Class Note for INDV 101 at UA
Popular in Course
Popular in Department
This 22 page Class Notes was uploaded by an elite notetaker on Friday February 6, 2015. The Class Notes belongs to a course at University of Arizona taught by a professor in Fall. Since its upload, it has received 12 views.
Reviews for Class Note for INDV 101 at UA
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 02/06/15
IN THIS ISSUE StateoftheArticle San Duanmu on Tone 3 lf English and Japanese are tone languages one wonders what language is not Recent Issues in Linguistics 7 Elan Dresher on The Untouchables 11 A strong wind was blowing Filling the sky were countless bits of what looked like confetti blowing in the wind They were facts linguistic facts I tried to catch some in my hand but they eluded my grasp Goodies Donald A Becker on Lee Hartman s PHONO 32 22 Letters and replies Luigi Burzio replies to Morris Halle 1 Dissertations The syntax of focus and topic in Mandarin Chinese 12 by Shuing Shyu reviewed by Rint Sybesma Books Particles and projections in Irish syntax 15 by Nigel Duf eld reviewed by Alain Rouveret Georgian A structural reference grammar 18 by BG Hewitt reviewed by Lea Nash Book notices 24 Conferences Glow Athens 23 by Marcel den Dikken International Conference on Syntactic Categories Bangor 23 by David Adger LETTERS amp REPLIES A REPLY TO PROF HAllE Luigi Burzio ln Glot International 1 910 Morris Halle com ments on my article The rise of Optimality Theo ry Glot International 1 6 henceforth ROT I am grateful to him for contributing to the discus sion In that article I expressed my modest views on OT and what seem to me to be good reasons for studying morphophonology from a parallel constraintbased perspective My goal was to be informative to the general public and did not expect to succeed in changing his speci c views I nd his remarks less than effective in challeng ing mine Prof Halle attempts to refute only one of the several arguments l outlined in ROT which I will consider below Much of his letter is devoted to constructing what he sees as an argument for rules based on the Elsewhere Condition The thrust of it so far as I can determine would be as follows There is an important generalization in phonology that requires the Elsewhere Condition ofAnderson 1969 Kiparsky 1973 or some variant thereof in turn the Elsewhere Condi tion requires a rulebased framework 1 do not believe either point is correct Starting from the rst the existence of the Elsewhere Condition my own judgment is that it is unlikely that such a speci c condition would be a theoretical pri mitive Other conditions of a similar sort seem equally imaginable and one sees no reason why this particular one should hold I therefore nd it of some signi cance that substantial elsewhere effects are reducible to a theorem of OT as Prince amp Smolensky 1993 show What they note is in essence that any sets of facts that show the work ings of two different and competing constraints will always entail that the more speci c con straint must be ranked higher than the more general one or else the more speci c one would just never be detectable Hence to maintain his rst claim the exis tence of an Elsewhere Condition Prof Halle would have to show that there are effects which are not captured by this logic To maintain his second claim rules are crucial he would have to show that constraints fail to express such effects So far as I can tell he does not carry out either task He notes that the English CiV lengthen ing environments eg canAdian caps long can be seen as a speci c subset of the trisyllabic environments with shortening applying to the complementary subset or elsewhere eg NaT URAL lower case short He characterizes this effect in terms of the Elsewhere Condition and of rules The CiV lengthening rule more speci c applies rst and blocks the shortening rule more general by disjunctive ordering But note that a constraintbased account seems straightforward The CiV lengthening constraint is ranked high er than the shortening constraint Note too that this gives the exclusive disjunction automatically while the rulebased Elsewhere Condition needs to stipulate it That is while it s enough to rank the constraints it is not enough to order the rules One must additionally impose the exclusive disjunction disjunctive ordering to prevent shortening from applying to the output of length ening Furthermore for constraints the ranking found is in fact the only one that could be found Under the opposite ranking one would just never know there was a CiV lengthening constraint at all and would then only postulate the shortening constraint 7 Prince amp Smolensky s point In sum there seems to be no argument for rules from this elsewhere effect nor does there seem to be any need for a speci c condition at least with constraints Turning to the one argument of ROT Prof Halle challenges it has to do with the vowel shortenings exempli ed in 1 Continued on page 8 g E O Z Volume 2 USSUE Monthly Magazine for Linguists April 1996 ISSN 138173439 Price single issue D 1450 Glot International appears monthly except June amp July Copyright 1996 by Holland Academic Graphics All rights reserved Reference Colophon Editor Rint Sybesma Holland Institute of Generative Linguistics address Sinological Institute Leiden University PO Box 9515 2300 RA Leiden The Netherlands phone 31 71 5272538 fax 31 71 5272615 email glotrulletleidenunivnl Conference reports and announcements editor J anWouter Zwart Groningen University Department of linguis tics Oude Kijkin t Jatstraat 26 9712 JK Groningen The Netherlands email zwartletrugnl Correspondents Regular contributors Eulalia Bonet Barcelona Marcel den Dikken Vrije Universiteit Amsterdam Paula Fikkert Konstanz Bob Frank John Hopkins Universi ty Eric Hoekstra PJ Meertens Instituut Am sterdam Helen de Hoop Groningen Sabine latridou University of Pennsylvania Lea Nash Paris Guido Vanden Wyngaerd Brussels Shohei Yoshida Yokohama National University JanWouter Zwart Groningen Columnists Elan Dresher Toronto Neil Smith London Information for contributors Linguists who would like to contribute a Stateof theAr39ticle or a Review ofa book or a dissertation are invited to contact the Editor Linguists who have recently nished their dissertation and want Glot International to pay attention to it are invited to forward a copy to the Editor Every body is welcome to contribute to Glot Internationr al But before you send in anything please con tact the Editor Glot International prints confer ence announcements calls for papers conference programs etc Contact the editor Subscription information Glot International appears every month except in June and July Prices for Volume 2 1996 D 95 for individuals D 135 for libraries and institutions D 1450 for single issues Special prices for bulk subscriptions minimally 10 Subscriptions at the individual rate must be or dered directly from the publisher Subscriptions at the institutional rate may be or dered from your bookseller or subscription agent or directly from the publisher Prices include postage Prices exclude 6 VAT EC only Orders must be prepaid Payments preferably by giro credit card Ameri can Express EuroAccessMaster Visa or inter national money order Credit card orders must include the complete cardholder s address There is an additional charge of D 1750 for payments by international bank cheque or bank transfer Publisher Holland Academic Graphics Julius Rbntgenstraat 15 2551 KS The Hague The Netherlands phone 31 70 3251057 fax 31 70 3231686 email 72113335com puservecom bank acc ABN AMRO 49 88 22 796 postal giro 6261107 VAT nr NL 058861798B01 Advertisements Contact the publisher Printing Printed in The Netherlands by ICG Printing Dordrecht Glot International Volume 2 Issue 4 April 1996 Page 2 and the title of his talk is Wh in English quot Well I changed the title a little bit Now it s called lVIiddles in Oldra anese Contributors to this issue David Adger University of York Department of Language and Linguistic Science Heslington York YOI 5DD UK Donald A Becker Departments of Linguistics and German Uni versity of WisconsinMadison 1220 Linden Drive MadisonWI 53706 USA Marcel den Dikken HILDepartment of Linguistics Vrije Universi teit Amsterdam Postbus 7161 1007 MC Am sterdam The Netherlands Elan Dresher Department of Linguistics University of Toron to Toronto Ontario M5S 1A1 Canada San Duanmu Program in Linguistics University of Michigan Ann Arbor Michigan 481091285 USA Lea Nash Universite Paris 8 Departement des Sciences du Langage 93526 SaintDenis Cedex 02 France Alain Rouveret Universite Paris 8 Departement des Sciences du Langage 93526 SaintDenis Cedex 02 France Shuing Shyu Department of Foreign Languages and Literature National Sun Yatsen University Kaohsiung Taiwan 804 Dissertations reviewed Shuing Shyu The syntax offocus and topic in Mandarin Chinese University of Southern Cali fornia Supervisors Audrey Li Joseph Aoun George Hyden Degree date August 1995 246 pp Available from the author address see above Books reviewed Nigel Du ield Particles andprojection in Irish syntax Kluwer Academic Publishers 1995 xiv372 pp With index ISBN 0792335503 Price H 220 US 160 UK 100 BG Hewitt Georgian A structural reference grammar John Benjamins 1995 xviii7l4 pp With index ISBN 90 272 3802 2 Eur 155619 7268 US Price H 250 US 143 Goodies How to obtain phono 32 Those with access to the Internet may obtain it electronically as follows p linguisticsarchiveumichedu cd linguisticsso waredos binary get phonozip Others may obtain their free copy by sending 350 to defray the cost of a diskette mailer and postage to Lee Hartman Department of For eign Languages Southern Illinois University Carbondale IL 629014521 USA FUTURE ISSUES StateoftheArticle Alana Johns on Ergativity Kyle Johnson on VP Ellipsis Pieter Muysken on Creole languages Dissertations Jonathan Boban ik reviewed by Ur Shlonsky Anna Roussou reviewed Stateofthe article Glot International Volume 2 Issue 4 April 1996 San Duanmu past two or three decades TONE AN OVERVIEW What is tone and how should we represent it What different functions does tone perform in different languages What is the relation between tone and intonation San Duanmu enumerates the developments of the 1 Introduction This article surveys developments and achievements in research on tone in the past twenty years or so primarily in the generative framework and outstanding issues that pertain to current research I begin with a discussion of what tone languages are and point out some similarities between tone and intonation 11 Tone languages and nontone languages It is not difficult to identify some languages that are clearly tone languages A textbook exam ple is Standard Mandarin Chinese which has four tones on full syllables weak or unstressed syllables do not bear tone and are sometimes said to bear the neutral tone shown in 1 1 mal mother ma3 horse ma2 hemp ma4 to scold The numbers 174 refer to the four tones tone 1 is a high tone 2 is a rise tone 3 is a low or lowrise in nal position and tone 4 is a fall The four words in 1 are phonologically identical in all respects except tone I forgo the issue of duration for the moment In other words tone or pitch contour is lexically contrastive in Standard Man darin Chinese for other functions of tone see section 314 Most tone languages are found in Asia and Africa In addition some European and native American languages have also been con sidered to be tone languages It is not difficult either to identify some languages that are nontone languages For ex ample English is typically considered a nontone language When the word kaet cat is said with different tones or pitch contours additional meanings are added such as surprise doubt disbelief and so on however the word still refers to the same animal Thus tone is not lexically contrastive in English However difficulties can arise when one tries to classify languages as either tonal or non tonal A wellknown example is Japanese in which every word has a xed tone pattern This can be seen in 2 taken from McCawley 1978 113 where syllables are separated by a hyphen and where H is a high tone and L is a low tone 2 lgkiga HLL oyster kaE ga LHL fence kakigg LHH persimmon These examples show that tone is lexically con trastive in Japanese as it is in Chinese On the other hand as noted by McCawley 1965 see also Block 1946 Haraguchi 1977 the tone pat tern of a Japanese word is predictable if we posit an abstract accent position for each word shown by underline in 2 This analysis accounts for the fact that for a given word form there are only as many possible tone patterns as there are syllables ignoring unaccented words For ex ample the form in 2 has three syllables so it has three possible tone patterns It will be noted that unlike stress accent is not necessarily ac companied by greater duration or amplitude Apart from its effect on pitch therefore accent is hardly felt by native speakers I will return to this point below Since tone or pitch is predict able from accent in Japanese it is often called a pitchaccent language But English also has pitch contours In addi tion the contours are quite predictable eg Gold smith 1974 published in 1981 Liberman 1975 Pierrehumbert 1980 For this reason English has been called a tone language too For exam ple Goldsmith suggests that in neutral intona tion English words have the tone pattern MHL where M is a mid tone and H is linked to the stressed syllable Some examples are shown in 3 3 Chig go MHL AErica MHLL M ton HL Qnada HLL Analyses like 2 and 3 can be extended to Afri can languages eg Goldsmith 1982 Kenstowicz 1987 Sietsema 1989 Kenstowicz amp Kisseberth 1990 American Indian languages eg Hinton 1991 and European tone or pitchaccent lan guages such as Lithuanian Halle amp Vergnaud 1987 and SerboCroatian lnkelas amp Zec 1988 Such analyses raise serious questions for a typo logy of tone languages as noted by Hyman 1978 and McCawley 1978 In particular the distinc tion between tone and nontone languages is no longer obvious In addition if English and J apa nese are tone languages one wonders what lan guage is not 12 Tone and intonation The pitch contour in English is often called intonation Indeed all nontone languages have intonation Although both tone and intonation are characterized by pitch for a discussion on the difference between pitch and the fundamental frequency see Beckman 1986 one may still won der whether they are different For example are tone and intonation produced by the same articu latory mechanism Are tone and intonation per ceptually the same Should tone and intonation be represented with the same phonological fea tures Phonetically there is no evidence that tone and intonation are made with different articulatory mechanisms cf Zemlin 1981 In addition as far as listeners are concerned tone in Chinese and intonation in English are interchangeable I will cite two examples First as Chao 1980 p42 noted when English words are borrowed into Cantonese a stressed English syllable usually becomes a high toned Cantonese syllable This is shown in 4 where underlining indicates stress in English 4 English Cantonese Eland Okl n HL Paci c phasiwik MHL Obviously the high pitch on a stressed English syllable is heard as a high tone by Cantonese speakers Second Standard Mandarin Chinese has a rule that changes tone 3 low to tone 2 rise before tone 3 or 3 gt 2i3 Cheng 1968 noted that in codeswitching tone 3 changes to tone 2 before an unstressed English syllable as well but not before a stressed one This is shown in 5 Page 3 5 a xia02 professor xiao3 professor b xiao3 lecturer xia02 lecturer small professor small lecturer xiao3 small is underlyingly tone 3 It changes to tone 2 in 5a but not in 5b The reason is that in 5a the following unstressed syllable pro has a low tone although Goldsmith marks it as M see section 31 this is perceived by Standard Mandarin speakers as tone 3 low and as a re sult the rule 3 gt2 is triggered ln 5b the follow ing stressed syllable lec has a high tone which does not trigger tone change Finally let us see whether there are reasons to posit different pho nological features for tone and intonation Tradi tionally tone refers to the pitch contour of a syl lable or a word and intonation refers to the pitch contour of a phrase or an utterance But as Gold smith 1981 pointed out intonation in English consists of a sequence of word tones similar to what one nds in a tone language In addition tonal behavior in English is similar to that in other tone languages For example when the word tone MHL falls on Japan we get MHL where HL on the second syllable is realized as a fall this is what one nds in tone languages see below We have seen then that tone and intona tion display certain similarities This point will recur in the rest of this article 2 Developments in the past 20 years In early works of generative phonology 1950s and 1960s there was very little mention of tone This is probably because a main issue then was distinctive features and tone did not seem to t in easily In particular distinctive features work well for the representation of seg ments but tone has traditionally been thought to be as a suprasegmental property a property that belongs to units larger than the segment eg Pike 1948 Firth 1957 Another difficulty is that some tones seem to require trajectory features such as rise or fall whereas Stateofthearticle For example in Standard Mandarin all full syl lables are long In particular tones 1 2 4 and the low version of tone 3 have two rime segments and are HH LH LL and HL respectively thus ma in 1 should be maa In addition the low rise version of tone 3 has three rime segments and is LLH This proposal is in my view essen tially correct I will return to it below The next development came with the works ofLeben and Williams Leben 1971 1973 sug gested that tones should be represented on an independent tier parallel to the segment tier Williams 1971 published in 1976 further sug gested that the two tiers are coordinated by tone mapping rules which later became known as association lines and association conventions This line of research culminated in the in u ential work of Goldsmith 1976 through which the term autosegmental phonology became widely known other equivalent terms are three dimensional phonology nonlinear phonology and multitiered phonology In the following decade a large number of autosegmental studies ensued Among works on tone most dealt with African languages eg Clements 1978 Clements amp Ford 1979 Odden 1981 Hyman 1981 Pulleyblank 1983 Clements amp Goldsmith 1984 Cassimjee 1987 Myers 1987 and many others Main issues of interest included the association conventions the representation of multiple tone levels default tones oating tones downdrift a H after a L is lower in pitch than the preceding H and down step a H is lower in pitch than a H immediately preceding it this turned out to be the same effect as downdrift after it was discovered that there is a oating L right before a downstepped H pre linked tones accent marking the OCP Obligato ry Contour Principle effect and the interaction between tone and morphosyntactic structure Generative works have also been done on Euro pean tone or pitchaccent languages such as Swedish Norwegian SerboCroatian and Lithu anian eg Bruce amp Garding 1978 Gardin 1979 Fretheim 1981 Withgott amp Halvorsen 1988 lnkelas amp Zec 1988 Halle amp Vergnaud 1987 There were fewer generative studies on Asian tones in the 1970s and 1980s with most of these focusing on Chinese languages Main works included those of Yip 1980 Wright 1983 Chan 1985 Shih 1986 and Chen 1987 Primary issues of interest were the representation of contour tones the representation of tone sandhi and domains of tone sandhi The Chinese dialects that received the most attention were Standard Mandarin in particular its tone 3 sandhi rule Southern Min