654 IX. Flexion

5. References Meder, Gregor & Mugdan, Joachim (1990), “Alle reden von Häufigkeit …: Anmerkungen zum COBUILD (1987) ϭ Collins COBUILD English Thema ‘Frequenz’ in der Morphologie”. In: Bas- Language Dictionary, Editor in Chief John Sinclair, sarak, A[rmin] & Bittner, D[agmar] & Bittner, A[n- Managing Editor Patrick Hanks. London, Glas- dreas] & Thiele, P[etra] (eds.), Wurzel(n) der gow: Collins Natürlichkeit: Studien zur Morphologie und Phono- Ingo, Rune (1978), Suomen kielen pluratiivit eli logie, Vol. IV. Berlin: Zentralinst. für Sprachwis- monikkosanat.A˚ bo (Turku): A˚ bo Akademi (Med- senschaft (Linguistische Studien A 208), 87Ϫ108 delanden fra˚n Stiftelsens för A˚ bo Akademi For- Mugdan, Joachim (1983), “Grammatik im Wörter- skningsinstitut 34) buch: Flexion”. In: Wiegand, Herbert Ernst (ed.), Karlsson, Fred (1985), “Paradigms and Word Studien zur neuhochdeutschen Lexikographie, Vol. Forms”. In: Laskowski, Roman (ed.), Studia III. Hildesheim etc.: Olms (Germanistische Lin- gramatyczne, Vol. VII. Wroclaw etc.: Ossolineum guistik 1Ϫ4/82), 179Ϫ237 (Prace Instytutu Je˛zyka Polskiego 61), 135Ϫ154 Plank, Frans (1981), Morphologische (Ir-)regulari- Karlsson, Fred (1986), “Frequency Considerations täten: Aspekte der Wortstrukturtheorie. Tübingen: in Morphology”. Zeitschrift für Phonetik, Sprach- Narr (Studien zur Deutschen Grammatik 13) wissenschaft und Kommunikationsforschung 39, Quirk, Randolph & Greenbaum, Sidney & Leech, 19Ϫ28 Geoffrey & Svartvik, Jan (1985), A Comprehensive Karlsson, Göran (1957), Suomen kielen nukuksissa Grammar of the English Language. London, New ja hereillä-tyyppiset paikallissija-adverbit. Helsinki: York: Longman Suomalaisen Kirjallisuuden Seura (Suomalaisen Soboleva, P[olina] A. (1979), “Defektnost’ para- Kirjallisuuden Seuran Toimituksia 250) digmy i semanticˇeskoe tozˇdestvo slova”. Voprosy Kiparsky, Paul (1974), “Remarks on Analogical jazykoznanija 1979.5, 37Ϫ47 Change”. In: Anderson, J[ohn] M. & Jones, Spencer, Andrew (1991), Morphological Theory. C[harles] (eds.), Historical Linguistics, Vol. II: The- Oxford, Cambridge/MA: Blackwell ory and Description in Phonology. Proceedings of ˇ ˙ ˇ the First International Conference on Historical Lin- Stejnfel’dt, Evi A. (1963), Castotnyj slovar’ sovre- guistics, Edinburgh 2ndϪ7th September 1973. Am- mennogo russkogo literaturnogo jazyka. Tallinn: ˙ sterdam, Oxford: North-Holland; New York: Naucˇno-issledovatel’skij institut pedagogiki Eston- American Elsevier (North Holland Linguistic skoj SSR Series 12b), 257Ϫ275 Vincent, Nigel (1987), “The Interaction of Periph- LDEL (1984) ϭ Longman Dictionary of the English rasis and Inflection: Some Romance Examples”. Language. Harlow/Essex: Longman In: Harris, Martin & Ramat, Paolo (eds.), Histori- cal Development of Auxiliaries. Berlin etc.: Mouton Maslov, Ju[rij] S. (1964), “Zametki o vidovoj defek- de Gruyter (Trends in Linguistics: Studies and tivnosti (preimusˇcˇestvenno v russkom i bolgar- Monographs 35), 237Ϫ256 skom jazykach)”. In: Larin, B[oris] A. & Safronov, G[erman] I. (eds.), Slavjanskaja filologija. Le- Wurzel, Wolfgang Ullrich (1988), “Derivation, Fle- ningrad: Izd. Leningradskogo universiteta, 82Ϫ94 xion und Blockierung”. Zeitschrift für Phonetik, Matthews, P[eter] H. (1972), Inflectional Morphol- Sprachwissenschaft und Kommunikationsforschung 41, 179Ϫ198 ogy: A Theoretical Study Based on Aspects of Latin Verb Conjugation. Cambridge etc.: Cambridge Univ. Press (Cambridge Studies in Linguistics 6) Fred Karlsson, Helsinki (Finland)

68. Periphrasis

1. Defining periphrasis 1. Defining periphrasis 2. Suppletive periphrasis I: paradigm symmetry 3. Suppletive periphrasis II: inflectional The term periphrasis (from Greek perı´phrasis generality ‘paraphrase, circumlocution’), in its most 4. Categorial periphrasis 5. Periphrasis in inflection, derivation, and general sense, refers to the use of longer, syntax multi-word expressions in place of single 6. Periphrasis forms words, or “circumlocution” (this Latin term 7. Conclusion is simply a loan translation of the Greek 8. References term). In a narrower philological context, pe- 68. Periphrasis 655 riphrasis is one of the canonical literary rhe- are expressed by single-word forms. Thus, torical figures (cf. Lausberg 21963: 69), e.g. there is no form with the features ‘future’ and English to be hit with Cupid’s arrow ‘to fall in ‘subjunctive’. The future participle in -urus love’, or German Elbflorenz ‘Florence on the with the copula in the present subjunctive Elbe [Dresden]’. For the purposes of this (e.g. facturus sit lit. ‘he be going to do’) fills handbook, a still narrower, grammatical this gap, as illustrated by the array of 3rd sense of the term is relevant. Periphrasis re- person singular forms of facere ‘do’ in (1); the fers to a situation in which a multi-word ex- periphrastic form is given in square brackets. pression is used in place of a single word in (1) indicative subjunctive an inflectional paradigm: “When a form in a present facit faciat paradigm consists of two or more words it is imperfect faciebat faceret periphrastic” (Matthews 1981: 55), e.g. more perfect fecit fecerit beautiful instead of *beautifuller (cf. (b) be- low). But unlike concepts such as ‘mor- future faciet [facturus sit] pheme’ or ‘auxiliary’, the concept of ‘periph- The future subjunctive is required in certain rasis’ has never been an important issue in subordinate clauses that are subject to a se- linguistics, and it has not been used as a cru- quence-of-tense rule, so that there is a real cial ingredient in any formal grammatical need to fill the gap in the paradigm. framework. In general, the term is used in de- (b) In English, many adjectives have an in- scriptive and typological work in an intuitive flected comparative formed by adding the sense, and attempts at clarifying the concep- suffix -er to the base form (e.g. warm ϳ tual content of the term have remained the warm-er), but other adjectives lack such a exception (cf. Zawadowski 1959; Thümmel form, and comparison is conveyed by a 1966: 157Ϫ164; Rose´n 1992). Historically, multi-word (or phrasal) expression