Harvard University
Tuesday, December 17, 2002 at 5:15 PM, Linguistics Department Seminar Room, Boylston 303, HarvardEric Eben
Visiting scholar from Cornell UniversityOne of the several metrical irregularities in the Homeric texts is the apparent lengthening of short vowels when followed by a resonant initial word, a metrical license generally termed "resonant lengthening". The idea of "license", unfortunately, is inaccurate and obscures the singers' actual practice.
The license is generally described as taking its start in resonant initial words known historically to have contained a second initial consonant, h- < *s- or *w-, which allow a preceding short vowel to make position. h- and w- were later lost and thus presented the singers with forms of the same word which could trigger lengthening or not. Later singers then forgot which words could lengthen for historical reasons and so spread the apparent license to all resonant initial words to be utilized according to the demands of the meter.
An investigation of the relevant facts from the Iliad and Odyssey reveals that the final stage of the above process, the categorical generalization of the license, is not true. It is, in fact, spread only to a very few, generally important, resonant initial lexical items, e.g. forms of meg-. The lack of random distribution suggests the lengthening occurred on a case by case basis according to the specific needs of the singers. The actual distribution of resonant initial words at the beginning of formulaic phrases demonstrates that the phrases originated in metrically regular positions. They were then moved or inflected into new positions, usually before a major caesura position, where they no longer closed a preceding short syllable. Apparently lengthened vowels are not to be explained by a phonological license but rather by the techniques of oral versification. In this they agree with other "lengthenings" before words that originally had two initial consonants (e.g. *dw- and *sw-).
Harvard University
GSAS Workshop on Comparative Syntax and Linguistic Theory
Friday, December 13, 2002 at 4 PM, Fong Auditorium, 1st floor, Boylston Hall, HarvardZeljko Boskovic
U. of Connecticut"Intermediate Spec IPs" I examine whether the matrix subject in raising constructions like (1) passes through the intermediate SpecIP on its way to its final landing site.
(1) a. Someone seems to be in the garden
b. There seems to be someone in the garden.
Based on a number of crosslinguistic arguments, it is argued that the intermediate SpecIP is created only in some raising constructions. Consequences of this claim for the EPP and the proper formulation of locality restrictions on movement will be explored.
MIT LING-LUNCH
Thursday, December 12, 2002 at 12:00, MIT E39-335 (conference room)Joey Sabbagh, MIT "Right Node Raising is sometimes before, and sometimes after, Spellout" In this talk, I present an explicit extraction analysis of Right Node Raising. The primary analytical problem posed by Right Node Raising is that it appears at once, unbounded [(1)], and at other times bounded [(2)].
(1) The doctors said that they think that the nurses examined, and the nurses claimed that they know that the doctors vaccinated, all the newly admitted patients.
(2) *Xander [[[bought __ from Star, and returned __ to Shaws] apparently], new drapes for the entire house].
To solve this problem, I propose that Right Node Raising can apply either prior to or after Spellout of a Phase (Chomsky 1998,etc.). Right Node Raising, (being simply an instance of (Across the Board) rightward movement, is constrained by the Right Roof Constraint (Ross 1967, Akmajian 1974, Johnson 1986, McCloskey 1999) when it applies prior to Spellout of a Phase. Right Node Raising after Spellout, on the other hand, is not constrained by the Right Roof Constraint. Rather, Right Node Raising after Spellout is constrained by the Linearization Principle proposed by Fox and Pesetsky (2002)-a constraint on operations applying across Phase boundaries that potentially affect the linear order or constituents.
The analysis covers a range of previously unanalyzed data. Additionally, it provides a direct analysis of the so-called Right Edge Restrictions on Right Node Raising-the condition that Right Node Raised element must either be at or get to the right edge of each conjunct in order for the overall Right Node Raising construction to be well-formed. This restriction is stipulated under most non-extraction accounts of Right Node Raising (e.g., Wilder 1997, McCawley 1982). Moreover, the analysis covers a wider range of Right Edge Restriction effects than Wilder's (1999) recent analysis, which is motivated specifically to account for this restriction.
If successful, the conclusion to be drawn from this talk is that Right Node Raising does not force an analysis in terms of construction specific mechanisms, which many recent analyses of the construction employ.
For more information, contact anevins@MIT.EDU.
UMass, Amherst: Linguistics Colloquium
Title TBA
Not available at this time.
MIT LING-LUNCH
Thursday, December 5, 2002 at 12:00, MIT E39-335 (conference room)Elena Anagnostopoulou
MIT"MOVEMENT ACROSS PHONOLOGICALLY VISIBLE CATEGORIES" In the recent literature, there has been a debate concerning the scope and correct characterization of Holmberg's Generalization (HG). According to one view ('limited HG'), HG describes consequences of V-raising. In particular, HG states that Object Shift (OS) is dependent on V-movement (Chomsky 1993; Bobaljik 1995, 2002; Bobaljik & Jonas 1996 and others). According to an alternative view ('extended HG'), HG subsumes blocking effects triggered by any intervening overt category: V, P, a particle, an argument XP. OS cannot take place across any phonologically visible category (Holmberg 1986, 1999).
In this talk, I provide two arguments against the extended HG:
(a) First, there are environments where the V-Raising requirement must be met, but OS takes place across phonologically visible categories. More specifically, I identify two types of OS in Scandinavian:order preserving and order reversing OS. Order reversing OS cannot be accounted for by Holmberg (1999). I argue that the Minimal Link Condition, incorporating Chomsky's (1995) equidistance (Ura 1996; Collins 1997), provides an accurate characterization of order preserving as well as for order reversing OS.
(b) Second, movement of an intervener to v, T and C enables long A-movement in passives, unaccusatives, raising and clitic constructions. The same strategy permits long OS. Holmberg (1999) fails to recognize that OS on the one side, and A-movement on the other side form a natural class w.r.t. obviation of locality effects.
MIT Linguistics Colloquium
Mark Hale
Concordia University"Galilean-style Phonology; or, What Is A Linguistic Object?" This paper explores the implications of taking a "Galilean-style" stance (Chomsky 2002:98-99) in response to a wide variety of "data-matching" trends in contemporary linguistics, including Stochastic Phonology (Pierrehumbert 2001), "corpus" and "usage"-based models, "phoneticist" phonological theories, & etc. A highly restrictive, "substance-free" conception of the nature of linguistic objects will be presented. The implications of this conception for the nature of diachronic linguistics, synchronic linguistics, and the relationship between the two will be examined.
For more information, contact Heejeong Ko or Marketa Ceplova.
UMass, Amherst: Linguistics Colloquium
Jon Nissenbaum Title TBA Not available at this time.
MIT LING-LUNCH
Thursday, November 21, 2002 at 12:30, MIT E39-335 (conference room)Laura Redi
MIT"Categories in intonational phonology:
Implications of data from production studies"In this talk I will discuss some current problems in intonational phonology and how phonetic data from a variety of studies may shed light on the issues. In the two decades since the development of the autosegmental-metrical (AM) model of intonation (Pierrehumbert 1980), considerable evidence has come to light which supports models that assume that the underlying phonology of intonation contours consists of discrete tonal elements that are aligned with particular syllables and/or segments. Much of this evidence comes from phonetic studies which have demonstrated consistent alignment patterns between certain discrete F0 characteristics, such as F0 maxima and minima, and particular positions in the syllable structure. Consistent alignment patterns have been found across a variety of languages, and within a language, alignment differences often correspond to lexical or semantic distinctions. Since the ability to align discrete F0 targets with syllables appears to fall within the purview of linguistic competency, developing a full account of alignment data will be a crucial task for theories of intonational phonology. While the AM model provides an account of much of this data, certain findings have come to light which cannot be easily reconciled with the existing model. These findings include alignment consistency for Greek prenuclear accents (Arvaniti, Ladd, and Mennen, 1998, 2000), behavior of F0 minima in the neighborhood of H* accents (Ladd and Schepman, forthcoming), and treatment of F0 maximum alignment by speakers in a production experiment carried out over the summer by this researcher. The implications of these findings will be discussed, including what directions are suggested for future empirical and theoretic research.
For more information, contact anevins@MIT.EDU.
MIT - Phonology Circle
Thursday, November 21, 2002, at 2:00 PM, room TBA
Bert Vaux
Harvard University"Why the phonological component must be serial and rule-based"
In this talk I provide general arguments for abandoning Optimality Theory and outline the requirements that an alternative should satisfy.
A. OT fails to account for several of the central phenomena of human language (i.e. those that occur in all or most known languages) which any adequate theory of phonology must be able to explain.
i. Opacity.
The problems of opacity for canonical OT have already been treated in detail (Idsardi 1997, 1998, Odden 1999, Kager 1999, Kiparsky 2000, McCarthy 2002, etc.). The fact that opaque interactions between phonological processes occur in all natural languages receives an appealing explanation in derivational models, where opacity is a straightforward product of process ordering. Opacity created by iterative rules creates even more profound problems for OT, since proposed patches such as Sympathy, level ordering, and output-output constraints cannot be brought to bear (Hyman and VanBik 2002, Wolfe 2000).
ii. Optionality.
All languages contain numerous optional processes, a fact that is not predicted by the fundamental architecture of OT (Kager 1999). OT patches such as underdetermination (Hammond 1994), cophonologies, tied constraints (Anttila 1995), and differential constraints (Horwood 2001) fail to account (without significant modifications) for a variety of optionality effects such as sequential iterative optionality (e.g. French schwa deletion, which freely applies to any subset of schwas in a prosodic phrase) (Vaux 2002).
iii. Exceptionality and unnatural processes.
A grammar arises from the confrontation of the human language acquisition device with the arbitrary linguistic data to which it is exposed. Since these data encode layers of historical change, the resulting phonological grammar will in part be "unnatural" (Bach and Harms 1972, Anderson 1981, Hyman 2000). OT in contrast allows only "natural" grammars, constructed by ranking universal and/or functionally-motivated constraints. It thus fails to provide an adequate account for how accidents of history are incorporated into synchronic systems.
iv. Ineffability.
Some derivations produce no output whatsoever, eg. schm-reduplication with words like schmo and Schmidt for many English speakers. One of the central tenets of OT, Violability, explicitly predicts that ineffability should not exist. Orgun and Sprouse 1999 show that the Null Parse account of this phenomenon (Prince and Smolensky 1993) does not work; their solution requires abandoning Violability, which seriously undermines the OT enterprise.
v. Derived environment effects.
These are found in many or all languages, and have been treated in OT with conjoined constraints (Lubowicz 2002). Conjunction however encounters a fatal number of problems (Kager 1999, Gouskova 2002, McCarthy 2002) and moreover fails to account for DEC effects in L2 acquisition, where OT is forced to posit an implausible intermediate stage that contains both the constraint ranking of the target language and additional conjoined constraints found in neither L1 nor L2 (Iverson 2002).
I show that a serial, rule-based model along the lines of SPE and Halle and Marantz 1993 accounts straightforwardly for each of these phenomena, whereas Classical Optimality Theory, wherein "there is no sequential phonological derivation in the sense of traditional generative phonology [and] there is no set of rules and operations applying in a certain order" (Ito and Mester 1997) is fundamentally unable to derive any of them in an insightful way.
B. OT predicts the existence of unattested phenomena. Steriade 1999 observes that some phonological constraints receive only one solution across languages, e.g. coda devoicing. One of the core tenets of classic OT, free ranking/factorial typology (McCarthy and Prince 1993:145), explicitly and incorrectly requires that a wide range of repair strategies be employed cross-linguistically.
C. OT fails to provide satisfactory solutions to the problems it identifies in Derivational Phonology, such as conspiracies (Idsardi 1998; see also Kiparsky's 1973 refutation of Kisseberth).
In short, OT has failed to surmount the problems its practicioners associate with Derivational Phonology and has instead created new unsurmountable problems. Furthermore, if one considers that the self-proclaimed OT successes in accounting for markedness, naturalness, and conspiracies are not an exclusive OT prerogative (cf. Calabrese 1995), one sees no reason to maintain this theory.
It is time for an alternative to OT. It is my belief that this theory must have the following features:
i. It must be serial, derivational, and rule-based.
ii. It must recognize the basic difference between processes that are identical across languages (which must be explained by rules) and processes which are functionally related in conspiracies (which must be explained by negative constraints).
iii. It must be able to differentiate between universal and language-specific processes.
iv. It must have fundamental restrictions on the power of its phonological manipulations.
For further information, please contact czoll@MIT.EDU
MIT RLE Speech Communication Group Seminar
Doug Whalen
Haskins Laboratories"Speech-Specific Areas of the Brain:
fMRI Measures of Typical and Sinewave
Speech and Non-Speech Perception"The posterior superior temporal gyrus (STG) is actively engaged in the processing of speech sounds. But many researchers report results that they claim make this region sensitive to any complex sound, not just speech. Here I present results from two fMRI experiments that indicate that these areas are much more speech-specific than has been claimed by others. In the first experiment, stimuli that differ in complexity (both in the speech and the nonspeech domain) result in different levels of activation in STG, and even in primary auditory cortex. The results are most compatible with an interpretation of speech-specificity in STG. The stimuli with normal speech, however, must always be rather different acoustically from the nonspeech, allowing other interpretations of the data. The second experiment overcomes that liability by using sinewave speech, in which the formants of a speech sound are replaced with time-varying sinusoids ultimately perceivable as speech. Nonspeech combinations of those exact sinusoids were also made. Listeners of varying abilities with the sinewave speech participated in the second experiment. It was found that STG was active with the speech versions but not the nonspeech versions, even for those listeners who did not successfully report the speech percept. The speech system successfully recognized the structure in the sinewave speech, even if the processing was not successful to the point of accurate reporting. Together, these results indicate that there is a large region of STG that is sensitive to speech signals and not to nonspeech signals, even if their acoustic complexity is equated.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
MIT Linguistics Colloquium
Barbara Partee
University of Massachusetts at AmherstTitle TBA
For more information, contact Heejeong Ko or Marketa Ceplova.
UMass, Amherst: Linguistics Colloquium
Thomas Ernst
Title TBA
Not available at this time.
MIT LING-LUNCH
Thursday, November 14, 2002 at 12:00, MIT E39-335 (conference room)Ora Matushansky
MIT/CNRS/Université Paris 8"The DP and the Deepest" Even in predicate position, superlatives in many languages require the presence of the definite article:
(1) This theory is the best.
However, we know independently that as a rule articles syntactically combine with extended NP projections (xNPs) and not with APs (barring nominalizations of various sorts):
(2) *the heavy
To resolve this seeming contradiction, I will suggest that superlatives can only appear in an attributive (modificational) position. Examples like (1) thus involve a null head noun:
(3) This theory is the best 0.
I will show that adopting this hypothesis explains the distribution of superlatives in Russian (the unavailability of short-form superlatives), French (superlative partitives), German and Dutch (predicate agreement facts), Hebrew (the appearance of superlatives in construct state), etc. Finally, Breton provides the most spectacular evidence for the nominal nature of superlatives in that it requires an overt nominal marker on superlatives, which is also what appears in predicate possessives like _mine_. We will then examine the semantic impact of this observation in connection with the fact that attributive superlatives as in (4) are necessarily interpreted non-intersectively, as in (4a):
(4) the tallest mountain
a. = the unique x such that x is the tallest among mountains
b. /=the unique x that is the tallest and is a mountain
For more information, contact anevins@MIT.EDU.
MIT - Phonology Circle
Thursday, November 14, 2002, at 2:00 PM, room TBA
Morris Halle
MIT"The Declension in Russian"
The inflection of Russian nouns is reviewed. It is shown that all Russian noun forms consist of a root and a theme plus a Case/Number suffix. The theme is either o/e or a short [-back, +high] vowel (yer). It is shown that the surface forms are derived straightforwardly from this underlying structure with the help of two phonological processes: a. Jakobson's (1948) rule pair that deletes a vowel before a vowel, and a glide before a consonant; and b. the yer rules, which lower a yer to mid /e/ or /o/ when followed by another yer, and delete it elsewhere. Minor additional rules are needed to derive the Plural forms. The adjective declension is then shown to require little additional machinery, mainly the positing of /oj/ as the adjective theme. The short forms of adjectives are dealt with by a simple further extension. The paper concludes with a brief survey of the adjective declensions in Latvian and Lithuanian and even briefer speculations about the evolution of the Russian forms from their Common Balto-Slavic.