dialects and Wu dialects Besides real tone languages tonerelated work has also been done on other languages such as English Japanese Dutch and German Liber man 1975 Bing 1979 Ladd 1980 Pierrehumbert 1980 Poser 1984 Beckman amp Pierrehumbert 1986 Gussenhoven 1988 F ry 1989 among oth ers In particular Liberman 1975 has argued that English intonation can be analyzed in terms of a stress system and a tone system where the tone system consists of a single sequence of Hs and Ls This approach now a standard in genera tive analyses of intonation languages suggests that tone and intonation are essentially the same phenomenon Parallel to the generative study on tone four other developments should be mentioned First there was an effort to nd universals of tone through crosslinguistic survey eg Fromkin 1972 Hyman amp Schuh 1974 and Maddieson 1978 Second there were works on tonogenesis in Asian languages ie the emergence of tone as the result of the loss of consonant features eg Haudricourt 1954 Maran 1973 Matiso 1973 Third some phonetic models on tone were pro posed eg Halle amp Stevens 1971 Hombert 1978 Finally there is a continuous ow of descriptive works by linguists in China mostly written in Chinese Of special interest are experimental studies on Standard Mandarin tone e g Lin Yan amp Sun 1984 Lin 1985 Yan amp Lin 1988 He amp Jin 1992 Wang amp Wang 1993 a series of works on Glot International Volume 2 Issue 4 April 1996 Tibetan eg Hu 1980 Hu Qu amp Lin 1982 Qu 1981 and a large number of studies on the Wu dialects especially the work of Xu Tang You Qian Shi amp Shen 1988 on Shanghai 3 Current issues relating to tone Despite the tremendous progress in tone research in the past two decades many issues remain quite open Consequently in discussing current issues I will often speculate on what I think is likely to be the right solution 31 Levels of tone There have been many discussions on how many tone levels are needed to describe all lan guages eg Chao 1930 Wang 1967 Halle amp Stevens 1971 Anderson 1978 Yip 1980 Clements 1983 Huang 1985 Hyman 1986 van der Hulst amp Snider 1993 but the issue is still quite open Phonetically pitch is the primary perceptual correlate of tone and in real speech there can be many pitch levels But if we are concerned with phonemic levels ie those that are distinctive the number of levels is quite small In particular if downstep or downdrift belong to phonetic implementation which need not be represented with separate levels far fewer phonemic levels are needed In African languag es two phonemic levels H and L are often suf cient although many more have been posited I return to depressed tones later The same is true for English For example Goldsmith 1981 posited MHL as the neutral English word tone where syllables before stress have a higher pitch level than syllables after stress But if we assume downstep or what Beckman amp Pierrehumbert 1986 call catathesis by which both H and L de scend gradually there is no need to posit M and the English word tone can simply be LHL For this reason Liberman 1975 and Pierrehumbert 1980 only assumed two levels for English H and L In Asian languages three or four contrastive levels are quite common ve contrastive levels have also been reported Chang 1953 Shi Shi amp Liao 1987 The widely used system of Chao 1930 adopted by the International Phonetic Association 1989 posits ve levels Anderson 1978 also posits ve levels primarily because of Chang s 1953 report that in certain MiaoYao languages ve levels are found on isolated mono syllables lf tones are to be represented by distinctive features one needs to convert multiple levels into binary features It is possible to translate any multilevel system into a binary system the challenge is to nd an argument to justify the translation For example English vowels have three heights for which SPE uses two binary features high and low The analysis is justi ed by the fact that there are rules that refer to high and mid vowels captured by flow and rules that refer to mid and low vowels captured by ihigh However many Asian tone language have few tone sandhi rules especially those that involve tone split and tone spreading which might provide insight into the composition of the tones An important tone model was suggested by Yip 1980 1993 in which four levels are posited The four levels are divided into two Registers upper and iupper Each Register is divided into two levels raised and iraised for conve nience I will write them as H and 7H and call them the Pitch feature in contrast to pitch which refers to phonetic pitch height Thus the four tone levels are represented by two binary features as shown in 6 6 Register Pitch upper H 1 7H H upper H Ill 7H IV The strongest argument for Yip s model in my view is that Register is independently related to Page 4 the voicing of the onset consonant In particular in those Chinese languages that have retained onset voicing in obstruents upper Register invariably goes with a voiceless onset and eup per Register invariably goes with a voiced onset The same effect is found in many African languages where a voiced obstruent consonant lowers the pitch of a neighboring tone Stateofthearticle skepticism among Chinese phonologists eg Yip 1989 Bao 1990 Chan 1991 Contour tones pose two problems for distinctive feature theory First if contour tones are basic units they require trajectory features such as rise and fall or a modi ed version of it as in 7 after Yip 1989 where TBU is the Tone Bearing Unit 7 rise TBU fall TBU J L H H L Duanmu 1994 argued that there is no compel ling evidence for contour tone units instead all Chinese tones are clusters of level tones But even if all contour tones are sequences of level tones another problem remains Since Leben 1971 and Williams 1976 it has been common ly assumed that two or more tones can be linked to a single short vowel which creates a short contour tone Now if a short rise has the feature 7H H then the vowel carrying it has two values of the same feature H A segment that contains two or more values of the same feature is a contour segment lf phonological theory admits contour segments many possible seg ments are predicted most of which are not found Anderson 1976 noted this problem and suggest ed that perhaps contour features can be allowed for suprasegmental features but not for segmen tal features However the distinction between segmental and suprasegmental features is un clear In addition as will be discussed below there are reasons to consider tone a segmental feature too Duanmu 1994 argued that in Chinese lan guages a contour tone always falls on a long syl lable as suggested by Woo 1969 Each rime segment therefore can take just one tone either H or L ignoring Register Duanmu 1994 also argued that based on available evidence short contour tones in African languages always occur on lengthened syllables For example the widely cited short contour tones in Mende Leben 1973 Goldsmith 1976 and subsequent works occur on lengthened vowels Aginsky 1935 Ward 1944 The same is true in English For example in neutral intonation black is HL fall and black bird is HL H on the rst syllable and L on the second It appears that the former has a short contour tone But as Goldsmith 1981 p299 pointed out and as is wellknown phonetically ae is lengthened in black but not in blackbird In other words real cases of short contour tones contour tones on vowels that are not lengthened are yet to be seen This is not surprising in the work of Greenberg amp Zee 1979 it was found that a rising pitch contour will be heard as a level tone if the vowel is short 80 ms or less There is therefore no compelling reason to posit contour tone features in distinctive feature theory for arguments against other contour features see references cited in Duanmu 1994 35 The tone bearing unit TBU Four kinds of units have been proposed as the TBU They are described in 8 and illustrat ed with the syllables man and ma in 9 where delimits a TBU 8 a The entire syllable or the voiced part of it b The rime portion ofthe syllable but not the onset por tion c The mora including the onset d The moraic segment the segment in the rime 9 man Ina a man ma b an 3 c ma 1 ma d a in a Chao 1930 Firth amp Rogers 1937 and Wang 1967 among others assumed 8a Howie 1974 proposed 8b Hyman 1985 proposed 8c Woo 1969 proposed 8d 8d is consistent with the Glot International Volume 2 Issue 4 April 1996 moraic theory of Hayes 1989 where the onset consonant is not dominated by a mora There is no evidence that the onset can be a TBU by itself The question is when the onset is voiced such as a sonorant does it carry part of the tone ln 8a and 8c it is assumed that it does in 8b and 8d it is assumed that it does not The rst assumption is often made but never argued for as far as I am aware In contrast Howie 1974 has argued for the second assumption and shown that the pitch contour of a voiced onset sonorant consonants or glides is often irregular and that the expected pitch contour does not start until the rime portion lf Howie is correct the remain ing choice is between 8b and 8d According to 8b both man and ma have one TBU According to 8d man has two TBUs and ma has one There is no question that 8d is true in some languages such as Luganda and Japanese That 8b describes the TBU is often thought to be true in Chinese languages e g Chao 1930 Firth amp Rogers 1937 Wang 1967 Yip 1989 Chan 1991 It seems then that the choice between 8b and 8d is a matter of language variation But as Duanmu 1994 argued the TBU in Chinese is also 8d as suggested by Woo 1969 see also Shih 1992 who argued that the TBU in Standard Mandarin is not the syllable but the moraic unit If this is correct there may be no crosslanguage variation in the TBU In stead the TBU is a moraic segment in all langua ges 36 Tonogenesis and consonant tone inter action Tonogenesis refers to the emergence of tone in a previously nontone language based on ear lier consonantal contrasts The term is rst coined by Matisoff 1970 Tonogenesis is com monly found in Asian languages A wellknown example is Vietnamese Haudricourt 1954 Matisoff 1973 in which the loss of coda conso nants created an initial set of three tonal con trasts and a subsequent loss of onset voicing split the contrasts into six In other languages such as Lhasa Tibetan Hu 1980 Duanmu 1992 the loss of onset voicing created a twoway tonal contrast Some researchers believe that Chinese also evolved from a nontone language through a similar process eg Baxter 1992 p7 Tono genesis is native American languages is some what controversial see de J ong and McDonough 1993 for a recent discussion I am not aware of studies on tonogenesis in African languages but see depressor consonants below The effect of coda consonants on tone will not be discussed here see Maran 1973 for some dis cussion The effect of onset voicing on tone can be generalized as voicelesshigh and voiced low a voiceless onset leads to a higher tone and a voiced onset leads to a lower tone see section 33 and below for the featural link between voic ing and tone 10a shows an example from Lhasa Tibetan which acquired tone relatively recently Hu 1980 and 10b shows an example from Shanghai Chinese which still retains onset voicing in addition to tone differences Xu Tang You Qian Shi amp Shen 1988 10 a Historical Tibetan Lhasa Tibetan kho gt kho high he go gt kho low hear b Shanghai Chinese se high umbrella ze low rise wealth ln 10a a historically voiceless onset gave rise to a syllable with a high tone and a historically voiced onset gave rise to a syllable with a low tone ln Shanghai Chinese there are two rising tones on smooth syllables those without a glottalized rime As seen in 10b the high rise occurs with a voiceless onset and the low rise oc curs with a voiced onset Apparent exceptions to voicelesshigh and voicedlow have been observed eg Kingston amp Solnit 1988 However such exceptions often re Page 5 sult from subsequent tone change For example the historical Chinese tone Ping has split into Yin Ping and Yang Ping in Mandarin dialects which have lost onset voicing Yin Ping occurs with historically voiceless onsets In Beijing Man darin Yin Ping is a high level tone but in Tian jing Mandarin Yin Ping is a low level tone It is reasonable to assume that Yin Ping in Beijing Mandarin is closer to the value right after Ping split than Yin Ping in Tianjing Mandarin ln oth er words at the time of tone split voicelesshigh and voicedlow seems to be a valid correlation The split of an existing tone by onset voicing can be viewed in various ways A possible analy sis of10b is shown in 11 see Hombert Ohala amp Ewan 1979 for a different proposal and Maddieson Stateofthe article occur in the same language This issue has gener ated considerable discussion There is little doubt that in languages like English both stress and tone patterns are determinable at least by trained linguists But perhaps English is not a rea tone language Indeed I am not aware of any real tone language in which native speak ers or trained linguists have a clear judgment for phonetic stress For example in Chinese languages native speakers do not agree on stress locations on full syllables nor do Chinese lin guists agree Similarly in African languages Luganda has been at the center of the debate yet there is no agreement on whether it has accent some consider it to have tone only some consider it to have accent from which tone is predicted and some consider it to require both tone and accent Hyman amp Katamba 1993 and references therein In metrical phonology both stress and accent refer to the head of the prosodic unit foot I ignore higher levels of stress It is likely that prosodic structures are present in all languages eg Selkirk 1981 McCarthy amp Prince 1986 For example Poser 1990 shows that despite the lack of phonetic stress Japanese has foot structure Similarly Shih 1986 and Yip 1992 1994 have argued that despite the lack of agreement on stress Chinese languages also have foot struc ture Nevertheless as Pulleyblank pointed out it is rarely shown whether the purported accent in tone languages is motivated on independent metrical grounds In addition if accent can be justi ed independently one would like to know how it interacts with tone In Chinese languages stress is determinable on independent metrical grounds A0 1992 1993 Duanmu 1993 1995 In addition there is a spe ci c relation between tone and stress which I will call the ToneiStress Principle and state in 12 12 Tone s tress Principle A stressed syllable is accompanied by an underlying tone pattern An unstressed syllable is not accompanied by an underlying tone pattern A stressed syllable need not carry the entire tone pattern by itself For example in Goldsmith s analysis of English a stressed syllable in neutral intonation is accompanied by MHL When it is the only syllable it carries HL and when it is surrounded by unstressed syllables it carries just H According to 12 when a syllable loses stress it will lose its underlying tone pattern This agrees with the fact that in fast speech where more syllables are destressed fewer tone do mains occur I will illustrate 12 with some data from Shanghai Chinese which has two tone patterns on monosyllables LH and HL ignoring Register Polysyllabic domains also have two patterns L H LL if the initial syllable is underlyingly LH and H LL if the initial syllable is underlyingly HL I ignore a more restricted third pattern Thus L H LL is an expansion ofLH over the rst two syllables with additional syllables re ceiving a default L Similarly H LL is an expansion of HL In normal careful speech poly syllabic foreign words form disyllabic tone do mains with a trisyllabic nal domain if there is an odd nal syllable indicated by parentheses in 13 where underlying tones are shown above surface tones syllables are separated by a hy phen z is syllabic in sz 13 LH LH HL LH LH LH HL LH LH LH LH L H H L L H H L L H L kek kha sz lu va kha ka li fo l ya Czechoslovakia California The underlying tone pattern of a syllable in a foreign name is that of the character used to represent the syllable ln hyperarticulated speech where every syllable is stressed every syllable can surface with its underlying tone Glot International Volume 2 Issue 4 April 1996 pattern 13 shows binary foot formation In addition 13 suggests that stress is initial in each foot This is con rmed by a comparison between 1 2 and 2 1 compounds numbers indicate the number of syllables in a word In normal careful speech a 1 2 compound forms one domain as in 14a but a 2 1 compound forms two as in 14b 14 a sa fega b raw tomato fagagtltthagt tomato soup The difference between 1 2 and 2 1 is predict ed if stress is leftheaded so that in 1 2 there is a stress clash namely two stresses occurring on adjacent syllables as shown in 15a ln metrical phonology stress clash needs to be resolved this can be achieved by deleting stress from the sec ond word which gives rise to a trisyllabic foot In contrast there is no stress clash in 2 1 as shown in 15b therefore both feet remain 15 a b x x x x XX X X XX In many other Asian tone languages such as Standard Mandarin Chinese Cantonese Chinese and Thai consecutive full syllables do not form disyllabic domains According to Duanmu 1993 the reason is that in those languages every full syllable is heavy and forms a bimoraic foot so that it is able to retain its underlying tones In contrast in languages like Shanghai there are no heavy syllables therefore a foot usually contains two or more syllables In other words the ToneiStress Principle applies to both kinds of Asian tone languages In English eg Pierrehumbert 1980 Norwe gian eg Withgott amp Halvorsen 1988 and Dutch eg Gussenhoven 1988 a tone pattern accompa nies each stressed syllable This is similar to the ToneiStress Principle It remains to be seen to what extent the ToneLStress Principle applies to African tone languages once accent is motivated on independent metrical evidence for some dis cussion see Kenstowicz 1987 Hyman 1987 1989 Sietsema 1989 Kisseberth amp Cassimjee 1992 Hyman amp Katamba 1993 The ToneiStress Principle is not the only source of tone assignment An obvious exception is boundary tones which occur at the edges of certain phonological units Liberman 1975 Pierrehumbert 1980 Beckman amp Pierrehumbert 1986 Another exception is the associative tone in some African languages which is a oating tone that occurs between two nominals in a geni tive relation eg Odden 1980 Williamson 1986 It is unclear whether the associative tone is a boundary tone and I will leave the issue open 38 Typology of tone languages I mentioned in section 1 that it is dif cult to establish a typology of tone and nontone lan guages and that in some sense all languages are tonal This view was in fact held by Beach 1924 p84 However there are obvious differences among languages like English Japanese and Chinese which ought to be captured Pike 1948 p3 offered a narrower de nition of tone languages according to which every sylla ble in a tone language can carry a contrastive tone By this de nition Japanese is not a tone language because only the accented syllable carries a contrastive tone Similarly Swedish and Norwegian are not tone languages because only the stressed syllable carries a contrastive tone Pike 1948 p 14 However by the same de ni tion Chinese languages will not be tone languag es because as discussed in section 37 only stressed Chinese syllables are accompanied by a contrastive tone pattern Consequently differences among languages like English J apa nese and Chinese are again left unaccounted for In fact if the ToneiStress Principle is correct no language would qualify as a tone language by this de nition Page 6 Goldsmith 1981 suggested that all languag es have tone but there is a distinction between tone and accentual languages in terms of tone linking In a tone