contain- the abstract noun periphrasis in its grammati- ing the adverb more (e.g. beautiful ϳ more cal sense is a back-formation from the adjec- beautiful). In some intuitive sense, the tive periphrastic. In traditional Latin gram- phrasal expression serves the same function mar, the term conjugatio periphrastica was as the inflected form with other adjectives, used to refer to infrequent combinations of and linguists have often felt the need to as- participial forms with the copula, e.g. fac- similate such “periphrastic expressions” to turus sit ‘is going to do [subjunctive]’ (cf. the single-word forms, so that it becomes Kühner & Stegmann 21914: 180). This usage possible to say that more beautiful is the seems to have been the starting point of the “comparative form” of beautiful, just as contemporary use of the term periphrastic. warmer is the comparative form of warm (cf. Moreover, in current usage the term peri- Matthews 1981: 54f.). phrastic is almost always applied to verbal (c) The French construction with the auxil- constructions although there is nothing in its iary aller ‘go’ as in je vais le faire ‘I’m going definition that would require such a restric- to do it’ is commonly referred to as futur pe´- tion. (Thus, it would be perfectly coherent riphrastique. Apparently, the reason for in- but distinctly unusual to say that German cluding it in the inflectional paradigm of the has a periphrastic instrumental case, using French verb is that it expresses a grammatical the preposition mit.) This restriction seems to meaning (cf. Art. 27). This sense of periphras- be a historical accident, reflecting the origin tic is reflected, for instance, in the definition: of the term in the Latin conjugatio periphra- stica. It is shared by the related term auxil- “periphrastic: denoting a construction, especially iary (which always refers to verbs, although one involving a verb, in which one or more auxil- “auxiliary nouns” or “auxiliary adverbs” cer- iary words are used to express grammatical distinc- tions, as opposed to the direct inflection of the lexi- tainly exist; cf. Art. 78), but not by the quasi- cal item involved” (Trask 1993 s.v.) synonym analytic. This latter term has its ori- gin in morphological typology (cf. Art. 115) However, it is not self-evident what should and is contrasted with synthetic. It appears count as an inflectional meaning, and the that analytic is more common in Europe and practice of grammarians is rarely rigorous particularly in Russian linguistics, whereas and consistent. For instance, few would de- periphrastic is more common in English. scribe the French construction je veux le faire There are three types of cases in which ‘I want to do it’ as a de´side´ratif pe´riphras- periphrastic forms may be included in inflec- tique, although it is quite similar to the “peri- tional paradigms: phrastic future” je vais le faire. The reason (a) Latin verbs are inflected for different for restricting periphrastic to the construction tenses and moods, but not all combinations expressing future time reference may be 656 IX. Flexion mainly the expectation that the future more than one morphological category is “should be” an inflectional category, whereas combined, e.g. tense and mood in Latin (1). no such expectation exists for the desiderative. Thus, a language in which verbs inflect only In many cases such expectations are obviously for tense or nouns only for number could not influenced by the model of Latin grammar, for have this kind of periphrasis. Another exam- instance if English is said to have a “future ple from Latin involves the interaction of tense” (cf. Huddleston 1995) or German a voice and aspect. Latin has monolectic forms “vocative case” (cf. Häcki Buhofer 1987 on for the passive voice only in the infectum as- older grammars; see also Ch. II). pect (e.g. in the present and imperfect Types (a) and (b) can be regarded as pe- “tenses”), but not in the perfectum aspect riphrasis in the narrower sense, which “can (e.g. in the perfect and pluperfect). In the lat- be recognized only where there is a clear gap ter case, a periphrasis involving the perfect in the inflectional patterns, which the phrases passive participle and the copula is used, as serve to fill” (Hockett 1958: 212; cf. Smir- the 3rd person singular forms of capere ‘take’ nickij 1956; Mel’cˇuk 1993: 355). In (a), the in (2) show. gap is filled in order to create paradigm sym- (2) active passive metry in the forms of a lexeme (cf. 2); in (b), present capit capitur it achieves inflectional generality across dif- imperfect capiebat capiebatur ferent subclasses of lexemes (cf. 3). Type (c), perfect cepit [captum est] where a multi-word combination expresses pluperfect ceperat [captum erat] some additional semantic distinction, is less directly relevant to morphology, but it must The situation in Russian is quite similar. The be taken into consideration here because nei- 3rd person singular forms of (s)delat’ ‘do’ in ther the boundaries between the different (3) show that only the imperfective aspect has kinds of periphrasis nor that between periph- bound expression in the passive, while the rasis and inflection are very sharp (cf. 4). For perfective passive is periphrastic: similar reasons, the notion of periphrasis (3) active passive need not be confined to inflection (cf. 5). imperf. past delal delalo-s’ There are no established terms for the present delaet delaet-sja three subtypes of periphrasis. In this article, perf. past s-delal [bylo s-delano] suppletive periphrasis will be used for types perf. future s-delaet [budet s-delano] (a) and (b), and categorial periphrasis for type Another example, again from Russian, in- (c); (cf. Aerts 1967: 3; Rose´n 1992: 18f. for volves the interaction of tense and aspect. somewhat similar terminological proposals). Only perfective verbs have monolectic forms Suppletive periphrasis shares with true sup- in the future, while the imperfective future is pletion (cf. Art. 52) the function of supplying periphrastic: forms for the inflectional paradigms that are not formed in the regular way (cf. Vincent (4) perfective imperfective 1987: 242; Börjars et al. 1997), but it is of present Ϫ delaet course a very different mechanism. For want past s-delal delal of a better alternative, the rather old-fash- future s-delaet [budet delat’] ioned term monolectic form will be used as This Russian paradigm is perhaps not a good the opposite of periphrastic form (cf. Aerts example of a gap in terms of paradigm sym- 1967: 3; Rose´n 1992: 11); an equivalent dis- metry, because it is not very symmetrical to tinction applicable to grammatical meanings begin with. An alternative analysis would re- is that between bound expression and peri- gard the perfective future form sdelaet as pre- phrastic expression (cf. Bybee & Dahl 1989: sent, thus restoring the formal symmetry. The 51). The term periphrasis form will be em- future meaning could be attributed to the in- ployed for inflectional forms which appear teraction between the aspectual and the tem- only in periphrastic constructions (cf. 6). poral interpretation (a perfective situation is not easily construed as simultaneous with the 2. Suppletive periphrasis I: moment of speech). On this analysis, there paradigm symmetry would be no gap in the paradigm. But it is more likely that although this alternative Gaps which are filled by periphrastic forms analysis is diachronically accurate, synchron- for the purpose of paradigm symmetry can ically (4) is a realistic description of Russian only arise in inflectional systems in which grammar. The diachronic scenario for the rise 68. Periphrasis 657 of the asymmetry thus helps to explain the tically, periphrastic forms should be analyzed need for a periphrastic gap filler. like monolectic forms, whereas formally they So far all examples have been from verbal should be regarded as syntactic phrases inflection, but it is easy to imagine a similar (Matthews 1974: 171). Periphrasis can thus situation in nominal inflection: A language be used as an argument for “separationist” might have a monolectic form of the instru- approaches to morphology (cf. Aronoff 1994; mental case in the singular, but not in the Beard 1995), which stress the mutual inde- plural, so that a periphrasis involving an in- pendence of the formal expression of a mor- strumental adposition must be used. No ex- phological element and the semantic contri- ample of this kind has come to my attention, bution it makes (cf. Börjars et al. 1997). but this may be purely accidental. An empirical problem with the gap-filling In these cases of gaps filled by suppletive periphrasis to restore paradigm symmetry, it view of periphrasis is that it may not always is clear that a good case can be made that the be clear whether there are enough monolectic periphrastic forms belong to the inflectional forms to constitute a paradigm that can be paradigm. However, this does not necessarily said to have gaps. In (1)Ϫ(4), the gaps are a mean that they are considered as morpholog- distinct minority of the paradigm cells, but ical entities. A classical structuralist textbook what if there are more gaps than filled cells? states this explicitly in discussing Latin peri- An example of this comes from Hungarian, phrastic passives (cf. 2): which has a monolectic form expressing ‘1st “These phrases are not part of the inflectional mor- person singular subject Ϫ 2nd person object’, phology of the Latin verb, because the structure of but no others that express both subject and phrases is syntax, not morphology”. (Hockett non-3rd person object. The paradigm thus 1958: 212) looks as in (5), where only singular object Similarly, a standard introduction to mor- forms are given for the sake of simplicity phology comes to the conclusion that seman- (ke´r- ‘ask’; engem ‘me’, te´ged ‘you’):

(5) subject/object 1st singular 2nd singular 3rd singular 1st singular Ϫ ke´r-lek ke´r-em 2nd singular [engem ke´r-sz] Ϫ ke´r-ed 3rd singular [engem ke´r-ø][te´ged ke´r-ø] ke´r-i 1st plural Ϫ [te´ged ke´r-ünk] ke´r-jük 2nd plural [engem ke´r-tek] Ϫ ke´r-itek 3rd plural [engem ke´r-nek][te´ged ke´r-nek] ke´r-ik

Only ke´r-lek ‘ask-1.sg.subj&2.sg.obj (I ask 1985: 5). If a certain inflectional pattern is you)’ is a monolectic form, all other combi- not applicable to some members of the word nations involving a 1st or 2nd person object class, a periphrasis may fill this gap. An ex- are “periphrastic” (though this term is never ample of this type is the English periphrastic used in Hungarian linguistics for these cases), comparative (cf. 1), which allows adjectives using the accusative forms of the personal that lack the bound comparative (*beauti- pronouns. It is debatable whether the exis- fuller) to have a comparative form (more tence of a single monolectic form is sufficient beautiful). to set up a morphological paradigm that con- 3.1. Examples sists largely of gaps filled by periphrasis. Per- A further example comes from Romanian, haps more naturally, one might conversely where most nouns have a bound oblique case regard (5) as a syntactic paradigm with a sin- form, but masculine proper nouns lack it. gle gap, which is filled by the “anti-periphras- The auxiliary word lui (originally ‘his, to tic”, “compacted” form ke´rlek (cf. 3 for fur- him’) is used to allow these proper names to ther instances of this problem). occur in the oblique case form: (6) base form oblique case 3. Suppletive periphrasis II: masc. common prieten-ul prieten-ul-ui inflectional generality ‘the friend’ fem. proper Ana Anei Almost by definition, inflectional forms are ‘Anna’ highly general, i.e. they apply to all or almost masc. proper Petre [lui Petre] all members of a word class (cf. Bybee ‘Peter’ 658 IX. Flexion