For further information, please contact czoll@MIT.EDU
MIT RLE Speech Communication Group Seminar
Elliot Saltzman
Boston University"Prosodic Boundary-Adjacent Lengthening Viewed as Clock-Slowing"
This work examines the relation between phrasal structure and the control and coordination of articulation within a dynamical systems model of speech production. In this context, we review how speakers modulate the spatiotemporal organization of articulatory gestures as a function of their phrasal position. We present computational simulations that capture several important qualitative properties of these phrase boundary effects, such as prosodically-induced local slowing. This slowing is generated by dynamical effects on the activation timecourse of articulatory gestures and is controlled by prosodic gestures or *-gestures, which share much with the familiar dynamical description of constriction gestures. Prosodic gestures, however, function at boundaries purely to temporally stretch or shrink gestural activation trajectories. This modulation of the 'clock-rate' that controls the temporal unfolding of an utterance near junctures is such that the clock slows increasingly as the boundary is approached and speeds up again as the boundary recedes. Viewing phrase boundaries as warping the temporal fabric of an utterance represents a promising confluence of the fields of prosody and of speech dynamics.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
One-day workshop at Harvard:
LIGHT VERBS IN CROSS-LINGUISTIC PERSPECTIVE
Monday, November 11, 2002.
The Harvard University Linguistics Department and the GSAS workshop on Comparative Syntax and Linguistic Theory will be holding a one-day workshop on light verbs in cross-linguistic perspective. We are soliciting offers of papers from graduate students in the Boston area and North-Eastern US.Presentations and questions will total 30 minutes. The papers should deal with the syntax, semantics and/or historical linguistics of light verb constructions and related phenomena (such as verb classification and types of complex predicates) in your favorite language(s).
Date: Monday, 11th NovemberTime: 9-5, + dinner after
Place: Harvard University
Cost: $10 (+ dinner)
Organizers: Claire Bowern (bowern@fas.harvard.edu) and Conor Quinn (quinn@fas.harvard.edu)
We are pleased to announce that Dr Miriam Butt from Universität Konstanz will be our invited speaker.
Selected papers from the workshop will be published by Harvard University Working Papers in Linguistics.
Please contact us if you would like to be included on our mailing list.
For further information email the organizers, or contact:
Department of Linguistics
305 Boylston Hall
Cambridge, MA, 02138phone: 617-495-5054
fax: 617-496-4447
NELS (North East Linguistics Society)
at MIT
Arnim von Stechow (Tubingen)
Special session in honor of Ken Hale: NonconfigurationalityInvited speakers:
Judith Aissen (UC Santa Cruz)
Mark Baker (Rutgers University)
Mamoru Saito (Nanzan University)
MIT LING-LUNCH
Thursday, November 7, 2002 at 12:30, MIT E39-335 (conference room)Junko Ochi and Masao Ochi
Osaka University/UConn,MIT"Reciprocity and Null Operator Movement" In this talk, we aim to explain some puzzling facts observed in reciprocal constructions in Madurese (a Malayo-Polynesian SVO language spoken in Madura and parts of East Java). Madurese reciprocals (a) require a gap in an argument position, (b) force the actor voice (AV) morphology on the predicate to be absent, and (c) impose a distributional restriction on the negative element. We propose a null operator movement hypothesis for Madurese reciprocals, where the Case checking mechanism entertained in the Minimalist Program plays a crucial role. A guiding idea of our analysis is that Madurese reciprocals involve the composite functions of distribution and reciprocation in the sense of Heim, Lasnik, and May (1991). Specifically, adopting NishigauchiÅfs (1992) analysis of Japanese reciprocals and extending it to Madurese, we propose that Madurese reciprocals involve the null reciprocal operator movement from the gap position to the position of the distributor. We argue that this operator movement is responsible for the obligatory absence of the actor voice, which is empirically supported by other Madurese constructions in which operator movement triggers the loss of the AV along its path. We further claim that this phenomenon is best viewed as a morpho-phonological reflex of the Case checking at the vP level in overt syntax, which gains support from the fact that only the movement of a nominal element affects the actor voice. Finally, we also demonstrate that the distributional restriction on negation falls out as an instance of the inner island. In short, the puzzles in Madurese reciprocals are attributed to a single factor, the null operator movement.
For more information, contact anevins@MIT.EDU.
MIT - Phonology Circle
Thursday, November 7, 2002, at 2:00 PM, room TBA
Andrès Pablo Salanova
MIT"Nasality in Mebengokre and Apinayè"
The process of "orality spreading" in ApinayÈ exemplified in (1) has figured in the phonological literature since Anderson (1976), where it is used to argue that distinctive features can span domains smaller than or non-coincident with segments (in this case, [-nasal] is shared between oral vowels and the inner half of contour segments in the syllable periphery).
(1) $_~V [m] [mž] to go (plural)
~V_$ [m] [mr~um] ant
$_V [mb] [mbotS] ox
V_$ [bm] [obm] dust
The paradigm is again discussed in Steriade (1993), in the context of an effort to characterize the feature [nasal] as privative. In Steriade's analysis, it is fully nasal consonants that are derived by the spread of [nasal] from nasalized vowels, whereas the nasal phase in contour segments is created by a rule inserting [nasal] linked to one of the aperture nodes of the segment. This analysis is right with respect to previous ones in that no role is given to the spreading of orality from vowels, yet it fails to explain why, if [nasal] is present in the consonants, it is only phonologically active in vocalic segments.
In our talk, we propose a new analysis based on Piggott (1992), where contour segments are taken to be surface realizations of sonorant stops. We discuss some conceptual and empirical advantages of this approach with data from ApinayÈ and from two dialects of the closely related language Mebengokre (KayapÛ) collected by us, and consider some remaining problems.
For further information, please contact czoll@MIT.EDU
MIT RLE Speech Communication Group Seminar
Andrew Oxenham
MIT"Perceptual Consequences of Cochlear Nonlinearity"
Physiological studies over the last decade have indicated that the mechanical response to sound in the healthy cochlea is highly nonlinear. This nonlinear behavior, which disappears when the cochlea is damaged, is thought to be responsible in part for the tremendous dynamic range and exquisite frequency selectivity of human hearing, both of which form vital constituents of our ability to function in the acoustic environment. This talk will explore the perceptual correlates of a nonlinear cochlea in people with normal hearing, as well as the perceptual consequences of a loss of nonlinearity in people with impaired hearing.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
Boston University Psychology Department HUMAN DEVELOPMENT PROGRAM COLLOQUIUM
Wednesday, November 6, 2002 at 3:30 PM in Room 150, 64 Cummington Street.Refreshments are served at 3:30pm with the talk immediately following and time for Q&A at the end.
Catherine Snow
Harvard University Graduate School of Education"Learning words: From 'bye-bye' to 'bituminous'" Not available.
Keynote speaker:Susan Goldin-Meadow (University of Chicago)Plenary speaker:
Bonnie Schwartz (U. of Hawaii)
See the BUCLD Web site for further information.
MIT LING-LUNCH
Thursday, October 31, 2002 at 12:00, MIT E39-335 (conference room)Michael Wagner
MITShoichi Takahashi
MIT"Linearization and Holmberg's Generalization in an OV Language" "More than Two Quantifiers" Abstract
This paper argues that Holmberg's Generalization (HG) affects both VO and OV languages. HG prohibits an object from shifting over (i) finite verbs, (ii) participles, (iii) particles, (iv) prepositions, (v) higher arguments. Holmberg (1999:15) summarizes the pattern as follows: "Object Shift cannot apply across a phonologically visible category asymmetrically c-commanding the object position except adjuncts."
While (i)-(iii) are not applicable in OV languages, the evidence presented shows cases of (iv) and (v) in Dutch and German, and serves to explain certain differences in word order options between the two languages. The crucial evidence comes from ditransitive verbs in Dutch and German.
(i) In German verb classes with DO IO order, the IO can shift across negation only when it is introduced by a preposition, but it can shift when introduced by a zero preposition or a post-position. (ii) In languages that lack dative morphology and introduce low indirect objects with prepositions (e.g. Dutch and Swedish) this is reflected in restrictions on pronoun shift. For a narrow class of Dutch verbs that have DO IO order but feature a postposition rather than a preposition the restriction is obviated. (iii) German verb classes that have neutral DO IO word order disallow shifting a IO pronoun over the direct object - this is analogous to restrictions that form part of HG in Swedish. The proposal tries to link HG to a more general linearization restriction on PF-Movement.
The proposal follows Fox and Pesetsky (in progress)'s account of HG in arguing that certain dislocations cannot violate linearization established in earlier phases of cyclic spell-out, and links it to other types of pf-movement reflected in phonological domain formation that seem to obey the same restriction.
Abstract
The puzzle that arises from comparative quantifiers (CQs), such as more than three books, is that they must take narrow scope with respect to the subject QPs if they are in the object position. If we treat CQs as unanalyzable generalized quantifiers, there seems to be no straightforward account for the contrast between CQs and other QPs such as universal QPs. I claim that more than three books is decomposed into a comparative operator -er than three and many books, both of which are generalized quantifiers. In this approach, wide scope of CQs is the result of the representation in which the subject QP is outscoped by the two decomposed QPs. However, I claim that optional Scope Shifting Operation (SSO) deriving the intermediate scope of the subject QP is prohibited by a general condition on optional SSO which says that it must have a semantic effect (Fox 2000). If the first instance of SSO is prohibited, subsequent QR is not allowed even if it has a semantic effect. This is because grammar cannot tolerate optional SSO without semantic effect. Thus, it is never allowed to apply optional SSO deriving the representation in which the two decomposed QPs take wide scope over the subject QP. Consequently, CQs must take narrow scope. I also argue that this approach can capture the fact that monotone decreasing CQs, such as fewer than three books, also must take narrow scope if they are c-commanded by other quantifiers.
For more information, contact anevins@MIT.EDU.
Harvard University
Special Series with Ray Jackendoff, Professor of Linguistics at Brandeis University, in in 202 Harvard Hall. Receptions to follow in the Dept of Linguistics, 3rd floor Boylston Hall, Harvard Yard."Foundations of Language:
Brain, Meaning, Grammar, Evolution"
For more information, contact Soo-Yeon Jeong
MIT Linguistics Colloquium
Daniel Buring
UCLATitle TBA
For more information, contact Heejeong Ko or Marketa Ceplova.
UMass, Amherst: Linguistics Colloquium
Jill de Villiers
Smith CollegeTitle TBA
Not available at this time.
MIT LING-LUNCH
Thursday, October 24, 2002 at 12:30, MIT E39-335 (conference room)Winfried Lechner
Tuebingen/MIT" Interdependencies Between Economy and Merge " The search for a unification of Merge and Move/Attract has received considerable attention in the recent syntactic literature (Bobalijk and Brown 1997; Chomsky 1998, 1999, 2001; Hornstein and Nunes 2001). But the two operations are - at least on standard assumptions - also subject to a number of inherently distinct restrictions. For one, the Minimal Link Condition (MLC; Chomsky 1995) is believed to regulate only Move/Attract, whereas the Linear Correspondence Axiom (LCA; Kayne 1994), which is widely held to determine possible applications of Merge, does not appear to have an effect on Move. Furthermore, while Move operates on symbols which are part of the syntactic tree, the domain of Merge is limited to the numeration. This talk presents evidence indicating that some of these imbalances can be eliminated, resulting in a more parsimonious subtheory of structure building. More specifically, it will be argued that
(i) Economy restricts Merge as well as Move
(ii) the MLC can be reduced to a phase-based version of the LCA and that
(iii) extending the domain of the Attract relation to include the numeration makes it possible to subsume Merge under Move.
Empirically, the presentation will focus on VP-ellipsis, VP-fronting and Superiority phenomena. (Part (i) is largely identical with Lechner 2001/to appear, which can be downloaded at: http://www2.sfs.nphil.uni-tuebingen.de/winnie/Scope.pdf)
For more information, contact anevins@MIT.EDU.
MIT - Phonology Circle
Thursday,October 24, 2002, at 2:00 PM, room TBA
John Frampton
Northeastern University"Cyclic effects in Bantu verb reduplication"
Under the influence of Prosodic Morphology, prosodic concerns dominated the discussion of reduplication for over a decade. Hyman and Mutaka (1990), Downing (1997,98), and Hyman, Inkelas, and Sibanda (1999), highlighted the effect of hierarchical word internal moprhosyntactic structure on reduplication in various Bantu languages. This idea has developed into Inkelas and Zoll's Morphological Doubling (MD) theory of reduplication. While applauding the attention this approach focuses on the morphological structure of complex words, I hope to show that it is a mistake to view the reduplicant itself as a complex morphological structure. The complex morphology exerts its influence only through the cycle and its effect on the choice of material which is copied.
The most thoroughly worked out analysis in the MD framework remains Hyman, Inkelas and Sibanda's (HI&S) paper on Ndebele verbal reduplication. In a previous Phonology Circle talk (Oct, 2001) I showed that the main features of Ndebele reduplication could be analyzed simply in the Distributed Reduplication (DR) framework (Frampton, 2001), which relies on phonological copying, not doubling a morphosyntactic structure. This showed that Ndebele provides no evidence that phonological copying is inadequate. In this talk, I will examine some of the subtler features of Ndebele reduplication, demonstrate that MD makes incorrect predictions, and show how these features of Ndebele verb reduplication can be explained in a phonological copying framework (DR).
HI&K discovered a striking contrast between reduplication of an applicativized passive and a passivized applicative. Ndebele imposes surface conditions on suffix order (a "template"), which the derivational morphophonology is designed to satisfy. It achieves template satisfaction by infixing certain suffixes in certain environments. This produces surface violations of the Mirror Principle.
(1) a. ukudla kw-a-pheka-el-w-a abantwana
`food was cooked for the children'
(applicativized passive)b. abantwana b-a-phek-el-w-a ukudla
`the children were cooked food'
(passivized applicative)kw/b a phek el w
Subject_Agr Past cook APP PASSAlthough the surface order of the passive and applicative suffixes is the same in both cases, because an affix is infixed in one case, the reduplicative possibilities reveal the underlying order. With unintensive reduplication, (1a) has three variants and (1b) has only two.
(2) a. ukudla kw-a-pheka-el-w-a abantwana
ukudla kw-a-pheke-el-w-a abantwana
ukudla kw-a-phekwa-el-w-a abantwanab. abantwana b-a-pheka.phek-el-w-a ukudla
abantwana b-a-pheke.phek-el-w-a ukudla
*abantwana b-a-phekwa.phek-el-w-a ukudlaHI&S realized that the possibility of /phekwa/ in (2a) is a consequence of the fact that PASS is inside APP syntactically in (2a), but not in (2b). But I will show that the MD framework prevents them from giving a coherent account of how the facts in (2) follow from the verb internal syntactic structure. These facts will be shown to be a straightforward consequence of cyclic morphology and reduplication via phonological copying.
Further topics in Ndebele and Kinande reduplication will be discussed as time permits.
For further information, please contact czoll@MIT.EDU
MIT RLE Speech Communication Group Seminar
David Ostry
McGill University"The Somatosensory Basis of Speech Production"
Although speech production is substantially dependent upon auditory information, the capacity for intelligible speech by deaf speakers suggests that somatosensory input may contribute to the achievement of speech targets. In this talk, I will describe recent studies that provide evidence that somatosensory input is indeed central to the speech target. This conclusion is based on studies in which a robotic device is used todeliver mechanical perturbations to the jaw. With this device we have succeeded in manipulating somatosensory feedback independent of speech acoustics. I will show that when mechanical perturbations that have no measurable nor perceptible effect on speech acoustics are delivered to the jaw, adaptation can be observed on the basis of somatosensory feedback alone. That is, even when the acoustic goal is achieved, subjects modify their motor commands in order to reach somatosensory targets. Thus far, the pattern appears to be specific to speech: adaptation to the force field produced by the robot does not occur in matched non-speech movements even following extended practice.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
UMass, Amherst: Linguistics Colloquium
Richard Larson "The Projection of DP Structure"
Following work by Szabolsci (1983,1987) and Abney (1987), many researchers have pursued the idea that clauses (CP/TP) and nominals (DP) are fundamentally parallel in structure. Despite its overwhelming popularity, however, this view is not well-supported by semantic analysis. Indeed, under generalized quantifier theory (Barwise and Cooper (1981), Keenan and Stavi (1986)), which provides the basis of nearly all recent work on quantification, C/T and D have little or nothing in common.