language such as Mende tones are linked to syllables or TBUs in a word from left to right In contrast in an accentual language such as Japanese and English a desig nated tone is linked to an accented syllable rst such as the H in the English MHL pattern shown in 3 and other tones are linked after wards However this proposal implies that there is no accent or st Stateofthe article 19 a Stress analysis b Tone analysis W W W W H H H H c Stress and tone analysis W W HL LH In 19a suggested by Halle amp Vergnaud 1987 an accent is assigned to either the rst or the second slot of a long vowel Then H is assigned to the accented slot and L is assigned to other vowel slots However 19a assumes that accent can occur on the second mora of a heavy syllable which is metrically controversial Kager 1993 1995 In 19b suggested by Blevins 1993 H is lexically linked to either the rst or the second slot of a long vowel A default L can then be as signed to other vowel slots As Blevins pointed out 19b avoids the use of accent therefore it appears more economical On the other hand 19b implies that Lithuanian has no metrical system which is doubtful Besides 19b does not simplify the overall theory since in languages like Chinese and English both stress and tone must be posited anyway see section 37 19c assumes that Lithuanian has accent and two lex ical tone patterns LH and HL Which analysis is correct depends on i whether Lithuanian has accent and ii whether accent can fall on the second mora of a heavy syllable If 19a is cor rect Lithuanian is a pitchaccent language If 19c is correct Lithuanian is a tone language If 19b is correct Lithuanian belongs to none of the categories in 16 Whether all languages can be subsumed under a single toneaccent theory is an interesting question For a recent proposal see Kim 1996 39 Tone domains The term tone domain has been used to refer to two diff erent things First it refers to a stretch of sounds over which an expected contour occurs For example in Shanghai Chinese the pattern HL can occur on one or more syllables as in 20 a monosyllable is long in isolation 20 HL H L H L L see se pe se du pe three three cups three big cups Domains like those in 20 have one stress there fore they are a stress domain or a foot A second use of tone domain refers to a stretch of sounds that are affected by a tone rule For example in Xiamen Chen 1987 every full syllable has two lexical tone patterns one used in the nal posi tion A and the other in non nal positions B An example is given in 21 where the pitch val ues of A and B tones need not concern us 21 A1 A2 A3 gt B1 B2 A3 pang hong ts e to y a kite A domain like 21 has been called a tone domain because every non nal syllable takes its B tones But a domain like 21 is not a single stress do main In 21 all the three syllables are stressed and bear their own lexical tones whether it is the A copy or the B copy Both A and B copies are lexically speci ed and totally unrelated to what tones other syllables have Both kinds of domains interact with mor phology and syntax in intricate ways For exam ples from Chinese languages see Shih 1986 Chen 1987 Selkirk amp Shen 1990 Zhang 1992 and Duanmu 1995 For examples from African languages see Clements 1978 and Odden 1987 310 Is tone a prosodic feature Tone has traditionally been considered a prosodic feature which belongs to a unit larger than a segment eg Chao 1930 Pike 1948 Firth Glot International Volume 2 Issue 4 April 1996 1957 Wang 1967 There are several reasons for this view First a tone pattern usually remains constant independent of its carrier see 3 and 20 Second tone can often survive vowel dele tion and relocate to another vowel Third tone is quite free to move or spread from one syllable to another in contrast segmental features are usually not so mobile On the other hand there are reasons to con sider tone a segmental feature First although a tone pattern can extend over several syllables each moraic segment carries just one tone H or L In this regard a tone feature H or 7H is like a segmental feature which ultimately re sides in a segment and which cannot occur twice within a segment see sections 34 and 35 Sec ond a suprasegmental feature is usually thought to be diff erent from a segmental one However in many Asian languages tones came from conso nant features For example in 11 slack is a voicing feature on the consonant and Register feature on the vowel If we are to treat a feature consistently we might consider slack to be a segmental feature in both cases Third tone is not unique in surviving segment deletion Fea tures like nasal iback and round are also able to survive segment deletion Finally let us consider the mobility of tone First it will be noted that tone is not unique in moving or sprea ding Vowel features back high low and round and consonant features like nasal and retro ex have all been found to spread in har mony processes In the current view of feature theory eg Sagey 1986 Halle 1995 every fea ture lies on an independent tier A feature F can move or spread when there is nothing in its way that is when the intervening segments are not speci ed for F Tone is more mobile only because unstressed syllables are not speci ed for tone cf the ToneiStress Principle in 12 as a result of which tone can spread onto and past them It is worth noting that in tone spreading it is usually the Pitch feature that spreads and not the Regis ter feature This follows if Register is the same as voicing sections 31 32 33 and 36 which is usually speci ed in consonants Since most sylla bles have a consonant Register rarely spreads out of a syllable We have seen then that tone features are not fundamentally diff erent from features like na sal round and back which are usually con sidered segmental features Of course one can de ne suprasegmental features by to their be havior in a given context For example one can de ne a suprasegmental feature as one that spreads across syllables Thus nasal is a supra segmental feature when it spreads but a seg mental feature when it does not However by this de nition tones are also segmental features when they do not spread and since in most Asian languages tones do not spread one would con clude that Asian tones are mostly segmental But the usefulness of this de nition is not obvious as discussed in section 37 and 39 whether a tone pattern spreads or not depends on stress When there are many stressed syllables tones appear to be inactive and when there are fewer stressed syllables tones appear to be mobile A de nition of suprasegmental features based on their mobili ty would therefore appear to be quite super u ous 311 Categorical and gradient values There has been some debate about the cate gorical vs gradient nature of linguistic features and tone bears on this issue In real speech pitch levels seem to be in nitely variable However no language uses more than a small number of contrastive tone levels In particular as dis cussed in section 31 there may be only four contrastive tone levels and if Register is cued by voice quality sections 32 and 33 there may be just two levels that are cued by pitch H and L In any case contrastive tone levels are strikingly small in number especially in view of the fact that in singing any ordinary person can use at least a dozen or so music notes at ease The pau Page 7 city of contrastive tone levels supports the view that linguistic features are categorical in nature 312 Tone and pitch values In real speech the same tone feature can occur at diff erent pitch levels This is due to sev eral factors First H has a higher pitch with greater stress than it does with lesser stress Second the same tone will gradually descend in pitch owing to the downdrift effect this effect was rst observed in African languages but was later found in other languages too such as Eng lish Japanese Korean Chinese and German Pierrehumbert 1980 Bec Stateofthearticle to indicate an intonation either by a boundary tone or by changing the pitch level of an entire utterance In the rst way intonation and word tones are in a linear sequence Chao s successive addition English uses this method throughout but Chinese uses it only sometimes In the second way intonation and tone are superimposed on each other Chao s simultaneous addition English does not use this method but Chinese sometimes does As a result English tone and intonation can be represented by a single se quence of Hs and Ls whereas Chinese tone and intonation can be represented this way only sometimes This proposal essentially admits simultaneous addition at least in some languages and calls for a representation that goes beyond a single sequence of Hs and Ls For example one may assume two levels of representation a lower level of Hs and Ls for word tones and a higher level of Hs Ls rises falls etc for phrase and sentence intonation The second proposal attempts to preserve a single sequence of tones and account for their undulations in other ways A motivation for this proposal is that if H and 7H are values of a distinctive feature there ought to be just one sequence of them since there is just one se quence of values of any other distinctive feature It is worth noting that Chao 1933 mentioned only two cases of simultaneous addition in Chi nese raised level of pitch and lowered level of pitch pitch range widening and narrowing were found to be correlated to pitch raising and lower ing In He amp Jin 1992 six intonational mean ings were designed statement expecting con r Glot International Volume 2 Issue 4 April 1996 mation question simple request command and exclamation but again just two cases of simulta neous addition were observed raised or lowered If these are the only cases of simultaneous addi tion besides the neutral intonation a possible solution is as follows The raised level may have an accented nal boundary H which whether it is actually produced or not is targeted at a high er pitch level as any H with a main accent does and so may have prevented downstep in the pre ceding syllables Similarly the lowered level may have an accented nal boundary L which is tar geted at a lower pitch level and so may have accelerated downstep in the preceding syllables Of course this proposal is highly speculative Whether it is the correct solution for simulta neous addition will be left open 314 Functions of tone A basic function of tone is lexical distinc tiveness However tone can serve other func tions such as syntactic distinction and attitudi nal meaning For example in Igbo a oating H can indicate the associative structure noun of noun Williamsom 1986 In Chinese a bound ary H can indicate a question and a boundary L can indicate a protesting statement Chao 1933 In English attitudinal meanings can also be expressed by word tones pitchaccents and boundary tones eg Liberman 1975 Pierrehumbert 1980 The fact that tone has multiple functions is again not unique among distinctive features For example palatalization 7back high is lexical ly distinctive but can also indicate the mimetic structure in Japanese Mester amp Ito 1989 Page 8 4 Conclusions Research on tone in the past two decades has tremendously increased our understanding of the subject It also prompted developments in many other areas of phonology For example the rise of autosegmental phonology resulting from of tone research has brought about progress in such areas as harmony processes templatic morpholo gy syllable structure feature geometry and metrical phonology Works on tone domains have also propelled research on the syntaxiphonology interface For example Clements 1978 work on Ewe amp Chen s 1987 written in 1985 work on Xiamen have lead to the in uential endbased theory of Selkirk 1986 on the syntax to phonology mapping Works on tonogenesis and tone features call for a reconsideration of the traditional distinction between segmental and suprasegmental features Finally works on con tour tones have direct implications for the status of contour features in distinctive feature theory As seen in section 3 a number ofissues are still open and will concern researchers for some time In addition although I have not discussed the large body of phonetic literature on tone it is clear that tone is a leading area in which phonet ic and phonological researches come together This trend is likely to continue Acknowledgments Thanks to Pam Beddor Lesley Milroy James Myers Bonny Sands and Rint Sybesma for their comments by Luigi Burzio A REPLY TO PROF HALLE Continued from page 1 1 BLASPHeMOUS MiNORITY GENERaT IVE The shortenings of 1 show variability and occur if and only if the vowel loses the stress it had in the stem as in 1 which contrasts with 2 where the ca pitalized Vs maintain both stress and length the last two items being attested idiolectal variants 2 desIrous mlnority generAtive As I argued in ROT rules preclude a unitary account of 12 and the trisyllabic shortening cases Recapping the reasons metrical structure is needed to distinguish the cases in 2 that do not shorten from the the trisyllabic cases that do entailing the rule order stress shortening but shortening is needed for the correct assignment of stress in 1 versus 2 entailing the opposite rule order Con straints yield no comparable paradox because they work in parallel computing stress and shortening at the same time Prof Halle in effect concedes this argument since he accepts that the shortening rule of blas phemous in 1 must be different from that of the trisyllabic cases He proposes that the rule at work with blasphemous and i one must presume i the other cases in 1 is the same as the one giving buy bought which operates in the morphology and hence before stress There is no plausibility to that view however The two phenomena have nothing in com mon they involve different vowel shifts they are triggered by different af xes only one exhibits variability On the other hand the phenomenon in 1 has everything in common with the trisyllabic short ening except for the noted variability which the constraintbased account explains as discussed in ROT Note too that the shortening of 1 fairly system atically exempts bisyllabic structures like fAmous a restriction that has a straighforward metrical ac count suggesting it would be beside the point to order the relevant rule before stress Prof Halle insists on a solution to English vowel length allomorphy that even aside from the cases in 1 is amply superseded by the one given in my Principles of English Stress Burzio 1994 and re ferred to in ROT His solution includes a rule of shortening that singles out heads of binary feet and a rule of shortening that singles out stress wells beside the noted rule of morphological shortening with nothing in common with either of the other two To my way of thinking which incidentally owes a lot to him and my training at MIT there are far too many accidents in that combination Prof Halle is free to deem these convincing solutions but while they have been important stepping stones they provide no basis for rejecting the solution described in ROT The latter is superior because it reduces all accidents tojust one shortening under al xation the exact distribution of shortening following from the rest of the grammar It is of some interest that even that single accident can now be removed given Prince amp Smolensky s discussion of markedness and invento ries chapter 9 Since it is clear that long vowels are more marked than short ones ie crosslinguistical ly shortening in wordformation reduces to phono logical regulariza Stateofthearticle Aginsky E 1935 A grammar ofthe Mende language Language Dissertations 20 Phil adelphia Linguistic Society of America Anderson SR 1976 Nasal consonants and the internal structure of segments Language 52 3267344 Anderson SR 1978 Tone features In Tone A linguistic survey VA Fromkin ed 1337175 New York Academic Press Ao BXP 1992 Metrical constituents as do mains of tone sandhi in Nantong Chinese Paper presented at CLS 28 Chicago April 23 A0 BXP 1993 Phonetics and phonology of Nantong Chinese PhD dissertation Ohio State University Columbus Bao Z 1990 On the Nature ofTone PhD dissertation MIT Cambridge Mass Baxter W 1992 A handbook ofOld Chinese phonology Berlin Mouton de Gruyter Beach DM 1924 The science of tonetics and its application to Bantu languages Bantu Studies 2nd series vol 2 757106 Beckman M 1986 Stress and nonstress accent Dordrecht Foris Beckman M amp J Pierrehumbert 1986 Intona tion structure in Japanese and English Pho nology Yearbook 3 2557309 Bing J 1979 Aspects ofEnglish prosody PhD dissertation University of Massachusetts Amherst Distributed by the Indiana Univer sity Linguistics Club Bloomington Indiana Blevins J 1993 A tonal analysis of Lithuanian nominal accent Language 69 2377273 Bloch B 1946 Studies in colloquial Japanese II Syntax Language 22 2007248 Bolinger D 1985 Two views of accent Journal ofLinguistics 21 797123 Bolinger D 1986 Intonation and its parts Stanford Calif Stanford University Press Bruce G amp E Garding 1978 A prosodic typo logy for Swedish dialects In Nordic prosody papers from a symposium E Garding G Bruce amp R Bannert eds 2197228 Lund Department of Linguistics Lund University Cao JF amp I Maddieson 1992 An exploration of phonation types in Wu Dialects of Chinese Journal ofPhonetics 20 77792 Cassimjee F 1987 An autosegmental analysis of Venda tonology PhD dissertation Uni versity of Illinios UrbanaChampaigne Cassimjee F amp C Kisseberth 1992 The tono logy of depressor consonants Evidence from Mijikenda and Nguni Proceedings of the Eighteenth Annual Meeting of the Berkeley Linguistics Society February 18 1992 Special Session on the 39Iypology of Tone Lan guages LA BuszardWelcher J Evans D Peterson L Wee amp W Weigel eds 26740 Berkeley Calif Berkeley Linguistics Society Chan 1985 Fuzhou phonology A non linear analysis of tone and stress PhD dis sertation University of Washington Seatle Chan 1991 Contourtone spreading and tone sandhi in Danyang Chinese Pho nology 8 2377259 Chang K 1953 On the tone system of the MiaoYao languages Language 29 3747378 Chao YR 1930 A system oftone letters Le maitre phonetique 45 24727 Chao YR 1933 Tone and intonation in Chi nese Bulletin ofInstitute oinstory and Philology Academia Sinica 4 1217134 Chao YR 1980 Chinese tones and English stress In The melody oflanguage LR Waugh amp CH van Schooneveld eds 41744 Baltimore University Park Press Chen M 1987 The syntax oinamen tone sandhi Phonology Yearbook 4 1097149 Cheng CC 1968 English stress and Chinese tones in Chinese sentences Phonetica 18 77788 Chomsky N amp M Halle 1968 The sound pat tern of English New York Harper and Row Glot International Volume 2 Issue 4 April 1996 A Tone Bibliography Clark M 1990 The tonal system ofIgbo Dor drecht Foris Clements GN 1978 Tone and syntax in Ewe In Elements of tone stress and intonation DJ Napoli ed 21799 Washington DC Georgetown University Press Clements GN 1983 The hierarchical repre sentation of tone features In Current ap proaches to African linguistics Vol 1 LR Dihoff ed 1457176 Dordrecht Foris Clements GN amp K Ford 1979 Kikuyu tone shift and its synchronic consequences Lin guistic Inquiry 10 1797210 Clements G N amp J Goldsmith eds 1984 Autosegmental studies in Bantu tone Dor drecht Foris de Jong K amp J McDonough 1993 Tone and tonogenesis in Navajo UCLA Working Pa pers in Phonetics 84 1657182 Duanmu S 1990 A formal study ofsyllable tone stress and domain in Chinese languag es PhD dissertation MIT Cambridge Mass Duanmu S 1992 An autosegmental analysis of tone in four Tibetan languages Linguistics of the TibetoBurman Area 15 65791 Duanmu S 1993 Rime length stress and association domains Journal of East Asian Linguistics 2 1744 Duanmu S 1994 Against contour tone units Linguistic Inquiry 25 5557608 Duanmu S 1995 Metrical and tonal phonology of compounds in two Chinese dialects Lan guage 71 2257259 F ry C 1989 Prosodic and tonal structure of Standard