Published online by Cambridge University Press: 22 April 2026
Whether or not he is familiar with the technical terms currently favored, or aware of the actual processes behind these terms, a speaker or student of Spanish, on some undefined level of consciousness, senses (or acts as if he sensed) the existence of such contrastable inflectional models as sent-ir ‘to feel’ : sient-e ‘(he, she, it) feels’ vs. ped-ir ‘to ask’ : pid-e ‘(he, she, it) asks’. The native speaker may even at some time or other have faced the dilemma of choosing, in the stressed syllable of such a verb form, between ie and i, and the foreign student is certain to have come to grips with such practical difficulties and to have been guilty, upon occasion, of gross errors. Preceptive grammarians and practitioners of language teaching, accordingly, have at all times paid attention to the hazards of the choice between the two conjugational categories.
By polishing this lengthy paper a few weeks before his untimely death, Bernard Bloch rendered me the last in a long series of editorial favors and attentions. The article has further profited from comments and queries by William Bright, Diego Catalán, and Marilyn May Vihman.
1 Comparable tangles are known to have occurred (some recently) in other corners of the Peninsula, under different conditions and with different results. For one remote parallel see M. J. de Moura Santos, ‘Remarques sur deux systèmes vocaliques anciens du portugais du nord’, BJR 9.5–12 (1964), and its critical summary by P. F. Dembowski in RPh. 19:1 (1965–66).
2 For the state of affairs in Modern Spanish there are many treatments, not infrequently-slanted in the direction of purism and with almost pervasive confusion of linguistic and merely orthographic complications (alternation of c and qu, c and z, and the like). One finds ample information in readily identifiable chapters and paragraphs of the Academy grammar throughout its successive editions, and in the standard-setting mid-19th-century grammars by V. Salvá and A. Bello, with weighty comments on the latter in R. J. Cuervo's Notas and in the same scholar's masterly A puntaciones críticas, which whet the reader's appetite for dialectal idiosyncrasies (see in the definitive edition §283 on cern-er/-ir, vert-er/-ir, and obsol. discern-er beside mod. -ir; §284 on hend-er/-ir vs. herv-ir/-er; and §300 on desleír, engreñir, re- and so-freír, son-reír). Among foreign Hispanists those deserving closest attention are G. Weigand, Spanische Grammatik für Lateinschulen 79–82 (Leipzig, 1922); J. Bouzet, Grammaire espagnole §§228–38 (Paris, 1946); C. Boselli, La grammatica spagnola del XX secolo 3 §§165–8 (Milan, 1948), who refers to radical-changing verbs as ‘semi-irregolari’; and M. M. Ramsay, A textbook of modern Spanish (1894), especially as revised by R. K. Spaulding (New York, 1956). To this day the most thorough guide to dialectal divergencies is Á. Rosenblat, ‘Notas de morfología dialectal’, attached to the author's annotated translation of A. M. Espinosa, Estudios sobre el español de Nuevo Méjico (BDHA 2; Buenos Aires, 1946), esp. §§220–35. One can also profitably consult several of the books here examined, for the sake of their historical flavor, in the survey of analyses.
3 This powerful pattern of diversification has clearly presided over the Hispanic development of es ‘(thou) art’ beside est ‘(he) is’, two forms which, contrary to expectation, did not collide. The peninsular languages kept them apart at heavy cost: Portuguese by breaking one of its most solid sound correspondences (és es vs. é est), Castilian by the unique device of borrowing a single form from the paradigm of the Latin future otherwise near extinction (eres eris vs. es, proto-Spanish also ye, yes, yet est). The dialects went their own ways, in part preserving the diphthongized (i.e. syntactically stressed) variants; for expertly distilled data see R. Menéndez Pidal, Orígenes del español 3 §73.3 (Madrid, 1950).
4 The exceptions comprise, on the one hand, secondary verbs such as ahuecar 'to hollow (out)', from hueco, and amueblar 'to furnish', from mueble(s), which have a very low degree of semantic autonomy; and, on the other, certain indefinite pronouns: OSp. otri(e) ‘somebody else’, nadie ‘nobody’, alguien (orig. alguién) ‘somebody’, which many untutored speakers avoid, no doubt as a result of the conflict between rising diphthong and unstressed syllable, preferring to resort to rival pronouns like ninguno and alguno or to remove the stumbling block through metathesis: dial. naide(s), -en. For a microscopic inspection of the record and a genetic explanation see my two papers in HR 13.204–30 (1945) and in UCPL 1:9.357–442 (1948). Note also certain hypocoristic derivatives such as abuelito ‘granddaddy’ (beside unorchestrated abolengo 'ancestry'), a thin layer of—predominantly not very old—'cultismos': suff. -icie, serie ‘set’, bilingüe ‘bilingual’, exangüe ‘anemie, exhausted’ (structurally as marginal as foreignisms would be), plus a few members of the paradigms of -iar and -uar verbs, e.g. averigüe ‘let him verify’.
5 The traditional division of Spanish verbs into three major classes, on the basis of -ar, -er, and -ir infinitives, is upheld here as one of several possible classificatory schemes.
6 To achieve maximum simplicity at this stage I mention but abstain from discussing such familiar marginal cases as adquirir ‘to acquire’ (OSp. adquerir) beside adquiero, discernir (Class. Sp. discerner) ‘to make out’ beside discernamos, recibir ‘to receive’ alongside concebir ‘to conceive’, jugar ‘to play’ (OSp. jogar) despite juego noun and verb. Some of these points will claim our attention later.
7 For a fuller treatment, with special attention to Asturian, see C. Blaylock, ‘Hispanic metaphony’, RPh. 18.253–71 (1964–65).
8 On pi-huela, -güela and satellite formations see my circumstantial account in Studies in the reconstruction of Hispano-Latin word families, UCPL 11.23–39, 96–154 (1954).
9 It is the -ī, not the -u-, of potuī and posuī that produced the stimulus, as is shown by the parallel development of fēcī and by the tendential loss of the semiconsonant in cō(n)s(u)ō 'I am sewing (together)' > coso, par(i)ēte 'wall' > pared, etc. That powerful -i was for a while preserved, witness the conservative usage of Gonzalo de Berceo (G. Tilander, ‘La terminación -i por -e en los poemas de Gonzalo de Berceo’, RFE 24.1–10 [1937]); it yielded to -e when the reduction of word-final vowels to a triadic arrangement (-a, -e, or -o) turned out to be the most economical (see my analysis in the article ‘Toward a unified system of classification of Latin-Spanish vowel correspondences’, RPh. 16.153–69 [1962–63]).
10 True, many normative grammars cite the paradigm of podrir ‘to rot’ as a reputed parallel of mentir: 1. pudr-o, 2. -es, 3. -e, 6. -en, 4. podr-imos, 5. -ís, 7. pudra, etc. But there is little reality in this schema. Finite forms in actual use, plus infinitive and gerund, to all intents and purposes, show only the pudr- stem; the former p. ptc. podr-ido ‘rotten’ stands apart as a semi-independent verbal adjective (receiving the support of the stereotyped phrase and famous book title olla podrida 'Spanish stew'). Note the inroad of podr-ecer, parallel to OSp. nodr-ecer ~ -ir 'to nourish' (Miseria 34d, 35c vs. 48c).
11 On the latter problem—for many years a notorious crux—there is now available the refreshingly original study by C. Blaylock, ‘The monophthongization of Latin ae in Spanish’, RPh. 18.16–26 (1964–65).
12 Except in piecing together details of the trajectories of individual verbs, such as AUDIRE ‘to hear’ (> Ptg. ouvir), gaudēre ‘to rejoice’ (> Ptg. [obs.] gouv-er, -ir), and trādūcere ‘to carry over, lead over, bring over’ (> OSp. troçir ‘to pass’ cf. NRFH 10.385–95 [1956]).
13 For the best formulation, and a digest of earlier statements by C. C. Marden (1896), F. Krüger (1914), and G. Millardet (1923), see R. Menéndez Pidal, Manual de gramática histórica española 6 §13.2 (Madrid, 1941), in reference to fl(u)eco, fr(u)ente, Bur(u)eva Borovia and cul(u)ebra ‘snake’ < col-ubra, *-óbra: 'Ocurre la reducción a e después de una l o r precedida de un sonido labial.' Estantigua stands apart; for storea 'rush mat' > estera and suff. -tōriu > -d(u)ero the author correctly reckons with the interference of suff. -āriu > -ero; to account for pestorejo ‘back of the neck’ (post + auricula 'little ear') he could have pressed into service the dissimilatory formula underlying rotundu 'round' > redondo (cf. HR 14.130–7 [1946]), the more properly as unstressed post yielded OSp. pos (cf. en pos de, etc.) rather than pues.
