Habitual Aspect in English

http://eng.sagepub.com

Journal of English Linguistics

DOI: 10.1177/0075424205286006 2005; 33; 339 Journal of English Linguistics

Robert I. Binnick The Markers of Habitual Aspect in English

http://eng.sagepub.com/cgi/content/abstract/33/4/339 The online version of this article can be found at:

Published by:

http://www.sagepublications.com

can be found at:Journal of English Linguistics Additional services and information for

http://eng.sagepub.com/cgi/alerts Email Alerts:

http://eng.sagepub.com/subscriptions Subscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://eng.sagepub.com/cgi/content/refs/33/4/339 Citations

by Sorin Ciutacu on October 31, 2009 http://eng.sagepub.comDownloaded from

http://eng.sagepub.com/cgi/alerts

http://eng.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://eng.sagepub.com/cgi/content/refs/33/4/339


10.1177/0075424205286006ARTICLEJEngL 33.4 (December 2005)Binnick / Markers of Habitual Aspect in English

The Markers of Habitual Aspect in English

Robert I. Binnick

University of Toronto

Contrary to the traditional account of habitual aspect in English, the only marker ofhabituality is will, with its past tense, would. Used to functions as a kind of anti-present-perfect tense, contrasting past states of affairs with present ones. In certain contexts, varioustenses, and not merely the simple ones, may receive habitual interpretations, but these are nottheir meanings. Generic readings are contextual interpretations; there are no markers of ge-neric aspect in English.

Keywords: habitual aspect; generic aspect; would; used to

Introduction

The generally accepted account of the expression of habituality and genericityin English is this: would (as in example 1) and used to (2) are markers of pasthabituality, will (3) marks present or timeless habituality, and the simple present (4)and past (5) tenses may mark habituality as well. The meaning of the habitual pasttense, as in (5), is pretty much the same as that of would in (1) or used to in (2). Allthree express a past condition, that is, a state of affairs holding of a past interval oftime but one that no longer obtains (2). Furthermore, habitual expressions may, ingeneral, also express generic aspect, as in (6), referring to a characteristic propen-sity and not to an actual series of eventualities. Examples may be ambiguous be-tween the two, habitual and generic, readings; (7) can refer either to the habits of aparticular tiger or to the traits of tigers in general.

(1) Over and over and over again, we would make these journeys, and some-times they would last as long as three hours at a stretch. (Charles Dickens,Great Expectations, chap. 12)

(2) They used to cater to the elite; they were the temples of high culture, tryingto raise the level of public appreciation of the arts. But nowadays, museumsare becoming more responsive: it’s the public, not the institutions, that callsthe shots. (http://www.gc.cuny.edu/about_the_grad_center/conversations/conversation_arts.htm)1

(3) The dress is kept in a bag but every now and then she will bring it out for re-view. (http://www.tntleague.com/misc/bettyhill.html)

Journal of English Linguistics, Vol. 33 / No. 4, December 2005 339-369DOI: 10.1177/0075424205286006© 2005 Sage Publications

339



(4) My dad always yells instead of talking, and he has for years. He yells at mefor doing stuff . . . and not doing stuff. (http://lost.crystal-mind.net/bio.htm)

(5) From time to time she had a “bit of a go” at me, and I suppose I was ratherresentful. (http://www.pentecostal.org.uk/testimony1.htm)

(6) He builds canoes and kayaks for a living. (http://www.joekaz.net/friends.html)

(7) The tiger ate many fish live and whole. . . . (http://www.bergen.org/Update/03academy.html)

This standard account is almost wholly incorrect. Used to is neither a past tensenor a marker of habituality (see the section “Used to” below). The simple tenses donot have habitual meanings, though they may receive habitual interpretations incontext (see “The Simple Tenses”), nor is a habitual interpretation restricted to thesimple tenses (see “Habitual Aspect in English”). The only marker of habituality inEnglish is the modal will, and would is simply its past tense (see “Will and Would”).There is a considerable difference in interpretation between expressions such asthey used to go, they went, and they would go. None refers to a past state (see“Habituality as a State”). That a past condition no longer obtains is a conversationalimplicature and no part of the meaning of the expression itself (see “The Habitual asa Condition That No Longer Obtains”). And genericity is a contextual interpreta-tion; there are no markers of generic aspect in English (see “Generic Aspect”).