German PhD dissertation Uni versity of Konstanz F ry C 1993 German intonational patterns Linguistische Arbeiten 285 Tubingen Max Niemeyer Firth JR 1957 Papers in linguistics 1931957 Oxford University Press Firth JR amp BB Rogers 1937 The structure of the Chinese monosyllable in a Hunanese dialect Changsha Bulletin of the School of Oriental Studies viii4 Reprinted in J R Firth 1957 76791 Fretheim T 1981 Nordic prosody II Trond heim Norway Tapir Fromkin VA 1972 Tone features and tone rules Studies in African Linguistics 3 47776 Fromkin VA ed 1978 ToneA Linguistic survey New York Academic Press Gandour J 1974 On the representation of tone in Siamese UCLA working papers in phonet ics 27 1187146 Department of Linguistics University of California Los Angeles Garding E 1979 Sentence intonation in Swed ish Phonetica 36 2077215 Goldsmith J 1976 Autosegmental phonology PhD dissertation MIT Cambridge Mass Reproduced by the Indiana University Lin guistics Club Bloomington Indiana Goldsmith J 1981 English as a tone language In Phonology in the 1980 s DL Goyvaerts ed 2877308 Ghent Belgium E Story Scientia Goldsmith J 1982 Accent systems In The structure of phonological representations Part I H van der Hulst amp N Smith eds 47763 Linguistic Models 2 Dordrecht Foris Greenberg S amp E Zee 1979 On the perception of contour tones In UCLA working papers in phonetics 45 150764 Department of Lin guistics University of California Los Angeles Gussenhoven C 1984 On the grammar and se mantics of sentence accents Dordrecht Foris Gussenhoven C 1988 Adequacy in intonation analysis the case of Dutch In Autosegmental studies on pitch accent H van der Hulst amp N Smith eds 957122 Linguistic Models 11 Dordrecht Foris Page 9 Halle M 1995 Feature geometry and feature spreading Linguistic Inquiry 26 1746 Halle M amp K Stevens 1971 A note on laryn geal features RLE Quarterly Progress Report 101 1987213 MIT Halle M amp J R Vergnaud 1980 Three dimen sional phonology Journal of Linguistic Re search 1 837105 Halle M amp JR Vergnaud 1987 A essay on stress Cambridge Mass MIT Press Haraguchi S 1977 The tone pattern ofJapa nese An autosegmental theory of tonology Tokyo Kaitakusha t Hart J amp R Collier 1975 Integrating differ ent levels of intonation analysis Journal of Phonetics 3 235755 Haudricourt AG 1954 De l origine des tons en Vietnamien Journal Asiatique 242 68782 Hayes B 1989 Compensatory lengthening in moraic phonology Linguistic Inquiry 20 2537306 Hayes B 1995 Metrical stress theory Princi ples and case studies Chicago The Universi ty of Chicago Press He Y amp Song J 1992 Beijinghua yudiao de shiyan tansuo Intonations of the Beijing Dialect An experimental exploration Yuyan Jiaoxue Yu Yanjiu 19922 71796 Herbert R K 1975 Reanalyzing prenasalized consonants Studies in African Linguistics 6 1057123 Hinton L 1991 An accentual analysis of tone in Chalcatongo Mixtec Papers from the American Indian Languages Conferences Held at the University of California Santa Cruz July and August 1991 JE Redden ed 173782 Occasional Papers on Linguistics 16 Carbondale Depar Stateofthe article Hyman L 1987 Prosodic domains in Kukuya Natural Language and Linguistic Theory 5 3117334 Hyman L 1989 Accent in Bantu An appraisal Studies in the Linguistic Sciences 19 1157134 Hyman L amp F Katamba 1993 A new approach to tone in Luganda Language 69 34767 Hyman L amp R Schuh 1974 Universals of tone rules Evidence from West Africa Linguistic Inquiry 5 817115 Inkelas S amp D Zec 1988 SerboCroatian pitch accent The interaction of tone stress and intonation Language 64 2277248 International Phonetic Assciation 1989 Report on the 1989 Kiel Convention Journal of the International Phonetic Association 19 67780 Jun SA 1990 The accentual pattern and prosody of the Chonnam dialect of Korean OSU Working Papers in Linguistics 38 1217140 Jun SA 1993 The phonetics and phonology of Korean prosody PhD dissertation Ohio State University Columbus Kager R 1993 Alternatives to the IambicTro chee Law Natural Language amp Linguistic Theory 11 3817432 Kenstovvicz M 1987 Tone and accent in Kizigua 7 a Bantu language In Certamen Phonologicum PM Bertinetto amp M Lopor caro eds 1777188 Torino Rosenberg amp Sellier Kenstovvicz M amp C Kisseberth 1990 Chizigula tonology The word and beyond In The pho nologyisyntax connection S Inkelas amp D Zec eds 1637194 Chicago The University ofChicago Press Kim M 1996 The tonal system ofaccentual languages PhD dissertation University of Chicago Kingston J amp D Solnit 1988 The tones ofcon sonants Ms University of Massachusetts Amherst and University of Michigan Ann Arber Kisseberth C amp F Cassimjee 1992 Tone and metrical structure in Shinjazidja Paper presented at CLS 28 Chicago Ladd DR 1980 The structure ofintonational meaning Evidence from English Bloomington Ind Indiana University Press Laughren M 1984 Tone in Zulu nouns In Autosegmental studies in Bantu tone G N Clements amp J Goldsmith eds 1837234 Dordrecht Foris Leben W 1971 Suprasegmental and segmental Representation of Tone Studies in African Linguistics supp 2 1837200 Leben W 1973 Suprasegmental phonology PhD dissertation MIT Cambridge Mass Liberman M 1975 The intonational system of English PhD dissertation MIT Cambridge Mass Lin MC JZ Yan amp GH Sun 1984 Beijinghua lianzizu zhengchang zhongyin de chubu shiyan Preliminary experiments on the normal stress in Beijing disyllables Fangyan 19841 57773 Lin T 1985 Tantao Beijing hua qingsheng xingzhi de chubu shiyan Preliminary experi ments in the exploration of the nature of Mandarin neutral tone Beijing yuyin shiyan lu Peking University working papers in phonetics T Lin amp L Wang eds 17 26 Peking Peking University Press McCarthy J amp A Prince 1986 Prosodic mor phology Ms University of Massachusetts Amherst and Brandeis University McCawley JD 1965 The accentual system of modern standard Japanese PhD disser tation MIT Cambridge Mass McCawley JD 1978 What is a tone language In Tone A linguistic survey VA Fromkin ed 1137131 New York Academic Press McLemore CA 1991 The pragmatic interpre tation of English intonation Sorority speech PhD dissertation University of Texas Austin Glot International Volume 2 Issue 4 April 1996 Maddieson I 1978 Universals of tone In Uni versals of human language Vol 2 Phonolo gy JH Greenberg ed 3357365 Stanford Calif Stanford University Press Maddieson I 1984 The effects on F0 ofa voic ing distinction in sonorants and their impli cations for a theory of tonogenesis Journal ofPhonetics 12 9715 Maran LR 1973 On becoming a tone language A Tibeto Burman model of tono genesis In Consonant types and tone L Hyman ed 977114 Southern California Occasional Papers in Linguistics 1 Depart ment of Linguistics University of Southern California Los Angeles Matisoff J A 1970 Glottal dissimilation and the Lahu high rising tone A tonogenetic casestudy Journal of the American Oriental Society 90 13744 Matisoff JA 1973 Tonogenesis in Southeast Asia Consonant types and tone Larry Hyman ed 71795 Southern California Occasional Papers in Linguistics 1 Depart ment of Linguistics University of Southern California Los Angeles Mester RA amp J Ito 1989 Feature predictabil ity and underspeci cation Palatal prosody in Japanese minetics Language 65 2587293 Myers S 1987 Tone and the structure of words in Shona PhD dissertation University of Massachusetts Amherst Newman P 1986 Contour tones as phonemic primes in Grebo In Phonological representa tions of suprasegmentals K Bogers H van der Hulst amp M Mous eds 1757193 Dor drecht Foris Odden D 1980 Associative tone in Shona Journal ofLinguistic Research 1 37751 Odden D 1981 Problems in tone assignment in Shona PhD dissertation University of Illi nois UrbanaChampaigne Odden D 1987 Kimatumbi phrasal phonology Phonology 4 13736 Odden D 1995 Tone African languages In The handbook of phonological theory J Goldsmith ed 4447475 Cambridge Mass Blackwell Pierrehumbert J 1980 The phonetics and phonology of English intonation PhD dis sertation MIT Cambridge Mass Pike K 1948 Tone languages Ann Arber Uni versity of Michigan Press Poser W 1984 The phonetics and phonology of tone and intonation in Japanese PhD dis sertation MIT Cambridge Mass Poser W 1990 Evidence for foot structure in Japanese Language 66 787105 Pulleyblank D 1983 Tone in lexical phonology PhD dissertation MIT Pulleyblank D 1986 Tone in lexical phonology Dordrecht Reidel Qu AT 1981 Zangyu de shengdiao ji qi fazhan The Tibetan tone and its development Yuyan Yanju 7 1777194 Redden JE ed 1991 Papers from the Ameri can Indian Languages Conferences held at the University of California Santa Cruz July and August 1991 Occasional Papers on Linguistics 16 Carbondale Department of Linguistics Southern Illinois University Sagey E 1986 The representation of features and relations in nonlinear phonology PhD dissertation MIT Cambridge Mass Selkirk E 1981 On prosodic structure and its relation to syntactic structure In Nordic Prosody II T Fretheim ed 1117140 Trondheim Norway Tapir Selkirk E 1995 Sentence prosody Intonation stress and phrasing In The handbook of phonological theory J Goldsmith ed 5507569 Cambridge Mass Blackwell Selkirk E amp T Shen 1990 Prosodic domains in Shanghai Chinese In The phonologyisyntax connection S Inkelas amp D Zec 3137337 CSLI Stanford University Stanford Calif Distribut ed by The University of Chicago Press Page 10 Shen XNS 1989 Interplay of the four citation tones and intonation in Mandarin Chinese Journal of Chinese Linguistics 17 61773 Shi E L Shi amp RR Liao 1987 An experimen tal analysis of the ve level tones of the Gaoba Dong language Journal of Chinese Linguistics 15 3357361 Shih CL 1986 The prosodic domain of tone sandhi in Chinese PhD dissertation Uni versity of California San Diego Shih CL 1992 Mandarin third tone sandhi and prosodic structure To appear in Studies in Chinese phonology J Wang amp N Smith eds Sietsema BM 1989 Metrical dependencies in tone assignment PhD dissertation MIT Cambridge Mass Stevens K amp SJ Keyser 1989 Primary fea tures and their enhancement in consonants Language 65 817106 Tsay J S 1991 Tone alternation in Taiwanese In Arizona Phonology Conference 4 76787 Department of Linguistics University of Arizona Tucson Arizona Wang J amp LJ Wang 1993 Puton Column Glot International Volume 2 Issue 4 April 1996 Elan Dresher The Untouchables RECENT ISSUES IN LINGUISTICS 7 It has been observed that history books are al ways somehow about the present This seems to be true with respect to histories oi linguistics at any rate which tend to re ect the concerns oi the authors time no matter what era is being discussed Because a work oi history is ostensir bly about another time however its relevance to the present may remain elusive to its contempor raries though its impact may be felt subliminally Sometimes one has to sleep to bringthese subliminal messages to the surface I discovered this in the course oi reading Ideology end Lmr gutsth TheoryJVuam Chomsky end the Deep Structure Debates by Geo rey Huck amp John Goldsmith 1995 London Routledge This book purports to be about events which occurred over twenty years ago but it is really about how line guistics should be done today In the words oi its dust cover the book presents a revisionist ace count oithe development ofideas about semanr tics in modern theories oilanguage iocusing particularly on Chomskys very public ri with the Generative Semanticists about the concept oiDeep Structurequot The standard account as Huck and Goldsmith characterize it p 2 sees Generative Semantics as an exemplary iailure a theory whose rapid dissolution was due to its havingbeen rationally rejected by the ling uisa tics communityquot Though Huck and Goldsmith claim to disagree with this story the second part oitheir book Chapter 4 and the Appendix actur ally supports it In Chapter 4 Huck and Goldsmith present a number oi reasons why Generative Semantics was not able to sustain a uni ed research pro gram in the 1970s Not least among these rea sons is that most oi its developers and support ers abandoned its central tenets and went on to other things The interviews at the back of the backoithe book with key gures in the debates con rm this Here is George Lako quot I m glad the Generative Semantics days are over Cog nia tive linguistics is far more interesting than gene erative linguistics and logic ever werequot And Paul Postal The bad thing is not that Generaa tive Semantics disappeared but that the other branch oitransiormational theory didn t disapr pearquot They and their colleagues and students presumably thought they had rational reasons ior abandoning Generative Semantics The book however is not really about what happened to Generative Semantics The rst half oithe book is devoted to demonstrating a slightly diHerent thesis which can be summed up as iollows Contrary to what some maythink Chomskys arguments against Generative Se mantics were not conclusive and did not show that his theory was superior In case after case Huck and Goldsmith attempt to show that Gena erative Semantics positions were misrepresent ed or misunderstood or could have been slightly modi ed to meet the objections raised against them Here it has to be said that Huck and Gold smith set the bar rather high in deciding whethe er an argument is conclusive or not For examr ple they suggest p28 that Chomsky iailed in quotRemarks on Nominalizationquot to show that the theory he was criticizing could not in principle account ior the data as elegantlyquot as his own approach Again they contend p 37 that Chomsky and others were incorrect to suggest that no restrictions in principle were imposed or could be imposed on the iorm and iunction oi global rulesquot It would be helpiul iiHuck and Goldsmith could give one example oi an arg ur ment in some empirical domain that meets this standard It appears to me that it is simply not possible except in rare cases to demonstrate in pzinciple that no satisiactory revision or devel opment oi a competing theory could ever be de vised Moreover in considering ways that a re search program could have developed we should take into account the ethos oithe particr ular research community On the matter oi globe al rules ior example Huck and Goldsmith presr ent no evidence that imposingconstraints on linguistic theory was on the Generative Semanr tics agenda Could they have tried to constrain their theory Sure they could have but it wouldn t have been Generative Semantics as we know it Anyway what does it matter ii Chomskys arguments against Generative Semantics were conclusive or not None oithe participants in this dispute seem interested in retrying the case now I have to coniess that I iound this part oi the bookto be rather slow going While trying to imagine how Georgia Green might have replied to Richard oehrle s 1977 critique oi her 1974 dissertation my eyelids began to get heavy I took o my glasses and turned the bookover When I woke upI saw ior the rst time the complete dust cover What had seemed to be an abstract design made up oivarious shades oi geyturned out to be words and a picture Along each margin oi the cover the word CHOMSKY is inscribed in massive gey blocklike letters Dominating the rest oithe cover is an ancient photo oi Chomsky magni ed to irightening dimensions The word Ideology is stamped in red under each eye Viewed against this back drop the title invites us to make an analogy ideologv is to linguistic theory as Noam Chomsky is to the deep structure debates And what is the relation between ideologv and line guistic theory Ideology especially ideology in red ink is somethingthat can distort linguistic theory ii it is allowed to The cover then ad vances a third subliminal thesis which is very relevant to the present Chomskys presence in linguistics has been too big and oppressive and is bad ior linguistic theory Ii Chomsky represents the negative ideal the positive alternative is also contained in the book In their Conclusion Huck and Goldsmith contrast the methodological values held by Chomsky and the Generative Semanticists For Chomskyquot they write no datum had status until it had been explained by theoryquot whereas Generative Semanticists thought that the Cons struction oi deeper theories could only proceed iithe theorist were prepared to take seriously the insights about data as yet not satisfactorily expla Dissertations Glot International Volume 2 Issue 4 April 1996 by Shuing Shyu THE SYNTAX OF FOCUS AND TOPIC IN MANDARIN CHINESE Reviewed by Rint Sybesma Summary by the author This dissertation investigates focus and topic constructions in Mandarin Chinese with particu lar emphasis on understanding their structural representation I argue for syntactic focus constit uent movement focalization and the existence of Focus Phrase in a strict preverbal position In addition I argue that basegenerated sentenceinitial NPs can be structurally distin guished from moved contrastive topics especial ly in complex clauses The former are similar to major subjects The syntactic analysis proposed here not only systematically accounts for the semantic interpretation of Zion dou ye even sentences but it also provides us with further insight in the grammatical symmetriesasymmetries between focus and topic 1 Introduction Structural focus involving overt focalization is manifested in sentences with lian dou ye even allalso as in 2a and in cases of object preposing without lion and don ye as exempli ed in 2b In this thesis I propose a uni ed account for both instances of object preposing 1 Zhangsan maile zheben shu Zhangsan buyLE thisCL book Zhangsan bought this book 2 a Zhangsan lian zheben shu donye maile Zhangsan LIAN thisCL book DOUYE buyLE Zhangsan bought even this book b Zhangsan zheben shu mai le Zhangsan thisCL book buyLE Zhangsan bought this book A lionphrase can also appear in sentence initial position 3 3 Lian zheben shu Zhangsan douye maile LIAN thisCL book Zhangsan DOUYE buyLE Even this book Zhangsan bought I show in this thesis that this sentence is struc turally ambiguous between basegeneration and movement the former denotes the whole sentence as the focus scope while in the latter the litm phrase is seen as a contrastive focalized topic 2 Focalization Let us rst concentrate on sentences involv ing focalization like 2 I propose that douye or a strong Focus feature heads the functional projection of Focus Phrase FP In focalization the lionobject is preposed to the strict preverbal FP because it is attracted by a projected lexical Focus head with a strong Focus formal feature that has to be checked in the sense of Chomsky 1995 The lionobject is moved and merges with F to form FP Adopting Chomsky s Bare Phrase Structure 1994 the speci er of a functional category is not projected unless some preposing derivation targets F otherwise F is labeled as a maximal projection 4 FP M NP F F MPAspP Focalization displays properties of Amovement First the object of idiom chunks V 70 compounds allowing prenominal modi ers can be moved 5 Mali zhanle Lisi de pianyi Mali takeLE Lisi DE advantage Mali took advantage of Lisi 6 Mali lian zhegdLisi de pianyi dou yao zhan Mali LIAN thisLisi DE advantage DOU want take Mali wants to take advantage even ofthisLisi Second it is sensitive to locality conditions eg complex NP in 8 7 Zhangsan taoyan INP Cp e2 kuajiang Mali1 de renZ Zhangsan dislike praise Mary DE person Zhangsan dislikes the person who praises Mary 8 Zhangsan lian Mali1 dou taoyan NP GP e2 kuajiang t1 de Zhangsan LIAN Mary DOU dislike praise DE renZ person Even for Malil Zhangsan dislikes the person who praises herl Focalization does not allow resumptive pronouns The ungrammaticality of 8 cannot be improved by adding to she as in 9 Moving an indirect object cannot leave a pronominal copy either 10 9 Zhangsan lian Mali1 dou taoyan NP GP e2 kuajiang tal Zhangsan LIAN Mary DOU dislike praise her de renZ DE person 10 Zhangsan lian Mali1 dou bu songgei tal shu Zhangsan LIAN Mali DOU