14 After stem-final j /x/ going back to x /š/, -ieron to this day yields to -eron: condujeron ‘they led’, tra-, dial, tru-jeron ‘they brought’, etc. Not so in learned -gir verbs: fingieron ‘they feigned, pretended’, surgieron ‘they arose’.
15 For further details see fnn. 22 and 112. The problem is complicated by numerous side issues. Pressa ‘pressure’ > OSp. priessa > prisa ‘haste’ seems to conflict with gypsu > yeso ‘chalk’ and tēnsu > tieso ‘taut, tight’, given the affinity of -ss-, -ps-, and -ns- (the second with tendential compensatory lengthening of the s). On the one hand, in pri(e)sa the r may have acted as a concomitant and accelerator (cf. mirlo, ristra); on the other, the y /j/ of yeso may be a reflex of /g/ before /e/, while what is usually called the loss of g before a front vowel in words like germānu ‘brother’ > ermano, gingīva ‘gum’ > enzía may, in the last analysis, represent the monophthongization of secondary /je/ in pretonic syllable. Finally, tieso ‘stiff, tight, tense’ seems a mere elaboration on teso 'id.' under the influence of tender 'to stretch' / tiendo, while tiesto ‘stiff, tight, stubborn’ shows additional contamination with testarudo ‘stubborn’, cf. Fr. entêté. On the peculiarities of the representatives of gypsu: Sp. yeso beside Arabicized algez, Cat. gix, Leon, yelso, see E. Alarcos Llorach, ‘Resultados de Ge,i en la Península’, Arch. 4.330–42 (1954), esp. 339–40.
16 Since Portuguese occupies an ancillary position in this paper, the information offered does not transcend the scope of a standard manual like E. B. Williams, From Latin to Portuguese 2 (Philadelphia, 1962).
17 For the purpose of this article I disregard the fact that Ptg. pôr ‘to put’ < OPtg. põer pōnere cannot be fitted into any slot of the traditional triadic array of conjugation classes.
18 On the provenience of iguaria and OJud.-Sp. yegüería ‘mess, dish’ see my papers in Lg. 20.108–30 (1944) and 21.264–5 (1945) and the various comments that these contributions provoked in the mid-forties. As regards the descendants of pertica, note the unexplained replacement of OSp. piértega—a perfect match of Fr. perche—by mod. pertiga ‘rod, staff’.
19 Some illustrations of the wealth of possibilities within the confines of the (a) -er and the (b) -ir class: (a) caib-/cab- ‘to fit’, requeir-/requer- ‘to demand’, faç-/faz- ‘to do’, poss-/ pod- ‘to be able’, vej-/ve- ‘to see’, perc-/perd- ‘to lose’, valh-/val- ‘to be worth’, OPtg. moir-/ morr- ‘to die’; (b) cai-/ca- ‘to fall’, venh-/ve-, vi-, vĩ- ‘to come’, etc. As shown by Ptg. morro vs. OPtg. moiro and by Sp. puedo, miento, and siento vs. Ptg. posso, minto, sinto, the ever-present leveling trend in Hispano-Romance usually sacrifices Forms 1 and 7–11 to a consensus of the others; but occasionally, on the level of dialect speech, the reverse hierarchy prevails: cf. the situation in the Asturo-Leonese subdialect of Sistema as analyzed, on the basis of J. A. Fernández's Madrid dissertation, by Marilyn May in RPh. 19.54 (1965–66). Where the subjunctive en bloc is opposed to the indicative, as is true of Sp. sea, Ptg. seja < sedea(m) and, in its wake, Ptg. esteja, as a rule suppletion has been at work (esse + sedēre).
20 See ‘Elena y María (Disputa del clérigo y el caballero); poesía leonesa inédita del siglo XIII’, ed. R. Menéndez Pidal, RFE 1.52–96 (1914), esp. Section 5: ‘Lenguaje’ §1. The editor reports (78 f.), on the evidence of faulty rhymes, that the poet, as distinct from the scribe, failed to diphthongize ǒ; he may or may not have diphthongized Ě.
21 At this point I note a regrettable contradiction; the text, as established by Menéndez Pidal, reads: E siruen los escuderos (p. 57, line 76); but in his grammatical sketch (83) the editor—inadvertently, I suppose—attributes to this passage the reading sieruen.
22 F. de B. Moll's conjecture ('Suplement català al REW, 1928–30), supported by the contrast between Sp. aprisco and Cat. (Mall.) après 'sheepfold', found its way into REW 3 §554a. I have tried to provide analyses of pri(e)s(s)a in ‘Apretar, pr(i)eto, perto: Historia de un cruce hispanolatino’, BICC 9.103–5 (1953–55); of prisco in ‘The Hispanic suffix -(i)ego’, UCPL 4:3.164–6 (1951); of ristra in ‘The etymology of Hispanic restolho, rastrojo, rostoll’, RPh. 1.209–34, esp. 228–9 (1946–47).
23 Thus, H. Tiktin, Rumänisches Elementarbuch §47 and §82 (Heidelberg, 1905), cites r + -īre (inf.) > -rî, rīma ‘worm’ > rîm ǎ, rīvu ‘river’ > rîu, rīdeō ‘I laugh’ > rîz, terra + -īna > Larînǎ ‘realm of earth’; also pretonically : tǎnrîşoarǎ, dim. of tarǎ ‘land’ < terra.
24 There exist two more marginal possibilities: (c) sierve, vieste may be strictly dialectal forms and thus make no links in the chain stretching from L. servit, vestit to Mod. Sp. sirve, viste; and (d) sierve, vieste may represent temporary highwater marks of a vigorously expanding diphthong, of which, through a twist of circumstances, we happen to catch a glimpse through the sparse corpus of early 13th-century texts.
25 On the incubation period of Spanish rising diphthongs see Menéndez Pidal, Orígenes del español 3 §§22–6 (Madrid, 1950).
26 E. Alarcos Llorach, Fonología española 3 212–5 (Madrid, 1961), recognizes the roots of the contrasts /ε/ : /e/, /ɔ/ : /o/—at the outset possibly restricted to the free syllable—in Late Latin conditions, but credits the extension of the contrast to the checked tonic syllable as well as the transformation of e to ie and of e to ue to a period of substratal and Latin bilingualism, especially after the historical ties to the metropolis had been cut. The passage also contains critical references to differently slanted writings by G. Straka, F. Schürr, and H. Weinrich and to some of the author's own parallel researches.
27 Among the Latinisms there is a neat cleavage: some have been cast in the mold of vernacular verbs, e.g. aderir ‘to adhere’, advertir ‘to notify’, repetir ‘to reiterate’ (all three behave like servir); others, including agredir ‘to attack’ and transgredir ‘to transgress, trespass’ (also denegrir ‘to blacken, revile’, prevenir ‘to forewarn, anticipate’, progredir 'to thrive'), follow the Castilian pattern of pedir—a residue of 16th-century bilingualism on the level of grammar, the more startling here as lexically Castilian either favors different slices of material (atacar or acometer, denigrar, progresar, etc.) or, where it makes use of the same word (as is true of prevenir), has recourse to a totally different inflectional model. Ptg. remir 'to redeem' < redimere, obviously not a straight Latinism, has been added to the agredir group, possibly in tribute to the learned garb of Sp. redimir (beside vestigial vernacular rendir; see Corominas, DCELC 3.102b, s.v. lindo).
28 It is doubtful that the ç of ouço audiō represents a phonologically normal reflex of /dj/, but discussion of the point might lead us far afield, especially since it would have to include the status of OSp. gozo gaudiu ‘joy’.
29 E. B. Williams, From Latin to Portuguese 2 §176.8B. Given the specific conditions of Portuguese morphophonemics, it is, in general, a matter of very minor concern whether in the arrhizotonic forms e or i, o or u is favored in conventional spelling: afligir ‘to sadden’, corrigir ‘to correct’ (as against Sp. corregir), dirigir ‘to direct’, fingir ‘to invent’ go with agredir; frigir with servir; acudir and fugir with dormir. Ordinarily the contiguity of -g- /ž/ seems to favor pretonic i over e, bidding us expect *eligir ‘to choose’ as the equivalent of Sp. elegir; but here, for once, the thread snaps, since the form actually encountered is eleger, probably influenced by 1er legere.
