TREND: A Trilateral Weekend of Linguistics
May 9, 2009
Stanford University
TREND is an annual workshop encouraging collaboration between students and faculty in linguistics from UC Berkeley, UC Santa Cruz and Stanford University.
Location: Terrace Room, on the 4th floor of Margaret Jacks Hall (building 460)
Campus Map
(you can park for free in any spot marked 'A', 'C' or 'P' on the weekend)
Schedule (tentative)
9:30-10:00 |
Coffee, Welcome
| Session 1 - Chair: Arto Anttila |
10:00-10:30 |
Infrequency vs. ambiguity - which is more costly?
Meghan Sumner and Seung-Kyung Kim, Stanford
Abstract ±
In systematic variation in speech (e.g., tapping), a single lexical item typically has a standout variant - one that is much more frequent than the alternative. Consider the word traitor. In American English, the medial /t/ is tapped. While this is the norm (about 96% of the time (Patterson & Connine, 2001)), it is also possible to produce a voiceless aspirated stop (e.g., I said the trai[t]or, not the trader!). This project brings together two independent lines of research to examine the role of variant frequency in word recognition: within-item variant frequency (where infrequent variants slow processing) and lexical ambiguity (where ambiguous items slow processing). On the one hand, a production without a tap should improve processing by eliminating ambiguity. On the other hand, that same production should result in a processing cost because it is extremely low in frequency. The results of a cross-modal semantic-priming task suggest that variant frequency trumps disambiguation: A frequent but lexically-ambiguous variant activates related targets better than an infrequent but unambiguous variant.
| 10:30-11:00 |
Is speech talker-oriented or listener-oriented? - Evidence from
phonological neighborhood effect in spontaneous speech
production
Yao Yao, UC Berkeley
Abstract ±
It has been shown in the literature of word perception that the existence
of similar-sounding words in the lexicon (i.e. phonological neighbors)
inhibits auditory word recognition (Luce & Pisoni 1998). However, the
effect of phonological neighborhood density on word production is less
clear. Previous studies have proposed (or suggested) two possible
accounts. The first one is a talker-oriented view, that is, words with
many neighbors are easier to produce (Vitevitch 1997, 2002) and therefore
reduced in speech. The second is a listener-oriented account, i.e. words
with many neighbors are harder to perceive and therefore hyperarticulated
(Wright 1997, Munson & Solomon 2004, Scarborough 2002). The present work
aims to distinguish the two accounts by examining the effect of
neighborhood density on word duration. The data we use are drawn from the
Buckeye speech corpus, which has 40 American English speakers, each
recorded for about an hour. A mixed-effect model is built to predict the
durations of all CVC monomorphemic content words in the corpus, with
speaker and word as random effects. The results show that, after
controlling for other linguistic and nonlinguistic factors, neighborhood
density has a small but robust negative effect on word duration, i.e. the
higher the density, the shorter the word is. The effect of average
neighbor frequency is in the same direction but less stable. Present
results provide unambiguous evidence for the talker-oriented account
concerning word duration. Theoretical implications of the findings are also discussed.
11:00-11:30 |
Consonant-to-consonant gestural overlap and place assimilation in
Korean
Paul Willis, UC Santa Cruz
Abstract ±
In Korean, labial stop codas optionally assimilate to the place of following dorsals, but not vice versa. This pattern has recently been argued (Son 2008) to provide direct evidence for the hypothesis that place assimilation has its roots in perceptual masking due to gestural overlap in consonant clusters. Perceptual studies (e.g. Kochetov & So 2007) have shown that labials are generally more recoverable in coda position than dorsals; however, Son (2008) shows that the assimilating labial-dorsal clusters in Korean are far more overlapped than the non-assimilating dorsal-labial clusters are. Thus, Korean place assimilation looks phonetically 'natural' once cluster-specific levels of gestural overlap are taken into consideration. In this talk I argue that, while crucial in explaining the naturalness of labial-to-dorsal place assimilation in Korean, these cluster-specific levels of overlap are synchronically unnatural. Theories of overlap optimization (e.g. Silverman 1995, Wright 1996, Chitoran et al. 2002) take natural overlap systems to be those that maximize overlap while also maintaining phoneme recoverability. Within an OT formalization, I show that Korean is not among the predicted typology of overlap systems that optimally balance these two functional pressures - it is a harmonically bounded system. I suggest that the problem is one of phonological opacity, and present a diachronic account in which clusters locally evolve to be optimally overlapped, but are not constrained to be part of a globally optimal synchronic overlap system.
| 11:30-12:00 |
Vowel variation in Elliott County, Kentucky
Rebecca Greene, Stanford
Abstract ±
This study examines dialect leveling in vowels among native residents of a poor, rural community in the foothills of the Appalachian Mountains. Phonemes considered were /ay/ (BITE) and /^/ (BUT). Results suggest that overall, younger speakers show movement from the local variant, monophthongal /ay/, toward the mainstream diphthongal /ay/. However, speakers still generally prefer to use the monophthongal variant, despite explicit linguistic discrimination from outsiders about this feature. Younger and more educated speakers also are moving away from the local, traditional raising and fronting of /^/, but at a more rapid pace. It seems that speakers are not conscious of variation in /^/, and so trends in this vowel are more gradient in nature. Additionally, monophthongal /ay/ appears to be almost a required symbol of local identity, whereas raised/fronted /^/ carries ‘old-fashioned’ or ‘hick’ associations. Possible reasons for the different patterns of these two vowels are discussed.