A periphrastic gap-filler may achieve both (9) affirmative negative paradigm symmetry and lexical generality awun t-awun ‘do’ simultaneously. In Classical Greek, the 3rd xˆun ta-xˆun ‘become’ person plural form of the middle perfect of a gun ta-gun ‘give’ verb like gra´pho ‘write’ does not have a cˇüxün [cˇüxün t-awun] ‘wash’ bound expression and can only be expressed Similarly, in Maltese only a small class of periphrastically, using the middle perfect par- (mostly inalienable) nouns allow the use of ticiple and the copula: possessive person-number suffixes, e.g. dar (7) singular plural ‘house’, id ‘hand’, but the large majority of 1st ge´gram-mai gegra´m-metha nouns require a periphrastic construction 2nd ge´grap-sai ge´graph-the with ta’/tiegØ ‘of’ (this has been called “ana- 3rd ge´grap-tai [gegram-me´noi eisı´] lytic genitive”; cf. Koptjevskaja-Tamm 1996): Taken in isolation, the paradigm in (7) seems (10) base form possessive form 1st singular to involve periphrasis for paradigm symme- dar dar-i ‘house’ try (cf. 2). But in this case, the form in ques- id id-i ‘hand’ tion is not lacking in all Greek verbs. Many ktieb [ktieb tiegØ-i] ‘book’ verbs whose stem ends in a vowel do have a The existence of a paradigm whose gaps monolectic form for this paradigm cell, so the are filled becomes even more questionable periphrasis here also serves to ensure inflec- when the monolectic forms are not only very tional generality: few, but also quite irregular. Thus, in Afri- (8) 3rd singular 3rd plural kaans only a handful of verbs have a bound pepaı´deu-tai pepaı´deu-ntai ‘educate’ past tense, and all of these are irregular (cf. tetı´me¯-tai tetı´me¯-ntai ‘honor’ Donaldson 1993: 222): ge´grap-tai [gegram-me´noi eisı´] ‘write’ (11) present past In (6)Ϫ(8), the lexemes that lack a particu- is was ‘be’ lar monolectic form are the clear minority, weet wis ‘know’ and there may be very specific reasons why an kon ‘can’ these forms are impossible (thus, in Classical werk [het gewerk] ‘work’ Greek the regular form *ge´graph-ntai is sim- begin [het begin] ‘begin’ ply phonotactically ill-formed). But in the Similarly, in the Romance languages only case of the English comparative, matters are four adjectives have bound comparative more complicated. Forms like *negativer forms, and these are all suppletive. All regu- ‘more negative’ or *beautifuller seem to be lar forms are periphrastic, e.g. in Spanish: phonologically well-formed, and the peri- phrastic comparatives are actually the major- (12) positive comparative ity (at least in terms of lexical types). More- bueno mejor ‘good’ over, the periphrastic forms are not restricted pequen˜o menor ‘little’ to those lexemes that lack a monolectic form: oscuro [ma´s oscuro] ‘dark’ In many cases, both forms may be used side caliente [ma´s caliente] ‘warm’ by side, e.g. likelier or more likely. Thus, fill- In the extreme case, only a single lexeme ing a gap in certain lexemes is probably not has a monolectic form of a certain category. the only motivation for the existence of An example comes again from Hungarian, these forms. where only the verb lenni ‘be’ has a bound Gap-filling as the motivation for periphra- expression for future tense, whereas all other sis becomes even more difficult to maintain verbs only have a periphrastic future (fog when only a small minority of lexemes in the ‘take’ plus infinitive; the forms given are 3rd word class allow the monolectic form. For in- person singular): stance, in Lezgian only eighteen verbs allow negation by means of a prefix in non-finite (13) infinitive future forms (e.g. the verbal-noun form). All other lenni lesz ‘be’ verbs require a periphrastic form using the ´ırni [fog´ ırni] ‘write’ auxiliary t-awun ‘neg-doing’, and all verbs felelni [fog felelni] ‘answer’ but two optionally allow this periphrastic As in (5), one has to ask whether so few non- form (cf. Haspelmath 1993: 133). Some ex- periphrastic forms justify calling all the other amples are given in (9). forms (suppletively) periphrastic. One might 68. Periphrasis 659 just as well call these few monolectic forms As a result, the semantically equivalent “anti-periphrastic” or “compacted”. würde-periphrasis is much more common Another doubtful case is that of the Ger- with these verbs in contemporary German, man hypothetical mood. Theoretically, every and in practice the old hypothetical mood German verb has a monolectic hypothetical- (the “past subjunctive”) survives only in mood form, but in the regular (weak) verbs, strong verbs (and even some of these sound this form is homonymous with the past tense. awkward in this form):

(14) infinitive past hypothetical kommen kam käme ‘come’ sein war wäre ‘be’ loben lobte [würde loben](?lobte) ‘praise’ schwimmen schwamm [würde schwimmen](?schwämme) ‘swim’

Thus, in some sense the German würde-pe- lence of the new periphrastic past and the old riphrasis is tied to gap-filling contexts, but past tense then leads to the loss of the old this link as fairly weak Ϫ the forms in paren- past tense, except in the case of very common theses in (14) are not outright impossible, verbs, where high frequency helps to preserve they are just awkward. Conversely, the the old forms. würde-periphrasis is not impossible with Such a process of gradual grammaticaliza- strong verbs like kommen, though it often tion is ultimately responsible for most of the sounds very colloquial. patterns examined in 3.1. Grammaticaliza- tion is a continuous, unindirectional, cyclic 3.2. Grammaticalization change that turns periphrastic expressions Since speakers are generally guided by a (i.e. categorial peripherasis; cf. 4) into syn- maxim of economy, there is a certain ten- thetic forms, which are further reduced until dency that favors complementary distribu- they are again replaced by new roundabout tion: When a monolectic form exists, this will expressions (cf. Art. 145). Such grammaticali- tend to block the periphrastic form, and zation processes may at some point result in when a periphrastic form becomes the usual a kind of complementary distribution which one, the monolectic form will tend to disap- creates the impression that the periphrastic pear. But this principle is only a tendency, be- expressions exist in order to fill a gap, but cause there are deviations in both directions. that is by no means necessary. For instance, Moreover, in the long run periphrastic con- colloquial French has gone through a process structions tend to replace monolectic forms of extension of the ‘have’-perfect to the per- quite independently of whether