30 Conservative forms with -i rather than -e (fiz-i and the like) are, I repeat, still characteristic of Berceo's Old Riojan dialect. Veno occurs in Razón de amor, line 99. In the paradigm of this verb, at least, we can be sure that the analogical addition of -o preceded the withdrawal of ven- before vin-. On a trace of vieno see p. 460, below.
31 The muero/morir group also suffered losses, including toll-ō/-ere ‘to take away’ > tuello/to-, tu-llir, for which the Gallicism quitar (in a sense alien to Modern French) was at length substituted; cf. the remnant tullido ‘cripple’. Menéndez Pidal, Manual 6 §114.2, ascribes to Old Leonese and Old Aragonese alone comparable forms of (en)cobrir ‘to cover’ < co(o)perīre, descobrir ‘to discover’, and nozir ‘to harm’ < nocēre.
32 See the classic treatment of this problem by Menéndez Pidal in Orígenes del español 3 §27.
33 The two well-known modern examples of reverse distribution, Ptg. cair : Sp. caer ‘to fall’ < cadere and Ptg. possuir : Sp. poseer ‘to own’ < possidēre, have no bearing on Old Portuguese. Ptg. tossir ‘to cough’ reflects tussīre; Sp. toser seems to move in the wake of coser ‘to sew’ < cōns(u)ere, while Fr. touss-er beside older touss-ir (Rutebeuf) shows an even stranger deviation.
34 The contrast between concebir ‘to conceive’ and recibir ‘to receive’ < OSp. reçebir (also percibir ‘to perceive, collect‘) is one minor rough spot which has not been removed from the standard. Hispano-Romance lacks any counterpart of Fr. dé-cevoir (> E. deceive).
35 For details see ‘Español morir, portugués morrer’, BH 57.84–128 (1955).
36 Cf. premidera ‘treadle of a loom’. On OSp. (a)premer vs. (a)premir, all four later replaced by aprem(i)ar, see fnn. 372, 375–6, 402, 405, 408, 420 to my article in BICC 9.1–139.
37 Portuguese shares with Spanish many Latinisms in -i-ir, -u-ir (say, distribuir, com-, o-, and re-primir, presumir, resurgir) and occasionally exceeds Spanish in its openness to their infiltration; contrast Ptg. corrigir, divorced from reger, with Sp. corregir, subordinated to regir. On the whole, it has fewer of such erudite inlayings: Ptg. recorrer semantically matches both Sp. recorrer ‘to traverse, go over’ and recurrir ‘to resort, have recourse’. In all probability, the Portuguese compounds in -mitir (rather than -meter) etc. were suggested by Spanish models during the century of cultural connubium.
38 In Pero López de Ayala's Rimado de palacio (ca. 1400), two key passages display gemer (inf.) in rhyme: N 23d and E 1523c; less conclusive as to conjugation class are N 1022a: gemiera, and E 1763c, 1769d: gemiendo (gemí and gemido, found elsewhere, are inconclusive). Pianga in E 1472a, 1477d proves nothing, but in N 1465b verse-medial plañer is good evidence, and two forms which that MS exhibits in rhyme—plañer (819a) and plañed (969c)—provide the clinching argument. MS E offers toller only in verse-medial position (914d); it also favors tremer where N prefers tremir (143c: tremirá). It is hazardous to posit, with M. A. Zeitlin (unpublished Vocabulary, 1931), sumer rather than -ir on the strength of sumió and sumería (N 46a, 1086d), especially in view of Ptg. sumir.
39 The copyist of the S manuscript of Juan Ruiz was fairly consistent in distinguishing stressed cwnpl- (274b, 504c, 768d) from pretonic compl-, while MS G shows more liberal use of cumpl- (note the contrast in 1134d). But in many texts the conflation of manuscripts produced considerable confusion. (G, S, T refer here and in the following footnotes to the Libro de buen amor; E and N, to the Rimado de palacio.)
40 Examples: adormió (E 106c), arrepentió (E 1454d), cobierto ‘covered’ (S 216d, 306c) beside cobierta ‘covering, protection’ (S 656c), comediendo (G, S 811b) beside comides (G, S 1005a) and comidas (G, S 976b).
41 The derivation of asir is understandable only on the assumption that the rights to the sound sequence asar were, so to speak, preëmpted by the semantic nucleus ‘to roast’ < assāre (Apuleius), a racy frequentative verb based on assus, in turn related to Ārsus, the p. ptc. of ārdēre ‘to burn’. See Ernout-Meillet, DÉLL 4 51–2 (Paris, 1959–60), and J. Corominas, DCELC 1.295–6 (Madrid, [1954]).
42 In siding with Menéndez Pidal's original verdict (1911), dēcidere ‘to fall down’, H. B. Richardson, An etymological vocabulary to the 'Libro de buen amor' 76 (New Haven, 1930), briefly recalls G. Körting's dēsīdere 'to sink down, subside' (1907), Meyer-Lübke's dēicere ‘to throw, hurl down’ (1911–20), and V. García de Diego's *discidere ‘to fall apart’ (1923). Menéndez Pidal's subsequent endorsement of discēdere ‘to walk away’ seems all the more felicitous as cēdere discernibly broadened its meaning in Late Latin: ‘to go, walk’; see ed. Cantar de Mio Cid 21222–3 (Madrid, 1944–46), with special credit given to J. Vallejo and A. Magariños. For additional comments and decisions by other grammarians and etymologists see below, fnn. 43, 69, 83, and 90.
43 Aduzir was very common in Old Spanish; J. Ruiz apparently favored enduzir (1490a: enduxo), a variant on which all three manuscripts agree, but S, the latest, also admits induzir (223c). For all its uncommonness, the blurring of the boundary between prefix and stem implicit in the loss of -d- in trādūcere has a parallel in the dual transmission of benedīcere ‘to bless’ in Old Spanish (bendezir) and elsewhere; cf. the p. ptc. bendi-cho, -to beside the proper name Ben(e)ito and cf. Ptg. benzer, Fr. bénir and Benoît, It. (borrowed) Benito alongside (native) Benedetto. The same blurring, with no equal loss of consonant, characterizes the growth of deçir discēdere, the one verb in which, genetically, not the stem vowel but the vowel of a former prefix became subject to metaphony (OSp. diçe ‘he descends’, diçiendo 'descending'). Troçir probably owes its c, rather than -z-, to the coexistence of deçir; early loss of -d- and rapid consonantization of u (*Trawkere) may have been contributing factors. The change -sc(e)- > -ç- was normal; cf. piscēs 'fish(es)' > peçes, dis + cingere > deçeñir 'to ungird' (Ruiz, 1114d).
44 Juan Ruiz used the forms arre-zido, -çido ‘stiff with cold’ (954c, 966a, 982b), aterida 'id/ (1349c: S, T; G: atordida), desatyriziendo 'getting over numbness, thawing up' (970a). The upshot of the difficulty is that rezio itself presumably owed its -z- to association with OSp. suzio ‘dirty’ < sūcidu ‘juicy’; in Portuguese a pressure in the reverse direction seems to have moulded sujo after rijo. Corominas, DCELC 1.318b–319b, supports integru, capitalizing on OSp. enterecer, explaining OSp. ter[r]ecerse ‘to be jittery’ through contact with the family of teurēre ‘to frighten’ and declaring OArag. atelerido, Cl. Sp. entelerido ‘in the throes of fright and cold’, etc. to be mere ‘ampliaciones fonéticas’, which I take to correspond to the more precise label ‘interfix’ (cf. fn. 61 below). The choice of the intercalary segment -el- awaits clarification, as does the possibility of contamination by tericia, a sickness prominently mentioned in the older literature.
45 For an excellent semantic analysis of OSp. de(s)leír, Ptg. delir, based in part on earlier researches by R. J. Cuervo and L. Spitzer, see J. Corominas, DCELC 2.143b-145a [1955]; the original meaning was ‘to weaken, soften, extenuate’, refl, ‘to languish’. The connection with dēlēre ‘to destroy’ is indeed more plausible than the links to dīluere ‘to dissolve’ or to dis + ligāre ‘to untie’, previously posited. But Corominas' idea (reiterated in his BDE) of explaining the segment -eír through reference to details of the preterite of dēlēre is nothing short of bizarre, since that tense is not otherwise known to have influenced the shapes of Spanish infinitives and the choices of conjugation classes. Reterere has been pressed into service as the alleged base of OSp. (de)rretir and of OPtg. (de)rreter ever since the days of Diez, and a controversial passage in Villasandino (Cane, de Baena, N°. 174: Si el sol rretiere el plomo) has figured prominently in the successive elaborations on this conjecture by C. Michaëlis de Vasconcelos, García de Diego, and Corominas (DCELC 2.129a–130a). The latter was commendably restrained in accepting reterere, but uninspired in preferring to operate with the ad-hoc reconstruction *retrīre. He did, in passing, draw the reader's attention to the Catalan and Provençal offshoots of dēlēre, but only for the sake of the semantic analogy. I visualize a divergent solution (to be presented elsewhere) which bypasses terere in any of its guises.
