University of Msschusetts - Amherst ScholrWorks@UMss Amherst Lgutics Deprtment Fculty Publiction Series Lgutics 2014 Modl Comprons: Two Dilettntes Serch of n Expert Ki von Ftel Angelik Krtzer University of Msschusetts Amherst Follow th nd dditionl works t: http://scholrworks.umss.edu/lgut_fculty_pubs Prt of the Arts nd Humnities Commons von Ftel, Ki & Angelik Krtzer (2014). Modl Comprons: Two Dilettntes Serch of n Expert. In Luk Crnic nd Uli Suerlnd (eds.): The At nd Crft of Semntics: A Festschrift for Irene Heim. MITWPL 70, 175-179. Th Article brought to you for free nd open ccess by the Lgutics t ScholrWorks@UMss Amherst. It hs been ccepted for clusion Lgutics Deprtment Fculty Publiction Series by n uthorized dmtrtor of ScholrWorks@UMss Amherst. For formtion, plese contct scholrworks@librry.umss.edu.
MODAL COMPARISONS: TWO DILLETANTES IN SEARCH OF AN EXPERT KAI VON FINTEL AND ANGELIKA KRATZER Msschusetts Institute of Technology nd University of Msschusetts t Amherst Abstrct We express our puzzlement over the role of the Germn comprtive form metlgutic nd modl comprons. We cn t do th lone. We hve dt, we hve observtions, we hve specultions, but we need you to tell us wht to do with them. We wnt to sit down with you nd work on th together. The Germn word cn express temporl comprons roughly meng erlier or before : 1 (1) Die The Schildkröte tortoe wr ws erlier m t-the Ziel gol The tortoe reched the gol before Achilles. Achilles. Achilles As hppens ll the time, hs cquired uses beyond the temporl dom. One strikg use seems to fll under wht Morzycki (2011) clssified s metlgutic comprons: (2) Me cr t flott cool schnell. fst cr on the cool side on the fst side. We hve been puzzlg over the role of metlgutic nd modl comprons for quite while. We k Chr Kennedy, Pete Klech nd Mrc Morzycki for converstions where we shred our bfflement with them. Elen Herburger nd Aynt Rubste hve dependently embrked on reserch project on (Herburger nd Rubste 2014). Th not the time to compre our prelimry conclusions ny detil but we will drw on their WCCFL hndout few plces. 1 Etymologiclly, ppers to be the comprtive to n old Germnic word meng erly, cf. Gothic ir, relted to the Greek ri erly, the morng (see http://www.dwds.de/?qu=). The Southern Germn nd Austr dverb eh ( nywy ) my be bckformtion. 2014 Ki von Ftel nd Angelik Krtzer. In: Luk Crnič nd Uli Suerlnd (eds.), The Art nd Crft of Semntics: A Festschrift for Irene Heim, vol. 1, MITWPL 70, pp. 175 179
176 Ki von Ftel nd Angelik Krtzer (3) Me cr wr ws lngsm slow schnell. fst cr ws on the slow side on the fst side. (4) Me cr wr ws lngsm. slow cr ws on the slow side. (5) Me cr wr ws lngsm slow de your Motorrd. motorbike cr s performnce ws clerer cse of slowness your motorbike s performnce. Accordg to Morzycki, metlgutic comprtives differ from ordry comprtives relyg on scles of imprecion or prgmtic slck the sense of Lsersohn (1999). There cler difference between ordry comprtives nd comprtives with : (6) nd (7) do not men the sme. (6) Ell Ell (7) Ell Ell t t grösser tller gross tll.. Ell clerer cse of tllness. (7) doesn t sound right if the two girls re both gts or dwrves (hence cler exmples or counterexmples of tllness), or if one dwrf nd the other gt, even if Ell deed tller. For (7) to be cceptble, both Ell nd hve to be borderle cses of tllness. Tht s necessry, but not sufficient condition. (7) o uncceptble if Ell nd re both borderle cses of tllness, but they re very close height. Under those circumstnces, (8) uncceptble, too: (8) Im In Vergleich compron zu to t Ell Ell gross. tll. In compron to, Ell on the tll side. If Ell nd re very close height, no resonble wy of mkg the predicte tll prece would plce Ell its positive nd its negtive extension. Dog so would, fct, violte the Similrity Constrt of Fr (2000), which Kennedy (2011) formultes s follows: The Similrity Constrt. When x nd y differ to only very smll degree the property tht vgue predicte g used to express, we re unble or unwillg to judge the proposition tht x g true nd tht y g fe. No such problem res with the ordry comprtive (6), which cn be used even if the height difference muscule. Kennedy (2011) suggests tht th feture of ordry comprtives throws