not give her book Zhangsan doesn t give books even to Mali Focalization is clause bound compare 12 and 13 11 Zhangsan renwei Icp Lisi hen xihuan Mali Zhangsan think Lisi very like Mali 12 Zhangsan lian Malii dou renwei GP Lisi hen xihuan ti Zhangsan LIAN Mali DOU think Lisi very like 13 Zhangsan renwei ICP Lisi lian Mali1 dou bu xihuan t1 Zhangsan think Lisi LIAN Mali DOU not like Zhangsan thinks that Lisi doesn t like even Mali In addition focalization lacks binding reconstruc tion effects and remedies weak crossover WCO effects Page 12 The symmetries of bare object preposing as in 2b and lionfocalization structures are also studied in this thesis and it is argued that object preposing is also an instance of focalization in volving the same FP as in sentences with kian dou ye After examining previous analyses of object preposing in the literature I conclude that the current uni ed proposal not only incorporates their insights but also avoids their problems 3 Sentenceinitial lianphrases Sentence initial lionphrases in complex clauses are exempli ed in 14 where dou or ye occurs in the embedded clause and in 15 with dou ye in the matrix clause I argue that lian Mali in 14 is in sentenceinitial position as a result of movement whereas in 15 it is base generated 14 Lian Mali1 Zhangsan renwei CP Lisi donye hen xihuan t1 LIAN Mali Zhangsan think Lisi DOUYE very like 15 Lian Mali1 Zhangsan donye renwei CP Lisi hen LIAN Mali Zhangsan DOUYE think Lisi very xihuan ecl like 31 Moved Sinitial lianphrase as focalized contrastive topic Longdistance movement contrastive topics are directly moved to the matrix SpecTopicP 16 is equivalent to 14 Example 17 indicates that this movement is sensitive to whisland condition 16 TOM LianNPI GP 11 Subj GP t 1 Subj I don v mm 17 7quot Lian zheben shul Zhang xiang zhidao shei dou kan le t1 LIAN thisCL book Zhang Wonder who DOU read LE As is shown in 18 moved Sinitial lionphrases display binding reconstruction effects as well 18 Lian g39lanyu taziji12de Wenzhang3 Zhang1 renwei Lisi2 LIAN about heselfDE article Zhang think Lisi dou pipingguo t3 DOU criticizeGUO Even about his ownlZ article Zhangsan1 thinks Lisiz has made criticism 19 a Lian Zhangsanlde pengyou ta1 zhidao Wo dou LIAN ZhangsanDE friend he know I DOU changchang piping often criticize b Lian Zhangsanlde pengyou Wo zhidao ta1 dou LIAN ZhangsanDE friend I know he DOU changchang piping often criticize Furthermore they observe WCO effects 20 a LianNPI Subj V CP NPta1 dou V t1 b Z LianNP1 mutalul V GP don V t1 21 a Lian Zhangsan1 Mali renwei IcP piping ta1 de zheben LIAN Zhangsan Mali think criticize him DE thisCL shu dou huile t1 book BOOK destroyLE b 7quot Lian Zhangsan1 piping ta1 de nage n ren renwei LIAN Zhangsan criticize him DE thatCL woman think OP Mali dou xihuan t1 Mali DOU likes 32 IPadjoined basegenerated Sinitial lianphrases The structure of sentences with basegener ated lianNPs is represented in 22 it corre sponds to 15 Since the lionNP is basegenerat ed no derivation affects the merged FP hence no speci er position is projected Chomsky 1994 The lionphrase is licensed if dou cf Cheng 1991 or the Focus feature moving to adjoin to I0 at LF occupies a position in the same checking domain Chomsky 1995 Dissertations 22 IP lianNPI IP NP I Examples 23 and 24 allow a pronominal copy ta in the gaps cf 9 and 10 respectively which indicates that we are dealing with base genera tion 24 Lian Mali1 Zhangsan dou bu songgei ta1 shu LIAN Mali Zhangsan DOU not give her book When dou occurs in the matrix clause of25 cf 15 the occurrence of ta is still possible 25 Lian MALI1 Zhangsan dou renwei GP Lisi bu xihuan ta1 LIAN Mali Zhangsan all think Lisi not like her Sentences such as 15 do not observe WCO ef fects as is clear from 26 26 a LianNP1 Subj douV CP NP ta1 Vecl b LianNP1 NP ta1 douV CP SubjV ecl Given the above data a basegeneration analysis is called for It is also clear that there are impor tant differences between topics resulting from movement and basegenerated ones This is true for all NPs whether lianNP or not 33 Basegenerated IPadjoined position for major subject in double nominative sentences This thesis distinguishes major subject from topic position The ungrammaticality of 27 does not stem from the fact that yizhi daxiang an ele phant is a topic that has to be de nitegeneric Rather an inde nite subjectmajor subject in a root individual level predicate is not felicitous unless it is contrastively focused or denotes cardinality in Chinese cf Japanese in Kuroda 1986ab 27 Yizhi daxiang bizi hen chang oneCL elephant trunk very long Moreover the inde nite major subject can occur in embedded contexts 28 28 Ruguo yizhi daxiang bizi hen chang na yiding if oneCL elephant trunk very long then de nitely henkeai lovely My analysis predicts that no genuine topic ap pears in embedded contexts like relative or con ditional clauses for the embedding validity of the complement of bridge verbs see the discussion in Hoji 1985 p208 n24 and Kuroda 1986ab see also Hooper amp Thompson 1973 29 a 1P Chezi 1P wo songle kache gei Zhangsan car I giveLE truck to Zhangsan b CP gei Zhangsan1 1P chezi 1P wo songle kache t1 to Zhangsan car I giveLE truck 30 a Ruguo gei Zhangsan1 1P chezi 1P wo song kache t1 if to Zhangsan car I give truck Lisi hui chicu Lisi will bejealous If to Zhangsan car is such that I give atr39uck Lisi will be jealous b Lisi kanjian Np naliang CP gei Zhangsan1 1P chezi Lisi saw thatCL to Zhangsan car 1P wo song t1 tzll de kachez I give DE truck Lisi saw the truck that to Zhangsan car is such that I gave Glot International Volume 2 Issue 4 April 1996 34 Gap licensing The ec related to the base generated IPad joined lianNP is licensed if a rough aboutness relation can be established and the pronoun ta is its overt manifestation Directly moved topics from gaps to SpecTopic in root contexts do not allow genuine socalled resumptive pronouns in gap positions Consequently the proposed struc ture provides a systematic account for the long standing debate of nonmovement of topic struc ture and the identi cation of empty categories and overt pronominal copies 4 Summary In short this thesis proposes a number of structures to systematically account for the asymmetries of postsubject vs presubject lian phrases directly moved contrastive topics vs base generated Sinitial lianphrases and topic vs major subject It also compares topicalization with Japanese scrambling and suggests that scrambling is not attested in Chinese Moreover my proposed syntactic structures naturally dis ambiguate lian dou ye sentences in the light of even focus scope and association with focus Consequently this study lends further support to the match between syntax and semantics References Cheng LLS 1991 On the typology ofwh questions PhD dissertation MIT Chomsky N 1994 Bare phrase structure Ms MIT Chomsky N 1995 Categories and transforma tions Ms MIT Hoji H 1985 Logical Form constraints and con gurational structures in Japanese PhD dissertation University of Washington Hooper JB amp SA Thompson 1973 On the applicability of root transformations Lin guistic Inquiry 4 4657497 Kuroda SY 1972 The categorical and the thetic judgment Foundations of Language 9 1537185 Kuroda SY 1986a Movement of noun phrase in Japanese In Issues in Japanese linguis tics T Imai amp M Saito eds Dordrecht Foris Kuroda SY 1986b What happened after the movement of noun phrases in La J olla In Working Papers from the First SDF Work shop in Japanese Syntax UCSD Review by Rint Sybesma Introduction The dissertation under review is limited in scope 7 it basically deals with one type of construction the lian dou construction meaning even henceforth the evenconstruction 7 but it man ages to investigate virtually all aspects relevant to this construction in an admirable way I espe cially like chapter 4 in which we are presented with a very useful overview of many issues that touch upon the differences and similarities be tween subjects base generated topics and move ment topics with a very interesting batch of facts involving the possibility of having resumptive pronouns in some cases while they are barred in others These issues are discussed to the back ground of Kuroda s 1986 treatment of waNPs and gaNPs in Japanese and quite a number of problems are clari ed considerably if not solved All through the earlier chapters and especially chapter 3 ample attention had already been given to a comparison between focus movement and topicalization in Chinese In short this work would certainly deserve to be included in an up date of the bibliography of Henriette de Swart and Helen de Hoop s Stateofthe Article on topic and focus Glot International 17 1995 it offers a serious contribution to the discussion on topic Page 13 and focus Another good point of this dissertation is that it is very rich in data and most issues are spelled out with considerable patience I would have liked to see more discussion of the other construction which is always called the Focus Construction in Chinese ie sentences with shi be illustrated in 1 1 a shi wo maile zheiben shu be I buyLE thisCL book Iit s me who bought this book b wo shi maile zheiben shu I be buyLE thisCL book I did buy this book Although Shyu mentions this or actually a related construction twice pp4 10 in the rest of the work evensentences are treated as the only way in which Chinese handles focus and while reading I was constantly bothered by the question as to how Shyu would deal with shi sentences and how these sentences would t into Shyu s sentential tree More generally even may be one way of focusing but to me it does not seem to be a good idea to let it monopolize focus ing in one s analysis As is apparent from Shyu s summary the even construction in Chinese may be seen to consist of three elements lian dou or ye and movement As is also clear Shyu assumes a FocusP which may or may not be the right name and dou and ye serve as possible heads of this projec tion The movement of the object is explained by appealing to the minimalist need for feature checking it moves to the speci er of FocusP As to lian Shyu puts it on a par with adjectives in case a nominal element is focalized and an ad verb in the case of nonnominal phrases calling it an adjective is unfortunate but the idea that it is adjoined to the phrase it forms a constituent with is correct In what follows I discuss lian and the even semantics brie y and yedou as heads of FocusP I will not say anything about the move ment part of the analysis Shyu has many good reasons for claiming that we are dealing with A movement see her summary and I will n Dissertations this link we return to it below Shyu s claim then that dou ye heads the FocusP is reasonable as is the rest of the story viz that the focused NP moves into SpecFocusP to check features The FocusP To recapitulate Shyu s claim is there is a FocusP headed by dou ye and focused material is moved into the SpecFocusP That explains for instance why focused objects end up in preverbal position As I just suggested I think this is very rea sonable and the arguments that Shyu brings into play to support this claim are generally well taken However Shyu then turns the claim around in two ways and there she overplays her hand The rst instance of this is that part of her origi nal claim which says that FocusP is headed by an element dou leads to the conclusion that all instances of dou are heads of FocusP The second instance is that she turns around the part of the claim saying that evenfocused objects end up in a position between the subject and the verb such that she is forcing herself to conclude that all objects which end up between the subject and the verb must be treated in exactly the same way as she treats the preposing of objects in the even construction Douye As to muchdiscussed dou most recently Cheng 1995 see references there Shyu claims that dou in sentences like 4 below is essentially the same as the dou in evensentences like her 2a 4 a tamen dou zoule they all leaveLE they all lett b zheixie shu ta dou kanle thisCLPL book he all readLE these books he read them all Without going into details dou all in 4 quanti es over an NP that precedes it This NP must be plural In case of evensentences the NP preced ing dou does not have to be plural as is clear from Shyu s 2a above For convenience let s follow the terminology that Shyu borrows from Gao 1994 and refer to dou in evensentences as focalizer dou and to dou in 4 as quanti ca tional dou There have been several efforts to unify both dou s although others have pointed out that there are differences Tsai 1994 p26 for in stance after having discussed brie y instances of quanti cational dou introduces his short treat ment of focalizer dou by saying that it is a total ly different story He continues loc cit Here dou certainly does not quantify over the object rather there seems to be universal quanti ca tion over the contrast preestablished set impli cated by the semantics of lian even Gao 1994 is another example Shyu however wants to treat them alike p37f 43 ff claiming that the differences be tween them is caused by the difference between the NPs they quantify over in case of quanti cational dou we are dealing with referen tial NPs in case of focalizer dou we have QPs she argues extensively that NPs preceded by lian are universal Quanti erPs I would like to point out that there are a number of differences between the two dous which need to be taken into account when one wants to treat them as one and Shyu does either not take them into account or treats them unsat isfactorily First of all there is an important difference that l have not seen discussed any where but I haven t read all the literature on dou so this statement may not mean very much more than that I haven t read all the literature on dou which has to do with the pronunciation of dou or the intonation of the entire sentence Quanti cational dou like in 4 is fully pro Glot International Volume 2 Issue 4 April 1996 nounced with a full edged rst tone it may be seen to be slightly stressed Focalizer dou on the other hand is not stressed at all Stressing dou in Shyu s 2a above for instance leads to un grammaticality Here is one more illustration 1 use the notation DOU to indicate quanti ca tional dou because it is stressed focalizer dou is represented as dou 5 Zhang San zheixie shu douDOU maile Zhang San thisCLPL book DOUall buyLE dou Zhang San even bought these books DOU Zhang San bought all these books What 5 is meant to illustrate is that it makes a lot of difference whether dou is pronounced with some stress or without This must be accounted for I don t see how a reference to the difference between the NPs involved can account for this phonological difference The second difference between the dous is that while focalizer dou alternates with ye also quanti cational dou does not This must some how be accounted for if all dous are treated alike In addition like dou ye when used as a focalizer is unstressed it can be stressed in which case it means also 6 Zhang San zheixie shu yeYE maile Zhang San thisCLPL book YEalso buyLE ye Zhang San even bought these books YE Zhang San also bought these books The third difference is that the distribution is not entirely the same For instance quanti cational dou can be separated from the preposed object NP it quanti es over while this is impossible for focalizer dou 7 Zhang San huasheng yijing Zhang San peanuts already DOUall Zhang San already ate all the peanuts not Zhang San already even ate the peanuts douDOU chiwanle eat nishedLE The fourth comment I would like to make does not address a difference between the dous It has to do with the fact that both dous can cooccur in one sentence as noted by Gao 1994 as quoted by Shyu p46 with my conventions added 8 lian tamen dou meiyou DOU mai zheben shu LIAN they DOU nothave all buy thisCL book even they did not all buy this book Shyu s effort to disqualify 6 as a counter example to her claim that quanti cational dou and focalizer dou are the same is not convincing at all especially because she does not take into account that the dous in 6 are really pro nounced in a different way The fth and nal point addresses the in terpretation of dou and the role dou plays in the evenconstruction If it is true that dou is respon sible for the evensemantics it is not immediately clear how it would completely lose this element of its semantics when it gets to quantify over refer ential NPs In other words for these and other reasons I think that although Shyu s case for focalizer dou as the head of some FP is strong I do not think that quanti cational dou must be treated in exactly the same way and be viewed as heading the same FP At this point she pushes her analy sis too far Object preposing The second aspect of Shyu s analysis where I think she pushes it too far has to do with object preposing treated at length in Ernst amp Wang 1995 As we mentioned above lian is optional even without lian the sentence still conveys an eveninterpretation 7 at least as long as either dou or ye precede the VP because if they are gone too we lose the evensemantics see Shyu s 2b and 9 Page 14 9 wo aodaliya ganlanqiu bu mingbai 1 australian rugby not understand Australjan rules football I don t understand Nevertheless Shyu wants to make the case that all object preposing to a position between the subject and the VP involves her FocusP which in cases like 9 has a null head Her only argument is based on the similar distribution visavis certain adverbs and modals which is displayed by preposed objects in sentenc es with and those without dou ye However she does not account in any way for the fact that the evensemantics is replaced by a contrastive mea ning which is something which is associated with topicalization rather than with scalar even A second objection is more technical Above we saw that a phrase preceded by lian cannot occur with out ye dou and we concluded that this may be interpreted such that we only move when we have an overt head Apparently this is not true So how Books Glot International Volume 2 Issue 4 April 1996 by Alain Rouveret DECONSTRUCTING IRISH SYNTAX Review of Particles and projections in Irish syntax by Nigel Duffied 1 Overview As the title announces the primary aim of Nigel Duf eld s book is to isolate the functional structure of Irish clauses and noun phrases and to specify the position and the role of particles in this structure The author is also anxious to demonstrate that Irish presents no intractable problems for a restrictive theory of Universal Grammar in which parametric options are rath er highly constrained The theoretical frame work he adopts is a combination of the late Prin ciples and Parameters theory with the basic tenets of the Minimalist Program and Kayne s 1994 Antisymmetry hypothesis Building on his 1991 USC dissertation as well as on previous research by other scholars 7 McCloskey Sten son Noonan Guilfoyle and others 7 Duf eld is lead to develop sometimes entirely new proposals concerning the major syntactic structures of Irish nite clauses chapter 2 direct and affir mative and negative indirect relative clauses chapter 3 SOVX in nitival structures in Ulster Irish and the progressive construction chapter 4 and nally noun phrases in particular con struct state nominals chapter 5 The syntactic results he achieves are impressive But what makes the whole enterprise particularly interest ing is that at the same time Duf eld introduces a general claim about the phenomenon of muta tion in Irish the Initial Consonant Mutation hypothesis which provides a powerful tool to choose between competing syntactic analyses 2 The ICM hypothesis While the descriptive grammars of Modern Irish maintain the View that triggering mutation is an arbitrary lexical property of certain ele ments which they list exhaustiver Duf eld argues that mutation should be analyzed