46 In view of the monotony with which one Hispanist after another has endorsed Diez's utterly implausible *ingredī, it is refreshing to watch Corominas, DCELC 2.285a–286a, attack the problem, for once, from a different angle. Observing in Calila e Dimna (orig. ca. 1250) the vacillation between 3 engríe and engree and, independently, the attestation of inchoative engreyecer in a 14th-century Pentateuch, Corominas argues that the verb was initially engre-er, not -ir, and represents a distortion of OSp. encreer ‘to lend’, refl, ‘to trust’ parallel to cancriculu > cangrejo ‘crab’ and to Cat. Prov. concriar > congriar. The phonetic analogy is palpably weak, the record of encreer spotty, and the appeal, on the semantic side, to mod. dial, creído ‘pampered, vain, presumptuous’ (Albacete, Almería; also, I understand, Madrid) less than convincing; but as a first attempt to loosen engreír from the grip of ingredī, Corominas' entry is impressive.
47 On Ptg. íngr-eme, -ime ‘steep, craggy, isolated’ > Amer. Sp. íngrimo ‘quite alone’ the opinions of etymologists clash violently; for an analysis of the views of J. Leite de Vasconcelos, A. d. R. Gonçalves Viana, Toro Gisbert, E. Gamillscheg, and L. Spitzer see Corominas, DCELC 3.999a–1000b (1956) and 4.1027ab (1957), who reckons with the cross of two borrowed Old French words, encri(e)me ‘wicked, perverse’ (Gallo-Latin) and engremi ‘angry, irritated’ (Gallo-Frankish). Whichever genetic hypothesis is correct, an early contact of íngrime and encreído somewhere in Western Spain would best account for the aberrancy of engreír(se).
48 I have casually observed this avoidance of radical-stressed forms in the case of enxerir (Ruiz).
49 I owe many definitions to M. A. Zeitlin's unpublished University of California (Berkeley) dissertation, A vocabulary to the ‘Rimado de palacio’ of Canciller López de Ayala (1931), and have also occasionally profited from his comments and textual emendations.
50 Some of these verbs pose serious etymological problems. The unorganic f- of fenchir—and of f-inchar ‘to inflate’ < inflāre—has been traditionally explained through contamination with fartar ‘to stuff’. I once subscribed to this conjecture; see ‘La F inicial adventicia en español antiguo’, RLiR 18.161–91, esp. 164 (1954). But the violent distortion of the -pl- and -fl- clusters at the confluence of two currents, I now recognize, may have been a powerful contributing factor; see 'The interlocking of narrow sound change, broad phonological pattern, ...', ArL 15.144–73 (1963) and 16.1–33, esp. 27 (1964). On troçir see my study in NRFH 10.385–95 (1956) ; cf. fn. 43 above. I am now more than ever convinced, despite widespread indifference or opposition to this hypothesis, that OSp. (d)espedir must not be divorced from Gal.-Ptg. (d)espir 'to strip', representing as it does a blend of the petere and ex-pedīre families ; for old and new arguments see ‘The coalescence of expedīre and petere in Ibero-Romance’, UCPL 11.40–54, 155–69 (1954), and the forthcoming article ‘Two semantic ingredients in etymological analysis’, to appear in RPh.
51 Poenitēre, the Late Latin (particularly Church Latin) spelling, through rapprochement with poena ‘punishment’, has influenced spoken Hispano-Latin as little as (judging from Sp. cielo) coelum ‘sky, heaven’, favored by the same circles, for caelum.
52 This remark applies to the -t- of (a)rrepentir, as against its voiced counterpart in Ptg. arrepender and ORioj. (Berceo) arrependir. We cannot yet accurately assess the reasons for the split between quer-er and ad-, con-, per- (pes-), and re-querir. Whatever the channels of transmission, the compounds unquestionably had a more erudite flavor, thus setting the pace, in postmedieval Spanish, for the characteristic cleavage of such racy primitives as correr ‘to run’, meter ‘to put’, romper ‘to break’, verter ‘to pour’ and certain compounds: incurrir ‘to become liable’, permitir ‘to allow’ (beside older otorgar), interrumpir ‘to interrupt’, invertir ‘to invert’, etc. In the process, adquerir was allowed to adopt an ultralearned shape: adquirir (with the vocalic orchestration of fingir, resistir, etc.), while conquerir, unlike its French cognate, gave way to conquistar, based on the nominalized p. ptc. conquista. Italian goes even farther in its recoil from -quīrere: ac- and con-quistare.
53 For a full biography of this verb and its offshoots see RPh. 6.121–72 (1952–53).
54 The reconstructed forms are due to the agency of recomposition. This tendency has been rife in the ranks of -ere verbs (cf. Sp. atañer ‘to concern’ beside attingere ‘to touch, reach to, have to do with’, through the catalytic interference of tañer < tangere) and has, upon occasion, led to striking hypercorrections (Ptg. constranger ‘to force’ and Fr. contraindre [whence E. constrain] constringere ‘to fetter, restrain’, from stringere ‘to draw tight together, bind’, recomposed on the ‘false analogy’ of tangere /attingere.)
55 The telltale occurrence of pitent in the Glosas Silenses (No. 196) will be commented upon in the concluding paragraphs of this section. On the emendation pie(r)de su sen in the Razón de amor, line 48, see G. H. London, RPh. 19.28–47, esp. 36 (1965–66) ; the meaning is ‘loses her head’.
56 Significantly, vīvere ‘to live’ (which has in general preserved its high front vowel in Romance: Fr. vivre, OProv. viure, It. vivere, Ptg. viver, etc.) came perilously close to changing from vivir to vevir in Old Spanish; in Pero López de Ayala's poem the spellings biv- and bev- are used almost interchangeably. The dissimilatory nature of the shift can best be gauged through comparison of (a) fīn-īre ‘to end’, which before long yielded feneçer ‘to finish, bring to a close, (come to an) end’, via *fen-ir (the stage which alone justifies radical -ī- > -e-); and (b) *fīmn-āre ‘to (come to an) end, die’ (a Late Latin circumlocution for morī) > OSp. finar, never *fenar. There is on record the exceedingly rare compromise form fin-e(s)çer. Freír ‘to fry’ (which no doubt superseded *friir), from frīgere, provides a further, almost perfect parallel to reír rīdēre and son-reír ‘to smile’ < sub-, sur-rīdēre.
57 On the intricate resolution of the sequence o–ó (either > a–ó, or > e–o) see my paper ‘The word family of Old Spanish recudir’, HR 14.104–59, esp. 130–7 (1946).
58 I owe many data to the books, to be discussed later, by A. Gassner (1897), R. Menéndez Pidal (1908, 1926–50), and F. Hanssen (1913); but I have checked the passages, wherever possible, against superior modern editions.
59 The obsolescence was no doubt gradual, with the unpredictable radical-stressed forms preceding by a margin of (say) a century their less perplexing suffix-stressed counterparts. After the elimination of the latter, there remained one isolated remnant, the nominalized p. ptc. ejido ‘commons, public land, communal farm’; on And. lejío see A. Castro, RFE 1.181 (1914).
60 Cf. Ptg. disse dīx-ī, -it (with counterparts in Berceo's Old Riojan subdialect) and, in pronunciation if not in spelling, trouxe *traxu-ī, -it. Tanso and tanxo ‘he touched’ also coexisted in Old Spanish; cf. Lg. 36.313 (1960).
61 For the definition of interfix, brought in here for the sake of its conjugational implications, see my article ‘Los interfijos hispánicos: problema de lingüística histórica y estructural’, Estructuralismo e historia 2.107–99 (La Laguna, 1958; ed. D. Catalán), and the various reactions to it. Cf. fn. 44 above. In Old Navarrese 〈ss〉 could have represented /š/.
62 I owe much information and my initial impetus at this point to Menéndez Pidal, ed. Cantar de Mio Cid 1.269 (Madrid, 1908), but in elaboration and analysis I have followed my own bent.
