus here on a case study of Wu Chinese high vowel fricativization (HVF) that appears to complicate this point as well. In many northern Wu dialects, Proto-Wu (PWu) *i rhymes develop into fricativized [iz] (Ling, 2009; Cai, 2011); in a few cases, lower, diphthongal or nasalized front reflexes of PWu *ien also develop into a higher, more monophthongized vowel such as [i] (Ballard, 1969; Qian, 1992). Taken together, these resemble a pull chain, given that very few languages that have undergone fricativization have also raised a lower vowel, but some have filled the vacuum around cardinal [i] created by fricativization.
I suggest that the phonologization of frication in, e.g., *i > [iz] is unlikely to occur without some reduced utility of canonically vowel-like cues to category for high vowels (Kirby, 2013). This in turn suggests a different sort of “encroachment” of reflexes of *ien on reflexes of *i, such that the change resembles a push chain again, but without a “crisis” and close overlap of categories. I thus suggest that “encroachment” can exert a force on the development and change of vowel categories at greater distances than typically considered and provide an outline of an exemplar-based model of phonological categories that generates “push” effects from partial category overlap at a distance.
- November 3: Tyler Lau (UC Berkeley, Linguistics)
The Role of Frequency in Noun Class Mergers and Shift: A Case Study in Romance (Practice talk for 28th Annual Meeting of WeCIEC)
I compare simulations modeling nominal class changes from Latin to Romance. The simulations take into account token frequency and sound changes. Without token frequency, the model better reflects the history of Romance, while the sound changes reveal what mergers could have come about purely via sound change. These results corroborate literature stating that token frequency weakens analogical models and provide insight into precursors existent in Latin that caused an overhaul of the nominal system in modern Romance languages. Full abstract here.
- October 27: Peter Jenks, Hannah Sande, Nico Baier (UC Berkeley, Linguistics)
The diversity of SAuxOVX in the Macro-Sudan belt
Güldemann (2008) proposes that SAuxOV(X) word order is a diagnostic for a linguistic area covering much of northern Sub-Saharan Africa, the Macro-Sudan Belt. However, superficially SAuxOV word orders across this area are structurally diverse. At the same time, one subset of SAuxOV languages do share the same core clausal structure. This group is clearly distributed in an area which is coextensive with other important syntactic variables. Structurally distinct cases of SAuxOV at the margins lack some of these features. We conclude that typological generalizations which use structural variables lead to more fruitful predictions than those based on word order alone.
- September 22: Larry Hyman (UC Berkeley, Linguistics)
On Reconstructing Tone in Proto-Niger-Congo
The purpose of this talk is to determine whether Proto-Niger-Congo, a language spoken perhaps 10,000 years ago, had tonal contrasts. While it is largely assumed that PNC had *H(igh) and *L(ow) tones, there has been little explicit discussion of the issue or demonstration that there are regular correspondences across the distantly related subgroups. Various methodological issues are first discussed. My starting point is the widely accepted reconstructed tones of Proto-Bantu lexical morphemes (Meeussen 1969/1980, Bantu Lexical Reconstructions) to determine how far outside of Narrow Bantu proper the tones correspond. Looking particularly at lexical verb tones (for reasons I'll discuss), I show that a contrast between *H and *L can be traced back relatively high up the Niger-Congo tree, at the same time raising questions about word structure and verb morphology: (1) Were PNC verb roots monosyllabic or longer? (2) Were there verb "extensions" (causative, applicative etc.) in PNC, and if so, did they contrast tone? The presentation will sort out the various issues involved in doing such long distance comparative work in a language family whose subgroups are so evolved that some are only controversially accepted as belonging to Niger-Congo (e.g. Mande, Ijoid).
- May 4: Claire Bowern (Yale University)
The phylogenetics of color systems
I present the first study of the phylogenetic evolution of color naming systems. The naming of colors has long been a topic of interest in the study of human culture and cognition. Color term research has sought to answer diverse questions about color perception and lexicalization, but no previous research has tested color system change within an explicitly evolutionary framework. We show, using data from Pama-Nyungan (Australian) languages, that there is broad support for the most influential theory of color term development (that most strongly represented by Berlin & Kay 1969); however, we find extensive evidence for the loss (as well as gain) of color terms. We also find evidence for alternative trajectories of color term evolution, beyond those considered in the standard theories. These results not only refine our knowledge of how humans lexicalize the color space and how the systems change over time; they illustrate the promise of phylogenetic methods within the domain of cognitive science.