In this talk, I discuss the syntactic projection of DP from the standpoint of generalized quantifier theory. I argue that, under the latter, the most appropriate analogy is *not* between between DP and CP/TP, but rather between DP and VP. Specifically, I suggest that (i) DP can be understood as projecting arguments according to a thematic hierarchy that is parallel (but different in role-content) to that found in VP, (ii) that Ds sort themselves into intransitive, transitive and ditransitive forms, much like Vs, and (iii) that nominal modifiers, including relative clauses, project in the DP very much like adverbial elements in VP. A surprising consequence of this view concerns the analysis of prenominal genitives, which have (since at least Chomsky 1970) been taken to be sentence-like in many examples. I suggest that, on the view argued for here, prenominal genitive constructions are fundamentally parallel to double object forms in the VP, and arise by a form of "genitive shift".
If correct, these views suggest that many of the putative parallels between DP and CP/TP claimed over the last 15 years may deserve rethinking.
MIT LING-LUNCH
Thursday, October 17, 2002 at 12:00, MIT E39-335 (conference room)Heejeong Ko
MIT"External-Merge of Why-in-situ" In the talk, I claim that the peculiar behavior of 'why' in the intervention effect context (Beck and Kim 1997) reveals the answers to those questions. Based on the contrast between local vs. non-local construal of 'why' in the intervention effect context, I argue that 'why' is externally-merged in [Spec,CP] in the overt syntax: External-Merge Hypothesis (EMH). Further, I argue that 'why' is subject to the same LF licensing condition as other wh-phrases.
This claim is further supported by the cross-linguistic prediction concerning a non-scrambling wh-in-situ language: Chinese. It will be shown that the contrastive behavior of 'why' between Chinese vs. Korean/Japanese follows from the EMH and the availability of long distance scrambling in each language. I also present a new set of Korean-speaking child acquisition data (2:00-2:11), which shows a robust contrast between 'why' and 'where' in word order restriction. It will be shown that this word order contrast is exactly what the EMH predicts on the earlier stage of Korean acquisition.
For more information, contact anevins@MIT.EDU.
MIT - Phonology Circle
Thursday,October 17, 2002, at 2:00 PM, room TBA
Maria Guskova
University of MassachusettsTitle TBA
For further information, please contact czoll@MIT.EDU
MIT RLE Speech Communication Group Seminar
Keith Johnson
Ohio State"The Limits of Phonetic Transcription
Phonetic transcription provides much of the data used in language description, phonology, speech pathology, and in speech technology. Researchers in pathology and technology have studied the reliability phonetic transcription, while linguists have tended to assume that transcription disagreements indicate ideolectal differences among speakers, or the moral degeneracy of the other linguist. In this talk I will describe the ViC project, a vertically integrated research program exploring the phonetics, phonology, and psycholinguistics of conversational speech. In the first stage of this project we have collected 40 hours of unmonitored conversational speech and are in the process of phonetically transcribing it all. The focus of the talk will be on the design and results of a transcriber inter-reliability study, looking at transcriber agreements as a function of broad and narrow phonetic categories, and, for vowel transcription, the acoustic properties of inconsistently transcribed segments.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
Arto Anttila
NYUASL/English interpreters will be present.
Harvard University
Comparative Syntax and Linguistic Theory Workshop
Friday, October 11, 2002 at 4:00, Harvard U., Boylston 303Claire Bowern
Harvard University" The Devolution of Noun Incorporation" While the two branches of the Nyulnyulan family are quite close, we are presented with anomalies in verb root number and structure between the two groups. It is a startling fact that around two-thirds of the closed class of Eastern Nyulnyulan roots are directly reconstructible to Proto-Nyulnyulan qua roots, but the figure is much lower for the Western languages. This arises in part because the 2 Western languages also have more than twice as many roots as the 2 Eastern languages (over 200 versus fewer than 100). What happened to the verb roots in the Nyulnyulan languages? Why do 4 languages with overall very similar verb structure and morphology show such differences in the number of verb roots (85-225+), the etymology of those roots, and the organisation of their verb classifier/light verb systems? I propose here that many of the `extra' roots in Western Nyulnyulan can be analysed as fossilised combinations of incorporated nominal + monosyllabic verb root. This is particularly interesting from the point of view of historical syntax because the same monosyllabic verb roots are still in use in Nyulnyulan languages as light verbs in complex predicates. We will examine the historical syntax of Nyulnyulan predicate formation to reveal some interesting stable and not-so-stable properties.
For more information, contact Soo-Yeon Jeong
MIT LING-LUNCH
Thursday, October 10, 2002 at 12:00, MIT E39-335 (conference room)Jim Harris
MIT"Post-syntactic movement of [+plural] in Spanish clitics and the Local Dislocation Hypothesis"
For more information, contact anevins@MIT.EDU.
MIT - Phonology Circle
Thursday,October 10, 2002, at 2:00 PM, room TBA
Zhiqiang Li
MITTitle TBA
Not available at this time.
For further information, please contact czoll@MIT.EDU
MIT RLE Speech Communication Group Seminar
John Kingston
UMass, Amherst"Lenition: When (Some) Sonority Differences Don't Matter"
Not available at this time.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
MIT Linguistics Colloquium
Stephen Crain
University of Maryland at College Park"The Return of the Nativist" In the normal course of events, children manifest linguistic competence equivalent to that of adults in just a few years. Children can produce and understand novel sentences, judge that certain strings of words are true or false, and so on. Yet experience appears to dramatically underdetermine the competence children so rapidly achieve, even given optimistic assumptions about children's nonlinguistic capacities to extract information and form generalizations on the basis of statistical regularities in the data they receive.These considerations underlie various poverty-of-stimulus arguments for the innate specification of linguistic principles. But apparently certain features of nativist arguments have not yet been fully appreciated. This talk focuses on three (related) kinds of poverty-of-stimulus argument, with relevant empirical evidence from psycholinguistic investigations of child language. The first argument hinges on the observation that children project beyond their experience in ways that their experience does not even suggest. The issue is not whether or not children project beyond their experience. The issue is how children project beyond their experience. Evidence in favor of the nativist perspective comes from experimental studies of child language showing that children's projections do not violate any core principles of Universal Grammar, even in cases where children might be tempted to violate such principles if they adopted general-purpose learning algorithms. A second poverty-of-stimulus argument is based on the kinds of non-adult constructions children produce. Children appear to follow the natural seams of natural language even when child language diverges from the local adult language. On an experience-dependent approach to language learning, the pattern of children's non-adult linguistic behavior would presumably look quite different from this. From a data-driven perspective, children's non- adult productions would be expected to be simply less "filled out" than those of adults in the same linguistic community. The UG-based approach, by contrast, is largely consistent with the continuity assumption, which supposes that child and adult languages can differ only in limited ways -- specifically in ways that adult languages can differ from each other. If so, children are expected to project beyond their experience in ways that are attested in natural languages. The non-adult linguistic behavior of children is not expected to match the input (as empiricist approaches to learning suggest); rather the input is seen to guide children through an innately specified space of hypotheses made available by Universal Grammar, until they hit upon a grammar that is sufficiently like that of other speakers of the local language; at that point, language change is no longer initiated by the input.A third argument is based on the gap between a typical child's experience and the linguistic principles that govern children's competence. The key observation here is that linguistic principles unify and explain (superficially) disparate phenomena. These principles are "specific contingent facts" about language -- which apply to a wide range of constructions across different linguistic communities. In so far as this aspect of linguistic competence is not plausibly a product of children's experience, it is presumably a product of their biological endowment.
For more information, contact Heejeong Ko or Marketa Ceplova.
Boston University
Barnes and Noble Bookstore
Steve Pinker
MITA lecture about his new book:
The Blank Slate:
The Modern Denial of Human Nature
MIT RLE Speech Communication Group Seminar
Michelle Gregory
Brown University"Informativeness and Prosody:
The Role of Predictability on
Duration and Pitch Accent Placement"In this study I investigate the role of predictability in language production. A long-standing observation in language production is that high frequency and highly predictable words tend to become truncated, or shortened (Schuchardt, 1885; Zipf, 1929). Indeed, recent research has demonstrated that many aspects of word pronunciation are influenced by predictability (Bell, et al., 2002, inter alia). Despite the growing body of literature that demonstrates that predictability affects the way words are produced during lexical production, the relationship between predictability and the production process is not fully understood. For example, does word predictability only refer to how predictable a word is to the speaker, or are there other aspects of the speech context that interact with the effects of predictability? Are only reductive processes influenced by predictability?
The goal of the present work is to expand our understanding of how predictability interacts with the production process. I report on experiments that demonstrate that predictability affects the production of both content and function words. I also report on experiments that demonstrate that the reduction of predictable words is in part influenced by factors in the linguistic context. Lastly, I report on experiments that investigate whether predictability plays a role in non-reductive production processes: the placement of pitch accent. In a corpus study of the prosodic characteristics of words, I demonstrate that the predictability of a word, as measured by its frequency, conditional probability given surrounding words, repetition, and its semantic relatedness to the context, has a significant effect on whether the word bears pitch accent. These results indicate that predictability aids in the construction of prosodic phrases during speech production. I discuss the implications of these results for models of speech production.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
MIT LING-LUNCH
Thursday, October 3, 2002 at 12:00, MIT E39-335 (conference room)Christopher Kennedy
NWU, MIT"Comparative and relative clauses compared"
For more information, contact anevins@MIT.EDU.
MIT - Phonology Circle
Thursday,October 3, 2002, at 2:00 PM, room TBA
Michael Wagner
MITCinque(1993), drawing on observations going back to Kiparsky(1966) and the theory of metrical phonology outlined in Lieberman & Prince (1977), argues that nuclear stress is universally assigned to the most deeply embedded constituent. His observations encompass nuclear stress (corresponding to the last pitch accented syllable in a declarative sentence) main stress in compounds. Under the assumption that there is only a single generative engine to form compose morpho-syntactic objects, be they words or phrases, (as proposed in Distributed Morphology), we can ask the question how syntax is mapped in a more general way: how are syntactic configurations in Words, Compounds, and Phrases mapped to prominence relations as encoded by the grid?
The evidence from word formation, compounding, and phrasal phonology presented indicates that there is a single mechanism negotiating stress across these morphosyntactic domains. The match between syntax and phonology is far greater than current approaches to phonological domain formation would let one hope. An new algorithm is sketched that maps trees in to grids, and derives all accents and accent domains, not only the `nuclear' one. It draws on the theory of stress and the distinction between cyclic and non-cyclic affixation in Halle & Vergnaud (1987).
For further information, please contact czoll@MIT.EDU
MIT Linguistics Colloquium
Tanya Reinhart
Utecht/Tel aviv University(work with Tali Siloni)
"Thematic arity operations: Parametric variations"
For more information, contact Heejeong Ko or Marketa Ceplova.
UMass, Amherst: Linguistics Colloquium
Gennaro Chierchia "More on scalar implicatures and NPIs"
Not available at this time.
MIT LING-LUNCH
Thursday, September 26, 2002 at 12:00, MIT E39-335 (conference room)Nora Boneh
Université Paris 8"Possessive and Locative Constructions in Modern Hebrew and Standard Arabic" The starting point of the talk is Freeze's (1992) analysis of the locative paradigm argues for the same underlying structure for predicative-locative, existential and possessive clauses. It will be claimed here that while Arabic and Hebrew are have-less, Arabic corresponds to Freeze's findings but Hebrew does not. The preposition in the Hebrew possessive construction is not the head of the predicate phrase, and the possessee is not the predicate phrase's subject. The alternative analysis for Hebrew will be suggested is in the spirit of Kayne (2000). The possessor DP does not form a constituent with the preposition; rather the preposition is merged above the predicate phrase (SC), and the possessor is then attracted to it. Finally, I will attempt to link the difference between the language's possessive constructions to differences that the two languages present in the syntactic representation of tense.
For more information, contact anevins@MIT.EDU.
MIT - Phonology Circle
Thursday, September 26, 2002, at 2:00 PM, MIT 36-144
Mary Ann Walter
MIT"Prominence effects in final positions"
For further information, please contact czoll@MIT.EDU
UMass, Amherst: Linguistics Colloquium
Carson Schütze This talk focuses on the syntax of dummy 'do'. An analysis of standard English is proposed that will be used to also account for historical and dialectal variation, the related phenomenon of periphrastic 'tun' in colloquial German and neighboring Germanic dialects, and facts about English acquisition. The acquisition facts include children's early non-adult use of "don't" with 3rd person singular subjects, where I will contrast my approach to INFL underspecification with that of Guasti and Rizzi (2002).
The central claim of the analysis is that 'do' does not belong under the Tense head, nor does it arise by adding structure in the Spell-Out component. Rather, 'do' is an expression of a Mood head, specifically an allomorph of Indicative Mood, whose other allomorph is zero. Its "support" function is executed by the conditions on Vocabulary Insertion (in the sense of Distributed Morphology), whose task is to determine which allomorph of Indicative to spell out in a given environment. The apparent "last resort" restriction on 'do' is argued to be an historical accident, and a model of INFL is sketched in which free variation of the sort found with 'tun' and in child English is expected.
MIT LING-LUNCH
Thursday, September 19, 2002 at 12:00, MIT E39-335 (conference room)Ingo Reich
Tuebingen/MIT"Pied Piping and the (syntax and) semantics of complex wh-phrases" In this talk, I want to investigate the so-called pied piping problem or more generally, the syntax and semantics of complex wh-phrases. Having discussed the most prominent previous approaches -- the reconstruction approach, the choice function approach and the presuppositional approach -- it will be argued that all interrogative wh-constructions -- wh-phrases and wh-interrogatives -- should be treated both syntactically and semantically in a uniform way. To this effect, an analysis is presented that generalizes the treatment of pronominal wh-phrases and wh-interrogatives developed in Reich (2002) which will be sketched during the talk to arbitrary instances of complex wh-phrases. I conclude that wh-phrases differ from wh-interrogatives only in that they are assigned an additional functional projection that accounts for their specific phrasal behaviour.
For more information, contact anevins@MIT.EDU.
MIT Linguistics Colloquium
Hilda Koopman
UCLA"Decomposing DPs" I will assume agreement reflects a Spec head relation, and will thus defend Spec head over Agree (Chomsky, 1998). My analysis yields new insights into structure of the DP (DPs are basically relative clauses D CP/IP structures, containing a nominal small clause; the variable of the noun is projected as the subject of the NP predicate) , the basic building blocks that make up the DP (case number and gender are merged as independent projections low in the DP), the movement processes that apply within DPs (there is no head movement; only movement to Spec position, either akin to NP movement, or predicate inversion (movement of a predicate over a Spec position to a subject (=Spec position) (Moro, 1997), will allow me to make sense of the quite extensive and asymmetric agreement patterns within the Maasai DP, and suggest a simple way in which English and Maasai DP differ (English is a ?subject? raising language, Maasai a predicate inversion one).
A preliminary version of this paper can be found on her website
http://www.linguistics.ucla.edu/people/koopman/koopman.htm
For more information, contact Heejeong Ko or Marketa Ceplova.
MIT LING-LUNCH
Thursday, September 12, 2002 at 12:00, MIT E39-335 (conference room)Michael Kenstowicz,
MIT"Tone in Buli" George Akanlig-Pare (University of Ghana) and Michael Kenstowicz (MIT)
In this paper we report some of the results of our study of the tonology of Buli, a Gur language of Northern Ghana. The research grows out of the Field Methods course (24.942) held last semester at MIT as well as the first author's doctoral dissertation. The presentation has four parts. First, we situate Buli in the more general Gur context by reviewing the features found in other Gur languages studied in the recent literature (tonal polarity, a high vs. low vs. zero lexical contrast, with low as the default tone and a process of H-tone Spread creating a downstep). Second, we document a quite different tonology for Buli: high vs. mid vs. low lexical contrasts as well as two general tonal processes: L-tone Spread and Rising-Tone Absorption. Next we examine the tone of epenthetic syllables and the likely historical origin of the Buli mid tone. Finally, we briefly survey the complex tonal alternations in the verbal inflection that reflect tense and aspectual distinctions and offer a tentative analysis of same.