| 12:00-1:30 |
Lunch Break
| Session 2 - Chair: TBA |
1:30-2:00 |
Non-feature-driven movement and the sentential position of
German weak pronouns
Judith Fiedler, UC Santa Cruz
Abstract ±
The pronoun system in Standard German has traditionally been viewed as consisting of two pronoun classes: strong and weak. The salient distinction between them concerns their clausal positions. Strong pronouns may occur in any position available to full DPs, while weak pronouns are normally restricted to a high position in the clause. This distributional restriction is often taken as evidence of a syntactic distinction between the two pronoun types, such that weak pronouns are deficient in syntactic features, and must be licensed by association with a functional head. However, the position of weak pronouns is not invariant; they may, under certain prosodic conditions, occur in non-canonical positions lower in the structure. Using the relationship between prosody and weak pronoun position as a diagnostic, I propose an alternative analysis of the German pronoun system. Weak and strong pronouns are syntactically identical, and the syntax places no movement requirement on them. Instead, the position of pronouns is determined by constraints holding post-syntactically at the interpretive interface. Weak pronouns are those which do not bear a [+Topic] or [+Focus] feature, and they cannot be interpreted in areas of the clause associated with these discourse functions. The canonical position of weak pronouns is derived via non-feature-driven movement out of the vP phase, the default focal region. I conclude that the movement of German weak pronouns is most naturally understood as motivated by information structural rather than by syntactic or prosodic constraints.
| 2:00-2:20 |
On what comes first in a verb-second language
Line Mikkelsen, UC Berkeley
Abstract ±
Most work on verb-second order in Germanic languages assumes that the choice of initial constituent in declarative verb-second clauses is a matter of textual organization which falls outside the domain of syntax. In this talk I give evidence that in at least one Germanic language there are intrasential syntactic principles that restrict what may occupy initial position in verb-second clauses. I further show how these principles bear on one of the central disagreements in the generative literature on Germanic clause structure, namely whether subject-initial verb-second clauses have the same structure as non-subject-initial verb-second clauses.
| 2:30-3:00 |
Mixed Gender Agreement and Word Order in Tigrinya
Laura Whitton, Stanford
Abstract ±
As early as Leslau (1941), it has been noted that for the majority of inanimate nouns in Tigrinya, speakers find both masculine and feminine agreement morphology acceptable. This lack of lexical specification of grammatical gender is rare - if not unattested - in the world’s languages. Even more interesting from a typological and theoretical standpoint is the fact that a noun may behave as masculine by one metric and feminine by another within the same clause. For example, a noun might have a feminine determiner while the corresponding subject or object marking on the verb for that noun is masculine. This talk presents the results of a study of judgments from native Tigrinya speakers in the Bay Area regarding both gender variation and the interaction of word order and "morphological mismatches." I will also address the interpretation of these results from a broader and cross-linguistic perspective on mixed agreement patterns.
| 3:00-3:30 |
Break
| Session 3 - Chair: TBA |
3:30-4:00 |
Meaning Targets in Syntax and Morphology: A Study of Hupa Agreement
Amy Campbell, UC Berkeley
Abstract ±
Verbs in Hupa (Pacific Coast Athabaskan) inflect for subject, direct object, and indirect object agreement. A discontinuous augmentative morpheme ya:- encodes the distinction between dual and plural in the first person, as shown in (1), and between singular and dual/plural in the third person.
(1) | na:-ya:-s-di-l-to'n
| | pvb-aug-pfv-1nsgS-thm-jump
| | 'we (pl) danced'
|
The augmentative appears at most once in any verb word, in a fixed position, but it can semantically modify any or all of the verb's arguments. To account for these facts I propose a meaning-driven model of Hupa grammar, in which meaning targets set up in the syntax compel and constrain morphological processes, deriving the distribution of the augmentative morpheme and accounting for its interpretational ambiguity.
| 4:00-4:30 |
Varieties of Distributivity: One by One vs. Each
Adrian Brasoveanu and Robert Henderson, UC Santa Cruz
Abstract ±
Example (1) below is compatible with events that unfold in very different ways. The addition of "one by one" in (2) constrains how the leaving events can proceed. Intuitively, "one by one" is an event modifier that targets a plural participant in the event: it breaks an event down into temporally sequenced subevents and it distributes the plural participant over these subevents.
(1) | The boys left.
| (2) | The boys left one by one.
|
The main goal of the presentation is to investigate the constraints on the distribution and interpretation of "one by one" and give a compositional semantics for "one by one" that captures them. The larger goal is to motivate two routes to establishing distributive quantificational dependencies - exemplified by "each" and "one by one", respectively. The first route is decomposition into sets of assignments: each n-tuple of quantificationally dependent entities is individually stored in a variable assignment and quantifiers are interpreted relative to the entire set of variable assignments (van den Berg 1996 and Nouwen 2003 among others). The second route is encapsulation into a function: functions, e.g., Skolem functions, store quantificational dependencies as a whole, mapping each entity to the (possibly non-atomic) entity that depends on it (Stone 1999, Bittner & Trondhjem 2008 and Dekker 2008 among others).
The two routes to distributivity enable us to capture the fact that "each", but
not "one by one", licenses internal readings of "different". That is, sentence (3) below -- but not (4) -- has a reading under which, for any two boys, the poem one of them recited is different from the poem the other one recited. With "each", we get this reading by comparing variable assignments in a pairwise way (see Brasoveanu 2008 for more details). The internal reading is not available with "one by one" because encapsulated quantification involves only one variable assignment.
(3) | The boys each recited a different poem.
| (4) | The boys recited a different poem one by one.
|
| 4:30- ?
|
Social!!
| |
|