- April 13: Zachary O'Hagan (UC Berkeley, Linguistics)
Negation in proto-Kampan
This presentation explores the diachrony and variation of negation systems in the Kampan Arawak languages of lowland Peru and reconstructs both the categories and forms of the proto-Kampan negation system. Kampan languages -- which include Nomatsigenka, Ashéninka, Asháninka, Caquinte, Matsigenka, and Nanti -- are largely homogeneous in their subsystem of main clause negation, but are largely heterogeneous with respect to other sorts of negation. As an example of the former point, all Kampan languages exhibit distinct preverbal particles that differ in whether they negate a notionally realis or irrealis clause. As an example of the latter point, Nomatsigenka, Matsigenka, and Nanti preserve a proto-Kampan pattern of expressing negative indefinite pronouns via a construction that consists of the realis negator followed by an interrogative pronoun. In contrast, Ashéninka, Asháninka, and Caquinte have innovated a negative indefinite pronoun that is distinct from all interrogative pronouns in those languages.
I review standard negation, negative imperatives, existential, copular, and metalinguistic negation, negative indefinite expressions, and negative interrogative pronouns in order to demonstrate how these subsystems have remained stable, or come into or gone out of existence, as grammaticized constructions in these languages, and how new forms have been innovated to express already extant categories. For example, the ancestral language of Matsigenka and Nanti innovated an irrealis main clause negator *gara and an existential negator *mameri -- two categories which reconstruct to proto-Kampan -- and in doing so it lost a reflex of the proto-Kampan existential negator *kaga. It also innovated a distinct category of metalinguistic negation in the form of *matsi. This work contributes to the understanding of synchronic variation among these languages, informs research on the reconstruction of grammatical subsystems and on the phylogeny of Kampan languages, and allows for finer-grained comparison of negation systems within the large and diverse Arawak family.
- March 30: Tom Güldemann (Humboldt University Berlin, MPI-SHH Jena)
What gender systems? Agreement classes vs. noun form classes in Niger-Congo with particular reference to Ghana-Togo-Mountain languages - with Ines Fiedler (Humboldt University Berlin)
This paper reviews the treatment of gender systems in Niger-Congo. It is based on a consistent methodological approach employing the four analytical concepts of gender (class), agreement class, noun (form) class, and declension (class). Gender systems in Niger-Congo tend to be analyzed on the assumption of a consistent alliterative one-to-one mapping of noun form classes and agreement classes, owing to the strong bias towards the reconstructed gender systems of Bantu and wider Benue-Congo. As a result, these two in principle independent concepts are conflated and a gender system is often deduced merely from the number mapping of noun form classes, aka declensions, rather than from the agreement behavior of nouns as the ultimate basis of gender. The paper shows that gender systems based properly on agreement and apparent gender systems based on nominal declension are systematically not identical, this even in Proto-Bantu whose analysis gave rise to the defective philological approach. The data are drawn from various sub-groups of Niger-Congo like Ghana-Togo-Mountains, Guang, and Gur. The paper's conclusions are relevant for the assessment of Niger-Congo gender systems by both historical-comparative linguists and typologists.
- February 24: Ruprecht von Waldenfels (UC Berkeley, Slavic)
Using a parallel corpus for diachronic research: Slavic and beyond
Functional similarities and differences of comparable linguistic items across a set of closely related languages provide valuable insights for comparative and diachronic linguistics. At the same time such differences are often not obvious and difficult and time-consuming to analyze. In the talk, I present an ongoing research project that uses a parallel corpus to investigate such items in a bottom-up fashion in Slavic, starting with an analysis of their distribution in different languages, rather than their semantic or functional profile.
As a data basis I use ParaSol, a word aligned, morphologically tagged and lemmatized parallel corpus of prose in all major Slavic languages I have been (co-)developing since 2006. The variables under consideration are evaluated both from a quantitative perspective using clustering algorithms as well as qualitatively using a web interface that visualizes the contrasting variables in context. Data aggregation and visualization is largely standardized and automated. It is currently being implemented as an online research instrument that will be accessible to the research community at large.
Variables under comparison include verbal aspect, derivational morphology, middle marking, the use of prepositions, and perception verbs. The approach is successful in revealing interesting differences in the grouping of the Slavic languages that arguably reflect both contact-induced and genetic factors in their historical development.
January 27: Mairi McLaughlin (UC Berkeley, French)
What can the Early French Press corpus tell us about linguistic change?
The Early French Press corpus was developed for my current project on historical news discourse in French. This project is the first systematic linguistic study of the language of the press before the twentieth century. It covers the period from the publication of the first periodical in 1631 to the French Revolution in 1789. The project has two halves. The first half focuses on the journalistic genre, exploring its origins and evolution and comparing it to the contemporary press. The second half concerns language variation and change. I will be presenting the first set of results from this latter part of the project in today’s paper.
The aim of my talk is to explore what the Early French Press corpus will add to our understanding of language variation and change. It is based on a series of case studies which treat a set of linguistic changes that took place in the seventeenth and eighteenth centuries. The approach is essentially comparative: these changes are already well documented for other text types so studying them in the Early French Press corpus letsus explore whether the changes took place in this text type at the same time; at the same rate; and whether they followed the same evolutionary stages. The features to be explored include verb morphology, negation, anaphora and word order. Together, the results from the case studies will let me address a series of larger questions about language variation and change.