j | i v SUzANNEFLESCAMAN Silva-Corvalén, Carmen. 1985, “Modality and semantic change”. In: Fisiak (ed) 1985: $47-72 tional Journal of American Linguistics 1; 200-217 Strang, Barbara M. H. 1968. Modern English structure. London: Edwatd ‘Contribution & étude du parler mignard,” Bulletin linguistique de Pars 7: th Clos. 1982, “From propositional to textual and expres- quelques remarques Mélanges de grammaire trenti¢me anniversaire du Bon usage, 343-66. Gembloux: Ducuot. Welmers, William E. 1973, African language sructures, Berkeley: Univer- ty of Cat use of imperfect and perfect tmerican Philolggical Associa 1967, Sanskrit grammar. Cambridge: Harvard Uni- th rept. of 2d ed. 1889), 1988. “A cross-linguisc survey of the grammaticalization Studies in Language 12: 517. Wimet, Mae 19669. °impaa t ypooisiqu” Le Frans Mod. Le pus compet nara ta Wine 1516: 6 976. tudes de morphosyniaxe verbale; Pars: Klincksick, Mare. 1983. “L'imparfait forain.” In: Verbe et phrase dans les langues romanes. Mélanges offens & Louis Mourin, 160-67, [Romania Gandensia 20), Woodbury, Anthony. 1986, “I study of Sherpa and English. In tons of tense and evientality: A. sf and Nichols (eds) 1986: 188-200, Sues i Language 1, 1-10 (199). ight sed ‘THE CREATION OF TENSE AND ASPECT SYSTEMS INTHE LANGUAGES OF THE WORLD JOANL, BYBEE & — OSTENDAHL University of New Mexico University of Stockholm Introduction | A characteristic and universal property of natural languages ‘Many terms used in grammar may be understood as ambiguous between referring to “notior2 JOAN L. BYBEE & OSTEN DAHL, ther denote certain meaning or context of use (which pre- ‘in any language) or a category, like the English Progres- certain meaning anda certain expression! Our ‘grams’ are the second kind. ‘general theory of grams with tense and aspect as a special case. 1. Comparing the meaning of grams across languages sis may be similar across languages has often been ics during the current century. React ism where grams were taken to form sets of semantic opposi- ike phonemes, were necessarily language-spec ‘developments in Ameri “THE CREATION OF TENSE AND ASPECT SYSTEMS but also that there isnot any intresting sem ‘matical and lexical morphemes (Chomsky 1 suggest that standing of grammatical meaning, oF more ambitiousl, ‘theory of grammatical meaning. "The impetus fr the current paper i the fact the pendently and using diJOAN L. BYBEE & OSTEN DAHL ese languages was available only through published mate- nce grammars, The test ofthe hypotheses required iden- infesion a belonging toon ofthe percents of ak “THE CREATION OF TENSE AND ASPECT SYSTEMS 5s scope, they both found that a large majority (between 70% and 80%) of the b.impertective, indicating that the situation is viewed as not bounded; popes (cal conavns in Bybee’ tu) indetng rudy) indicating that a sit- vant atthe moment of speech56 JOAN L. BYBEE & OSTEN DAHL. “THE CREATION OF TENSE AND ASPECT SYSTEMS 7 3. Form/meaning correlations Both of our studies show that it is possible to make substantive the semantic content of grams of tense and ly characterizable. In addition, however, both studies unco- lation between meaning and form even among grammatical ally have periphrastic expression, whi tive usually have bound expression.38 JOAN L. BYBEE & OSTEN DAHL Table 2. Leticl sures for ese and spect grams tet source Grammatical eteory LanguagevExamples Sear Tintendon fore ‘movement toward intention ure eel huvefte + inf obligation ture invebe + perfects PastParipe "pastor prtstive sh perfects pastor perectve throw sways perfect movement fom» perfect» vouree Pato perfective > prowenire Iperectve movement» progresive ‘THE CREATION OF TENSE AND ASPECT SYSTEMS© JOAN L. BYBEE & OSTEN DAHL ‘have 10, want 10, need to. try to start to, ee. x closed or ase, as well sin the case of incorporated nouns, the lexical classes, the more dificult it isto determine if they are open or close. (On the other hand, many closed classes consist of » single member, or 4 single overt member contrasting with zero. The English Perfect gram, hhave + Past Partciple and the Progressive, be + ing, ate each the sole ‘THE CREATION OF TENSE AND ASPECT SYSTEMS 6 ‘pic: each member of a semantic category may belong to a different ‘lass asthe tenses past (-d), future (wil and perfect (have + past partic ple) do, or single clase may express meanings from diferent semantic categories, a the modal auxliris, which express tens (will and shal), ‘eon modality Come readings of must and should) or epistemic modality ts meaning, ut teas n part by creole cinience. 