For more information, contact anevins@MIT.EDU.
MIT
CP Doherty, WC West*, C Evans,
S Shattuck-Hufnagel, D Caplan,Harvard Medical School & MIT, MA
QUESTION-INTONATION PROCESSING: AN FMRI STUDY Intonation refers to the use of suprasegmental phonetic features to convey sentence-level pragmatic meanings. We examined changes in fMRI BOLD signal associated with question-intonation compared to wh- questions and statements to investigate the notion of a phonological role for intonation. Stimuli consisted of digitized recordings of a male voice enunciating 150 sentences in English intermixed with white noise bursts in an event-related paradigm. Sentence types included pairs of lexically similar stems, with a falling intonation for S and rising intonation for Q [e.g., "She was talking to her father." (S) vs. "She was talking to her father?" (Phonetic Q)] and a 3rd type, with a falling intonation with a wh-word beginning [e.g., "Was she talking to her father?" (lexical Q)]. Functional EPI scans were collected (Seimen's 3.0T scanner) from 11 normal subjects who listened to stimuli through masked headphones and made Q/S judgments with a button press. Results suggest that the processing of semantic features of Q over S preferentially activate left middle and inferior temporal gyri and the anterior cingulate gyrus (ACG). Phonological differences between phonetic and syntactic Qs (questions with identical semantic features) caused preferential activation in the right inferior frontal lobe, which has been implicated in pitch processing of tones. The left ACG was also active in this comparison. A hypothesis driven anatomical region of interest (ROI) analysis further indicates that areas in the left hemisphere known to be involved in the phonological processing of sublexical and lexical items are also implicated in the processing of question-intonation.
This information received from lisa@speech.mit.edu.
American Association of Teachers of French (AATF)
GSAS Workshop on Comparative Syntax and Linguistic Theory at Harvard
Carol Neidle
Boston University"Focus and Wh Constructions in ASL" Wh-question constructions in American Sign Language (ASL) containing in situ wh phrases and moved wh phrases appearing at the right periphery of the clause will be compared. A syntactic account involving distinct projections for focus and wh phrases will be proposed. The differences in interpretation of the two constructions will be argued to follow from Relativized Minimality. Evidence in support of this analysis comes from (1) certain restrictions on the occurrence of wh-movement, and (2) the distribution of an indefinite focus particle in ASL.
The findings to be discussed have emerged from joint work with Ben Bahan, Sarah Fish, and Paul Hagstrom, supported in part by grants from the National Science Foundation.
There will be ASL/English interpreters present.
For more information, contact Conor McDonough Quinn
MIT RLE Speech Communication Group Seminar
Helen Hanson
MITTitle TBA Not available at this time.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
MIT - Phonology Circle
Friday, May 10, 2002, 204 PM, MIT 56-180
Dr. Bushra Zawaydeh
BBN Technologies"The Phonetics and Phonology of Gutturals in Arabic" In this talk I will review results of the research I conducted for my dissertation work, which was done at Indiana University. I investigated two issues that deal with the physical attributes of a class of speech sounds made with a constriction in the back part of the vocal tract, called the gutturals (they include the uvulars, uvularized, pharyngeal, and laryngeal sounds). The first issue revolves around why gutturals comprise a natural class despite their apparent articulatory dissimilarities. To answer this question, an articulatory (endoscopic) experiment and an acoustic experiment were conducted. The endoscopic experiment indicated that there is a constriction in the pharynx during the articulation of all the sounds except the laryngeals. This commonality explains grouping in the Salish languages of the Pacific Northwest, where all the gutturals but not the laryngeals pattern as a natural class. The acoustic experiment indicated a consistent acoustic effect on the lowest vocal tract resonance when the preceding consonant is a guttural. Hence, it seems that Arabic uses an acoustic feature [high F1] for grouping gutturals together. The second issue investigated in this dissertation deals with the spreading of uvularization from a subgroup of the gutturals: the uvular and uvularized segments. Contrary to what has been previously assumed, the spreading of uvularization is a phonetic process, not a phonological one. This is because no high segments block leftward or rightward spreading, the spreading is gradient in the rightward direction, and the strength of spreading is systematically affected by numerous phonetic factors. Both experiments highlight how important auditory features are in the phonology and phonetics of Arabic, and emphasize the importance of conducting phonetic fieldwork to test phonological phenomena. Hence, this research highlights the importance of the integration of phonetics and phonology.
For further information, please contact kenstow@MIT.EDU
MIT Linguistics Colloquium
Barry Schein
University of Southern California"Coordination and Number Agreement in Lebanese Arabic"
For more information, contact Michael Wagner or Heejeong Ko
GSAS Workshop on Comparative Syntax and Linguistic Theory at Harvard
Miguel Vazquez Larruscain
Harvard (Visiting Scholar, Linguistics)/Real Colegio"A Transderivational non-metrical approach to inflectional stress: the Russian nominal declension"
For more information, contact Conor McDonough Quinn
MIT LING-LUNCH
Thursday, May 9, 2002 at 12:00, MIT E39-335 (conference room)Gianluca Storto
UCLA"Possessives in context"
It has been argued already many times in the literature that Williams's (1982) contention that "the relation between the possessive NP and the following N0 can be any relation at all" is not correct. Research by Barker (1995), Partee (1997), Partee and Borschev (2000a; 2000b; to appear), and Vikner and Jensen (1999), among others, has shown that there are grammatical constraints on what possessives can mean. Not all (potentially conceivable) interpretations are available for every specific instance of a possessive NP, which suggests the hypothesis that these interpretations--or better, classes of them--have distinct sources in the semantic composition of possessives.In previous work (Storto, 2000a,b) I have argued that grammatical constraints on the interpretation of possessives are to be found not only in the lexical choice of possessor and possessum, but in the syntactic/semantic properties of the whole possessive NP as well. Even keeping the context of use and the lexical choice of possessor and possessum constant, indefinite possessive NPs license only a subset of the interpretations licensed by definite possessive NPs. In particular, the relation holding between possessor and possessum in indefinite NPs seems to be restricted to the relation of ownership: indefinite possessives do not license interpretations whose semantics seems to be provided by their context of use. I used this as evidence to claim that ownership interpretations and contextual interpretations have distinct sources (i.e. are encoded in distinct ways) in the semantic composition of possessive NPs.
In this talk I revise, expand and strengthen the arguments for these conclusions on the basis of the following observation: even those interpretations for possessives that--at first sight--derive from their context of use do not constitute a homogeneous class. Two types of interpretations are easily distinguished on the basis of their distribution. E.g., consider the fact that the Italian (1c) can be used in the context given in (1a) but is quite infelicitous in the context given in (1b). The indefinite NP due cani di Gianni can be interpreted as denoting two entities satisfying the property in (2a) but cannot be interpreted as denoting two entities satisfying the property in (2b). Interpretations of the first type can be characterized as involving relations that can be described as control : the possessor has some sort of control of the possessum or of his bearing the relevant relation to the possessum. Conversely, interpretations of the second type involve relations that can!!not be construed as instances of control. The distribution of interpretations of the first type--control interpretations--is essentially unrestricted. Interpretations that cannot be construed as involving a relation of control--contextual interpretations--have a more restricted distribution (see the table in (3)). This suggests (i) that the relation of ownership which is usually taken to be the meaning of possessive constructions is just a special case of control, and (ii) that the role of the context of use in the derivation of control vs. contextual interpretations is somewhat different.
(1)a. Ieri a Gianni e Paolo sono stati affidati due gruppi (distinti) di
yesterday to Gianni and Paolo were entrusted two groups (different) of
cani...
b. Ieri Gianni e Paolo sono stati attaccati da due gruppi (distinti) di
yesterday Gianni and Paolo were attacked by two groups (different) of
cani...
dogs...
unfortunately twoudogs of Gianni had the rabiesabbia.
b. LAMBDAx[dog'(x) & attack'(x,g)]to'(x,g)]
(1a) the dogs of G two dogs/each dog of G two/each of the dogs of G
the dogs of G two dogs/each dog of Gianni two/each of the dogs of G
I argue that the difference in the role played by context in the derivation of the two types of interpretations is a matter of degree, and that this difference is a consequence of the different syntactic/semantic encoding of the relation holding between possessor and possessum in the two cases. In control interpretations the relation is encoded in the semantic composition of the possessive NP by a lexical constant Cctrl denoting a general control relation, whose interpretation is further specified as a pragmatic inference from the context of use. In contextual interpretations the relation is encoded by a variable R whose meaning is not specified at all in the semantic composition of the possessive NP and is entirely determined by the context of use, again through a process of pragmatic inference. I furthermore argue that the semantics of the definite article is crucial in deriving an interpretation for this relational variable. R is not a pronoun-like category and assignment o!!f a value to it is not akin to (discourse-)binding: R cannot be directly bound to an antecedent in the context or by a process of text-level existential closure as in the case of individual (type e) variables. Binding of R takes place at the level at which the definite article is composed with the property denoted by its NP complement.
This analysis--whose consequences for the taxonomy of possessive interpretations and for the more general issue of information-tracking in discourse will be discussed in the talk--accounts for the restricted distribution of contextual interpretations: the variable R cannot get bound in indefinite and quantificational possessive and the derivation of contextual interpretations crashes (an LF containing an unbound variable is not well-formed).
References
Barker, Chris 1995. Possessive Descriptions. Stanford, Cal.: CSLI Publications.
Partee, Barbara 1997. Genitives - A case study. In Johan van Benthem and Alice ter Meulen (eds.), The Handbook of Logic and Language, pp. 464-470. Dordrecht: Elsevier. Appendix to: Theo Janssen, Compositionality.
Partee, Barbara and Vladimir Borschev 2000a. Genitives, relational nouns, and the argument-modifier distinction. In C. Fabricius-Hansen, E. Lang, and C. Maienborn (eds.), Approaching the Grammar of Adjuncts, ZAS Working Papers in Linguistics 17, pp. 177-201. Berlin: Zentrum f"ur Allgemeine Sprachwissenschaft.
-------- 2000b. Possessives, favorite, and coercion. In Rebecca Daly and Anastasia Riehl (eds.), ESCOL 99 Proceedings. Ithaca: CLC Publications.
-------- to appear. Some puzzles of predicate possessives. In R. Harnish and I. Kenesei (eds.), Festschrift for Ferenc Kiefer. Amsterdam: John Benjamins.
Storto, Gianluca 2000a. Double genitives aren't (quite) partitives. In Arika Okrent and John Boyle (eds.), Papers from the Thirty-Sixth Regional Meeting of the Chicago Linguistic Society, vol. 36-1, pp. 501-516. Chicago, Ill.: CLS, University of Chicago.
-------- 2000b. On the structure of indefinite possessives. In Brendan Jackson and Tanya Matthews (eds.), Proceedings of Semantics and Linguistic Theory X. Ithaca, NY: CLC Publications, Cornell University.
Vikner, Carl and Per Anker Jensen 1999. A semantic analysis of the English genitive: Interaction of lexical and formal semantics. Manuscript, University of Copenhagen and University of Kolding.
Williams, Edwin 1982. The NP cycle. Linguistic Inquiry, 13:277-29
For more information, contact anevins@MIT.EDU.
GSAS Workshop on Comparative Syntax and Linguistic Theory at Harvard
Salikoko Mufwene
University of Chicago, and Visiting Professor, Harvard"Some consequences of thinking of language as a species"
For more information, contact Conor McDonough Quinn
MIT RLE Speech Communication Group Seminar
Louis Goldstein
Haskins Lab"Gestural Action Units and their Emergence in Phonological Development" Articulatory phonology has hypothesized that the primitive units of phonological structure are units of vocal tract action during speech production ("gestures"). There has been relatively little evidence, however, for the decomposition of articulatory activity during speech production into units of a phonetic grain-size. In contrast, several approaches treat execution in speech production as continuous process of navigating a smooth trajectory through the sequence of goals specified in a plan (which is composed of units). In this talk, I will present the results of some recent experiments on the kinematics of elicited speech errors that support the view that production of speech (not just its plan) involves units of constriction action. In addition, I will show how the dynamical interaction of between such units may account for some of the observed patterns of errors (e.g., asymmetries) that arise in both elicited and spontaneous speech errors. Finally, I will discuss how constriction gestures capable of bearing phonological contrast could emerge in the child, and how that emergence can account for some generalizations about phonological development.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
GSAS Workshop on Comparative Syntax and Linguistic Theory at Harvard
(Note: postponed from April 30.)
Judy Kegl
University of Southern Maine"Are there Serial Verbs in Nicaraguan Sign Language?"
For more information, contact Conor McDonough Quinn
Workshop on the Semantics/Syntax of Possessive Constructions
UMass, Amherst
Eleventh Annual Workshop on Formal Approaches to Slavic Linguistics
UMass, Amherst
MIT Linguistics Colloquium
Anoop Mahajan
University of California, Los AngelesRemarks on Non-nominative "Subjects"
For more information, contact Michael Wagner or Heejeong Ko
MIT - Phonology Circle
Friday, Aprl 26, 2002, 1-3 PM, MIT 56-154
Adam Albright
UCLA"A Restricted Model of UR Discovery:
Evidence from Lakhota"A fundamental tenet of generative phonology is that learners compare surface variants of morphemes, combining unpredictable information to establish URs that distinguish all surface forms. In this talk, I pursue the more restrictive hypothesis that learners base URs on a single surface alternant, even if this means that some unpredictable properties must be handled by mechanisms other than the UR (Hayes 1999; Albright et al. 2001). I present evidence for this approach from Lakhota. I also provide a computationally implemented algorithm that selects the optimal UR under this restriction, learns rules to derive the rest of the paradigm, and identifies exceptions, which must be lexically listed. In Lakhota, some verbs display an invariant final vowel and final reduplication (1a), while others exhibit an [e]~[a] alternation and penultimate reduplication (1b):
(1) a. 'good' 'spotted' b.'fat'
3sg. washte gleshka chepe
3pl. washte-pi gleshka-pi chepa-pi
REDUP. washte-shte gleshka-shka chep-chepa In a traditional approach, learners could compare these forms and posit URs that distinguish them; following Boas & Deloria (1941) and Shaw (1980), we might say that 'good' ends in /e/ and 'spotted' in /a/, while in 'fat' the final [e]~[a] is absent underlyingly (/chep/), and is inserted by a later epenthesis rule. If learners are uncertain about the UR and must guess, there are many possible errors: they may assume that the final [e] of chepe is underlying and produce 3pl. *chepe-pi, that the final [e] of washte is epenthetic and produce 3pl. *washta-pi, and so on. Erroneous assumptions about the UR should always lead the learner to produce a "valid" paradigm, however; if the final vowel is underlying, the verb should behave as in (1a), and if not, as in (1b).
What we find in fact is that two additional, innovative paradigms have arisen: the first has [e]~[a] alternations but final reduplication: hanske/hanska-pi/hanska-ska 'tall'. The second has invariant final [a] but penultimate reduplication: thokca/thokca-pi/thok-thokca 'different'. Neither of the new paradigms is compatible with any UR in the old system. In addition, both of these innovations affected words which originally had invariant [a], and never invariant [e] or variant [e]~[a]. These changes appear quite mysterious, but I argue that they make sense under a system in which learners must choose a particular surface form (chepa or chepe) as the UR. In this circumstance, the best the learner can do is to choose the form that distinguishes the greatest number of words, and suffers from the fewest (and least serious) neutralizations. When we consider the neutralizations that affect each part of the Lakhota paradigm, it turns out that the "most informative" form in this language is a second person form; this form maintains almost all distinctions, except the difference between invariant [a] and variant [e]~[a] verbs. I show that when this form is used as the UR for Lakhota, only two errors are predicted for partially known words: the [e]~[a] alternation may incorrectly be extended to invariant [a] verbs, and penultimate reduplicate may also incorrectly apply to invariant [a] verbs. The restricted UR approach correctly predicts the two attested errors, and no others. Furthermore, because [e]~[a] alternations and reduplication patterns are learned as separate rules, with their own lists of exceptions, we can interpret the decoupling of these processes, which were once predictably linked.