as a syntactically conditioned and constrained phe nomenon more precisely as a property of lexical ized functional heads This approach Views muta tion as the phonology of functional categories it is the syntactic context and the host category rather than the lexical item itself which is re sponsible for the obligatory mutation effect asso ciated to certain particles The Mutation Hypoth esis formulated in 1 goes beyond this general claim since it speci es which functional head triggers which effect 1 a A lexicalized C node triggers Eclipsis b A lexicalized T node triggers Lenition Once 1 is integrated into the grammar a deci sion about the functional structure of a given syntactic con guration or about the realization site of some particle will end up committing one to predictions about the presence and the nature of mutation effects in the same con guration The considerable interest of an approach based on 1 lies in the great predictive risks it takes in the expectations it generates about how facts in apparently independent domains will correlate Duf eld s research program is precisely to deter mine to which extent such expectations are borne out by the major syntactic structures of Irish The preverbal particles go positive complemen tizer an positive question particle complementizer nach negative question parti clecomplementizer 1N indirect relative com plementizer which are usually analyzed as ele ments appearing under C trigger eclipsis But others do not trigger the mutation that is expect ed from them according to 1 at least if follow ing standard practice they are analyzed as com plementizers Duf eld s research strategy in these cases consists in showing that the relevant elements are not realized in C but in a lower position or occupy such a position at some point in the derivation For example the various parti cles and complementizers in the past tense incor porating the r past morpheme gar or m or mir which uniformly trigger lenition on the following verb result either from a late Clower ing operation attaching the content of C to the past morpheme in T or 7 in the case of the sen tential negation Mar 7 from a headto head raising process adjoining the negative element to T The negative marker gan which appears in non nite clauses and small clauses and triggers no mutation is shown to remain in its original position the speci er of the NegP projection The leniting particle 1L introducing direct relative clauses is argued to be attached to T thus lower in the structure than the eclipsing particle 1N heading indirect relatives in C Duf eld also extends the syntactic treatment of mutation to noun phrases In the construct state structure the leniting effect of the head noun on any pos sessor to its right is traced back to the fact that it has raised to and lexicalized the functional head D which in nominal domains is the closest equi valent of T in clausal domains 2 capall mhac e in horse son Se n Se n s son s horse The picture is somewhat clouded however by the author s acknowledgment at a rather late point in the book chapter 5 that Irish mutation is not an homogeneous phenomenon Functional Mu tation must be sharply distinguished from Lexical Mutation a distinction which is ortho gonal to the lenitionversuseclipsis contrast Only the former can be characterized as syntac tic since it targets syntactic objects and its trig gers are sensitive to syntactic rather than to linear locality for example only syntactic mutation can give rise to the spreading effect illustrated in 2 We are told that all the cases of ICM triggered by grammatical particles in stantiate Fmutation which is fairly obvious in some cases much less in others In what follows I will concentrate on the syntactic proposals which Duf eld develops for a series of syntactic issues and puzzles that have attracted theoretical attention from generative linguists abstracting away from the fact that some of these proposals have been prompted by the desire to make Irish syntax compatible with 1 If the various mutation patterns can help to Page 15 choose between competing syntactic analyses when interpreted in the light of 1 the latter can also be used to validate 1 ie to decide to what extent it is correct I regret that in this cursory survey I had to ignore a range of interesting analyses touched on in this very rich book In particular I will say nothing about chapter 5 which presents a bril liant and utterly convincing account of the func tional organization of Irish noun phrases based in part on a comparative study of Irish Maltese and Hebrew construct states 3 The derivation of the VSOX order It is generally agreed that the verbinitial order in Celtic languages results from the move ment of the nite verb to the highest in ectional head rather than to C But there is no consensus concerning the necessarily lower speci er posi tion which the subject occupies Much recent work on Irish assumes that it is realized in a VP internal position at all Visible levels of represen tation cf Chung amp McCloskey 1987 Koopman amp Sportiche 1991 Guilfoyle 1990 1993 Duf eld 1991 An alternative View presented in Bobaljik amp Carnie 1992 and in Rouveret 1991 for Welsh and also defended on quite different grounds by McCloskey 1996 holds that the subject raises out of VP into the speci er position of one of the lower in ectional projections The claim that verb raising is to the highest in ectional head and the hypothesis of the short movement of the subject are both called into question by Duf eld Availing himself of the machinery of the minimalist pro gram and crucially relying on Kayne s 1994 Linearity proposal he develops a minimalist and antisymmetric analysis of the functional struc ture of Irish nite clauses in which a there is a landing site within VP for derived objects which is the speci er of a Books favor of this analysis is considered below It is clear that steps i7iii suffice to derive the verbinitial order found in nite clauses although curiously not much is said about why this order arises except that in Irish the V fea ture of AgrS is strong and its Nfeature weak It is for step iv 7 raising of the TP projection into the speci er of WP 7 that we need strong empir ical evidence The evidence comes from two syn tactic characteristics of Irish which in other perspectives could seem rather eccentric adver bial placement and the distribution of weak ob ject pronouns In Irish it is impossible to place any constituent including sentential and tempo ral adverbials between the complementizer and the nite verb But adverbials are also barred from intervening between a nite lexical verb and the subject NP 3 a D irt si gur inne chuir si an cheist air said she C yesterday put she the question tohim She said that she asked him the question yesterday b Chuala inne an fear an tamhran sin heard yesterday the man that song The man heard that song yesterday These two restrictions Duf eld takes as evidence for the raising of the entire verbal complex in cluding the subject NP to a leftperipheral posi tion If the entire TP domain not just the verb raises to SpecWP there is no way to derive the ungrammatical adverbmedial order in 3b pro vided that two additional assumptions are made i the relevant adverbs are necessarily adjoined to TP adjunction to WP is not available ii the speci er movement of TP targets the original TP category not the extended one the one which results from the adjunction of the adverb Interpreted in the light of Kayne s Anti symmetry proposal the distribution of weak pro nouns provides additional evidence in favor of raising to SpecWP Contrary to object noun phrases and contrastive or emphatic pronouns object and small clause subject weak pronouns usually appear at the right periphery of the clause ie more to the right than expected 4 a Leigh se go c ramach e read he carefully it He read it care llly b Chuala me ag ceol iad heard I PROG singVN them I heard them singing Given the ban on rightward adjunction the only remaining option Duf eld argues is rst to raise the pronoun to W and then move the residue of the clause to SpecWP According to him the pronouns in W occupy the same syntactic position at SpellOut as their Dutch and French counter parts This last claim I nd dubious whatever the situation in Dutch or in European Portu guese for that matter clitics in French are cer tainly not adjoined to a leftperipheral head at least in nite nonimperative clauses It seems to me that more information about Irish weak pro nouns would have been welcome Does their weakness make them clitic pronouns hence ele ments adjoinable to a head viz W Note that Welsh which shows the same restrictions on the distribution of adverbials as Irish does not instantiate the phenomenon of pronounpostposing it allows object pronouns to be realized either as independent elements in situ or as clitics inserted between complementi zers and clauseinitial verbs The idea that the verbinitial constituent moves into the speci er of WP provides a uni ed and highly elegant account of various phenomena which prima facie implicate the rightperiphery of the clause It is in fact dif cult to think of an alternative solution if both rightward movement of pronouns and rightward adjunction of adver bials are impossible But it is still an open ques tion whether these two operations can be entirely dispensed with or whether a theory that does like Kayne s 1994 one is too restrictive for an interesting discussion cf Manzini 1996 Glot International Volume 2 Issue 4 April 1996 Only one question does Duf eld s treatment leave unanswered If what moves in each case is pretty clear we would like to know what moti vates the movement It does not seem possible to characterize it as a de focusing strategy leav ing behind the focus of the sentence Reinhart 1995 and Nash 1995 both characterize some leftward movements along these lines because in 4 at least nothing is left behind the clause nal pronoun has been adjoined to W before the movement of TP to the speci er of WP Accord ing to Duf eld movement to SpecWP can be viewed as a type of generalized topicalization operation related to the type of VP topicalization to be found in German verbsecond contexts Related questions are Does a relation exist be tween step iv 7 raising of the verbinitial TP constituent to the speci er of WP 7 and step i 7 short movement of the nite verb to AgrS 7 which produces the verbinitial order Does TP raise into SpecWP only in verbinitial languages Although Duf eld rightly emphasizes the dan gers inherent in the assumption that there is a uniform account of VSO wordorder in Celtic still another question might be asked does TP raise into SpecWP in other verbinitial languag es Welsh for example shows the same restric tions on the distribution of adverbials as Irish but does not instantiate the phenomenon of pro nounpostposing More precisely it allows objects pronouns to be realized either as independent elements in situ or as clitics inserted between complementizers and clauseinitial verbs Irish allows neither option Suppose that adverbial distributions are taken as an indication that TP raises obligatorily to SpecWP The fact that independent pronouns can occupy the direct argument position can be reconciled with this analysis if one assumes that they are raised with the rest of the TPdomain not moved indepen dently On the contrary the existence of the C7Cl7V con gurations suggests that object clitics move to W but that the rest of the clause does not A situation which looks paradoxical if interpreted in the light of Duf eld s analysis 4 Negation complementizers and Tense The particles which function as markers of sentential negation in root declaratives or as negative complementizers appear in clause initial position and show a pastnonpast distinction as do other sentential markers and complementizers m nach are nonpast Mar and mir are past There is however a difference between senten tial negation and the negative complementizer the former triggers lenition on the following verb the latter eclipsis Duf eld rst suggests that two types of nite clauses should be distinguished unselected in particular matrix clauses which are WPs and selected embedded ones which are CPs Adopting Zanuttini s 1994 claim that the negative elements occurring in sentence initial position should be analyzed as heads he proposes that in negative matrix clauses in past tense the negative element originates in the head of a NegP projection intervening between TP and AgrSP while in embedded clauses it is directly inserted into C which renders the projec tion of NegP super uous In the rst case Neg raises to T and attaches to the past marker in the second the negative element and the past marker occupy different positions at Spellout C and T respectively Nevertheless the marker in T can take the form r in both cases since the nega tive element suffices to support it With respect to ICM this syntactic analysis has the desired property since it provides distinct syntactic char acterizations of the constructions in which nega tive particles give rise to distinct mutation ef fects we get eclipsis in the only situation in which C is lled and T empty ie in negative root clauses in the present tense Although appealing this account is not fully convincing It assigns quite different derivational origins to elements which look very similar the sentential negation and the negative complemen tizer in the past tense The former m or is Page 16 derived by a standard syntactic headraising operation the latter ndr by Books analyses which this approach leads to are origi nal and break new grounds The comments and questions I made above concerning some of the analytic choices made by Duf eld should not obscure the real appreciation I have for his work They certainly do not affect the general assump tion on which it is founded and as the interested reader can check by himself would require only minor adjustments of the speci c hypothesis it explores namely 1 References BenMamoun E 1989 Negation minimality and in ectional morphology Ms USC Bobaljik J amp A Carnie 1992 Aminimalist approach to some problems of Irish word order In Proceedings of the 12th Annual Harvard Celtic Colloquium Chung S amp J McCloskey 1987 Government barriers and small clauses in Modern Irish Linguistic Inquiry 18 1737237 Cottell S 1995 The representation of tense in Modern Irish GenGenP 32 Demirdash H 1989 Nominative NPs in Mod ern Standard Arabic Ms MIT Duf eld N 1991 Particles and projections PhD dissertation USC Guilfoyle E 1990 Functional categories and phrasestructure parameters PhD disserta tion McGill University Guilfoyle E 1993 Non nite clauses in Modern Irish and Old English In Proceedings of CLS Kayne R 1994 The antisymmetry ofsyntax Cambridge Mass MIT Press Koopman H amp D Sportiche 1991 On the posi tion of subjects Lingua 85 2117258 Koster J 1994 Predicate incorporation and the word order of Dutch In Paths towards uni versal grammar Studies in honor of Richard S Kayne G Cinque et al eds Washington DC Georgetown University Press McCloskey J 1992 On the scope of verbmove ment in Irish Ms University of California Santa Cruz McCloskey J 1996 Subjects and subject posi tions In The syntax of the Celtic languages R Borsley amp I Roberts eds Cambridge Cambridge University Press Manzini MR 1996 Adjuncts and the theory of phrase structure Ms University College London Muller G amp W Sternefeld 1992 Improper movement and unambiguous binding Lin guistic Inquiry 24 4617508 Nash L 1995 Portee argumentale et marquage casuel dans les langues SOV et dans les langues ergatives L exemple du georgien Doctoral dissertation Universit Paris8 Noonan M 1994 VP internal and VP external AgrOP Evidence from Irish In Proceedings of WCCFL 13 S Preuss ed Stanford CSLI Ramchand GC 1993 Aspect and argument structure in Modern Scottish Gaelic PhD dissertation Stanford University Reinhart T 1995 Interface strategies OTS University of Utrecht Rizzi L 1990 Relativized minimality Cam bridge Mass MIT Press Rizzi L 1995 The ne structure of the leftpe riphery Ms Universit de Geneve Rouveret A 1991 Functional categories and agreement The Linguistic Review 8 353387 Rouveret A 1994 Syntaxe du gallois Principes generaux et typologie Paris CNRSEditions Siloni T amp MA Friedemann to appear AgrObject is not Agrpamciple The Linguistic Review 14 Travis L 1991 Derived objects inner aspect and the structure of VP NELS 21 Zanuttini R 1994 Re examining negative clauses In Paths towards universal gram mar Studies in honor of Richard S Kayne G Cinque et al eds Washington DC Georgetown University Press Glot International Volume 2 Issue 4 April 1996 Page 18 by L a Nash GEORGIAN IN THE MIND Review of Georgian A structural reference by B C Hewitt Georgian A structural reference grammar aims to present the essential structural characteristics of Georgian for the bene t of general linguists It represents the sum total knowledge acquired by its author BG Hewitt over the last twenty years Indeed this long period of patient and extensive study of Georgian admirably shows itself through each of the 700 pages of the book Not only does the author succeed in lucid descrip tion of the general grammatical structure of the language but also in minute presentation of abso lutely every single departure or exception from the general rule As irregular morphemes and suppleted roots are extremely numerous in Geor gian and as none are ignored by the author who himself admits that he has endeavoured to give too much rather than too little information the book is de nitely not an easy read and will best reward a patient and absorbed scholar whose goal goes far beyond grabbing the essentials of the language through a quick scan Although the book has many merits I would like to stress two important aspects which make it particularly invaluable for a general linguist i in describing a given construction or morpheme the author presents previous linguistic analyses pertaining to the subject this strategy gains much more value as most of the discussed work is only avail able in Georgian ii throughout the entire study the description of each structure is accompanied with a wealth of examples carefully glossed and translated both from Modern and Old Georgian either taken from literary texts or made up by the author and subsequently checked with na tive speakers This fact brings to life many opaque constructions and gives the reader the feeling that he is dealing with a real pulsating language instead of with abstract and possible constructions Besides the introduction in which Hewitt presents some general facts about Georgia and Georgian as well as introduces his methodologi cal strategies the book comprises chapters on the soundsystem Ch 2 a chapter on nonverbal ie nominal pronominal postpositional adjectival adverbial etc morphology Ch3 a chapter on syntax Ch5 However true to the old tradition in Kartvelian philology HBG devoted most of the book the 400iodd pages that make up chapter 4 to the wonder ll complexitiy of the verbal mor phology The book also contains a very use il chapter of glossed texts from Contemporary Classic and Old Georgian Ch6 as well as sam ple vocabularies in semantic elds Ch7 a bibli ography and an index Although Hewitt does try to provide the fullest possible information about Georgian syntax a domain notoriously avoided by Georgian linguists his chapter on syntax is unfortunately not as strong and informative as the one on verbal morphology Yet as someone who has spent quite a few years on the study of Georgian syntax I can witness that its wonder ful complexity such as agreement patterns with main arguments the structure of headinitial and head nal relatives the interaction of word order and interpretation anaphoric binding nominal or subjunctive complementation to name just a few does not fall short from the wonderful complexity of verbal