the latter fective past function (similar to the cases of have a defective distribution or not. The Afri- Afrikaans and colloquial German), and the kaans example (11), for instance, shows the result of a change that has been going on in a old perfective past form (the passe´ simple) similar form in varieties of German for many has been lost from this register of French. centuries. Standard German still has bound But unlike the cases of Afrikaans and Ger- past tense forms for every verb, but in the man, the passe´ simple does not even survive colloquial language (and particularly in in a few isolated cases, and there was appar- many southern dialects) only the haben-past ently never anything resembling a comple- is used with most verbs. This replacement mentary distribution over different verb had nothing to do with any kind of difficulty types. in forming the bound past tense form. It re- Even if there is a clear phonological reason sulted from the continued grammaticaliza- for the nonexistence of certain inflected tion of the haben/seinϩparticiple construc- forms, the periphrasis need not owe its exis- tion, which originally had a special perfect tence to its gap-filling function. For instance, meaning, but generally came to be basically Latin comparatives are formed by a suffix equivalent to the simple past tense (cf. Bybee -ior, which cannot, however, be attached to et al. 1994: 51Ϫ105). The functional equiva- vowel-final stems for phonological reasons. 660 IX. Flexion

For these stems, a periphrasis with magis is meanings that are more specific than the used: meanings already expressed grammatically in the language at the time” (Bybee et al. 1994: (15) positive comparative 133). These authors do not link the use of longus longior ‘long’ periphrasis in any way to inflectional para- felix felicior ‘happy’ digms. A periphrastic expression is simply arduus [magis arduus] ‘steep’ one which expresses a grammatical meaning idoneus [magis idoneus] ‘suitable’ in a multi-word construction. A periphrasis Could it be that the periphrastic compara- can be identified if there is a conventional tive, which became prevalent in the Romance construction in a language which expresses a languages (cf. (12)), began its existence as a grammatical meaning, and where there is a filler of phonologically conditioned gaps? particular lexical item (an auxiliary word) Perhaps, but another scenario is at least as that regularly combines with all members of likely: At the time of Classical Latin, from a word class to express this meaning. which the data in (15) are taken, the magis- The question, of course, is: What is a periphrasis was already widely used in the grammatical meaning (cf. Art. 27)? In the colloquial language. The written norm still worst case, all and only the meanings that are required the comparative in -ior for most ad- clearly grammaticalized in a language that jectives, but where this was difficult to form, linguists happen to know well (e.g. Latin or the colloquial periphrasis was admitted into English) are counted as grammatical. Gram- the written language. If so, suppletive periph- marians rarely justify their descriptive deci- rasis is not as different from categorial pe- sions explicitly. For instance, in one descrip- riphrasis as it might have seemed. From a tion of Welsh, two methods for forming verbs strictly structural and descriptive point of are presented, the “inflected” method (by view, it might be desirable to restrict the in- adding endings to a verb stem) and the “peri- cursion of phrasal forms into the domain of phrastic” method (by using an auxiliary verb inflection as tightly as possible, but this in combination with the verbal noun to form hardly does justice to the reality of languages a compound tense), without any discussion and their speakers. (cf. King 1993). Similarly, in a grammar of Lezgian, the chapter on verbal inflection in- cludes a section on “periphrastic tense-aspect 4. Categorial periphrasis categories” (Haspelmath 1993: 146Ϫ148), in which the Periphrastic Habitual, the Peri- Examples of categorial periphrasis are the phrastic Future, and the Hearsay Evidential English have-perfect (explicitly excluded from are described. However, no justification is the inflectional paradigm by some authors, given for this particular choice of categories. e.g. Wallis 61765: xxvϪxxvii, 102Ϫ111; Hock- This does not necessarily mean that such jus- ett 1958: 212; cf. Art. 62), the French aller- tification is not possible, but grammatical de- future (je vais chanter ‘I am going to sing’), scriptions usually assume that their choice of and the Spanish estar-progressive (estoy can- periphrastic constructions is unproblematic. tando ‘I am singing’). There exist no mono- Three specific criteria and one more general lectic forms of any of these categories in the criterion for recognizing categorial periphra- languages in which they occur, so these peri- ses are worth considering (cf. Bertinetto 1990 phrastic forms do not fill a gap defined by a for a longer list of possible criteria). system of monolectic forms, i.e. “real” inflec- The first specific criterion asks whether the tional forms. Hence, these constructions can- kind of meaning expressed by the periphrasis not be properly said to be “circumlocutions” is expressed by monolectic forms elsewhere in for anything, i.e. if they are periphrases, this the language. Thus, if a language has bound can be understood only in an extended sense tense forms, then a complex construction ex- of the term, e.g. relative to the monolectic pressing tense (e. g. the English will-future) forms of another language. However, the use will count as a periphrastic form (cf. Smir- of the term periphrasis for such constructions nickij 1956; Mel’cˇuk 1993: 355 on “analytic is widespread in the literature. Statements forms”). The problem here is that assigning such as the following are typical: “[‘Go’] oc- specific grammatical meanings to broader curs in periphrastic futures in English and grammatical categories is often not at all various Romance languages” (Lehmann 1995: straightforward. For instance, is the Spanish 29); “New periphrases develop to express estar-progressive an aspectual category on a 68. Periphrasis 661 par with the imperfective/perfective past dis- affixes are not always clear-cut. For instance, tinction? Is the English have-perfect a tense? the Polish past tense czytałem ‘I read’ looks The second specific criterion is semantic like an ordinary monolectic form czyta-ł-em non-compositionality (cf. also Art. 82). Ordi- ‘read-past-1.sg’, but occasionally the 1st per- nary phrases must be interpretable composi- son singular marker -(e)m may occur else- tionally, but periphrastic constructions are where in the clause, e.g. co-m czyta-ł? ‘what- often non-compositional. For instance, while 1.sg read-past (what did I read?)’