For further information, please contact kenstow@MIT.EDU
MIT Linguistics Colloquium
Chris Collins
Cornell University"Remnant Movement in Quotative Inversion" (1) "Where are we going?" asked John.
In Collins (1997), I argued that the subject John stays in-situ in Spec vP, the verb raises to Infl, and the quote (or quotative operator) moves to Spec IP. Given these assumptions, it is possible to account for a number of generalizations, including the transitivity constraint.
In this talk, I will discuss the distribution of particles in QI, and show how the facts are inconsistent with an analysis where the verb moves to Infl. Rather, I show that QI involves movement of a VP or vP to a position higher than the subject.
I will explore the implications of my analysis for theories of remnant movement. Lastly, based on a general similarity between the passive and QI, I will outline a remnant movement analysis of the passive construction in English.
For more information, contact Michael Wagner or Heejeong Ko
MIT LING-LUNCH
Thursday, April 25, 2002 at 12:00, MIT E39-335 (conference room)Hisatsugu Kitahara "A Phase-Based Notion of Chain" Within minimalist perspectives (advanced in Chomsky 2000, 2001a,b), this paper seeks a new notion of chain. Specifically, it develops a phase-based notion of chain by examining the following data:
(1) a. he seems to be expected to be smart
b. he thinks that he is smart
(2) (guess) what John read t
(3) a. *what did you meet John [before you read t]
b. why did you meet John t
(4) a. what did you file t [before you read e]
b. *what did you file t [before you met John [without reading e]]
c. *who t met you [before you recognized e]
d. *who t filed what [before you read e]
For more information, contact anevins@MIT.EDU.
MIT RLE Speech Communication Group Seminar
Rahul Sarpeshkar
MITTitle TBA Not available at this time.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
Boston University Applied Linguistics Spring Speaker Series
Joe Pater
University of Massachusetts, Amherst
"Phonotactic knowledge and morphophonemic acquisition" In Optimality Theory, morphophonemic alternations are driven by the same markedness constraints that shape the phonotactics of the language. This leads to a prediction about the ease of acquisition of different types of alternations: An alternation that serves to meet a phonotactic target should be easier to learn than a similar alternation that does not satisfy a phonotactic requirement. While claims that phonotactic learning can aid morphophonemic acquisition have been made in recent OT learnability work, there appears to be no empirical evidence on this issue. In this talk, I present an experiment that tests this prediction by studying the acquisition of two different alternations by adult subjects in a laboratory setting. Preliminary evidence suggests that the phonotactically driven alternation is in fact easier to learn than a comparable non-phonotactically motivated one.
ASL/English interpreters will be present.
For more information, contact linguist@bu.edu
MIT - Phonology Circle
Friday, April 19, 2002, 3-5 PM, E-39, 335 MIT
Daniel Harbour
MIT"A Metrical Account of Tanoan Tonology" My phonology circle talk (Friday 19th) is an account of Kiowa tone using metrical grids.
Kiowa is a Tanoan language of Oklahoma, with some several hundred older speakers. The data for the talk come from Laurel Watkins' grammar (1984) and my fieldnotes of last summer.
The framework is that of Purnell 1997, who extends metrical grid theory (that we know and love from first year phonology) from stress to tone languages. I will show that this extension to tone grammar is very natural in the case of Kiowa: several tone interactions (lowering / raising in compounds, inflected verbs, possessives, ... ) and interplay between length and tone are elegantly captured in the model.
Some trickier cases remain to be accounted for and audience members are requested to bring thinking caps.
For further information, please contact kenstow@MIT.EDU
GSAS Workshop on Comparative Syntax and Linguistic Theory at Harvard
Len Babby
Princeton University"Dative Subjects and Nominative Objects: Infinitives in Russian:
A Diacrhonic Boost for Synchronic Syntactic Analysis"
For more information, contact Conor McDonough Quinn
MIT LING-LUNCH
Thursday, April 18, 2002 at 12:00, MIT E39-335 (conference room)Francesca Del Gobbo
Harvard/UC Irvine"What Appositives Really Are" Appositive relative clauses have been analyzed as independent sentences, or "assertions-to-follow" (Sells 1985, Demirdache 1991), given that a certain parallelism between appositives and main clauses seems to hold:
(1) a. John, who teaches Italian, came to the party.
(1) b. John came to the party. He teaches Italian.(2) a. *Every/*Any/*No professor, who teaches Italian, came to the party.
(2) b. *Every/*Any/*No professor came to the party. He teaches Italian.But notice that the parallelism breaks down when a wider range of quantifiers are taken into consideration:
(3) a. *Many professors, who had a good time, came to the party.
(3) b. Many professors came to the party. They had a good time.(4) a. *Most professors, who had a good time, came to the party.
(4) b. Most professors came to the party. They had a good time.Moreover, quantifiers such as 'many' and 'most' do not exhibit a uniform behavior:
(5) They invited many professors, who had a good time. (cf. (3a))
(6) They invited most professors, who had a good time. (cf. (4b))
The contrasts above suggest that appositives cannot simply be analyzed as anaphora across discourse, and that such contrasts cannot be explained in terms of type-driven interpretation (Partee 1976, Dayal 1996).
I propose to analyze appositive relative clauses as independent sentences of type t, where the appositive relative pronoun is an E-type pronoun. In the spirit of Heim (1990), I propose that the appositive pronoun is coindexed with the 'head' of the relative clause and:
1. If the 'head' is definite, the pronoun is replaced by a copy of the 'head'/antecedent;
2. If the 'head' is not definite, the pronoun is re-written according to the rules in Heim (1990).
In order to capture the contrast between (3a) and (5), I postulate a Restructuring process that can only apply before Spell-Out and cannot undo the given linear order. Text Formation (Heim 1982) applies once the E-type pronoun is interpreted.
My proposal accounts for the data in (1) through (6) and it makes the strong empirical prediction that prenominal relative clauses cannot be appositives. The prediction is fulfilled in Chinese: I provide evidence from binding, long-distance anaphora, sentential adverbs and presupposition tests to show that Chinese relative clauses are only restrictive. A question arises, namely what about relatives that modify proper names in Chinese? I'll conclude with the two possible directions I think we should take.
Selected references:
Dayal, V. (1996). Locality in Wh-Quantification. Kluwer Academic Publishers.
Demirdache, H. (1991). Resumptive Chains in Restrictive Relatives, Appositives and Dislocation structures. Ph.D. dissertation, MIT, Cambridge.
Heim, I. (1982) The Semantics of Definite and Indefinite Noun Phrases. Ph.D. dissertation, University of Massachusetts, Amherst. Published 1987, Garland Press, New York.
_____ (1990). "E-type pronouns and Donkey Anaphora", Linguistics and Philosophy 13, 137-177.
Partee, B. (1976). Montague Grammar. Academic Press, New York.
Sells, P. (1985). Restrictive and Non-Restrictive Modification. CSLI Report No.CSLI-85-28. Stanford, California.
For more information, contact anevins@MIT.EDU.
MIT RLE Speech Communication Group Seminar
Harold Cheyne
MIT"Estimating Glottal Voicing Source Characteristics by Measuring and Modeling the Acceleration of the Skin on the Neck " In the clinical management of voice patients, quantifying vocal function is becoming increasingly important both for corroborating clinicians' subjective impressions during a voice evaluation and for assessing the effectiveness of surgery or voice therapy. Current devices for quantifying vocal function measure acoustic, aerodynamic and electrical parameters during short tasks such as reading. One technique that has shown potential for measuring vocal function but has been mostly used to quantify speech-related behaviors besides phonation is measuring the acceleration of the skin near the larynx.
The acceleration of the skin on the neck between the cricoid cartilage and the sternal notch arises from the airflow pulses that result from vocal-fold vibration. At least two sets of structures play a role in the transformation of the acoustic source resulting from vocal-fold vibration into the measured acceleration: the subglottal system, and the tissues between the subglottal airspace and the accelerometer (e.g., tracheal cartilage, skin, etc.). Advantages of measuring acceleration over current techniques include 1) the structures that filter the glottal pulses vary less over time than the vocal tract and thus they may be adequately modeled as time-invariant, making signal processing potentially easier; 2) environmental acoustic noise has a minimal influence on the measured acceleration; and 3) the accelerometer's size and placement make it more unobtrusive and comfortable for extended recordings than the current techniques.
This thesis work investigates the potential of using the measured acceleration for quantifying vocal function. Simultaneous acceleration, acoustic and aerodynamic recordings on ten subjects with normal voices are made to examine relationships between the acceleration signal and the acoustic or aerodynamic signal. A vocal system model is also developed to provide insights into these relationships. Specifically, estimates of Sound Pressure Level (SPL) and Maximum Flow Declination Rate (MFDR) are calculated from the acceleration using the vocal system model and show high correlation (r2>0.71 and r2>0.75 respectively) with their respective values as derived from the acoustic and airflow signals for 9 out of 10 subjects. These results demonstrate the potential of the acceleration signal to provide an alternate, non-invasive means of obtaining measures of vocal function.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
MIT Linguistics Colloquium
Luigi Rizzi
Università degli Studi di Siena"Cartography, Locality, and Asymmetries" The first part of the talk deals with selective locality effects with left peripheral movement of adverbial elements. Movement to the left periphery of the clause may target at least three distinct positions: Focus, Topic, and a dedicated adverbial position, Mod(ifier), hosting preposed adverbials which are neither focus nor topic. Each targeted position defines a type of chain which shows selective locality effects: Focus chains are sensitive to the intervention of positions with quantificational properties (Wh island effects, Negative Islands, quantificational adverb intervention, etc.); chains targeting Mod are sensitive to intervention effects triggered by any intervening adverbial position in the IP space; chains targeting Top enjoy a surprising freedom, at least in languages and constructions permitting multiple topics, such as Romance Clitic Left-dislocation: they are apparently not sensitive to the intervention of a lower Top position. We will develop an analysis of these selective effects in terms of a featural approach to Relativized Minimality / Minimal Link.
The second part of the talk tries to capitalize on these results to revisit a classical topic of the theory of locality: argument / non-argument asymmetries. We will argue that extractable Wh elements have topic-like qualities, which make them suitable candidates for movement to a Top(-like) position, thus allowing them to (partially) enjoy the special freedom of topics. Various consequences of this approach will be explored.
References
Belletti, A., ed. (2002) Structures and Beyond. The Cartography of Syntactic Structures, vol. 3, OUP.
Chomsky, N. (1995) The Minimalist Program, MIT Press, Cambridge, Mass.
Chomsky, N. (2000) "Minimalist Inquiries: The Framework", MIT.
Cinque, G. (1999) Adverbs and Functional Heads: A Cross-linguistic Perspective, Oxford University Press.
Cinque, G., ed. (2002) The Structure of DP and IP - The Cartography of Syntactic Structures, vol. 1, OUP.
Rizzi, L. (1990) Relativized Minimality, MIT Press, Cambridge, Mass.
Rizzi, L. (1997) "The Fine Structure of the Left Periphery", in L.
Haegeman (ed.), Elements of Grammar, Kluwer, Dordrecht.
Rizzi, L. (2002) "Locality and Left Periphery", in Belletti, ed. (2002)
Rizzi, L. ed., (2002) The Structure of CP and IP, The Cartography of Syntactic Structures, vol.2, OUP.
Starke, M. (2001) Move Dissolves into Merge, doctoral dissertation, University of Geneva.
For more information, contact Michael Wagner or Heejeong Ko
Harvard Linguistics Department
Santeri Palviainen "The Development of the Nominative Singular
of the Ja-stem Masculine Nouns in Gothic"
For more information, contact Patrick Taylor.
MIT - Phonology Circle
Friday, April 12, 2002 at 3:00 PM, MIT E39-335 (conference room)
Kie Zuraw
USC"Near- and pseudo-reduplication" Pseudo-reduplicated words are sometimes treated by the grammar as though they were morphologically reduplicated, suggesting that a word can be construed as reduplicated even in the absence of appropriate morphosyntax. Moreover, words that are only near-reduplicated (i.e., contain two relatively similar substrings) sporadically undergo enhancement of internal similarity (orangutang < orangutan; smorgasborg < smorgasbord; Okeefenokee < Okefenokee). Such enhancement is rare, however, and probably restricted to low-frequency words, because of the faithfulness violations incurred. I propose that there is a general drive (whether in learning or in production is debatable; both possibilities will be considered), for all words to be construed as reduplicated; once such a construal is in place, base-reduplicant correspondence constraints can enhance a word's internal similarity, if the violations of faithfulness are not too great.
A reduplicative construal can also preserve (rather than outright enhance) similarity, by blocking alternations that would disrupt it. In Tagalog, mid vowels are normally allowed only in final syllables. Final-syllable mid vowels, therefore, normally raise under suffixation (abiso 'information', abisu-han 'notify'). Many loanwords, however, have mid-vowel penults, and many do not undergo raising. Statistical analysis of loans with mid-vowel ultimas shows that (i) nonraising is much more common among stems with a mid-vowel penult, and (ii) given a mid-vowel penult, the probability of nonraising increases significantly if penult and ultima have similar onset shape, onset place, or vowel backness. This suggests that loanwords with sufficient internal similarity are treated as reduplicated, blocking raising, because it would disrupt vowel similarity. Reduplicative construals are thus more freely applied when they do not disturb input-output faithfulness.
For further information, please contact kenstow@MIT.EDU
MIT LING-LUNCH
Thursday, April 11, 2002 at 12:00, MIT E39-335 (conference room)Liina Pylkkanen
MIT"Applicatives and Depictive Secondary Predication" English indirect objects are mysterious in that they constitute the only case in which a depictive secondary predicate cannot be predicated of a bare DP that is inside a VP (Williams, 1980; Baker, 1997).
(1) a. I gave Mary the meat[1] raw[1].
b. *I gave Mary[1] the meat hungry[1]. (= Baker, 1997: ex 23c,d)
In this talk I
(i) argue for a complex predicate analysis of depictives and
(ii) show that the impossibility to have depictives modify English
indirect objects is predicted by the Pylkkanen (2000) analysis of
applicative/double object constructions.Pylkkanen (2000) argues that applicative constructions crosslinguistically divide into those where the applicative head denotes a relation between an event and an individual (= high) and to those where the applicative head denotes a relation between two individuals, the direct and the indirect object (= low). Once we have a concrete proposal about the syntax and semantics of depictive phrases, low applied arguments, which English indirect objects exemplify, are predicted to be unavailable for depictive modification. Interestingly, the opposite prediction is made for high applicatives: they *should* be available for depictive modification. In other words, the following generalization should hold:
(2) If a language has an English type depictive secondary predicate, the depictive should be able modify an applied argument only if the applied argument is high.
This prediction has been tested for six languages (English (low), Japanese(low), Korean(low), Luganda(high), Venda(high) and Albanian(high)) and so far all data collected are consistent with it.
I conclude with two puzzles for the generalization that low applied arguments are unavailable for depictive modification. One of them is easy and has to do with light verb constructions. The other one is hard and has to do with movement.
Refs:
Baker, Mark, 1997. Thematic roles and syntactic structure. In: Liliane Haegeman (ed.), Elements of grammar: Handbook in generative syntax, 73-137. Dordrecht: Kluwer.
Pylkkanen, L. 2000. What Applicative Heads Apply To. In Minnick, M., A. Williams and E. Kaiser (eds.), Proceedings of the 24th Annual Penn Linguistics Colloquium, UPenn Working Papers in Linguistics, Volume 7.1. Available at http://web.mit.edu/liina/www/ Williams, Edwin, 1980. Predication. Linguistic Inquiry 11,203-238.
For more information, contact anevins@MIT.EDU.