63 A similar situation obtained in the case of the cluster fl-. It was infrequent in terms of incidence and lexical representation, and involved a word-initial consonant which, alone among those preceding Lat. l, was not obstruent. As a result fl- oscillated, in its transmission into Hispano-Romance, between joining pl-/cl- or else bl-/gl-, both courses being clearly delineated and copiously represented. For details see my article in ArL 15. 144–73 (1963), 16.1–33 (1964).
64 The assignment of impelleke ‘to drive on’, absorbed as a Latinism, to impeler rather than (as might otherwise have been expected against the background of learned transmission) to *impelir may be due precisely to the speakers' reluctance to experiment with such forms as *impielo or *impilo, which ‘Systemzwang’ would have foisted on them. The preëxistence, in Old Spanish, of vernacular empeller impellere may have been a potent concomitant.
65 Grammatik der romanischen Sprachen 2.151–2; 2d ed., 2.168–70; 3d ed., 3.180–2. Later editions are mere reprints.
66 In 1838 Diez was in doubt as to the accuracy of heñir fingere, supplied no etymon for derretir, and inadvertently operated with obsol. ad-, in-querir. The second edition shows some progress: the earlier doubt has been dissipated, derretir is traced to dēterere, adquerir is paired off with inquirir (‘das einzige mit radikalem i‘), OSp. con-, cun-tir conti(n)- gere and no-, nu-cir nocēre are thrown in for good measure on the side of dormir and morir. In the final revision the author ends the section by remarking that the late medieval paradigm of the strong preterite shows sporadic extensions of the -e-/-i- alternation: dixe, dexiste; fice, j'eciste [sic]; quise, quesiste. (What one actually witnesses here is varying degrees of resistance to the inroads of the -i- from the 1st sg. and from the plural.)
67 ‘Beiträge zur lateinischen Lautlehre: Umlaut (eigentlich Vokalsteigerung) im Romanischen’, ZRPh. 3.481–517, esp. .507 (1879).
68 For early criticism note Meyer-Lübke's remark in his comparative grammar: ‘Allein damit ist keine Erklärung gegeben und, wie Förster selber bemerkt, auch das tatsächliche Verhältnis wenig treffend dargestellt, da servir trotz r zur zweiten Klasse gehört.‘
69 ‘Zu W. Förster's romanischer Vokalsteigerung’, ZRPh. 4.113–23, esp. 121. That same year P. Förster, Spanische Sprachlehre 325, 328–9 (Berlin, 1880), touched upon the subject with scant originality, except that he questioned the derivation of engreír from ingredī and brought in desleír ‘to dilute, dissolve’, [Ast. Leon. Pal. Sant.] espurrir 'to stretch out (one's limbs)' < exporrigere, muir 'to milk'. P. Förster, probably the first Hispanist to recognize the genetic link between Ptg. despir and Sp. despedir, went astray in etymologizing desleír (liquēre 'to be fluid, clear') and OSp. decir (dēsīdere 'to sink down, subside, settle').
70 ‘Melanges espagnols: Remarques sur les voyelles toniques’, Rom. 13.285–97, esp. 296–7 (1884). J. Wiggers's Grammatik der spanischen Sprache 2 (Leipzig, 1884) was not particularly impressive on the side of inflectional analysis (148–52), but proved strikingly original and ahead of its time in details of etymological classification; see the remark (150) on the contact between ingerir ingerere and inserir (enjerir) inserere and the derivation (151) of (d)espedir from expedīre, in harmony with P. Förster's hunch.
71 ‘Die spanische Sprache’ §72, in G. Gröber's Grundriss der romanischen Philologie 1 (Strassburg, 1888). For the diphthong in hied- foet-, hiend- find- Baist implicated the f- > h- series: herir, hervir, herrar ‘to shoe, brand’. In 1894 Meyer-Lübke took exception to several of Baist's claims. Can one maintain, he argued, that sintamos shows influence of sintió if the only possible corollary is that temió ‘he feared’ shows influence of temamos? If digo : decimos attracted, as regards the vowels, pido : pedimos, why did it leave siento : sentimos intact? And how can one account for the local shift of dīcere to the -ir class?
72 Altspanisches Lesebuch mit Grammatik und Glossar 146–53, esp. 147 (Leipzig, 1890).
73 Historische Formenlehre der spanischen Sprache 40–5 (Murrhardt [Würt.], 1894). Clearly in the wake of Keller's monograph, to its distinct disadvantage, was the treatment by E. Gorra, Lingua e letteratura spagnuola delle origini 133–5 (= §174), esp. the amorphous concluding section. Ill-advisedly the author chose pedir as his model, slipped up (like most of his contemporaries) on the etyma of derretir and engreír, misfitted (unlike the best among them) freír, and showed eloquence only in bemoaning his perplexity: ‘Troppo complessi sono i problemi che questi verbi suscitano, perchè si debbano affrontare qui; nè d'altro canto i glottologi sono riusciti finora a mettersi d'accordo nello spiegarli’ (135).
74 One finds, amid much chaff, scattered useful nuggets on jogar ‘to joke, toy’ > jugar ‘to play’ (§57), on d(i)ezmar ‘to tithe’ and impacientar ‘to make impatient’ (§58), etc. Verbs with a long-drawn-out split in their paradigmatic preferences, such as po- ~ pu-drir ‘to rot’ and ergu-ir (yergo, irgo) were, of course, grist to Keller's mill. The untenable belief in an all-pervasive ‘doppelte Konjugation‘—as distinct from Diez's two tidily delimited groups—was formulated in §59.
75 In handling such archaic, pre-Alfonsine texts as the Alexandre, the Apolonio, and the Fuero Juzgo, Keller either was unaware of their strong dialectal nuancing or drew no inferences from any awareness.
76 Yet he probably overshot the mark by dragging in not only pedir, seguir, servir, vestir, but also gemir (without asking himself whether gemer might be involved) and the two clearcut Latinisms impedir and regir.
77 Grammatik der romanischen Sprachen: Formenlehre §191 (Leipzig, 1894).
78 Cf. yerto ‘stiff, rigid’ < ēr(e)ctu via *ěr(c)tu. One remote parallel: rēte mán-ica ‘fisherman's net cast by hand’ > OSp. red maniega beside ORioj. redmanga (UCPL 4:3.167 [1951]) and the other offshoots of man-icu, -ica discussed in the same monograph (see Index).
79 The passage appears in the Concionero de Baena: Pues quien poco sabe, co(n)viene que se rryenda (A. Álvarez de Villasandino).
80 As a parallel Meyer-Lübke adduces tepidu ‘lukewarm’ > [*]tievio > tivio (sic), for which he refers the reader to his Lautlehre (1890). The point, however, is that recorded Leon, tebio marks the actual midway point of the trajectory tepidu > Sp. tibio: the early loss of -d- reduced trisyllabic *te-bi-o to /tebju/, with /ε/ concomitantly closed to /e/, and since /j/, unassimilable to the preceding labial, failed to disappear in the process, it could again, at a later date and originally in a narrower zone, act on /e/, raising that vowel to /i/; cf. Berceo's Latinism lídio beside mod. tedio. For details see my analysis in Rom. 74.145–76 (1952). There is no compelling reason for reconstructing any underlayer with a rising diphthong.
81 Over against -e-/-ie- Meyer-Lübke places -o-/-ue- (dormir/duermo), pitting on the other hand -o-/-u- (as in po-, pu-drir/pudro) against -e-/-i-; but he refrains from explaining the asymmetry of the ensuing rearrangement. The author misses any trace of -e-/-i- before -ll-, a point at which he unwittingly comes nearest to implicating monophthongization. He concludes on an anecdotal note, correctly interpreting the evolutionary vagaries in the Spanish offspring of levāre ‘to lighten, lift up’: liev- (> llev-) / lev-, then generalization of llev- ‘to carry’.
82 Das altspanische Verbum §§42–59 (front vowels) and §§60–71 (back vowels) (Halle a.S., 1897). Following in Meyer-Lübke's footsteps and bowing to the current fashion, Gassner used the term ‘Ablaut’ in reference to morphophonemic variants—a sense frowned upon by most present-day historical linguists.