. This shows the meaning of I want to break it can be de- that at least in these cases the form is not rived from the component parts want and to completely bound yet, although grammati- break, the meaning of I have broken it cannot calization is advancing and these split forms be derived from the components have and are becoming rarer in the modern language. broken. Non-compositionality is always pre- It is impossible to draw clear lines between sent when the main verb is in a periphrasis the various stages of grammaticalization pro- form (cf. 6). This criterion is perhaps easier cesses; there are often cases that are indeter- to apply than the first, but it is only a suffi- minate not just for linguists, but also for the cient, not a necessary condition for peri- speakers. phrastic status. Finally, grammaticalization helps us to The third specific criterion concerns the understand why certain grammatical cate- range of forms of the auxiliary element that gories are very often expressed periphrasti- occur in the periphrasis. In an ordinary com- cally in the world’s languages, while others bination of a finite and a non-finite verb, strongly tend to be expressed as monolectic there are no restrictions on the forms of the forms. For instance, “perfect and progressive finite verb, but in a periphrasis sometimes usually have periphrastic expression, while only a subset of the forms are allowed. For past, and perfective and imperfective usually instance, in the German werden-future only have bound expression” (Bybee & Dahl present indicative (and perhaps subjunctive) 1989: 56; cf. also Dahl 1985; Bybee et al. forms of werden are allowed, but not past 1994: 104Ϫ124). In the nominal domain, it tense forms (e.g. wird kommen [lit. becomes has been noted for temporal markers that an- come] ‘will come’, but not *wurde kommen terior-durative (‘until’) and posterior-dura- [lit. became come]; cf. Vincent 1987 for dis- tive (‘since’) markers tend to be bound, cussion of a similar constraint on the Italian whereas anterior (‘before’) and posterior (‘af- venire-passive). ter’) are almost always periphrastic, i.e. ex- More generally, we need a comprehensive pressed adpositionally, in the world’s lan- theory of grammaticalization in order to guages (cf. Haspelmath 1997: 145). The ex- understand periphrasis. Grammaticalization planation for these form-meaning corre- is a very complex, multi-faceted phenome- lations is that the meanings of the bound non, and there are indefinitely many possible categories arise only after a longer process of degrees of grammaticalization (cf. Lehmann grammaticalization, so by this time the ele- 1995 for a systematic treatment of the vari- ments have also undergone a significant ous parameters of grammaticalization). Once amount of formal reduction and agglutina- we have such a theory, the definition of peri- tion. The periphrastic categories usually ex- phrastic construction is easy: The more gram- press younger, less grammaticalized mean- maticalized a construction is, the more it can ings. claim to have periphrastic status (unless gram- maticalization has proceeded far enough to turn it into a monolectic form, of course). If 5. Periphrasis in inflection, derivation, we recognize this, it does not necessarily be- and syntax come easier in practice to identify a peri- phrastic construction, but we have reduced Inflection is the prototypical domain for sup- this problem to another, independent prob- pletive periphrasis (as it is for suppletion), be- lem for which solutions have to be proposed cause only inflection is organized in tight, (cf. Bertinetto 1990 for a similar conclusion). symmetrical paradigms in which gaps can be- The grammaticalization perspective also come salient. But insofar as derivation may helps us to solve the problem of delimiting also be regular, periphrasis within derivation periphrastic forms against monolectic forms. is certainly not unimaginable, even though it In some cases, this is not an easy task, be- is not normally called periphrasis (cf. also cause the boundaries between free words and Art. 52 on suppletion in derivation). The 662 IX. Flexion

English derivational suffix -ology, for exam- Somewhat more interesting are cases of ple, productively forms nouns denoting a sci- periphrasis in syntax: At first it might seem ence from nouns denoting a possible subject that this is an incoherent notion, because matter for a science, e.g. climate ϳ climatol- there are no “syntactic paradigms”. How- ogy, Egypt ϳ Egyptology, volcano ϳ volcanol- ever, it is not difficult to find syntactic phe- ogy. However, the Latinate suffix -ology does nomena that provide a striking analog of in- not combine felicitously with nouns like com- flectional paradigms, gaps, and periphrasis in puter, so for the science of computers the morphology. Again, a good example comes “periphrastic” term computer science must be from English, where only a small subclass of used. Another possible example are multipli- verbs can occur without complications in in- cative numerals in English, e.g. on-ce, twi-ce, terrogative and negative clauses. In (16), this thri-ce. For higher numbers, “periphrastic” well-known pattern is represented in such a numerals are required (four times, five times, way that the similarities with morphological etc.). suppletive periphrasis become apparent.