morphology and 77 so should merit the same respect In what follows I will not summarize or criti cally evaluate the entire content of the book Instead I shall invoke the particular treatment of certain wellknown or lesser known grammati cal phenomena which struck me as particularly challenging and thoughtprovoking for a general linguist and therefore requiring a further sepa rate investigation 1 Nonverbal morphology 11 In discussing noun plural formation Hewitt notes an interesting difference between Modern and Old Georgian The agglutinative strategy is used in Modern Georgian the plural form of a nominal contains the pluralizing suffix eb and the same Caseending as in singular e g Kacis manGEN 7 Kacebis man PL GEN In Old Georgian however the oblique Caseendings cannot be agglutinated to the plural morpheme n rather a portmanteau morph ta is employed which comprises information plural obl Case eg Kacta vs Kacnis manPL GEN The situation in Old Georgian is reminiscent of Slavic or Latin declensional sytems where plural infor mation and Case information are fused into one morpheme The possibility to agglutinate Case endings to the modern plural suffix eb might thus reveal an interesting property of this mo Books English where both prenominal and postnominal genitive NPs are available prenominal s geni tives may be stranded by gapping but post nominal of genitives may not eg Mary s father writes books and John s plays the piano vs The bombardment ofSarajer upset many people while of Baghdad was almost ignored In fact the phenomenon of double Case is reminiscent of another phenomenon described by Hewitt later in the chapter ln Georgian prenominal adjectives do not bear Caseendings however when these are unaccompanied by nouns and hence are used as reduced relatives the green one the one that is green they appear with Caseendings These facts taken together may suggest that prenominal genitive NPs in Georgian acquire their Case inherently and are not Casemarked structurally by government or Specihead agree ment by the head noun and therefore may fur ther acquire another Case structurally Alterna tively we may think of the genitive Case not as a Case per se but as a type of a predication marker signaling the fact that the two nominal elements the possessor and the head noun stand in the relation of predication for a similar idea concern ing English s see Kayne 1993 If the prenominal genitive is viewed in this particular light it be comes clear why genitive NPs and adjectives may share certain grammatical properties such as morphological compatibility to appear in gapped and reduced NPs with a Caseending that should have appeared instead on the missing head noun 13 In discussing inde nite adjectives Hewitt enumerates the inde nite suffixes me and gac which are added to interrogative pronouns to yield the corresponding inde nite adjectives some any For ease of exposition I will limit the discussion to so called animate inde nite adjectives What is the difference between these two morphemes The suffix me is de ned as creating nonspeci c inde nites whose identity is quite unknown whereas gac is used for specif ic inde nites yielding respectively Uinme Uin gac Now of course any linguist who has dealt with general aspects of inde niteness or non speci city is fully aware of the notorious difficul ties concerning the exact de nitions of speci c and nonspeci c inde nites The subsequent discussion in the section indicates that gac ad jectives are to be translated as some and me adjectives as any 2a b These translations seem to me to be more helpful than the some what murky speci cnonspeci c division At the same time it should be stressed that Uinme does not always function as the English any Uinme cannot be employed as a negative polarity item 2c Moreover in conditional clauses Uinme is preferred over Uigac 2d 2 a am k lassi vinmevigac moc apes tu this classin anysome pupilDAT if icnob 2 SUBJknowPRES Do you know any pupil in this class b vigacvinme moc ape gedzebda someany pupilNOM 20BJsearchIMPERF Some pupil was looking for you c me ar vicnob vinme mocapes I not ISUBJknowPRES any pupilDAT I do not know any pupil d vinmevigac moc apem tu dagirek os anysome pupilERG if 20BJcallSUBJ ras et q vi whatDAT ZSUBJsay If some pupil calls you what will you say to him 14 In the section on re exive pronouns the re exive tavi lit a noun meaning head is pre sented Hewitt mentions that this pronoun is either used by itself or in association with the appropriate possessive adjective The reader is left with the impression until the relevant sec tion in chapter 5 which will be discussed below that the two strategies are identical Unfortu nately Hewitt does not make it explicit what factors condition the use of tavi alone and wheth er the addition of the possessive adjective modi Glot International Volume 2 Issue 4 April 1996 es the interpretation In this respect it is inter esting to note that the use of possessive adjec tives is sometimes mandatory and sometimes unfelicitous much depending on the inherent re exive semantics of the verb 3a as well as on the thematic proximity of the re exive argument to the verb 3b c 3 a Vanom Wtavisi tavi moik la daixrco VanoERG selfGEN self killAOR drownAOR Vano killeddrowned himself b Nino tavis tavs seek itxa Nino NOM selfGEN selfDAT askAOR Nino asked herself a question c Ninom tavis tavze c aik itxa st at ia Nino ERG selfGEN selfon readAOR ar ticleNOM Nino read an article about herself 2 Verbal morphology 21 In the introductory section of the most im portant chapter in the book Hewitt presents an overview of the morphological complexities of the Georgian verb The following phenomena among others are dealt with a the fact that the verbal root may be augmented by various suffixes and pre xes some of which function as tenseagreement markers and others modify the verb semantics and valency b the notion of screeves tenseiaspectimood forms and the effect of moodiaspect distinctions on the morpho logical shape of the verb c the unpredictable character of valencymodifying morphemes some verbs which are formally ie morphologically trivalent are in fact bivalent and fail to lexicalize one of the two internal arguments One such verb is the trivalent scems strike which can appear only with the indirect object corresponding to the recipient of the strike the unrealized but under stood argument being the object used to hit the recipient with eg the hand d multiple agree ment Georgian verbs can agree not only with the subject but with the direct and indirect objects e pre xation of lexical aspectual preverbs in perfective tenses and the dif culty to predict despite their lexical meaning which aspectual preverbs will co occur with a given verbal root f numerous cases of root suppletion triggered by moodiaspect considerations or by animacy and number of the direct object 22 A remark is in order concerning Hewitt s treatment of agreement morphemes The seg mented forms of transitive verbs taking a 3rd person direct object systematically contain a null pre x glossed as the third person object agree ment marker Furthermore the author assumes that Georgian also instantiates the third person indirect object agreement marker which is almost always null but in rare cases is spelled out as the h or s pre x The order of agreement mor phemes is supposed to be subject markeriobject markeriindirect object marker It is a general fact of Georgian pre xal pronominal morphology that stacking of agreement pre xes is impossible only one pronominal morpheme may appear on the verb in the cases where the language disposes of two nonnull subject and object pre xes eg the 1st person subject agreement marker U and the 2nd person object agreement marker g Therefore Georgian dyadic or triadic verbal forms with only one pre xal pronominal slot cannot really provide any morphological evidence for any order of agreement affixes One might object to this claim as sequences of two agree ment morphemes seem to exist these are con structions containing the 1st person subject and a triadic verb with the so called third person indi rect object marker s eg U sc er I write to him But it is not at all clear whether s should be considered as a third person agreement marker As it never cooccurs with version vowels which license an indirect object in the clause see be low one might conversely argue that this com plementary distribution is not accidental s could be thus taken as a kind of a versionizer which Page 19 triggers the presence of a dative argument in the clause Although Georgian has nonnull rst and second person object agreement morphemes which should be more accurately analyzed as pronominal clitics corresponding to French me te clitics this fact does not guarantee in itself that Georgian also instantiates a null homologue of French le and of dative lui The entire grammar of Georgian clearly splits the rst and second person p Books presence vs absence of the plural indirect object in the clause or whether the plural indirect ob ject precedes or follows its ver p 137 Such a conclusion and the examples discussed in the section should challenge anyone who is interested in the investigation of syntactic and semantic correlates of plural agreement to take a closer look at the phenomenon at hand 24 In the extremely informative section on aspectual preverbs Hewitt raises an interesting issue concerning the optionality of aspectual preverbs on the aorist and on the non nite ver bal forms The presence of the aspectual preverb where it is possible 7 the unmarked option 7 expresses telicity of the action depicted lts ab sence however does not entail that the action is atelic but rather that the action although aspectually nondurative does not reach its natural conclusion Hewitt remarks that this generalization albeit accurate is only valid for nal forms and only for certain constructions which stress that the action depicted in the sen tence can not be accomplished easily and the agent has made several attempts to bring it to a successful end Furthermore the absence of the preverb can carry other more subtle meanings especially on the non nal or subjunctive forms selected by control verbs For example the ab sence of the preverb can be triggered by the in de niteness of the time when the whole action was carried out or by the nonconcrete nature of the selected direct object The following example illustrates this It is expected that the presence of the preverb on the subjunctive form is not felici tous in the universal gnomic statements such as 5 where actions due to the universal avor of the statement are not timespeci c 5 mq velam tavisi sakme unda ak etos everyoneERG selfGEN business must doSUBJNCT gaak etos PREVdOSUBJNCT Everyone should do his own business Unexpectedly the same preverbless ak etos can be encountered when referred to quite speci c business Hewitt provides an interesting explana tion based on the internal semantics of the direct object business this noun does not neccesarily describe one speci c act but can be comprised by a whole series of individual actions the ful ll ment of each one of which will not neccesarily result in the completion of the series overall hence the lack of the preverb At the same time Hewitt remarks that this type of explanation does not always hold there are minimal pairs where non nite forms are combined with con crete and speci c objects but the presence of the aspectual preverb is still nonobligatory eg daic q o daamtavra c ignis targmna gada targmna he started nished translatingPREV translating the book Hewitt does not provide an explanation for these minimal pairs and indeed the explanation is dif cult to provide Another intriguing point brought up in the book is that aspectual preverbs show the abovedescribed uctuations only on transitive verbs with intran sitive verbs they are almost always obligatory whenever possible In conclusion the author succeeds in showing that aspectual preverbs display quite complex properties and merit fur ther extensive research which might contribute to the general understanding of the interaction of aspect with niteness valency and speci city 25 The phenomenon of Version constitutes one of the touchstones of Georgian verbal morpholo gy Hewitt de nes Version as a system whereby the language employs a range of pre xal vowels to indicate certain types of relation holding be tween the subject and the object or between the direct and indirect objects of the verb Basically three types of Version are distinguished which modify the number of and relation between the main arguments of the verb Subjective Objec tive and Locative When the Subjective Version Glot International Volume 2 Issue 4 April 1996 vowel i appears on the verb the meaning of the verb is modi ed7the subject is understood to be acting upon himself or in his own interest In a way then the Subjective Version bears striking resemblance to the lndoEuropean middle voice as has often been pointed out by Georgian gram marians In the section on the Subjective Version vowel Hewitt provides interesting remarks on the relation between this morpheme and direct objects depicting bodyparts He notes that verbs which describe the performance of some act on some part of the subject s body will naturally take the Subjective Version vowel The relevant 77 u expressions are clear out one s ears sprain one s wrist shave one s moustache combcut one s hair etc However not all expressions where the direct object is a bodypart trigger the af xation of the Subjective Version on the verb for example the verb in Georgian equivalents of poke one s nose in someone s business or openclose one s eyes occurs without i Notice that the same expressions in French do not re quire the appearance of the re exive clitic se The above asymmetry raises an interesting question about factors governing the distribution of the subjectivere exive morphemes on verbs select ing bodyparts as direct objects One possible answer might be sought in lexical semantics of the verbs involved it could be the case that verbs describing involuntary or quite natural actions implying bodyparts do not require the re exive morpheme because such actions when performed one someone else s body will not be conceived in the same way as when performed on one s own body To make the point clearer when one closes one s eyes one does not perform some externally detectable action in order to achieve the result whereas when one shaves one s moustache the action performed and the utensils used in order to achieve the result is identical to shaving some one else s moustache Therefore the presence of the re exive morpheme in case of shaving modi es the meaning of the the expression by restrict ing the transitive action to the re exive act whereas in case of closing eyes the re exive morpheme seems super uous in the inherently re exive expression 26 The Objective Version is sensitive to the person speci cation of the indirect object and occurs as i homophonous and probably structur ally identical with the Subjective Version vowel with 1st and 2nd person indirect objects and as u with third person pronouns and full NP indirect objects Hewitt de nes it as the morpheme indi cating that the direct object either belongs to or is designated for an indirect object Such a de ni tion implies that the indirect object cooccuring with the Objective Version is perceived either as a benefactor or a possessor of the direct object However Hewitt discusses some data which pose a problem to this de ntion Certain control verbs selecting non nite verbal complements can sometimes bear the Objective Version vowel which triggers the dative Case on the direct object of the non nite gerundival form thus formally transforming it into the indirect object of the main verb Without there being the Ver sion vowel on the main verb the direct object of its non nite complememnt would have borne the genitive Case The following pair illustrates this transformation 6 a k acma daic q o unagiris mt vreva manERG PREVbeginAOR saddleGEN breakNONF b k acma unagirs dauc q o manERG saddleDAT PREVOBJVERSbeginAOR mt vreva breakNONF The man started breaking the saddle As a result of the transformation the indirect object is understood as pro ting being affected by or as a possessor of the process described by the non nite form It is important to note that such a transformation is severely restricted in Georgian For example if saddle is replaced by pencil in 6 the Objective Version cannot occur Page 20 on the control verb begin Furthermore not every control verb is a good candidate for such a transformation In fact Hewitt enumerates only Goodies Glot International Volume 2 Issue 4 April 1996 by Donald A Becker HISTORICAL LINGUISTICS AS A HACKER S PARADISE Review of PHONO 32 by Lee Hartman If we were playing a word association game with a colleague who spontaneously responded to the prompt computer with the term Lautgesetz our rst thought might be that he or she had been working too hard and had lost touch with reality for the notion of sound law is rooted in a scholarly tradition established in the nineteenth century long before the dawn of the computer age Yet if we paused to ponder the historical accounts of the development of the phonology and morphology of various languages that are enshrined in the many handbooks we have in herited from that bygone era we could not fail to notice the striking resemblance that such works bear to computer programs 7 programs whose input consists of the reconstructed forms of a proto language and whose subroutines or functions are the chronologically ordered diachronic rules which map that input step by step into the desired out put the attested forms of a particular daughter language Despite this resemblance however most historical linguists are even today obliged to run their programs in their heads rather than on the personal computers that virtually all of them have on their desks as de facto dedicated word proces sors for 7 having been trained in a traditional philological mode 7 they lack the experience in computer programming needed to convert their historical descriptions into executable code This is truly a pity for lacking computer models with which to test the rigor of their diachronic descrip tions these linguists can easily overlook inaccura cies and inconsistencies in the formulations and chronologies they have posited for various sound changes Worse yet their audience is deprived of the possibility of seeing their analyses in ac tion and must be content instead to read about them The scholars and students who constitute this audience ought to feel no less disappointed than a software buyer who upon removing the shrinkwrap from a newly purchased program nds that the box contains a printout of the program s source code but no diskettes Lee Hartman of the University of Southern Illinois has attempted to rectify this lamentable situation by creating a software package enti tled PHONO that frees linguists wishing to run their historical descriptions on an MSDOS com puter from the necessity of rst learning to pro gram and he has generously made it available free of charge to all interested persons see page 2 The centerpiece of Hartman s package is the PHONOEXE program which permits the user to create new models of regular historical sound change and to run existing ones such as those for Spanish and Pig Latin that are distributed with it The Spanish model is actually extensive enough that some students of Romance philology may wish to acquire PHONO solely in order to obtain it see page 2 Hartman assumes that each model created with PHONO will consist of the following three components 1 a redundancyfree classi catory matrix in which up to 70 segments are assigned dis tinctive and redundant speci cations for as many as 23 binary features of the user s choice 2 a set of contextsensitive historical