Stress in English Long Verbs

Stress in English Long Verbs Q. Dabouis1, J.-M. Fournier1, I. Girard2, N. Lampitelli1 1Laboratoire Liǵrien de Linguistique (UMR 7270) – Universit́ François-Rabelais, Tours 2Universit́ du Littoral – Côte d’Opale Background and data THE TWO PARAMETERS Primary stress assignment in English verbs has been analysed in different ways. All analyses acknowledge that certain suffixes determine the position of stress (e.g. -ate, -ify). If we leave out suffixed words, compounds or constructions with semantically transparent prefixes, two parameters have been argued to determine the position of primary stress. Both are found in early generative phonology (Chomsky & Halle 1968; Halle & Keyser 1971; Liberman & Prince 1977) but have since been pursued separately in different theories. Syllable weight (Burzio 1994; Giegerich 1999; Halle & Vergnaud 1987; Hammond 1999; Hayes 1982) If the final syllable is heavy, it is stressed (e.g. moĺst, ob́y) If it is light, stress falls on the penult (e.g. ast́nish, dev́lop) Opaque prefixation Fournier (2007), Guierre (1979) and Trevian (2003) Primary stress may not fall on a semantically opaque prefix. ➢ Stress is therefore almost systematically root-initial (e.g. beǵn, comḿt, àppreh́nd, c̀ntrad́ct, emb́llish, env́lop, inh́bit) WHICH PARAMETER CAN BEST ACCOUNT FOR THE LOCATION OF PRIMARY STRESS IN VERBS? Previous work: dissyllabic verbs A previous study (Dabouis & Fournier, in prep.) on 2,544 dissyllables shows that a purely weight-based approach misses the 101 verbs with stressed light ultimas, which are often common words (e.g. admit, become, discuss, forget, permit…). 4 101 823 30 226 27 6 8 Light ultima Heavy ultima Light ultima Heavy ultima Prefixed /–1/ /1–/ Simplex Methodology Most studies on English stress actually rely on way too few examples to seriously argue in favour of any given theory. For this study, using the Laboratoire Ligérien de Linguistique’s Dictionary Database, we extracted 1976 verbs from Jones (2006) – syntactic categories from the Macquarie Dictionary. Converted nouns and syntactic structures Syntactic structures: 386 Converted Nouns: 348 Using semantics, dates of first appearance in the Oxford English Dictionary (online) and word frequency, we identified no less than 348 verbs that are actually converted nouns, and always preserve the stress pattern of the noun. A few examples: advantage, autograph, barbecue, condition, engineer, honeymoon, marinade, remedy. This constitutes an important part of the data: 17.6% Semantically compositional constructions whose leftmost formative is a prefix or an adverbial particle tend to behave as two distinct phonological domains, as evidenced by stress clashes (e.g. d̀pŕgram, m̀sḿnage, ̀utb́lance) or morphological geminates (e.g. di[ss]atisfy). We found 263 constructions with transparent prefixes and 123 with an initial adverbial particle (among which only undergo, undertake and understand are semantically opaque – included below). Primary stress is always on the base, on the same syllable as when it is used as an independent word. Five exceptions only: ćuntercharge, ćunterpoise, ćuntersign, ćuntersink, ́versew Suffixed verbs Suffix-induced stress: 615 Neutral suffixes: 337 -(i/e)fy: 89 words, all stressed /(-)100/, e.g. div́rsify, gĺrify, ŕrefy, soĺmnify, v́lify Primarily words in -ize + aẃken, iḿgine, ḿltiply, pŕphesy. -ate: 497 words, stressed /(-)100/, regardless of derivation (e.g. ́rigin > oŕginate; mat́re > ḿturate) or morphological structure (e.g. d́monstrate, eqúvocate, hydŕgenate, śbjugate, ćlebrate). Very few exceptions: ➢ /(-)10/: detŕncate, ̀quiĺbrate and insṕssate ➢ /1000/: ́xygenate, ṕregrinate, t́rgiversate ➢ and 7 potential /(-)1000/ with medial hiatus: ́lienate, aḿliorate, det́riorate, ́rientate…  Analogical (?) extension: 14 words in -ute or -ite, stressed /(-)100/ (e.g. ćnstitute, ́xpedite, ṕrsecute) except 3 in -tribute which vary between /(-)10/ and /(-)100/: attribute, contribute, distribute. -esce: 9 words, all stressed /-1/ (e.g. àcquísce, ̀ffloŕsce, l̀mińsce) + r̀mińsce ➢ 322 words with no stress shift (e.g. ćnon > ćnonize, exp̀riḿntal > exp̀riḿntalize), including 75 stressed /(-)1000/ (e.g. ch́racter > ch́racterize, ćpital > ćpitalize, ǵneral > ǵneralize) and 1 stressed /-10/ : pr̀paǵnda > pr̀paǵndize (all others are stressed /(-)100/). Only 12 exceptions with stress shifts (e.g. cańl > ćnalize, adv́rt > ́dvertize, moŕle > deḿralize, ́pilogue > eṕlogize, ǵlatin > geĺtinize, v́latile > voĺtilize, Galv́ni > ǵlvanize, hypńsis > hýpnotize, imḿne > ́mmunize, ap̀thésis > aṕtheosize, pànegýric > ṕnegyrize + ́mage > iḿgine) (24 verbs in which -ize is attached to a bound root are treated below) Other: b̀ccańer, el̀ctiońer, apṕrtion, d̀silĺsion, env́sion