Modl Comprons 177 doubt on semntics for such comprtives tht bsed on theories of vgueness (such s the propos by Kmp (1975), Kle (1980), vn Rooij (2011)). We tke no position on th qurrel. The pot here tht comprtives differ strikgly from ordry comprtives, nd there no question tht mechnms like vgueness resolution or prgmtic slck mngement ply role for the uses of illustrted (2) to (8). The difference between comprtives nd ordry comprtives o shows up with non-grdble predictes. Predictes like schwnger (pregnnt) or tot (ded) do not pper the ordry comprtive. The bsence of schwngerer (pregnnt-er) my be prtly becuse of the er -doublg but tht cn t beg to expl the severe ungrmmticlity of toter (ded-er): 2 (9)?Ell Ell t (10) *Me schwngerer pregnnt Ptient ptient t toter ded. dr. yours. The -comprtive fe even with non-grdble predictes nd s before, there metlgutic compron terprettion. (11) Ell Ell (12) Me t schwnger pregnnt. Ell clerer cse of pregnnt womn. Ptient ptient t tot ded dr. yours ptient clerer cse of ded person me. In ddition to temporl nd metlgutic terprettions, cn o express wht seem to be modl comprons. In fct, ll sentences we hve looked t so fr cn hve modl compron terprettions. Consider (7), for exmple. Suppose you got to know Ell nd s little girls nd hven t seen either one of them mny yers. You re wonderg wht they my look like. Given tht Ell s prents were tll, but s prents were not, you my sy (7). Wht we hve here n eptemic compron: the evidence supports the hypothes tht Ell tll better the hypothes tht tll. In other words: Ell likely to be tll. While the temporl terprettion feels clerly dtct from both the metlgutic nd the modl compron terprettion of, there question whether the metlgutic nd modl compron terprettions correspond to two genuely dtct redgs. Th, then, our first question: Question 1. Is genuely mbiguous between metlgutic nd modl compron redg? Our second question reltes to wht hppens when ppers sentence tht lredy conts modl expression. Here we re up for surpre: (13), (14), nd (15) re very close 2 Herburger nd Rubste mrk sentences like (9) with str.
178 Ki von Ftel nd Angelik Krtzer meng, nd seem to be truth-conditionlly equivlent, even though (14) hs possibility modl, (15) hs necessity modl, nd (13) hs no overt modl t ll. (13) Ds Th (14) Ds Th (15) Ds Th t jpnche Jpnese chesche Chese Th likely Jpnese Chese mche. knn might jpnche Jpnese chesche Chese Th likely Jpnese Chese mche. muss must jpnche Jpnese chesche Chese Th likely Jpnese Chese mche. Msche. mche. Msche mche Msche mche se. be. se. be. Question 2. How exctly does terct with overt or covert mod so s to deliver the truth-conditionl equivlence of (13) to (15)? Our third nd lst question reltes to subtle differences between modl comprons with nd modl comprons with grdble djectives like likely, desirble, or fesible. (16). Her movg to Pr likely h levg Rome (). b. Her movg to Pr likely you thk (it ). c. H levg Rome desirble (it ) fesible. (17) Jockl Jockl (18) Jockl Jockl (19) *Jockl Jockl knn cn Aufhusen Aufhusen gewesen been se, be Jckl Jckl Mdelheim. Mdelheim. Jockl likely to hve been Aufhusen Jckl Mdelheim. knn cn den the Mord murder begngen committed hben hve der the Richter judge denkt. thks. Jockl likely to hve committed the murder the judge thks. wollte wnted diesen th Mord murder begehen commit er he konnte. could. Jockl ws desirous cpble of committg th murder. We hve strred (19) to dicte tht it cnnot hve modl compron redg. It does hve temporl redg ( the time such tht Jockl wnted to commit the murder t tht time ws erlier the time t which he ws cpble of commitg it ). The puzzle why the modl redg bsent (19). (16c) shows tht, contrst, nlogous cross-sclr modl comprons re perfectly well-formed with grdble djectives.
Modl Comprons 179 Question 3. Why re modl comprons with restricted ordry modl comprons with grdble modl djectives? We hve no nswers, which why we need help from n expert on degree semntics. We could speculte nd revel our suspicions, but tht my be even embrrssg. Plese jo us nd set us stright. References Fr, Deli Grff. 2000. Shiftg snds: An terest-reltive theory of vgueness. Philosophicl Topics 20:45 81. Herburger, Elen, nd Aynt Rubste. 2014. Is possible possible Germn? Hndout from tlk t WCCFL 32, USC, Mrch 9, 2014. Kmp, Hns. 1975. Two theories bout djectives. In Forml semntics of nturl lnguge, ed. Edwrd Keenn, 123 155. Cmbridge University Press. Kennedy, Chrtopher. 2011. Vgueness nd compron. In Vgueness nd lnguge use, ed. Pul Egré nd N Kledst. Plgrve Mcmilln. Kle, Ewn. 1980. A semntics for positive nd comprtive djectives. Lgutics nd Philosophy 4:1 45. Lsersohn, Peter. 1999. Prgmtic hlos. Lnguge 75:522 551. Morzycki, Mrc. 2011. Metlgutic compron n lterntive semntics for imprecion. Nturl Lnguge Semntics 19:39 86. vn Rooij, Robert. 2011. Implicit versus explicit comprtives. In Vgueness nd lnguge use, ed. Pul Egré nd N Kledst. Plgrve Mcmilln.