rules that operates on segments composed of distinctive feature bundles and 3 a statement of the linear order in which the rules are to be applied Some rules may be designated as persistent ie as applying whenever another rule creates an environ ment that meets their structural descrip tion Persistent rules that capture segment internal redundancies make it possible to simplify the formulation of the ordered rules by obviating the necessity of including re dundant feature speci cations in their structural changes In short each model closely resembles a syn chronic analysis written by a generative phonol ogist working in the late 1960s and early 1970s Hartman justi es his choice of this descriptive framework entirely on the basis of its precision and versatility and cautions PHONO users not to read into this choice any theoretical claims on his part concerning the psychological reality of such descriptions This disclaimer notwithstand ing users familiar with classical generative phonology will feel quite at home with this framework and PHONO could be put to good use in classes in phonological theory in which stu dents are given a chance to try their hand at writing descriptions conforming to this model These three components are transformed into a set of three ASCII les with names ending in the extensions ALF MAK and ORD with the aid of the alphabet editor the makeup edi tor and the order editor which are accessed from PHONO s runedit menu Hartman has clearly exercised considerable ingenuity in creating a makeup editor exible enough to handle the full range of rule types encountered in classical gen erative phonology Although the use of this edi tor is explained adequately in the PHONO manu al found in the README le Hartman s writ ing style is succinct and most PHONO users will bene t from studying the form of the numerous rules contained in the MAK le of his Spanish model These three les are all that is required to run a model for a given language in PHONO s socalled Interactive mode This mode allows the user to enter ancestor words from the key board for which the computer then generates multiline diachronic derivations of their descen dants in the language under study Whereas the interactive mode is the only one of immediate interest to the individual who has acquired PHO NO as a means of studying the historical devel opment of a particular language with the aid of an already existing model Hartman has also Page 2 2 provided for a batch mode which automates the task of testing the integrity ofa model under development Running in its batch mode PHONO checks the accuracy with which the model con verts a set of representative inputs into their desired outputs This set of paired inputs and outputs must be listed in a fourth type of data le identi ed with a name ending in the exten sion PAR that the model developer has created with a word processor Frequent batch runs make it possible to discover and eliminate newly introduced bugs before their pernicious effects are compounded by additional ones The PAR le for Hartman s Spanish SPANIPAR model will serve as a handy reference for those who are interested in using his Spanish model in the interactive mode but who are a bit shaky in their knowledge of the history of the Spanish language These users would be well advised to print out this le with the DOS command copy SPAN LPAR lpt1 They will presumably also want to consult the two printed works on which the model is largely based Otero CP 1971 E voluci n y revoluci n en romance Barcelona Hartman SL 1974 An outline of Spanish historical phonology Papers in Linguistics 7 1237191 Although it seems a bit churlish to nd fault with a valuable piece of software that is distrib uted free of charge I would be remiss if I failed to mention that PHONO is marred by what many users will regard as a serious aw This is PHONO s inability to display any symbol not in cluded in the DOS extended character set Lin guists who a decade ago exchanged their DOS computers for Apple Macintoshes in order to avoid being limited in their computing to fewer than 255 symbols will hardly be eager to turn back the clock and to make do once again with this highly restricted transcriptional arsenal In fairness it should be noted that Hartman has devised an ingenious workaround that allows unavailable characters to be represented on the computer s screen as highlighted versions of other Conferences Glot International Volume 2 Issue 4 April 1996 GLOW 19 by Marcel den Dikken by David Adger CONFERENCE REPORTS International Conference on Syntactic Categories GLOW 19 Athens 17719 April 1996 Linguists like extremes it seems Alter last year s visit to the world s northernmost university Tromso last April the GLOW circus moved to the southernmost venue in its history the Uni versity of Athens The wellorganized conference with Asymmetry as its theme was held at the Divani Palace Hotel close to the Acropolis 7 an excellent base for sightseeing trips but the qual ity of the program ensured high attendance throughout the conference In this report I will highlight some of the papers which I think deserve special attention which is not to say of course that the others were uninteresting Viviane Deprez Rutgers addressed subjectobject asymmetries in the domain of the distribution of negation words also noting the parallel between these and the ones found in the distribution of bare nominal phrases in the Ro mance languages The core of her proposal rested on the pair of claims that i SpecDP and D may not be simultaneously lled and ii an empty Dhead must be properly licensed Languages may choose one of three ways of licensing D Nraising to D or Spec7Head agreement with an NP raised to SpecDP or somewhat unfortunately given current theoretical developments government The different ways of licensing D were shown to correlate with different interpretations of nega tion words and different negative concord proper ties The proposal was illustrated extensively with data from a wide variety of Romance languages and Romancebased Haitian Creole being extend ed to the local language Greek towards the end of the presentation Alter Gisbert Fanselow Potsdam had revisit ed scrambling from a minimalist perspective and had come to the conclusion that it can only be accurately analyzed in terms of basegeneration this chestnut again gured prominently in the excellent paper by Artemis Alexiadou Berlin and Elena Anagnostopoulou Tilburg They extended the results of their earlier work on subject licens ing and the checking of the EPP feature pre sented at CGSW 11 at Rutgers University see the conference report in GLOT 1 910 into the realm of objects The central claim of the objects sec tion of their paper was that clitic doubling is the counterpart of scrambling something which they gave shape in a way that perfectly mimicked their story about subjects In the case of both ArgS and AgrO feature checking can take place either via NPmovement to SpecAgrP or via headmovement to Agr Just like there is parametric variation between languages featuring subject movement to SpecAgrSP and languages resorting to V movement to AgrS we also nd a dichotomy between languages with scrambling NPmove ment to SpecAgrOP and clitic doubling clitic movement to AgrO The paper constituted a strong plea for an Agrbased system hence against Chapter 4 section 10 especially in view of the observed correlation between the ways of subject licensing and object licensing Caroline Heycock Edinburgh presented the results of ongoing research conducted together with Anthony Kroch Pennsylvania on connectiv ity effects in pseudoclelt constructions In stark contrast to the strong pleas for predicate inversion in the GLOW papers by Bennis Corver amp Den Dikken and Moro they concluded on the basis of an impressive body of facts that there can be no such thing as predicate raising to subject position Instead they analyzed speci cational inverse sentences as equatives and went on to accommo date the connectivity effects found in pseudoclelts by transferring the binding conditions onto an interpretive component forming the interface be tween LF and discourse models On the second day Gisbert Fanselow made his second appearance this time seconded by Anoop Mahajan UCLA Their paper addressed the intricacies of partial whmovement and whcopying phenomena in German eg was glaubst du wen Hans gesehen hat what believe you who Hans seen has and wen glaubst du wen Hans gesehen hat who believe you who Hans seen has resp and Hindi The focus was on partial whmovement for which the authors argued that the scope marker was is an expletive element associated with the entire embedded clause 7 in effect the whcounterpart of the CPrelated exple tive es They opted for an expletive replacement approach according to which the CP associate raises to the whexpletive in SpecCP at LF One of the interesting consequences of Chomsky s Chapter 4 is that Merge can in princi ple add structure to the root alter spellout Phil Branigan Newfoundland exploited this conse quence of stateoftheart minimalism in an ac count of the difference between root and embed ded whquestions with respect to verb fronting in Germanic He argued that while embedded claus es can only feature whmovement to the Spec of a wh CP root whclauses involve movement of the whphrase to the Spec of a Topic phrase the wh CP being added to the root in covert syntax By proposing a restriction on the application of intermediate trace deletion Branigan moreover nessed the results of Muller amp Sternefeld s Prin ciple of Unambiguous Binding Kayne s Linear Correspondence Axiom severe ly limits the amount of phrasestructural variation in language and thus works as a restrictor on initial representations But as Andrea Moro Mi lan argued the LCA can also be used as a trigger for movement something which he dubbed dy namic antisymmetry Suppose we have a struc ture of the type K M P where M and P are both maximal projections This structure is ruled out on account of the fact that there is a mutual ccom mand relationship between the two constituents of K For Kayne this automatically leads to rejection of the structure But Moro suggested that we al low this and other kinds of initially illformed syn tactic structure and resolve the antisymmetry problem by having movement create a derived structure which is in conformity with anti symmetry He illustrated his proposal on the basis of two sets of facts copular inversion and English root whmovement On the last day of the conference Elena Anagnostopoulou took the oor once more this time taking turns with Martin Everaert Utrecht Page 23 in a talk on asymmetries in binding in particular between thematic and con gurational constraints on binding They showed that while English bind ing can be accounted for on the basis of a con gu rational approach to binding the facts of Greek fall out only from an analysis in terms of Brole hierarchies This difference is related to the in ternal syntactic structure of the anaphoric ele ments of the languages concerned 7 the Greek anaphor undergoes incorporation the English one does not Incorporation is an alternative to chain formation as a way of re exivemarking a predi cate in the sense of Reinhart amp Reuland s binding theory The generalization that now emerges is that in English where chain formation applies binding is not sensitive to thematic prominence while in Greek where there is no chain formation but re exivemarking via incorporation themat ic prominence is observed In an intriguing but unfortunately not easy tofollow presentation Lea Nash and Alain Rouveret Paris proposed a new approach to checking theory in which all formal features whether strong or weak must be checked in overt syntax The central idea of their talk was that a functional head can be involved in precisely one checking relationship in its checking domain and that whenever a feature of F cannot be checked in F s own checking domain an additional functional projection of what they called a proxy head is erected onto which the feature still to be checked is copied t Conferences phrasal complement is dealt with by assuming that it is de cient in its status as a phrase an assumption defended for a range of other construc tions in French by Anne Abeill and Danielle Godard Universit de Paris VII TALANA and CNRS They showed that there was a need for a separate minigrammar to regulate the distribu tion of phrasally de cient categories This is a direct challenge to Bare Phrase Structure type theories which de ne phrasal status relationally Andreas Kathol GroningenBerkeley exam ined agreeing complementizers and verb second in Germanic arguing that there was a need to distin guish between two types of category the familiar categories with syntactic and semantic features and the Pcategory which has positional features Positional features determine the placement of elements within an ordering domain mimicking the effects of V toC movement and are crucially accessible to the morphological mechanisms deter mining the form of agreement morphology 7 which is why fronted verbs whphrases and complementizers all appear to display agreement On a broader theme within HPSG Carl Pollard Ohio State argued that phrase structure itself was purely epiphenomenal He showed that the mechanisms available within HPSG allow apparently hierarchical generalizations to be read off the dependencies independently set up between the heads in a structure thus doing away with the daughters attribute in HPSG The result ing framework is akin to a more sophisticated version of the Uni cation Categorial Grammar developed by Zeevat Klein and Calder in the late eighties and bears affinities to Word Grammar Dick Hudson UCL taking the minimaliza tion theme further argued that there was no evi dence in English to posit separate categories for D and C Turning Postal s insight on its head he argued that determiners were simply a subclass of pronoun which are speci ed as taking a noun com plement He then argued that complementizers are syncategorematically introduced into the grammar so that that if for etc do not share a category at all The general conclusion is that within dependency grammar there is no need to posit any class of functional categories Diametrically opposed to this view was Ronnie Cann Edinburgh who argued mainly on psycholinguistic grounds that two separate lexica exist 7 one for functional and one for conceptual elements The latter lexicon enters into an essen tially categorial syntax structured by central cog nitive processes such as inference while the for mer partakes of an encapsulated modular UG Pursuing the theme that inferential processes are crucial for understanding apparently grammatical phenomena Ruth Kempson SOAS argued for a strictly incremental analysis of WHextraction and crossover where the incorpo ration of new information word by word into the interpretation process is constrained by an infer ential proof system WHfeatures are reinterpret ed as instructions to recover information in a goal directed proof rather than being generalizations over classes of phrases Kempson showed how the range of variation found in the world s languages was exactly what was expected given the limits of the proofsystem Another theme that emerged during the con ference was how the grammar should deal with cases of squishiness is there a way of maintaining a discrete computational system in the face of phenomena that seemingly call for con tinuumlike generalizations Fritz Newmeyer Se attle strongly argued that there is and showed how a range of phenomena that have lead func tionalist grammarians to propose that cognitive messiness pervades deep into core grammatical processes are actually best analyzed in terms of an autonomous discrete syntax interacting with independently needed pragmatic principles This technique of dissolving the need for squishiness in the grammar by modularization was carried to its logical conclusion by Jerry Glot International Volume 2 Issue 4 April 1996 Sadock Chicago who proposed that the grammar consists of a large number of independent hierar chies whose interaction results in the empirical phenomena Sadock made a strong plea for the efficacy of lowlevel generalizations comparing much current theorizing with looking for quarks from the point of view of 15th century chemistry This same theme was explored by Bob Borsley Bangor and J aklin Korn lt Syracuse who showed that gerundlike phenomena were perva sive in the world s languages They argued for a relaxation of Grimshaw s system of extended pro jections so that the projections of nominal func tional categories can dominate verbal lexical ones but not vice versa Jane Grimshaw Rutgers was unfortunately unable to make the conference but her paper was read by Eric Bakovic Rutgers This paper argued that the unexpected alternation of the forms of clitics in certain combinations and its variation across Romance languages is the result of competing morphotactic constraints The unex pected clitic form arises because the expected form violates some more highly ranked constraint Rita Manzini UCLFlorence and Anna Roussou Bangor examined whether empty cate gories were motivated within Minimalism and argued that the phenomena usually dealt with by PRO and NPtrace were best seen as the result of chain formation between a Casechecking head and a thetaassigning aspectual head Minimal Distance Principle violations and Superraising are uni ed as violations of Relativized Minimality They conclude that there is no need to posit empty categories in a Minimalist Grammar The nal three papers all addressed checking theory Anders Holmberg Tromso argued that the Extended Projection Principle is essentially a phonological requirement rather than one on cate Page 24 gorial features and that purely phonological fea tures may move to satisfy it thus accounting for the phenomenon of stylistic fronting in Icelandic A consequence of this theory is that expletives consist of purely phonological features deriving Chomsky s conclusion that they check neither Case nor Categorial features David Adger York also argued that greater attention should be paid to the PF wing of the grammar His paper showed that agreement argument placement and adja cency effects in Irish and Scottish Gaelic could all be collapsed under the assumption that syntactic ally distinct heads could be merged in the morpho logical component and that this eliminates any strong feature on either head Checking theory becomes then only one way to eliminate strong features In the nal paper of the conference Ian Roberts Bangor went further and argued for the elimination of Checking theory entirely He pro posed that head attraction is a result of the affixal nature of the higher head and that XP chains may be de ned in terms of the heads that the XP is related to These heads may choose to head a trivial or a nontrivial chain in the latter case the head attracts the XP Roberts then explores the consequences of this position for head placement and agreement effects In addition to these papers there was also a demonstration of an interactive syntax tutor by Ana von Klopp and Chris Dalton Bangor which uses principles of internet design to teach simple ruletree correspondences This is available on httpwwwlinguisticsbangoracukjavaltLingTu torhtml On the whole this was an extremely valuable and wellorganized event It is to be hoped that other potential conference organizers will be simi larly brave The proceedings are expected to be published as part of the Syntax and Semantics series Academic Press BOOK NOTICES AseBerit and Rolf Strandskogen Norwegian An essential grammar Routledge Essential Grammars 1994 First published in Norwegian 1986 Translated by Barbara White London Routledge 202 pp with index ISBN 0415109795 Price 1299
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'