MIT RLE Speech Communication Group Seminar
Winifred Strange
CUNY"Cross-Language Differences in Vowel Acoustics and Perception" Current theories of Second Language (L2) Speech Learning invoke the concept of "similarity" of first language (L1) and L2 phonetic segments in predicting which L2 segments/contrasts will be difficult to perceive and pronounce by adult L2 learners. However, L1/L2 contrastive analysis at the level of (context-free) phonological segments or features has often failed to predict L2 perception/production patterns. Our research seeks to describe cross-language phonetic similarities of vowel systems, using detailed acoustic analysis of vowels produced in sentence level materials. The effects of consonantal context and, more recently, sentence prosody on the phonetic realization of vowels is explored. Cross-language perceptual similarity patterns are assessed directly using the Perceptual Assimilation paradigm. Results to date have shown language-specific variability in vowels that result in rather striking differences in L1/L2 acoustic similarity across different phonetic and prosodic contexts. In addition, tests of perceptual assimilation of L2 vowels to L1 categories also show systematic effects of contextual variables, and perceptual similarity is not always well predicted by acoustic similarity. These results have implications for our theories of phonological representation as well as practical application in the assessment and instruction of L2 speech learning by adults.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
GSAS Workshop on Comparative Syntax and Linguistic Theory at Harvard
Audrey Li
University of Southern California"Variations in Relativization"
For more information, contact Conor McDonough Quinn
Eunice Kennedy Shriver Center - Psychology Colloquium Series
Refreshments served 10 minutes prior to talk
Laura Petitto "How Yong Babies Begin Language: Evidence for Cortical Plasticity and Cortical Specificity in Early Brain Development"
For more information, see http://www.umassmed.edu/shriver/Research/Psychological/Colloquium/
MIT NON LING-LUNCH
Thursday, April 4, 2002 at 12:00, MIT, E25-401David Caplan
A talk about the neural correlates of syntax
For more information, contact MIT Department of Brain and Cognitive Sciences
Boston University Linguistics Lecture
sponsored by the James Geddes, Jr., Lecture Series
Thursday, April 4, 2002 at 7:30 PM - College of Arts and Sciences (725 Commonwealth Ave.) room B12.
Note: This talk is intended for a general audience; no previous knowledge about linguistics is presupposed.
Cheryl Zoll
Associate Professor of Linguistics at MIT and Radcliffe Institute Fellow
"Melodies and Mapping:
Understanding African Tone Languages"One of the hallmarks of language is the creativity of its speakers. When we speak, we produce original sentences that may never have been uttered before. This creativity extends to a language's sound patterns. A speaker of English, for example, knows not only that questions are pronounced with a rising intonation, but also exactly where in the sentence the rise should begin and where the melodic peak should occur. This knowledge cannot be the result of memorization of every possible sentence with properly placed intonation, since most sentences we express are original creations. Rather, it reflects complex and abstract linguistic knowledge.
The goal of linguistics is to discover what this abstract knowledge is. The methodology of linguistic discovery involves detailed examination of linguistic phenomena across a variety of languages. In her lecture, Zoll will focus on the use of pitch in a number of sub-Saharan African languages as a case in point. She will present a novel framework for understanding the sound systems of these languages and demonstrate the universality of the principles governing their tonal patterns.
Refreshments will be served. ASL/English interpretation will be provided.
For more information, contact carol@bu.edu
MIT RLE Speech Communication Group Seminar
David Ostry
McGill"The Jaw's Mechanical Behavior and Its Relation to Speech Production" I will describe two sets of studies that examine the relationship between the jaw's mechanical behavior and its role in speech production. Both involve the use of a robotic device to deliver mechanical perturbations to the jaw. In the first set of studies, we have assessed the jaw's impedance and its relation to kinematic variability in speech. We will show that the pattern of kinematic variability in speech production can be accounted for by directional asymmetries in the stiffness of the jaw. In the second set of studies, we explore the basis of adaptation to a motion dependent force field applied to the jaw during speech production. We have designed small amplitude mechanical perturbations for these studies that enable us to alter the motion path of the jaw in proportion to its velocity and hence manipulate proprioceptive input without producing any associated changes to speech acoustics. We show that under these conditions changes to proprioceptive input due to the perturbation result in adjustments by the speaker that restore normal jaw movement even though the acoustics are unaltered. Comparable patterns of adaptation are observed when perturbations are delivered during silent speech, when there is no acoustic signal at all. This indicates that learning about dynamics in speech production is based on proprioceptive information and suggests that proprioceptive input is a primary factor in the achievement of speech goals.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
MIT - Phonology Circle
Friday, March 22, 2002, at 3:00 PM in Room E39-335
Victor Manfredi
Boston University"A fonosyntactic parameter in Benue-Kwa and its consequences for Edo" There is a large literature on "incomplete tonification" (Hyman 1982) or "predictable tone" (Odden 1988) in Bantu languages, which are otherwise considered to be "tonal" at least to some degree. Typically, in verbs, the location of H need not be learned, because it can be computed from a combination of syllable structure and inflectional category. In fact, similar phenomena are pervasive throughout Benue-Kwa, and motivate a reduction of tone to metrical structure, based on the parameter expressed in (1) or (2).
(1) Tone is (just) the option of lexically prelinked-i.e. phonologically unpredictable-metrical structure. In acquisition, this option is blocked wherever metrical structure can be assigned on a predictable basis, i.e. on the basis of distinctive contrasts in syllable weight and/or by position within complex word domains.
(2) Prosodic linking parameter. "The units which project metrical structure attach are: {tones}, {rimes}" (Manfredi 1993, 177).
In this talk I will analyze the pitch realization of genitive phrases in Edo, following earlier proposals about Igbo and Yoruba. The only extant, tonal analysis of these data, by Amayo (1983), requires "global" or diacritic rule application, and a tone-free analysis arguably fares better.
References
Amayo, A. [1983]. Tone rules and derivational history in Edo phonology.
Current Approaches to African Linguistics 1, edited by I.R. Dijoff, 185-95. Foris, Dordrecht.
Hyman, L. [1982]. Globality and the accentual analysis of Luganda tone. Journal of Linguistic Research 2, 1-40.
Manfredi, V. [1993]. Spreading and downstep: prosodic government in tone languages. The Phonology of Tone; the Representation of Tonal Register, edited by H. v.d. Hulst & K. Snider, 133-84. De Gruyter, Berlin.
Odden, D. [1988]. Predictable tone systems in Bantu. Autosegmental Studies on Pitch Accent Systems, edited by H. van der Hulst & N. Smith, 225-51. Foris, Dordrecht..
For further information, please contact kenstow@MIT.EDU
GSAS Workshop on Comparative Syntax and Linguistic Theory at Harvard / Ling 204r: special guest lecture!
The Department of Linguistics at Harvard is pleased to announce a special guest lecture as part of Prof. C.T. James Huang's Topics in Syntax (Ling 204r) seminar. Interested persons are invited to sit in.
A reception will follow immediately thereafter in the Department of Linguistics lounge.
Prof. Mamoru Saito
Hanzan University, Japan"Movement and Theta-Roles: Case studies
with resultatives and complex predicates"
For more information, contact Conor McDonough Quinn
MIT LING-LUNCH
Thursday, March 21, 2002 at 12:00, MIT E39-335 (conference room)Ken Hiraiwa "Indeterminate Agreement and Varieties of Raising" There has been much controversy as to whether the so-called Raising-to-Object (RTO) construction in Japanese involves movement or control. In fact, the fact that Japanese freely allows a null pronoun leads a number of linguists to claim that RTO in Japanese is a control/major object construction (cf. Saito 1985, Takezawa 1987, Oka1988, Hoji 1991, Takano 2001)
In this talk, I will demonstrate, building on what I call Indeterminate-Agreement (cf. Kuroda 1965, Nishigauchi 1990, Kishimoto 2001), that RTO in fact involves 'raising' (cf. Kuno 1976, Ura 1994) not control. Furthermore, I will show that the derivation of RTO in Japanese is Agree (v, DP) + optional Move (v, DP), arguing against Kuno's (1976) obligatory raising analysis.
The proposed theory brings a number of consequences: Agree suffices for Case (Chomsky 2001, Hiraiwa 2001; contra Kishimoto 2001); the ban on 'super' long-distance Agree; the 'Phi-over-Phi' generalization that Agree (P, G1) forces Move (P, G1) when G1 is in [Spec, G2], where G1 and G2 has matching phi-features for a probe P.
A more detailed abstract can be found at: http://ling.ucsc.edu/~wccfl-21/abstracts/hiraiwa.pdf.
For more information, contact anevins@MIT.EDU.
Boston University Human Development Colloquium
Harriet Tenenbaum
Harvard University"Parent-child science talk: Socialization of gendere inequities?"
MIT RLE Speech Communication Group Seminar
David Gow
MIT"Processing Assimilated Speech: What Listeners Need to Know" Given the observation that listeners regularly discriminate between words based on single feature differences (e.g. right, ripe) it is somewhat surprising that productive phonological processes that appear to change surface features do not disrupt spoken word recognition. For example, English coronal place assimilation may make the phrase right berries sound very much ripe berries, and yet listeners do not confuse the two (Gow, 2002). Work on this problem over the last few years has demonstrated that listeners rely on the context that triggers modification to recognize modified items. Indeed, both regressive and progressive context effects have been found in the processing of assimilated speech. In this talk I examine several types of knowledge that might be used to produce these context effects and resolve the difference between right berries and ripe berries. These include knowledge that would allow listeners to make inferences at the levels of phonological or phonetic representation, or articulatory or perceptual organization. I will present new evidence from phoneme monitoring studies of the processing of Hungarian voicing assimilation and Korean labial to velar place assimilation by native and non-native speakers suggesting that these context effects rely more on the acoustic realization of feature cues than on phonological or phonetic regularity. I will also discuss a series of acoustic analyses, online behavioral studies, and a pilot magnetoencephalography (MEG) study that suggest that listeners resolve assimilated speech through a process of feature cue parsing, in which feature cues recovered from the speech stream are associated with other cues or abstract segmental representations based on perceptual grouping principles.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
MIT: SPECIAL EECS SEMINAR
Friday, March 15, 2002, MIT AI Lab Playroom, 8th floor, NE43Refreshments at 2:45 PM, Talk at 3:00 PM
Sandiway Fong
Principal Research Scientist
NEC Research, Princeton, NJ"WordNet: Three Views from Semantic Inference WordNet is perhaps the most widely used, freely available lexical resource today in natural language processing. Designed originally as a model of human semantic memory, it consists of a large network of synonym sets for nouns, adjectives, verbs and adverbs connected by semantic relations such as hypernymy (isa/aka) and antonymy. In this talk, I will investigate the performance of WordNet and suggest directions for improvement for three problems in semantic inference: (1) the computation of Semantic Opposition (Pustejovsky,2000), a particular case of the Frame Problem, (2) Semantic Bleaching (Kiparsky,1997), for predicting the dilution of meaning for related noun/verb pairs, and (3) telic, i.e., purpose/function, role recovery for cases of Logical Metonymy (Pustejovsky,1995).
Host: Professor Gerry Sussman
For more information, contact anevins@MIT.EDU.
GSAS Workshop on Comparative Syntax and Linguistic Theory at Harvard
A reception will follow immediately thereafter in the Department of Linguistics (3rd floor, Boylston Hall).
Ellen Woolford
University of Massachusetts"Nominative Objects and Case Locality" There is an extensive literature on the question of how objects get nominative Case in dative and ergative subject constructions. Many of the competing hypotheses can be eliminated based on an examination of certain constructions in Icelandic, Faroese, and Nez Perce where nominative licensing to objects is blocked. The locality restrictions manifested in these constructions show that a closer DP can block nominative licensing to a further DP, even when that closer DP is not itself nominative. These and other Relativized Minimality effects on Case licensing are consistent with the view that heads license Case under c-command (m-command) (e.g. Chomsky 2000). In contrast, these locality restrictions are puzzling under other several approaches to nominative objects, including (covert) NP Raising, allowing a lower head to license nominative on objects, or treating nominative as simply the absence of Case.
This paper develops and refines ideas on Case Locality from Chomsky 2000 and Rizzi 1990. It will be argued that Case is subject to two sorts of Relativized Minimality effects, one involving blocking by a closer potential source (head) and the other involving blocking by a closer potential target (DP). Closer heads with different features (e.g. +V/-V) always block Case licensing (P always blocks V), but there is evidence that closer heads of the same extended projection never block Case licensing (V never blocks I). As for the effect of a closer DP, in some languages Case licensing of a further DP can be blocked by any closer DP, while in other languages, only a DP with a similar or identical Case counts as a closer potential target.
Chomsky, N. 2000. Minimalist Inquiries: The Framework. In R. Martin, D.
Michaels, and J. Uriagereka eds. Step By Step. Cambridge, MA: MIT Press.
Rizzi, L. 1990. Relativized Minimality. Cambridge, MA: MIT Press.
For more information, contact Conor McDonough Quinn
MIT - Phonology Circle
Friday, March 15, 2002, 3:00-5:00, Room E39-335
John Frampton
Northeastern University"The Nature of Reduplicative Copying" I will talk about various reduplication phenomena in Chaha. There are a few reasons to talk about Chaha reduplication. First, Kenstowicz and Banksira (LI 30.5, 1999) have claimed that certain aspects of Chaha reduplication are "impossible for the standard theory to express in terms of rules" and I have been engaged in a program of systematically refuting such claims. Second, and more importantly, Chaha reduplication has a lot to tell us about the structure of the phonological representations involved in reduplication and the nature of reduplicative copying.
In his well-known paper "OCP Effects: Gemination and Antigemination" (1986), McCarthy proposed that biconsonantal roots filled tri- and quadri-consonantal templates by spreading the final consonant over multiple "C slots," creating what came to be called "long distance geminates." In the recent OT literature, this analysis has been rejected in favor of an analysis in terms of reduplication and Correspondence Theory. I will agree that the phenomenon is a reduplication phenomenon, but that McCarthy was correct in proposing that long-distance geminates play a central role. The mechanism which McCarthy proposed for their formation is not correct. The long-distance geminates are produced by reduplicative copying, not spreading. This, in fact, is what reduplicative copying always produces. Reduplicative copying, which I call "transcription," copies only timing slots and their associations, not melodic elements.
The distinction between copying and spreading is important>because the evidence for No Crossing Constraint (NCC) intervention effects is limited to evidence that the NCC blocks spreading in certain configurations. I will argue that the NCC>is not a constraint on representations, but a constraint on>spreading. Transcription can and does produce representations with multiple NCC violations. The long-distance geminates produced by transcription are the source of many of the most puzzling reduplication overapplication effects. This includes the overapplication effect in Chaha which Kenstowicz and Banksira identify as fatal to derivational phonology as well as the famous Malay nasalization example which Prince and Smolensky similarly identified as impossible to analyze in derivational terms.
For further information, please contact kenstow@MIT.EDU
MIT LING-LUNCH
Thursday, March 14, 2002 at 12:00, MIT E39-335 (conference room)Hisa Kitahara
Keio U., Japan"A hard question confronting the minimalist program" In this paper, after the brief review of the landscape of generative grammar, I highlight a hard question confronting the minimalist program, the latest version of the generative approach (Chomsky 1995, 2000, 2001). I then discuss some recent work on acoustic perception in nonhuman animals (Hauser 2000, 2001), which provides a different perspective on this question.
Since the advent of generative grammar, it has been assumed that there is a faculty of language (FL), which undergoes state changes from an initial state to a stable state, where a state of FL is understood to be a language. The language L includes a cognitive system that stores information (about sound, meaning, and structural organization), and that information is accessed by sensorimotor systems and systems of thought, and they do so by means of two distinct levels of representations, Phonetic Form (PF) and Logical Form (LF), respectively. L is then understood to be a recursive procedure that generates linguistic expressions EXP = <PF, LF>.
For EXP = <PF, LF> to yield right sound-meaning relations, EXP=<PF,LF> must satisfy the legibility conditions (holding of PF and LF). Assuming that elements interpretable at PF are not interpretable at LF, L must split a generative procedure into two parts: one forming PF and the other forming LF. Thus, at some point, the phonological component (forming PF) is separated from the syntactic component (forming LF). To the extent the legibility conditions are clarified, we can further ask how well L is designed to satisfy them. Formulating and studying questions of this kind, the minimalist program, the latest version of the generative approach, seeks to discover the nature of the genuine theory of L.