42 Loss of lexical meaning and fixed position grammatically defined: ‘conditions, in questions, ity of more than one postion may india JOAN L. BYBEE & OSTEN DAHL. quences for developing grams, which we will discuss in two groups. Under the heading of loss of semantic autonomy’ we wil diggus inthis section the changes in the types of semantic relations grams may enter into, In the next Section under the heading of ‘generalization of meaning’ we wil discuss the properties derivable from the wider appicabty of meaning which lacks specific lexical content. The fixing of a gram’s postion reflects the loss of semantic autonomy {nthe following way: grams differ from lexial morphemes in that the posi- “THE CREATION OF TENSE AND ASPECT SYSTEMS 6 ‘which may be modified in it ability sense: (1) He can easily swim a mile. However, the more developed grams do not take modification. Compare the following, in which the adverb can only be interpreted as modifying the rain verb: @) He will easly swim a mile, @) Hes easly swimming the mile. 43. Generality of meaning ‘Whereas lexical meaning i specific and referent {ng is highly general and relational in qual ‘ulated a difference between material andre to characterize the difference between lex6 JOAN L. BYBEE & OSTEN DAHL“ ‘mon uses, which makes it applicable'tosuibjectson {ora discussion ofthe development of will Infact, inflectional grams such as perfective, past tobe applicable to all the verbs of a language. ‘THE CREATION OF TENSE AND ASPECT SYSTEMS 6 is much more difficult, and perhaps unnecessary, to apply the notion, of obligatorinss to periphrasic grams. As periphrastic grams develop they are gradually becoming obligatory, but there is no one point which can be singled out asthe point at th a set of grams comes to be obligatory, since this often entails the creation of an unmarked category — a meaningful unit which hhas no overt marker. ‘The notion of obligatoriness also has what might be called a semantic side: the national domain encoded by a set of o must be touched on whenever the appropriate gram languages with obligatory evidential grams, nin every sentence. In languages ker may chose when to suppl Sequence, but may be omitted on subseq ward 1953).6 JOAN L. BYBEE & OSTEN DAHL ‘extent upon semantic factors. That is, the phonological reduction necessary {or afixation moves hand in hand with the rediction of semantic content in grammaticization. Asfxation is not a discrete event with clear before (unbound) and ater (bound) the Dahl evaluat citerion was chosen to determine afithood: whether the gram ‘bound by the analyst or author of the grammar. It is important to under- stand, however, what the prerequisites are fora gram to be considered an affix and written bound, (One requirement is that the gram appear in fixed position — mova be and permutable elements are not considered afixes and are hardly ever written bound. Another requirement is that no open cass items may inter- ‘yene between the gram and its head, ic. the noun or verb stem the English articles, she and a/an, although they appear are not considered affixes because an adjective may come ‘noun. However, a gram could meet both of these l be written separately. For instance, the Mandarin hich occurs directy after the verb, meets the riteria ‘gram is not ‘Once the gram loses the stem, it becomes very a8 does extreme reduction in size and phonologi the stem. Thus all the factors that the development ams tends to occur at approxim: verb stem, Such ion would imply ad mantic and formal developments in grammati ‘we have mentioned above shows a clear corel ‘the agglutination wi nite connection betwe zation, Of course, the “THE CREATION OF TENSE AND ASPECT SYSTEMS point of speech is said to preced: to the simple past which also correspond to common diachronic sources for perfect a