(16) decl., affirm. interrogative negative You are here Are you here? You are not here You saw her [Did you see her?][You did not see her] (*Saw you her?) (*You saw not her)

Clearly, “periphrastic do” is periphrastic in 6. Periphrasis forms much the same way as the cases of morpho- logical periphrasis, but the filled gaps in (16) The term periphrasis form is introduced here are not morphological monolectic forms. Did for inflectional forms of lexemes which com- you see is a syntactic phrase which replaces bine with auxiliary elements to form periph- the impossible syntactic phrase *saw you. rases and which have no other function in the What this case shares with the cases of mor- language. An example of a periphrasis form phological suppletive periphrasis is the re- is the Modern Greek form in -i (derived from striction of the non-periphrastic forms to cer- the aorist stem) that combines with the auxil- tain high-frequency verbs. In French, too, in- iary e´xo (‘have’) to form a perfect periphrasis verted interrogative clauses (e.g. viens-tu? (e.g. dhe´no ‘I tie’, periphrasis form dhe´-s-i, ‘are you coming?’) tend to be superseded by periphrastic perfect e´xo dhe´si ‘I have tied’). circumlocutions with est-ce que (lit. ‘is it In Swedish, the perfect periphrasis consists of that’, e.g. est-ce que tu viens?), and these are ha (‘have’) plus a periphrasis form (called often called “periphrastic questions” (cf. “supine” in Scandinavian linguists) in -t/-tt/ Behnstedt 1973). Thus, the general principles -it (e.g. skriva ‘write’, periphrasis form of periphrasis remain the same in syntax, in- skrivit, periphrastic perfect jag har skrivit ‘I flection and derivation (which is in accor- have written’). dance with the notion of a syntax-inflection- The concept of periphrasis form is not yet derivation continuum; cf. Bybee 1985: 81Ϫ generally recognized in theoretical linguistics, 110). but it is necessary for a complete theory of Finally, lexical substitution, i.e. the use of grammar. It appears that in the majority of a different lexeme, may make up for missing periphrases, the form of the lexeme is iden- inflectional forms. For instance, English tical to some inflectional form that occurs in- modal auxiliaries like must and can lack non- dependently in the language, e.g. an infinitive finite forms, and when these are required by or a participle (cf. (1)Ϫ(4), (7)Ϫ(8), (13)Ϫ the syntactic environment, they are usually (14)). This reflects the diachronic origin of replaced by the quasi-synonyms have to and periphrases in ordinary combinations of a be able to. This has been described as periph- main verb plus a subordinate verb. But as the rasis (cf. Westney 1995), but in other respects periphrasis is grammaticalized, the connec- it is more like suppletion. Unlike standard tion between the non-finite form in the pe- suppletion, however, the “periphrastic mod- riphrasis and in other parts of the grammar als” have to and be able to are not old rem- may be severed. This separate development nant forms, but they have full paradigms and may concern only the semantics (as in Eng- can occur also in finite forms (e.g. she can ϭ lish a broken heart vs. I have broken), or it she is able to; cf. also Vincent 1987: 242; may concern the morphological form as well Börjars et al. 1997: 168). (as in Swedish skrivit, contrasting with the 68. Periphrasis 663 past participle skrivet; originally the two the view that it is not possible to separate forms were identical), or the non-periphrastic morphology and syntax neatly anyway: The use of the non-finite form may disappear en- two are linked inextricably through the con- tirely from the language (as in Modern Greek tinuous and ubiquitous process of grammati- dhe´si, going back to the old infinitive, which calization. In fact, most inflectional forma- has fallen into disuse elsewhere). In the latter tives arise through grammaticalization in the two cases, the result is a special periphrasis first place, so periphrasis is in a sense the ba- form. And even the first case (English broken) sis of inflection. In this perspective, it appears could be described as a periphrasis form, legitimate to extend the notion of periphrasis which (for diachronic reasons) happens to be even further to semantic categories which are homonymous with the adjectival past partici- never expressed by monolectic forms, but ple. which show a sufficiently high degree of Like the non-finite verb forms from which grammaticalization to be described as part of they develop, periphrasis forms may be vari- the verbal paradigm rather than only in the able. For instance, the Lezgian negative pe- syntax (i.e., to categorial periphrasis). riphrasis illustrated in (9) consists of a nega- tive auxiliary plus a preceding periphrasis form whose shape depends on the aspectual 8. References stem of the auxiliary, e.g. rax-un tawuna ‘not Aerts, Willem Johan (1967), Periphrastica: An In- having talked’, but rax-an tijiz ‘not to talk’ vestigation into the Use of eı˜nai and e´khein as Auxil- (Haspelmath 1993: 134). Synchronically this iaries and Pseudo-Auxiliaries in Greek from Homer looks like an unusual kind of agreement, and up to the Present Day. Chicago: Argonaut [orig. nothing is known about the diachronic origin 1965: Proefschrift, Universiteit van Amsterdam] of the construction. An analog of these verbal periphrasis Aronoff, Mark (1994), Morphology by Itself: Stems and Inflectional Classes. Cambridge/MA, London: forms can be found in the nominal