83 Thus, it seems gratuitous to aver (§43) that the original distribution of e and i vars. in the paradigm could not, for some inherent reason, survive indefinitely (for evidence to the contrary cf. Portuguese). Not only is Gassner's etymology of deçir (dēcidere) untenable, but, worse, he states quite hazily (§45) the relation of that doomed verb to dezir dīcere. Reír rīdēre should, for consistency's sake, have elicited a comment in §56, not in §46. ORioj. impie ‘fills’ < implet is anything but Latinizing (§46). Being a late learned importation makes fingir 'to feign, fake'—in contrast to its rustic counterpart heñir 'to knead' (brought in belatedly in §58)—ineligible for the company of ceñir, co(n)streñir, reñir, and teñir-, distinguir, on all counts, stands even farther apart (§48). From the strictly learned group -mitir, -sistir, dividir, eximir, im- and reprimir, the author might here have advantageously segregated ORioj. red-emir, -imir, especially in view of Ptg. remir (§49). For learned impedir im-pedīre (behind which he fails to sense the vernacular near-doublet empecer), Gassner posits the influence of pedir petere (§50), a situation paralleling the impact of ferir < ferīre on in-, re-ferir, from ferre. If this is so, why not go one step further and account for certain idiosyncrasies of the paradigm of pedir (Sp. pido, Ptg. peço) by invoking the influence of medir and, independently, positing an early contact with ex-pedīre (cf. des-, es-pedir) rather than charge the erratic behavior of pedir solely to the pressure of its near-synonym cupiō -ere, unproductive in Hispano-Romance? In the archaic eastern Glosas Silenses siegat = Sp. sea (Nos. 9, 11, 36) unmistakably reflects sedeat, not sequāt(ur) (§50); reverberations of Gassner's error reach as far as Pellegrini's textbook (1950). Regir, in the company of corregir, should chronologically be placed at the very head of learned words attracted into the orbit of the -e-/-i- group (§51). F firid (imp. pl.) does not plausibly owe its pretonic i to association with firades (§53); it is reminiscent, rather, of (ben)dizir, diz-imos, -ides, -it which Gassner inexplicably declares ‘wahrscheinlich halbgelehrt’ (§56) and should also be classed with pidimos, siguimos (§50) : diz-, -quis-, and viv-, with their etymological i, clearly precede the dez-, -ques-, vev- variants throughout. In yscamos ‘let us leave’ the intrusion of the inchoative pattern (or simply of the velar; cf. fir-g-amos) has been overlooked; exeāmus falls short of qualifying as a starting-point (§54). For OSp. fiedo/jeder faet- furnishes a smoother base than foet-ēre (§55). Why should derretir and desleír, of all verbs, rank as nonvernacular (§58), while patently erudite formations like adherir, di-, in-, su-gerir and con-, dis-cernir are far too hesitantly labeled ‘nicht ganz volkstümlich’ (§59)? If the shift gem-er > -ir required collateral semantic support, plañ-ir would have provided a more suitable partner than the fairly remote reír and reñir (§58); above all, why grammatically separate gem-er > -ir from rend-er > -ir (§51)? The whole treatment abounds in such qualifiers as ‘naturgemäss’, ‘selbstverständlich’ and in platitudinous passages: ‘Dass sich *rendere zu dieser Klasse geschlagen hat ist bei der im Altspanischen herrschenden Zerfahrenheit gleichfalls verständlich’ (§58). Further details are set straight in Cornu's devastating review, LGRPh, vol. 18, cols. 202–6 (1897).
84 As regards diphthongization in the ranks of other conjugation classes, Baist credited friego ‘I scrub’, pliego ‘I fold’, and riego ‘I water’ (from fricō, plicō, and rigō respectively) to the pressure of ciego ‘I blind’, niego ‘I deny’, and siego ‘I mow’, while estr(i)ego ‘I rub hard’ (from strigō ‘I draw furrows, make a halt’ [in ploughing]) and entr(i)ego ‘I deliver’ (from metathesized integrō ‘I renew, begin afresh‘) wavered throughout the waning Middle Ages. We find here heder traced to faetēre, hender attached to defender, and nevar associated with l(l)evar.
85 Manual elemental de gramática histórica escañola (Madrid, 1904).
86 Particularly commendable is Section (b) of §114.1, in which the author sets off (a) ō Ǔ stems: bullīre ‘to well up, bubble up’, confundere ‘to flow together, mingle’, excurrere ‘to run out’ (> Sp. escurrir 'to drain, ooze, trickle'), fugere 'to flee' (> OSp. foyr), ōrdīre 'to warp (cloth before putting it on the loom), begin (weaving)', putrēre 'to be rotten, decayed', recutere ‘to strike back’, subīre ‘to go, come, dive under’ beside ‘to approach, advance, mount, climb to’, suf-ferre, *-ferīre ‘to suffer’, from (b) ⊖ stems: abhorrēre ‘to shrink back from’, complēre ‘to complete, (ful)fill’, c⊖⊖perīre ‘to cover entirely, envelop’, ex + cō(n)sp(u)ere ‘to spit’, mollīre ‘to soften, make pliable’, monēre ‘to admonish’ (> Sp. muñir 'to summon') and (c) ū stems: addūcere 'to draw to oneself, bring' and mūgĩre ‘to bellow, low, roar’ > mugir (as against mulgēre ‘to milk’ > Arag. muir). But note that pūtridus and putrēre coexisted on the level of Latin and that c⊖⊖pmust, on that same level, have been contracted to *cōp-, as cōnsp- was to [*] cōsp-, so that the blurring of borderlines started long before the crystallization of pro to-Spanish. The author speaks as witness and participant in noting the growing prevalence of u in pudrir and particularly in repudrir ‘to rot completely’, (coll.) ‘to vex, irritate’. The conspicuous failure of oír to advance to *uir, an aberrancy on which he rightly insists without offering any explanation, may have been caused by avoidance of the homonymie collision with (h)uir fugere, the more so as in substandard speech huigo ‘I flee’ (= stand, huyo) drew perilously close to oigo ‘I hear’.
87 The etymologies of (de)rretir and engreír have been cursorily examined in §12 above.
88 Yet in §104 Menéndez Pidal contrasts vistes, vistamos with sientes, sintamos merely to exemplify the leveling vs. the diversification of the verb stem. Among the reviewers of the original Manual A. R. Gonçalves Viana, RH 10.608–14, esp. 612, noted that ei, not unlike ie, was occasionally reduced to i, citing ar(e)isco ‘churlish, surly, shy’ and iglesia ‘church’ beside Ptg. eigreja (but neglecting OSp. eglesia); while A. Morel-Fatio, Rom. 33.270–2 (1904), adduced coll. supusiendo for suponiendo, paving the way for a later argument of Garcia de Diego's (1914).
89 In the 2d ed. (1905), the 3d ed. (1914), and the 4th ed. (1918) only the introductory remarks in §114 were somewhat expanded, despite the author's new exposure to the highly idiosyncratic language of Elena y María. Among the reviewers of Manual 4 J. Ronjat, RLaR 62.435–6 (1923–24), commented on analogical duerma vs. ideal *durma, while J. Judand A. Steiger, Rom. 48.137–49 (1922), mistakenly denied the learned status of fingir (139), attributed floccu > fleco to recoil from near-homonymy with *clocca > llueca ‘brooding hen’ (142), and traced OSp. cuntir to pret, contigit, alleging the inordinate frequency of that tense (148).
90 Cantar de Mio Cid 264–9. In the revised ed. (Madrid, 1944–46) the etymon of degir, I repeat, was replaced (1229–30) : discēdere instead of dēcidere.
91 Thus, Manual 5 (1925; reprinted in 1929) adds to Leon. Arag. cuebre the compounds descuebre and encuebre (§114.2). Manual 6 (1941; frequently reprinted) emphasizes, in the preamble to §114, the contrasting treatment of -ti- in nouns (vitiu ‘fault, blemish’ > OSp. vezo ‘custom’, pǓteu ‘well, fountain’ > pozo) and verbs (mētiō > mido, recutiō > recudo), explaining it by discrepant degrees of analogy involved, not by two dissimilar sets of rules for actual sound change; on the theoretical possibility of the latter course see W. S. Allen, RPh. 17.173–4 [1963–64], on W. P. Lehmann, Historical linguistics: An introduction. Also, the author stresses (§114.4) the late consolidation of preferences currently valid (A. de Nebrixa's 1492 grammar, we are reminded, still favored nonmetaphonic variants in pretonic position: recebir, regieron, sentiendo, seguiente, sofrir). The masterly Orígenes del español (1926) documents early shifts of conjugation class (implir) and vestiges of weakening resistance to such shifts (top. Benbiber [a.d. 978], Benvivre [a.d. 1125] = mod. Belbimbre bene vīvere), traces to proto-Spanish 10th-century glosses and to other archaic texts such salient verb forms as 6 sierben, 11 bestan, 6 pitent, and 6 tingen, follows the velar interfix /k, g/ in its inroads, and supplies many other details (§72.1, §73.1–2), yet stops short of deepening the author's own earlier analyses (1904, 1908) of the key problems; the revised editions (1926, 1950) supply no remedy in this respect.