The substantive thesis of the minimalist program is that language design may really be optimal in some respects, approaching an optimal solution to the legibility conditions. However, there is a serious problem. The phonological component (forming PF), unlike the syntactic component (forming LF), appears to be not optimal in many ways. As Chomsky himself notes, to what extent the phonological component is an optimal solution to the task of forming legible PF is a hard question, not yet seriously addressed.
The failure of the phonological component to meet the minimalist criteria calls for an explanation. In this context, one might find intriguing Hauser's recent work on acoustic perception in nonhuman animals. Based on his new findings, Hauser concludes that most of the phonological resources that we bring to the task of speech perception are derived from a nonhuman animal ancestor, but nonhuman animals are significantly limited with respect to certain key syntactic resources. One possible interpretation of this conclusion is that the phonological component, unlike the syntactic component, did not evolve uniquely for the human faculty of language. Under this interpretation, the contrast between these two components of FL (with respect to the minimalist criteria) can be attributed to their different evolutionary paths.
For more information, contact anevins@MIT.EDU.
MIT RLE Speech Communication Group Seminar
Seppo Ahlfors
NMR-MGH"Magnetoencephalography: Detecting Spatiotemporal
Patterns of Human Brain Activity"In magnetoencephalography (MEG), extra-cranial magnetic fields reflecting electrophysiological activity in the brain are recorded. The high temporal resolution of MEG and EEG (scalp potentials) makes it possible to determine patterns of activation, including latencies, temporal sequences, and oscillations, in networks of cortical areas related to sensory and cognitive processing. The interpretation of MEG data is limited by the difficulty of determining the locations of activity generating the measured magnetic fields. Complementary information provided by hemodynamic measures (functional MRI, PET) of the brain activity can be of help in resolving the spatial distribution of activity underlying MEG and EEG signals. Results of a study of visual motion related activity demonstrate the usefulness of fMRI information in suggesting likely source locations for the MEG data. With the combined MEG-fMRI approach it is possible to identify activity in a network of areas and to measure the relative timing of each.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
Boston University Applied Linguistics Spring Speaker Series
Malcolm Coulthard
University of Birmingham, England
"The Linguist as a Detective" In this talk I will focus on Qs of disputed authorship. Using real examples from Court cases I will present some of the techniques linguists use to reach opinions on how and by whom a particular text was produced My data will include two examples of suspected police forgeries, a letter from a case where a wife may have committed suicide or been murdered by her husband and one where student plagiarism was suspected.
ASL/English interpreters will be present.
For more information, contact linguist@bu.edu
MIT Linguistics Colloquium
Elan Dresher
University of Toronto"The Contrastive Hierarchy in Phonology" The role of contrast in phonology is the subject of renewed interest, though approaches vary widely (Flemming 1995, Avery 1996). How contrast should be incorporated into linguistic theory remains unclear, as does its relationship to markedness (Chomsky & Halle 1968, Kean 1980, Calabrese 1995) and underspecification (Steriade 1987, Archangeli 1988, Itô, Mester & Padgett 1995). In this talk I will argue that contrast in linguistics must be understood in terms of a CONTRASTIVE HIERARCHY. The contrastive hierarchy is a ghostly presence in the history of phonological theory: seldom acknowledged explicitly, and misunderstood on the rare occasions when it was, its influence can nevertheless be discerned in the central works of phono-logical theory from the Prague School to early generative grammar. I will trace some of this history, and show how the contrastive hierarchy is relevant to current approaches to markedness and underspecification, in derivational phonology as well as in Optimality Theory (OT).
In any inventory characterized in terms of distinctive features, it is necessary to specify the relative SCOPE, or ORDERING, of each feature. For example, in his analysis of the vowel system of Polabian, Trubetzkoy (1969: 102-3) observed that the feature [back] has wider scope than, or is ordered ahead of, the feature [round]. Thus, contrasts based on the feature [round] are relevant only to [-back] vowels; the [+back] vowels /u/ and /o/ are phonetically round, but not contrastively so. Many of Trubetzkoy's examples presuppose partial hierarchies of this kind, though he did not consistently adhere to the principle of a contrastive hierarchy.
Jakobson & Halle (1956) refer to the hierarchy as the 'dichotomous scale', and adduce 'several weighty arguments' in support of it. Their argument that the scale governs the sequence of acquisition of contrasts took hold in the field of phonological acquisition (Dinnsen 1992, Rice & Avery 1995). The contrastive hierarchy appears again in Halle 1959. By this time, however, the hierarchy seems to have become associated with certain notions of information theory (Cherry, Halle & Jakobson 1953), and its potential to account for empirical facts of phonological patterning was not exploited. Stanley (1967) argued against the 'branching diagrams' of Halle 1959 as part of his general argument against underspecification in phonology, and the contrastive hierarchy disappeared from phonological theory.
Theories of markedness and underspecification arose to fill gaps left by the demise of the contrastive hierarchy. However, neither theory adequately replaces it, for neither gives a central role to language-specific contrasts. I will show that the hierarchy is required by any theory that refers to the distinction between contrastive and redundant features. Indeed, much of the controversy about whether redundant features play a role in the phonology has suffered from a lack of clarity as to what features are actually redundant in any given situation (Archangeli 1988, Steriade 1987, 1995, Kiparsky 1985).
Finally, I will show that OT does not do away with the need for a contrastive hierarchy. Though the contrastive hierarchy can be encoded by means of interleaved markedness and faithfulness constraints (Kirchner 1997, Bakovic 2000), only a small subset of possible constraints and rankings correspond to well-formed contrastive hierarchies, and thus, arguably, to possible phonologies.
References
Archangeli, Diana. 1988. Aspects of underspecification theory. Phonology 5. 183-207.
Avery, Peter. 1996. The representation of voicing contrasts. Doctoral dissertation, University of Toronto.
Bakovic, Eric. 2000. Harmony, dominance and control. Doctoral dissertation, Rutgers University.
Calabrese, Andrea 1995. A constraint-based theory of phonological markedness and simplification procedures. Linguistic Inquiry 26. 373-463.
Cherry, E. Colin, Morris Halle, and Roman Jakobson. 1953. Toward the Logical Description of Languages in their Phonemic Aspect. Language 29. 34-46.
Chomsky, Noam and Morris Halle. 1968. The sound pattern of English [SPE]. New York: Harper & Row.
Dinnsen, Daniel. A. (1992). Variation in Developing and Fully Developed Phonetic Inventories. In Phonological development: Models, research, implications, ed. Charles A. Ferguson, Lise Menn & Carol Stoel-Gammon, 191-210. Timonium, MD: York Press.
Flemming, Edward. 1995. Auditory representations in phonology. Doctoral dissertation, UCLA.
Halle, Morris. 1959. The sound pattern of Russian: A linguistic and acoustical investigation. The Hague: Mouton. Second printing, 1971.
Itô, Junko, Armin Mester, and Jaye Padgett. 1995. Licensing and underspecification in Optimality Theory. Linguistic Inquiry 26. 571-613.
Jakobson, Roman, and Morris Halle. 1956. Fundamentals of language. The Hague: Mouton.
Kean, Mary-Louise. 1980. The theory of markedness in generative grammar. Ph.D. dissertation, MIT. Reproduced by the Indiana University Linguistics Club, Bloomington, Indiana.
Kiparsky, Paul. 1985. Some consequences of Lexical Phonology. Phonology Yearbook 2. 85-138.
Kirchner, Robert. 1997. Contrastiveness and faithfulness. Phonology 14. 83-111.
Rice, Keren, and Peter Avery. 1995. Variability in a deterministic model of language acquisition: A theory of segmental elaboration. In Phonological acquisition and phonological theory, ed. John Archibald, 23-42. Hillsdale, NJ: Lawrence Erlbaum.
Stanley, Richard. 1967. Redundancy rules in phonology. Language 43. 393-436.
Steriade, Donca. 1987. Redundant values. CLS 23.2. 339-362.
Steriade, Donca. 1995. Underspecification and markedness. In Handbook of phonology, ed. John A. Goldsmith, 114-174. Oxford: Blackwell.
Trubetzkoy, N. 1969. Principles of phonology. Translated by C. Baltaxe. Berkeley: University of California Press. Translation of Grundzüge der Phonologie. Göttingen: Vandenhoek & Ruprecht, 1939.
Department of Linguistics
University of Toronto
Toronto, Ontario
Canada M5S 3H1
For more information, contact Michael Wagner or Heejeong Ko
Eunice Kennedy Shriver Center - Psychology Colloquium Series
Refreshments served 10 minutes prior to talk
Laura Pettito
Professor, Department of Psychological and Brain Sciences
Cognitive Neuroscience Center, Dartmouth College"How Young Babies Begin Language: Evidence for Cortical Plasticity and Cortical Specificity in Early Brain Development"
For more information, see http://www.umassmed.edu/shriver/Research/Psychological/Colloquium/
MIT LING-LUNCH
Thursday, March 7, 2002 at 12:00, MIT E39-335 (conference room)Ora Matushansky
CNRS/MIT"What is it a beauty of?"
A beauty of a constructionThe problem: The syntax of 'N of an N' constructions in (1a) is subject to some controversy.
(1) a.[a beauty]i of a t i problem
(1) b.[how beautiful]i (of) a t i problemBennis et al. 1998, den Dikken 1995, 1998, and Español-Echevarría 1998 have argued that (1a) results from predicate inversion inside the DP. We will also adopt a movement approach but argue that 'N of an N' is a modificational structure combined with DP-internal degree movement on a par with the adjectival construction in (1b). This approach explains the semantic properties of the construction and provides a trigger for the movement, as well as resolves the question of what the head of the construction is.
Scalarity: The nouns that appear before of in cases like (1a) are scalar, i.e. their lexical entry contains a degree variable slot (Matushansky 2001). Non-scalar predicates such as French or doctor are subject to scalarity coercion in degree-sensitive environments. The meaning of the adjective changes to 'having the properties typically associated with being French', and this meaning shift is paralleled in the nominal domain when coercion is allowed. Contrasts between scalar and non-scalar predicates, nouns in (2) and adjectives in (3), show that scalarity is indeed the relevant dimension.
(2) a.Miranda seems an idiot/*a doctor.
(2) b.Miranda is even more of an idiot/?of a doctor than I thought.
(2) c.Miranda is such an idiot/a doctor.= 'idiot to a high degree'/'typical doctor'(3) a. Belinda seems intelligent/French.
(3) b.Belinda is even more intelligent/?French than I thought.
(3) c. Belinda is so intelligent/French.= 'intelligent to a high degree'/'so typically French'
Movement: Examples like (4) (though not accepted to the same degree by all native speakers) show that the surface position of the extended AP in (1b) is due to movement.
(4) a. Miranda is a more capable doctor than anyone I know.
(4) b. Miranda is more capable a doctor than anyone I know.
(4) c.More capable a doctor Miranda has never met.This movement is restricted to APs containing a degree operator, and licenses non-scalar DPs in the complement of seem even when covert, as in (5).
(5) a. * Belinda seems a trained doctor.
(5) b.Belinda seems a more trained doctor than anyone I know.In other words, DPs containing a degree-modified AP, as in (5b), behave like the scalar DP in (2a). Matushansky 2001 argues that seem requires QR of degree in its complement, in the same way verbs such as ask force wh-movement in their complement.
Having established the existence of DP-internal degree movement, we see that the analysis extends to scalar nouns. Indeed, not only does the fronted noun in (1a) license a non-scalar head noun, as in (6a), but its structure parallels that of degree modification in (6b).
(6) a. This seems ??(a beauty of) a problem.
(6) b. This seems too much of a problem.Moreover, as shown in (1b), in certain dialects DP-internal degree movement of the extended AP triggers of-insertion, which suggests that movement is to the same position.
Predication vs. modification: Bennis et al. 1998, den Dikken 1995, 1998, and den Dikken' Lipták 1997 propose that the fronted element in 'N/A of an N' constructions starts out as the predicate. We will argue that the similarity to copular inversion that they note is due to independent factors. The fact that the DP in (1a) denotes a kind of a problem and the recursivity of 'N of an N' constructions, as in (7a), which are puzzles for the predicative analysis, ensue naturally from the attributive analysis.
(7) a.that asshole of an idiot of a doctor
(7) b. This doctor is an idiot an asshole.Other results: The prohibition against any determiners except a, as in (8a), also follows from parallelism with APs: since DP-internal degree movement in (8b) can be argued to be the first step of QR to the clause level, degree extraction out of a DP will naturally be blocked by the presence of a higher quantifier.
(8) a.* a beauty of every/the problem
(8) b. * how beautiful (of) every/the problemThe prohibition against mass nouns and plurals in 'N of an N' construction in (9a) is paralleled in the adjectival domain in (9b).
(9) a. these idiots of *policemen/*populace/a police force.
(9) b.how clean *glasses/*water/a glassThe fact that adjectives appear in the fronted position in the Romance languages and in Hungarian is also expected under the modification approach.
Definite 'N of an N' constructions: Examples such as (10) can be shown to be cases of non-restrictive modification where the entire DP functions as an epithet.
(10) I went to see my dentist, but the idiot (of a doctor) didn't know what was wrong.This claim is supported by the fact that the nouns forming epithets are necessarily scalar, as (11) shows (Milner 1978, but see also Ruwet 1982).
(11) I went to see my dentist, but the idiot/the butcher/*the doctor didn't know what was wrong.If the entire 'N of an N' construction seems to have the same distribution as the fronted NP, neither the predicative nor the modificational analyses can be right. But the purported derivational connection is an illusion, since French in (12) and some dialects of Spanish (Español-Echevarría 1997) disallow the definite article in 'N of an N' constructions, while allowing definite epithets and indefinite 'N of an N'.
(12) * l'diot/imbécile de médecinMoreover, in (13a) and (13b), the possessor cannot be part of the fronted NP because my/mon idiot are not legitimate epithets while the second noun is a relational noun, which requires a possessor.
(13) a. mon/*l'diot de cousin/médecin
(13) b. my idiot of a brother-in-law/doctor
(13) c.the [big bear/mountain] of a manAlso, in cases such as (13c), the fronted NP big bear/mountain is not a legitimate epithet.
Conclusion: Similarities between epithets and definite 'N of an N' constructions are due to common pragmatics (which necessarily involves scalarity) rather than common syntax. The proposed parallels to adjectival modification extend to Hungarian, Romance and Germanic languages other than English.
References
Bennis, Hans, Norbert Corver & Marcel den Dikken (1998): Predication in nominal phrases. The Journal of Comparative Germanic Linguistics 1, pp. 85-117.
den Dikken, Marcel (1995): Copulas. Presented at GLOW 18, Tromsø, Norway.
den Dikken, Marcel (1998): Predicate Inversion in DP. In: Alexiadou, Artemis, & Chris Wilder (eds.): Possessors, Predicates and Movement in the Determiner Phrase. John Benjamins Publishing Co., Amsterdam/Philadelphia.
den Dikken, Marcel, & Anikó Lipták (1997): Csoda egy nyelv Nominal-internal predication in Hungarian. In: Coerts, J., & H. de Hoop (eds.): Linguistics in the Netherlands 1997, pp. 177-214. John Benjamins Publishing Co., Amsterdam/Philadelphia.
Español-Echevarría, Manuel (1998): Two Aspects of the Sentential Syntax of N/A of a N DP‚s: Predicate Raising and Subject Licensing. In: A. Schwegler, B. Tranel and M. Uribe-Etxeberría (eds.): Romance Linguistics: Theoretical Perspectives, p. 67-80. John Benjamins Publishing Co., Amsterdam/Philadelphia.
Matushansky, Ora (2001): Obligatory Scalarity (a sliding scale). K. Megerdoomian and L. A. Bar-el (eds.): WCCFL 20 Proceedings, pp. 400-413. Somerville, MA: Cascadilla Press.
Milner, Jean-Claude (1978): De la syntaxe à l‚interprétation. Editions du Seuil, Paris.
Ruwet, Nicolas (1982): La Grammaire des insultes et autres études. Editions du Seuil, Paris..
For more information, contact anevins@MIT.EDU.