domain: MIT Press (Linguistic Inquiry Monographs 22) Russian has a nominal case form that occurs only in combination with certain prepositions Beard, Robert (1995), Lexeme-Morpheme Base (sometimes called “prepositive case”, some- Morphology: A General Theory of Inflection and Word Formation. Albany: State University of New times called “locative”, reflecting its origin), York Press e.g. v Rim-e ‘in Rome’ (*Rim-e alone does not occur). One could say for the sake of Behnstedt, Peter (1973), Viens-tu? Est-ce que tu viens? Tu viens? Formen und Strukturen des direkten consistency that v Rime is the periphrastic in- Fragesatzes im Französischen. Tübingen: Narr (Tü- essive case of Rim ‘Rome’, and that Rime is binger Beiträge zur Linguistik 41) its periphrasis form. Bertinetto, Pier Marco (1990), “Perifrasi verbali italiane: criteri di identificazione e gerarchia di pe- rifrasticita`”. In: Bernini, Giuliano & Giacalone Ra- 7. Conclusion mat, Anna (eds.), La temporalita` nell’acquisizione di lingue seconde. Milano: Angeli, 331Ϫ350 Periphrasis (in the narrower sense of supple- tive periphrasis) presents an important chal- Börjars, Kersti & Vincent, Nigel & Chapman, Carol (1997), “Paradigms, Periphrases and Pro- lenge to theories of inflection in that it shows nominal Inflection: A Feature-based Account”. that there are potentially two distinct notions Yearbook of Morphology 1996, 155Ϫ180 of an inflectional paradigm (cf. Art. 62) Bybee, Joan L. (1985), Morphology: A Study of the which need not coincide: The paradigm may Relation between Meaning and Form. Amsterdam, be construed Philadelphia: Benjamins (Typological Studies in (a) as the set of all word-forms belonging to Language 9) a lexeme, or Bybee, Joan L. & Dahl, Östen (1989), “The Cre- (b) as the set of all elements filling the cells ation of Tense and Aspect Systems in the Lan- defined by the inflectional categories that guages of the World”. Studies in Language 13, Ϫ can be expressed for the lexeme. 51 103 Bybee, Joan & Perkins, Revere & Pagliuca, Wil- On the second interpretation, periphrastic liam (1994), The Evolution of Grammar: Tense, As- forms are admitted as members of the para- pect and Modality in the Languages of the World. digm, but at the price that the paradigm is Chicago, London: Chicago Univ. Press no longer a purely morphological notion. But Dahl, Östen (1985), Tense and Aspect Systems. Ox- arguments were provided in this article for ford, New York: Blackwell 664 IX. Flexion

Donaldson, Bruce (1993), A Grammar of Afri- Mel’cˇuk, Igor (1993), Cours de morphologie ge´ne´- kaans. Berlin: Mouton de Gruyter (Mouton Gram- rale (The´orique et descriptive), Vol. I: Introduction mar Library 8) et Premie`re partie: Le mot. Montre´al: Presses de Häcki Buhofer, Annelies (1987), “Die Kasus des l’Universite´ de Montre´al; [Paris]: CNRS Deutschen Ϫ wissenschaftsgeschichtliche und me- Rose´n, Haiim B. (1992), Die Periphrase: Wesen und thodologische Überlegungen”. Deutsche Sprache Entstehung. Innsbruck: Univ. Innsbruck (Innsbruk- 1987, 137Ϫ150 ker Beiträge zur Sprachwissenschaft, Vorträge und Haspelmath, Martin (1993), A Grammar of Lez- Kleinere Schriften 57) gian. Berlin: Mouton de Gruyter (Mouton Gram- Smirnickij, Aleksandr I. (1956), “Analiticˇeskie mar Library 9) formy”. Voprosy jazykoznanija 1956.2, 41Ϫ52 Haspelmath, Martin (1997), From Space to Time: Thümmel, Wolf (1966), Das Problem der periphra- Temporal Adverbials in the World’s Languages. stischen Konstruktionen. München: Fink (Forum München: Lincom Europa (Lincom Studies in Slavicum 5) Theoretical Linguistics 3) Trask, R[obert] L. (1993), A Dictionary of Gram- Hockett, Charles F. (1958), A Course in Modern matical Terms in Linguistics. London, New York: Linguistics. New York: Macmillan Routledge Huddleston, Rodney (1995), “The Case Against a Vincent, Nigel (1987), “The Interaction of Periph- Future Tense in English”. Studies in Language 19, rasis and Inflection: Some Romance Examples”. Ϫ 399 446 In: Harris, Martin & Ramat, Paolo (eds.), Histori- King, Gareth (1993), Modern Welsh: A Comprehen- cal Development of Auxiliaries. Berlin etc.: Mouton sive Grammar. London: Routledge de Gruyter (Trends in Linguistics: Studies and Koptjevskaja-Tamm, Maria (1996), “Possessive Monographs 35), 237Ϫ256 Noun Phrases in Maltese: Alienability, Iconicity Wallis, John ϭ Iohannes (61765), Grammatica ling- and Grammaticalization”. Rivista di Linguistica 8, vae anglicanae. Londinum: Bowyer [11653 Oxford: 245Ϫ274 Lichfield; reprinted in: Wallis, John (1972), Gram- Kühner, Raphael & Stegmann, Carl (21914), Aus- mar of the English Language, with translation and führliche Grammatik der lateinischen Sprache, 2. commentary by J. A. Kemp. London: Longman] Teil, Vol. II. Hannover: Hahn [reprint 1992] Westney, Paul (1995), Modals and Periphrastics in Lausberg, Heinrich (21963), Elemente der literari- English: An Investigation into the Semantic Corre- schen Rhetorik. München: Hueber spondence between Certain English Modal Verbs and Lehmann, Christian (1995), Thoughts on Grammat- their Periphrastic Equivalents. Tübingen: Niemeyer icalization. München: Lincom Europa (Lincom (Linguistische Arbeiten 339) Studies in Theoretical Linguistics 1) Zawadowski, Leon (1959), Constructions gramma- Matthews, P[eter] H. (1974), Morphology. Cam- ticales et formes pe´riphrastiques. Krako´w: Ossoli- bridge etc.: Cambridge Univ. Press neum (Prace je˛zykoznawcze 18) Matthews, P[eter] H. (1981), Syntax. Cambridge etc.: Cambridge Univ. Press Martin Haspelmath, Leipzig (Germany)