92 While in his early monograph ‘El dialecto leonés’ the author merely exemplified the local survival of -ie- (RABM 3:14.146 [1906]), his Orígenes §27.3 (starting with the 1926 ed.) supply important paleo-toponymic data: Liestra ~ Listra, Asieso ~ Asiso, Agierbe ~ Agirbe (with g /j/), Xavierre ~ Scavierri ~ Exavirr, beside mod. Javierre, Espierre, Espiarlo, perhaps not going far enough in affirming the authenticity of -i- spellings.
93 Spanische Grammatik auf historischer Grundlage §27.6 (Halle, 1910). Useful collateral information will be found in §25.3 (dialectal distribution of -er/-ir), §26.2 (reduction of -ie- to -e-), and §27.3 (remnants of inflectional -io [pres, ind.] and -ia etc. [pres, subj.] in Old Portuguese, Old Aragonese, and Modern Asturian). The range and accuracy of Hanssen's bibliographic information were exemplary. Among the reviews of the book the most thorough, by H. R. Lang (RR 2.331–47 [1911]), concerns itself only with points of collateral interest: igl-, igr-eja ‘church’ and OPtg. per-go, -co, -ço ‘I lose’ (337).
94 In view of the importance of these ‘Vorstudien’ (in part merely descriptive, in part explicative as well), and because, being difficult of access, they have—unlike Gassner's inferior monograph—been virtually ignored by all successors except Menéndez Pidal, it may be useful to extract some points from them. In ‘Sobre la conjugación de Gonzalo de Berceo’, An. Chile 90.231–80 (1895), Hanssen sees rend-er torn between the models of vend-er (-e-) and defend-er, entend-er (-ie-), quite apart from the pressure exerted by -ir (244); reconstructs the asymmetric paradigms (a) 1 sirvo, 2–3 sierv-e(s), 6 -en, 7 sirva, 12 *servi (?) beside (b) duermo, duerm-e(s), -en, duerma, duermi (246); discovers traces of -ie- in the present tense of (des)servir and vestir (248); notes fluctuations such as pe- ~ pi-dimos, se- ~ si-guir, ve- ~ vi-stir (249); precedes others in attributing the adventitious u of pudiendo to imitation of pudieron (252); and posits a regression, ingeniously, to account for sentiāmus > [*]senzamos > [*]sentiamos > sintamos (258). But, in analyzing metaphony, which he unaccountably calls ‘suavización’ (251–8), he stumbles, like other experts before him, over the reputed contradiction debiendo ~ pidiendo and, by way of emergency solution, offers an untenable hypothesis. The ‘Suplemento a la conjugación de Berceo’, ibid. 90.773–81, contains an unrealistic reconstruction of the paradigm of vestir in proto-Castilian (*vexo, viest-es, -e, -en, vestimos, *vexa, etc., in criticism of the author's own earlier [*]veszo, p. 257, equally implausible) alongside a well documented comment on OSp. es = eres (779–80; cf. fn. 3, above). ‘Sobre la conjugación del Libre de Apolonio’, ibid. 91.637–65 (1895), barely inventories forms extracted from a mediocre edition later superseded by C. C. Marden's; from the survey of the key verbs: pedir, seguir, vestir, complir, nozir, etc. (648–52) one gathers the early ascendancy, in the underlying Old Aragonese subdialect, of í and ú; requerir is flanked by 1 requiro, from which mod. requiero may have sprouted through contamination with quiero. ‘Estudios sobre la conjugación aragonesa’, ibid. 93.391–409 (1896), parade noteworthy -ir infinitives, some of them characteristically eastern: conplir, con- and re-querir (again 1 requiro), constreñir, enclodir, exir (7 esca), possedir, and reír (402–3). In the finest of the five monographs, ‘Estudios sobre la conjugación leonesa’, ibid. 94.753–807 (1896), Hanssen inspects with microscopic precision the rivalry of such typically western -er verbs as aduzer, changer plangere, combater, costrener, encher, escriver, erger, morrer, parter, premer, reñer (but fiñir), rier, viver with their respective -ir variants (including both rustic constrennir and learned constringir), reconstructs such sets—placing them at the threshold of the literary tradition—as sigo, siegu-es, -e, -en, seguimos, siga, Isiegue! beside si(e)rvo, sierv-es, -e, -en, servimos, si(e)rva, ¡sierve!, and pieces together for sofr-er/-ir (781) a triple development almost as complicated as that of exir (779): 1 su-, sue-, so-fro, 4 sofr-imos, -emos, 7 su-, sue-, so-fra, etc.
95 Hanssen here refers the reader to an earlier passage in his book (§10.2) in which he adduced tinea ‘moth, bookworm’ > tiña ‘beehive spider, ringworm; stinginess’ and cuno ‘wedge’ > cuño, but seems to forget his own important qualification: ‘Veränderungen ... welche jedoch schwer in Regeln zu fassen sind und dialektisch schwanken’. What we are facing here is plainly a weak phonetic change, as defined in Lingua 11.263–75 (1962) and in Ibérida 6.127–71 (1961 [–63]), in part on the basis of findings originally presented in Lg. 36.284–91 (1960). But this time the cause of the weakness is hardly to be sought in dialect mixture; it is not impossible that, contrary to Hanssen's surmise, the change occurred first, analogically, in the verbal paradigm (*feñe > fiñe), then spread haltingly to other form classes.
96 This rubric marks the one inconsistency in Hanssen's otherwise almost flawlessly executed design: to begin with, there is no evidence of any earlier ei in the progeny of exīre, and further, even if the existence of the falling diphthong were demonstrated, there would be no safe way of projecting it onto the phase of Latin to which the remaining root vowels invariably refer.
97 No subsequent monographic treatment of this side issue has come to my attention. A quarter-century ago M. A. Zeitlin wrote a splendid note, ‘Unstressed ascending diphthongs in Spanish’ [pretonic ie, ia, io, and iu reduced to i], MLF 24.84–90 (1939), a paper which unfortunately contained a misleading incidental statement on s(i)eglo. Note OFr. siegle, the variant favored by Chrétien de Troyes (Perceval 1669).
98 Gramática histórica de la lengua castellana §200 (Halle, 1913).
99 He extracted sieguen from Brutail's collection of Navarrese charters and sierven from El libro de Marco Polo; see §13 above. The other addendum, constrenga = Sp. constriña, which now flanked OLeon. receba = Sp. reciba, was a philologist's reward for scanning the Documentes para el estudio de la historia de Aragón.
100 From our angle it is regrettable that Hanssen's Gramática (and the monographs underpinning it) should have elicited major critiques in a frame predominantly either phonological (A. Castro, RFE 1.97–103, 181–4 [1914]) or syntactic (O. J. Tallgren, NM 18.138–56 [1917], and F. Krüger, RFE 8.311–8 [1921]); even L. Spitzer, while recognizing, in the wake of E. Staaff (RDR 2.427–8 [1911]), the chapter on inflection as Hanssen's most meritorious, centered his attention on syntax and stylistics (LGRPh. 35 [1914), cols. 206–12).
101 ‘Spanish etymologies’, MP 7.49–60, esp. 55–9 (1909).
102 Étude sur l'ancien dialecte léonais d'après des chartes du XIIIe siècle 298–302 (Uppsala, 1907). The author omits the present tense from his purview, but incidentally discusses the conjugation class of constrener, contradicer, and correger (314–5).
103 Altspanisches Elementarbuch §§130–31 (Heidelberg, 1908). Though the book is dedicated to the author's direct teacher Meyer-Lübke, the paragraphs at issue show estrangement from the latter's doctrine where it was indefensible (1890) and adherence to Menéndez Pidal's trail-blazing Manual elemental. The 2d, revised ed. of the Elementarbuch (1921) shows the transfer of the discussion to §§124–5, but no reappraisal of the substance of the problem.
104 The two paragraphs are marred by several blemishes. Unless ‘analogy’ means simply ‘extension, diffusion’ without any intervening change in circumstances, I find no excuse for applying it (§130) to the effect of metaphony in the 3d sg. impf. : sintié, murié, udié. Confusingly enough for the tyro whom the book purports to guide, the following paragraph mentions only the less frequent variants unaffected by metaphony: sentié, pedié, complié. If tondir hints at the predecessor of tundir ‘to shear’, the base is tondēre, not tunděre. An inaccurate etymon is supplied for arrepentirse.