Opaque prefixed verbs Bound bases: 104 Free bases: 90 Semantically non-compositional prefixed constructions with a free base. Stress is always the same as in the base: adḿasure, conf́gure, enćurage, r̀asśre, r̀preśnt, s̀perṕse, ̀nderst́nd… except ŕtrograde. This includes 4 parasynthetic formations: aggŕndize, accĺmatize, emb́lden, enĺven In most cases, stress is indeed on the first syllable of the root: àppreh́nd, c̀ntrad́ct, ̀nterćpt, ̀nterḿt, r̀trogŕss, aḿrtize, cont́nue, deḿlish, eĺcit, inh́bit, reḿmber Only 13 words (12%) do not have primary stress on the root: ćrcumcise, ćrcumscribe, ćmpliment, disćmfit, ́xorcize, ́mplement, ̀mport́ne, ́mprovise, int́rpret, ́ccupy, ŕcognize, ŕconcile, śpervise Overall, in opaque prefixed verbs, these exceptions represent 7% of the data. A weight-based approach would not only fail to account for all these cases except disćmfit, ̀mport́ne and int́rpret, but also ̀nterḿt, ̀ntroḿt, pr̀terḿt, aḿrtize, imṕverish and r̀trogŕss (and hardly applies to free base cases). Other verbs Compounds (and assimilated): 61 Simplex and bound base + suffix: 35 Diverse root + root structures with various stress patterns. 17 bound base + -ize stressed /(-)100/ (e.g. fŕternize, ́rganize, śnitize) except et́rnize Examples: t́rbocharge, b́bysit, cr̀ssex́mine, d̀uble-ṕrk, v̀viśct, g̀nufĺct, fĺbbergast, j́xtapose, d́cuple, qúntuple 8 simplex stressed /(-)10/: bamboozle, canoodle, finagle, malinger, manoeuvre, sequester, skedaddle, solicit 10 simplex stressed /(-)100/: analyse, damascene, gallivant, manacle, manifest, massacre, minister, monitor, orient, paralyse (some of these are arguably not simplex, but their structure is particularly opaque) Our theory predicts /(-)100/, which leaves 9 unexplained cases. Discussion Long verbs are morphologically complex ➢ If we discount the 348 converted nouns, our corpus contains 1628 actual verbs of 3 syllables or more (which means probably all frequent ones and a number of rarer cases) ➢ Long verbs are essentially complex (versus simplex): only a few of them can be analysed as simplex, and even some of those few could be challenged. Stress is morphology-driven References ➢ ≈ 60% of the data have a stress pattern determined by their suffix (directly or through stress preservation) ➢ Nearly all other verbs (≈ 36%) are prefixed: following the model set by transparent (syntactic) structures (2/3 of all prefixed verbs), all of them assign primary stress on the root. Syllable weight isn’t better than morphology ➢ If both approach attain similar efficiency in the case of prefixed bound bases, the morphology approach yields a far more coherent understanding of the behaviour of the whole class. ➢ This is crucially confirmed by the inability of the weight-based approach to account for the predominantly final stress of dissyllabic prefixed verbs with light ultimas. Opaque prefixation: a plausible parameter? ➢ Often thought to be “unlearnable” or “not real morphemes”, prefixed words which are semantically opaque have often been analysed as if they were morphologically simple. ➢ There is evidence from a variety of combinatorial, phonological and psycholinguistic facts that these opaque structures can be accessed by speakers, and our studies imply they must be visible to the phonology (Dabouis 2017). Burzio, L. . P i iples of E glish St ess. Ne Yo k: Ca idge U i e sit P ess. Cho sky, N., & Halle, M. . The Sou d Patte of E glish. Ca idge, MA, Lo do , E gla d: MIT P ess. Da ouis, Q. . Se a ti all Opa ue P efi es a d E glish Pho olog . Ha dout fo the th https:// .a ade ia.edu/ /Se a ti all _Opa ue_P efi es_a d_E glish_Pho olog Da ouis, Q., & Four ier, J.-M. i p epa atio . Dis lla les a d S lla le Weight. I J. Du a d, A. P ze oz , & E. Ya ada Eds. , E glish Wo d St ess: Theo ies, Data a d Va iatio . Four ier, J.-M. s ste to a Ro a e e sus Ge a i o pholog -d i e d a i s: i ho ou of Lio el Guie e. La guage S ie es, 2 , – . Giegeri h, H. J. . Le i al St ata i E glish: Mo phologi al Causes, Pho ologi al Effe ts. Ca idge: Ca idge U i e sit P ess. Guierre, L. . Essai su l’a El e ts pou u e s th se. Ph.D. disse tatio . U i e sité Pa is-VII. Halle, M., & Keyser, S. . E glish St ess: Its Fo , Its G owth, a d its Role i Ve se. Ne Yo k: Ha pe & Ro . Ha o d, M. . The Pho olog of E glish: A P osodi Opti alit -Theo eti App oa h. J. Du a d, Ed. . O fo . Ca idge E glish P o ou i g Di tio a th ed . Ca idge: Ca idge U i e sit P ess. Li er a , M., & Pri e, A. . O St ess a d Li guisti Rh th . Li guisti I ui , , – . Trevia , I. . Mo phoa e tologie et p o essus d’affi atio de l’a glais. Be : Pete La g. OCP o fe e e i Düsseldo f. URL: . F o a Lati s lla le-d i e st ess e tuatio e a glais o te po ai : d: O fo d U i e sit P ess. Jo es, D.