MIT - Syntax/Semantics Reading Group
Friday, March 7, 2002, 5:15, Room E39-335
Shin Ishihara
MIT"The Syntax-Phonology Interface of
wh-questions in Japanese"We will explore the interaction of the prosodic and the syntactic properties of wh-questions in Japanese. First, it will be shown that Japanese wh-questions have a particular intonation pattern: wh- phrases receive phonological prominence, and the pitch contour of the following material is all reduced until the question particle _ka_ that agrees with the wh-phrase. This prosodic property is also shared by another construction involving wh-phrase, namely, so- called "indeterminate construction" (cf. Shimoyama 2000), where a wh-phrase and a particle _mo_ together constitute a universal quantifer or a NPI even when they are syntactically remote.
On the basis of the observation above, we will next revisit several syntactic constructions involving wh-phrases that have been discussed in the literature, e.g., Takahashi's (1993) data for overt wh-movement in Japanese, Radical Reconstruction (Saito 1989), Anti- superiority (Watanabe 1992), etc.. It will be shown that the intonation pattern discussed above is responsible for these apparently syntactic effects.
For further information, please contact rgajews@MIT.EDU
MIT RLE Speech Communication Group Seminar
Liina Pylkkanen and Alec Marantz
MIT"The Neuromagnetism of Lexical Access:
MEG Indices of Activation and Competition"Virtually all linguistic computation involves retrieval of lexical items from their mental store. Therefore, understanding the brain activity associated with basic lexical access is crucial for unpacking the neurobiology of language. In this talk, we describe a series of experiments which use magnetoencephalography (MEG) in order to identify a neural response component reflecting the activation of lexical representations in the mental lexicon. Our results show that the first MEG component showing sensitivity to lexical stimulus factors peaks at approximately 350ms after the presentation of a visual word and is generated in the left superior temporal cortex. Importantly, there is evidence that this component, the M350, does not reflect a task-related process. Stimuli which are high in phonological neighborhood density, and thus invoke intense competition among activated lexical entries, elicit delayed lexical decision times but shortened M350 latencies. Thus, the data suggest that the M350 indexes initial activation of the lexicon, prior to competition. An index of automatic lexical activation constitutes a valuable tool for the study of language processing and structure. In the second half of this talk we describe how we have used the M350 to investigate morphological decomposition, the nature of lexical entries as well as the neural mechanisms of inhibition in spoken word recognition.
For current details of our entire spring seminar series, please refer to http://web.mit.edu/speech/www/seminars.html
MIT Linguistics Colloquium
Tim Stowell
University of California, Los Angeles"(Parenthetically)" Parenthetical material is perceived intuitively to stand apart from its immediate syntactic environment, as though it were an interloper in the structure of the sentence in which it appears. It typically forms an autonomous intonational phrase, with a distinctive low pitch. Furthermore, it behaves with respect to various syntactic tests as though it occupied a higher, peripheral, or external syntactic position. For example, McCawley (1982) observed that parenthetical adverbs and appositive relative clauses are interpreted as being absent from the surrounding sentence for the purposes of computing variable binding and ellipsis.
Marga Reis has appealed to this interloper status of parentheticals in arguing that various types of finite clausal complements of bridge verbs in German are really main clauses in disguise. By taking the apparent main clauses in these sentences to be parenthetical, she seeks to reanalyze the central data involving embedded main-clause-like behavior (embedded verb-second, cases of partial wh-movement, etc.). Analyses of parentheticality have sought to capture their interloper status in various ways. Ross (1973) proposed that parentheticals originate as adjuncts to the entire sentence, forming the basis for their , and that they are inserted into their surface position transformationally. McCawley advocated an analysis allowing for novel devices such as discontinuous constituent structure. Safir (1987) suggested that parenthetical relative clauses are merged into the constituent structure at a remote level of representation (LF').
These analyses of parentheticality have generally been based on an incomplete survey of the potentially relevant data. I will examine a number of distinct parenthetical types, including parenthetical DPs, adverbials, relative clauses, as-clauses, main clauses, and conjoined constituents, as well as constituent types that cannot occur parenthetically, seeking to draw some syntactic generalizations about parenthetical placement. The syntax of parenthetical main clauses in German and English prove to be particularly informative. The gist of our approach is that parentheticals are, in an important sense, not as parenthetical as has been generally assumed. In this respect, I will argue for an analysis that bears important similarities to that of Emonds (1976).
Recent proposals that remnant movement operations play a major role in deriving surface constituent structure provide the basis for a new approach to parenthetical placement. I will advocate an analysis along these lines, according to which both the parenthetical material and portions of its immediate syntactic context undergo movement.
For more information, contact Michael Wagner or Heejeong Ko
Boston University Applied Linguistics Spring Speaker Series
Norvin Richards
MIT
"And God said, 'Weeqayuch': Resurrecting Wampanoag from 17th-century documents" ASL/English interpreters will be present.
For more information, contact linguist@bu.edu
MIT LING-LUNCH
Thursday, February 21, 2002 at 12:00, MIT E39-335 (conference room)MaryAnn Walter
MIT"Pharyngealization Effects in Maltese Arabic" The work presented concerns the behavior of pharyngealized consonants in Maltese Arabic, based on the analysis of vowel alternations in the Maltese lexical corpus. After the process of pharyngeal spread operative in Maltese is described, apparent exceptions to it are accounted for by a guttural dissimilation process with significant implications for the feature geometric representation of the consonants involved.<p> The phonological behavior of the set of Arabic consonants conventionally known as the emphatics has recently attracted considerable attention in the phonological literature, garnering no less than three articles in Linguistic Inquiry in the past six years (Davis 1995, McCarthy 1997, Watson 1999). Despite the fact that, alone among the modern dialects, it has lost the contrast between emphatic and non-emphatic consonants, Maltese Arabic provides a unique opportunity for the study of their phonological effects. As several scholars have observed, the former emphatics are detectable through their effects on the development of the Maltese vowel inventory (Cowan 1966, Borg 1977, Aquilina and Isserlin 1981). Moreover, the romanized orthography of Maltese includes the usual five-vowel complement, rendering the effects of the former emphatics readily observable in lexical corpus data.
In the following study, I examine the extent and nature of emphasis spread as it must have applied in Maltese Arabic at some date, based on and resulting in the historical changes in vowel distribution that we see today. After establishing how this process applied, I account for apparent irregularities in its application due to a guttural dissimilation process similar to that previously documented for Palestinian Arabic (Davis 1995). Because the Maltese process involves three interacting classes of consonants, rather than the two classes active in Palestinian, it has important ramifications for the feature geometric representation chosen for these consonants and the Lower Vocal Tract as a whole. Recent competing theories of representation are evaluated in this light (Davis 1995, Shahin 1997, Zawaydeh 1999, Halle, Vaux & Wolfe 2000). Potential implications for the affiliation of Maltese among the Arabic dialects are also discussed with reference to the guttural dissimilation process.
For more information, contact anevins@MIT.EDU.
MIT Linguistics Colloquium
Charles Kisseberth
Tel-Aviv University"Prosody and the Phonology-Syntax Interface: Chimwiini Revisited"
The two most essential phonetic manifestations of prosodic systems across the world's languages are pitch and duration. In Chimwiini (a dialect of Kiswahili formerly spoken in the town of Brava in southern Somalia), there are two quite distinct prosodic systems governing the location of pitch and the distribution of vowel length. Particularly interesting from the point of view of the theory of prosody, both of these coexisting systems are ones that are typically referred to as "stress" systems. High pitch falls on the penult in the default case, but on the final syllable in grammatically defined contexts. Vowel length is contrastive in the language, but can appear only in what would be the "stressed" syllable given the "Latin stress rule".The most striking aspect of these coexisting prosodic systems is that they have as their domain a unit that is larger than the word but smaller than the sentence or the intonational phrase. In fact, they operate in exactly the same domain. This domain cannot be equated with any syntactic phrase or set of syntactic phrases, but nevertheless reflects aspects of syntactic structure. Chimwiini is one of the example languages that Selkirk (1986) used to motivate the "Edge-Based, Indirect Reference" model of the phonology-syntax interface. This model proposes that phonology operates not by making direct reference in a rule/constraint to syntactic phrases, but rather by constructing "prosodic/ phonological" phrases which then may be specified as the domain in which particular rule(s)/ constraint(s) apply. These phrases are constructed by locating a Left (or Right) edge of a phonological phrase at the Left (or Right) edge of some level of X-bar structure. In particular, for Chimwiini, she proposed that the R edge of a phonological phrase is aligned with the R edge of a maximal projection.
Selkirk's discussion of Chimwiini was based on a very incomplete account of the language in Kisseberth and Abasheikh (1974) and as a consequence she could not draw all the lessons that she might have from the Chimwiini data. In particular, since both the quantity and the pitch systems operate inside the same phrases, Chimwiini represents the only known case where more than a single phonological phenomenon can be shown to operate in the proposed phonological phrases. In every other case discussed in the literature, phonological phrases are motivated by a single aspect of pronunciation.
It is also the case that further examination of Chimwiini reveals significant data that either do not follow from the identification of the right edge of a phonological phrase and the right edge of a maximal projection, or do so only if the syntactic analysis of specific constructions has the appropriate (but not superficially obvious) character. In this paper, we will discuss a number of these complexities: specifically, emphasis, the indefinite/definite NP contrast, negative morphology, subject postposing.
The last issue that we address involves cases where "accent" seems to have a domain larger than the phonological phrase. We utilize the constraint Wrap-XP proposed by Truckenbrodt, along with his appeal to recursive structure, to solve the problem posed by these accentual domains. However, facts about focus/emphasis lead to a recasting of Truckenbrodt's "alignment" formulation of the focus constraint.
For more information, contact Michael Wagner or Heejeong Ko
MIT LING-LUNCH
Thursday, February 14, 2001 at 12:00, MIT E39-335 (conference room)Marta Luján
University of Texas, Austin"Anaphoricity of the Determiner " The approach entails a general descriptive simplification, as it exploits the ability of the D, as a pronoun, to be bound to an antecedent. In addition, it leads to the rethinking of the antecedent-pronoun relation, the status of (in)definite categories, the interpretation of generic nominals, the discourse linking function of D's, and the existence of the Binding principles. The study basically addresses the question of how the computational system of language handles the reference of nominals and their interpretation in the clause and discourse domains.
For more information, contact anevins@MIT.EDU.
GSAS Workshop on Comparative Syntax and Linguistic Theory at Harvard
A reception will follow immediately thereafter in the Department of Linguistics (3rd floor, Boylston Hall).
Kyle Johnson
UMass-Amherst"Towards an etiology of certain islands"
For more information, contact Conor McDonough Quinn
Boston University Linguistics Lecture
Jonathan Barnes*
Ph.D. candidate, UC Berkeley"The Phonology of Duration:
Evidence from Turkish and Elsewhere"I will present the results of an experiment conducted on segment durations in Turkish, a language in which vowels in closed syllables are, somewhat disturbingly, systematically longer than comparable vowels in open syllables. I will argue that the mora-timing hypothesis is insufficient to deal with the complexity of the facts of Turkish, and further makes a number of incorrect predictions about syllable and word duration in Turkish and other languages. Instead, I propose an isochronic constraint mandating uniform durations of syllable rhymes, rather than of their individual subcomponents, which both accounts for the typologically aberrant vowel durations in Turkish, as well as for other data previously adduced in support of moraic isochrony. Additionally, having shown the Timing Tier to be ill-equipped to account for, at the very least, timing, I will suggest we might in fact begin to reevaluate the mora-as-autosegment more generally, in connection with the other phenomena it was originally proposed to explain.
Note: ASL/English interpreters will be present.
For more information, contact carol@bu.edu
MIT - Phonology Circle
Friday, February 1, 2002, 11:00 AM (note change in time), Room E39-335
Cheryl Zoll, MIT and the Radcliffe Institute of Advanced Studies "Toward a more restrictive theory of autosegmental association" Patterns of autosegmental feature association are thought to be affected by three independent factors: the morphological category of the autosegments, the identity of the features themselves, and a (language-specific) setting of the directionality parameter. Taking tone as an example, the linking of tones to a string of tone-bearing units is determined partially by the tones' morphological category (e.g., prefix vs. suffix), partially by the quality of the tones themselves (e.g., as this translates into special rules or constraints that refer specifically to H tones or L tones) and otherwise by a directional association specification, which is in most cases left-to-right. While the first two factors seem indispensable to a theory of tone mapping, the directionality parameter is more problematic. Rather than act as a restrictive and unifying principle, as it was intended, the compulsion to associate tone directionally has spawned a variety of unrelated devices to accommodate non-conforming data. In this talk I present the evidence against a directional view of tone mapping, and illustrate that ostensibly directional patterns such as those found in Kukuya, Kikuyu, Tiv and other languages are better characterized in terms of tone quality and morphology alone. The more restrictive theory of autosegmental association that results from the elimination of directionality as a factor provides a typology more in line with the cross-linguistic distribution of tonal patterns.
For further information, please contact kenstow@MIT.EDU
Boston University Linguistics Lecture
Wednesday, January 30 at 7:30 PM: CAS (725 Commonwealth Ave.), room 427
Adam Albright*
Ph.D. candidate,UCLA"The Identification of Bases: A Computational Approach" In this talk, I suggest that the problem of base identification is more fruitfully approached as an acquisition problem, rather than as a typological problem. In particular, I propose that base identification is part of a strategy employed by language learners to develop a system that will allow them to project inflected forms that they have not encountered before. I present a computationally-implemented model of paradigm acquisition that attempts to use a single surface form to project the rest of the paradigm, using stochastic morphological rules. It compares the predictive power of various forms within the paradigm, and selects as a base the form that allows the remaining forms to be projected as confidently and accurately as possible. As evidence for this approach, I discuss two cases in which an unexpected, marked form served as the base of an analogical change: Yiddish present tense paradigms (in which all forms were remodeled on the 1st sg), and Latin noun paradigms (in which nominatives were remodeled on oblique forms). In each of these cases, I show how the model correctly selects the typologically marked form as the base, and also correctly predicts the direction of subsequent paradigmatic changes.
Note: Thus far, we have been unable to find ASL/English interpreters for this talk. Any new information will be posted at this site. SORRY.
For more information, contact carol@bu.edu
Boston University Linguistics Lecture
Bruce Morén*
Visiting Assistant Professor, Boston University"Manner of Articulation in Two Modalities:
A Unified Approach"The work presented here is a part of a larger project, which attempts to untangle the complexities of ASL phonetics, phonology and morphology and to demonstrate direct parallels across modalities. The ultimate goal is to provide a unified model of grammar.
This talk looks at one area in particular -- Manner of Articulation (MOA). I provide a brief description of the phonetics of oral MOA, show that there are remarkable parallels among consonants and vowels of different types, and propose a new, more economical model based on a hybridization of the Aperture model of Steriade (1991) and the Constriction model of Clements (1991). I then turn to ASL and show that some aspects of handshape (HS) (normally considered a modality-specific element of sign) are actually the sign realization of MOA. Finally, I show that HS may be analyzed straightforwardly using the Aperture-Constriction hybrid.
One of the basic claims is that simple, complex and contour segments in both modalities are the result of particular segment-internal structures consisting of a combination of a C-manner node, a dependent V-manner node and three terminal features [open], [closed] and [lax]. For example,
Simple segments
- [t] and the handshape in YES have a C-manner of [closed]
- [s] and the handshape in DEAF have a C-manner of [open]
- [i] and the handshape in AGAIN have a V-manner of [closed]
- [a] and the handshape in MINE have a V-manner of [open]
Complex segments
- The lateral fricative and the handshape in ELEVATOR have a single C-manner node with both [open] and [closed] features
- [e] and the handshape in COUSIN have a single V-manner node with both [open] and [closed] features
One interesting result of this analysis is the elimination of the feature [lateral] and the major class features.
This talk is designed for an audience with some knowledge of phonology, and although helpful, knowledge of ASL is not necessary.
ASL/English intrpreters will be present. For more information, contact carol@bu.edu
Boston University Linguistics Lecture
Sam Rosenthall*
Assistant Professor, Oakland University"Glide Distribution in Arabic Verb Stems" ASL/English intrpreters will be present.
For more information, contact carol@bu.edu