105 Elementos de gramática histórica castellana §166 (Burgos, 1914).
106 The author traces quiriendo to Álvarez Gato, Santa Teresa, and Cervantes (citing the var quisiendo as proof of contamination with the preterite); trimió to Hernán Mexía; tiniendo to Santa Teresa, arguing that mod. coll. virtió may be due to the pressure of the compounds in -vertir. In his analysis of adquirir and conveniente he underrates the power of Latinization.
107 As expected, the treatment of the problem in García de Diego's Gramática histórica española 197–203 (Madrid, 1951; reprinted 1963) shows a lamentably retrograde attitude. The collateral pressure of such ‘strong’ preterites as pude, quise is now brushed aside, and appeal is made instead to a ‘weak’ (i.e. subphonemic) metaphony to account for pudiendo—an inopportune retreat, by implication, to the stand of M. A. Colton (1909), impugned with such vigor by T. Navarro (RFE 10.26–56 [1923]) and made even more vulnerable by the advent of structuralism. García de Diego reconstructs the proto-Castilian paradigm sirvo, sierv-es, -e, -en, serv-imos, etc. (200), allows for agredió ~ acridió, su-mergió ~ -mirgió as coexistent standard forms, and credits the e of concerniente, from concernir [as against the -i- of (say) sirviente, from servir], to the old wavering cern-er ~ -ir.
108 Nothing better characterizes the prolonged spell of apathy than the fact that É. Bourciez's superficial statement in his Éléments de linguistique romane §360 (Paris, 1910)—to the effect that mid- spread from mid-iendo, at length causing mido to replace *miedo *metō (sic)—was left unchanged in the 2d ed. (1923), the 3rd (1930), and even the 4th (1946; reprinted 1956), the last revised by J. Bourciez. Yet our ignorance, in the author's own words, cried out for remedy: ‘Toutefois, pour des raisons qui restent obscures, cette influence n'a pas uniformément prévalu’.
109 ‘Le présent dans la conjugaison castillane’, Annales de l'Université de Grenoble 34.339–66, esp. 352–62 (1923), a paper apparently left incomplete except through its link with the author's later Études (1929); see fn. 111 below.
110 The monograph operates throughout with modern instead of medieval spellings (jibia ‘cuttlefish’, luci(e)llo ‘tomb’, rubio ‘blond‘); indulges in unwarranted reconstructions (*īmplō, *cōmplīre, [*]recutīre, *rubbium, [*]concīpīo, etc.); leaves out of the reckoning, in establishing chronological sequences, the ever-present possibility of wholesale replacement, under analogical pressure, of individual forms and partial sets (thus, VLat. *mītō for mēttō is superfluous: 353); and misses or, worse, glosses over all critically difficult points—the characteristically glib treatment of the spread of mid- to mide(s), etc. One suspects misprints at crucial junctures (354 : medide), and the asterisk is absent where most needed (ibid.: pedo, piedes ...).
111 ‘Études de philologie hispanique: 2. miento, visto, duermo’, RH 77.25–44 (1929).
112 The five categories set up (31–9) include: (a) before [λ], [Ŋ], [š]: OSp. castiello, Yéñego, Guadiex; (b) under the distant impact of [j]: vestiō > *viestio (sic) > visto; (c) before s + Cons., including lengthened s: OSp. avi(e)spa, ni(é)spera, pri(e)sco, pri(e)ssa, ri(e)stra, vi(é)spera—as against fiesta, hiniesta, siesta, tiesto; (d) in hiatus: Dios ‘God’, judio ‘Jew’, mío beside mió ‘my’ (-io < *-ieu); (e) in isolated words: mi(e)rla merula, remilgo *remellicu, pingo ‘I drip’ < pendicō, si(e)glo saeculu. Criticism: remilgo presupposes *remiellgo (cf. collocō ‘I place’ > *cuellgo > cuelgo 'I hang'), hence should be transferred to (a); the change sieglo > siglo must not be too heavily charged to the word's relative lack of stress in, say, siglo XV 'fifteenth century', since at the start the preponderant meaning was ‘world’; not unlike pretiu (see RLiR 21.1–55 [1957]), sieglo blends features of vernacular and learned development in Old Spanish (-ie- vs. -gl-), while OGal.-Ptg. segre adds one Gallicist overtone (cf. Sp. milagro beside Ptg. milagre, OGal. miragre mīrāculu). Most important : Fouché telescopes events by failing to discriminate between very old instances of monophthongization, like those under (d), and relatively late adjustments, amenable to direct observation. For Group (c) he leans on F. Krüger, Studien zur Lautgeschichte westspanischer Mundarten 80–3 (Hamburg, 1914), J. D. M. Ford, Old Spanish readings xiii (Boston, 1911), and G. Millardet's review of Ford's reader in RLaR 57.123–7, esp. 124–5 (1914), but fails to adopt the reviewer's plausible explanation of Sp. viejo ‘old’ (vetulu x veter-e, *-u, as in OSp. viedro, cf. topon. Murviedro, Pontevedra) and offers instead a completely unrealistic conjecture. Sieglo > siglo and eglesia > iglesia might have deserved joint consideration on account of the -gl- segment they share.
113 Thus, the documentation available for dialectally colored forms like vieste(n), highlighted in the earlier treatises by Keller, Gassner, Menéndez Pidal, Hanssen (1895, 1913), etc., did not prevent Fouché from having recourse to a gratuitous asterisk.
114 In mentir and sentir, for example, syllable-final n was credited with the braking power (without any independent corroboration) ; for the -ie- of fier- the metathesis of [rj] in feriō was made responsible; erguir and hervir were declared analogical, etc.
115 It is amusing to read Fouché's hair-splitting insistence on *engriedio rather than *engrie(d)io as the inferred midway point between Sp. engrío and VLat. *ingrediō if the underlying derivation itself is worthless, as Corominas rightly observes. For derretir 'to melt' Fouché relies on García de Diego's desperate attempt (RFE 9.152 [1922]) to salvage an old hypothesis weak beyond recovery (dēterere ‘to rub off, wear out’ x reterere ‘to rub again and again‘).
116 Straight historical grammars, from J. Alemany Bolufer, Estudio elemental de gramática histórica de la lengua castellana 3 §§225–7 (Madrid, 1911), to G. B. Pellegrini, Grammatica storica spagnola §149 (Bari, [1950]), have been quite unenlightening; the latter follows in Gassner's footsteps, to the point of repeating the erroneous analysis of siegat, and throws in a few misprints for good measure: r. rijo, erguir. Even B. Pottier, Introduction à l'étude de la philologie hispanique: 2. Morphosyntaxe 87 ([Bordeaux], 1958) is satisfied with declaring that mentir, dormir probably split off from pedir, escopir on account of the heavier use of radical-stressed forms, a peculiarity not only undemonstrated but unlikely, if proved correct, to have acted in isolation.
The authors of even the best-known language histories—a genre traceable to the thirties—tend to sweep the problem under the rug; this is especially true of W. J. Entwistle, The Spanish language together with Portuguese, Catalan, and Basque 164 (New York, 1938; orig. London, 1936), and of R. K. Spaulding, How Spanish grew 84–5, 111–2 (Berkeley and Los Angeles, 1943). R. Lapesa, in his monograph Asturiano y provenzal en el Fuero de Avilés §35 (Acta Salm. 2:4; Salamanca, 1948), has assembled carefully distilled data on the local pres. subj. (sparing use of metaphony: compla; preservation and spread of -ia: feria, partiant, podia; extension of the velar interfix: esca, perga); he gives the barest minimum of attention to the fringes of our problem, and apparently none to its kernel, in the five successive editions of his Historia de la langua española (Madrid, 1942–62). Equally disappointing, if not more, is V. Šišmarëv, Očerki po istorii jazykov Ispanii 105–6 (Moscow and Leningrad, 1941).
Among the monographic treatments one might have expected M. Křepinský's Prague dissertation (1918; translated and annotated by García de Diego : Inflexión de las vocales en español [Madrid, 1923; reprinted 1962]), to shed light on the development at issue. But while many of its pages are devoted to an adjective like te-, ti-bio, and to the stems of certain verbs (on exir see 85), the all-important interplay of stem and conjugational suffix elicits no comment worth noting. The same bias has spilled over into the attitude of D. Alonso, La fragmentación fonética peninsular 105–54 (Madrid, 1962).