Habitual Aspect in English

It is usually assumed that such sentences as (1) through (5) express a grammati-cal category of habitual aspect (Comrie 1976; Freed 1979; Dahl 1985; Leech 1987;Brinton 1988), which indicates a series of events or episodes, viewed as a whole(Lyons 1977, 716; Leech 1987, 5; Tagliamonte and Lawrence, 2000, 326), and dis-tributed over an explicitly or implicitly given interval of time (Dahl 1985, 97).

Following Comrie (1985, 39), Harrison (2002) sums up by saying that “the ha-bitual aspect refers to a situation that is protracted over a long period of time, or asituation that occurs frequently during an extended period of time, to the point thatthe situation becomes the characteristic feature of the whole period,” even if the sit-uation in question does not literally hold at a particular time (Cutrer 1994, 150).

As noted above, English has traditionally been viewed as having no singlemarker of habitual aspect but as having several: would, used to, and will. The simplepast and present tenses may receive habitual interpretations as well, especiallywhen combined with temporal adverbials such as usually, often, always, from timeto time, and so on. Indeed, eventive (or episodic) expressions in the present tensenormally have either habitual (4) or generic (6) interpretations and cannot refer toan actual eventuality ongoing at present.

340 JEngL 33.4 (December 2005)



(4) My dad always yells instead of talking, and he has for years. He yells at mefor doing stuff . . . and not doing stuff.

(6) He builds canoes and kayaks for a living.

While other tenses have not traditionally been identified with habitual aspect,habitual readings of the pluperfect (e.g., had had in (8)) are possible, and becausefuture-marking will can receive a future-habitual interpretation (9), so can the con-ditional (future-in-the-past), as in (10). Hence, habituality is not limited to the sim-ple tenses, but we will restrict ourselves to such examples here and continue to referto the simple tenses in the discussion of the habitual interpretation of sentenceslacking would and used to, since the habitual interpretations of tenses other than thepast, present, and future are derivative of these (e.g., the pluperfect in example (8) isa habitual past in a past context).

(8) He was mostly out of town on his business, but from time to time, she hadhad the opportunity to observe him and as always admired his gentle andpleasant disposition. (http://www.austen.com/derby/radhika4.htm)

(9) After you join you will occasionally receive our newsletter in your email.(http://www.all-free-samples.com/newsletter.php)

(10) He predicted that he would lose any case that he initiated. The CanadianCharter of Rights and Freedoms’ guarantees of religious freedom wouldprobably nullify the Province’s marriage act in these cases. (http://www.religioustolerance.org/mar_bibl.htm)

Would is formally the past tense of will, and in its future-in-the-past meaning, theso-called conditional tense is clearly just the past counterpart of the future tense.The putative habitual marker would is likewise nothing more than the past tense ofthe habitual will found in examples such as (11). As simply the past tense of a modalauxiliary, habitual would in a sentence such as (12) is just like a modal such ascould, when that modal is used of past time, as in (13) (as opposed to nonpast uses,for example, would in example (14)). Indeed, habitual would occasionally is foundin connection with another past modal in a context that suggests a parallel betweenthe two (e.g., (15)).

(11) From time to time he will yell that he doesn’t “want to be managed,” but over-all, I am the one who is more frustrated. (http://www.marriedtomommy.com/_disc5/0000006c.htm)

(12) She would never simply answer a question with a yes or a no.(13) She could never simply answer a question with a yes or a no.(14) You would think differently if you were me.(15) The King suffered from hallucinations. He would often hear footsteps and

voices when nobody else could. (http://www.geocities.com/Paris/LeftBank/4080/report.htm)

Binnick / Markers of Habitual Aspect in English 341



It is often the case that would and the simple past can freely, and while seeminglypreserving the meaning of the sentence, replace one another—compare would seeand saw in examples (16) and (17), respectively—and sometimes they alternate in aseries of clauses, as in (18). We shall see below that there are, however, subtle butnonetheless significant differences in meaning between the two (and between usedto and would).

(16) For the next month, from time to time, she would see her father’s face in dif-ferent objects or pictures, and she would hear voices saying: “You’ll neverget away with it. . . .” (http://www.deikman.com/depression.html)

(17) From time to time she saw him as he hurried around the village, but theywere only able to smile at each other in passing. (http://mujweb.atlas.cz/www/Pepa.Josef/pribehy/cl_zvz04).html)

(18) It was reported by those who encountered them in their Sunday walks, thatthey said nothing, looked singularly dull, and would hail with obvious re-lief the appearance of a friend. (R. L. Stevenson, The Strange Case of Dr.Jekyll and Mr. Hyde, chap. 1)

Habituality as a State

A habit has generally been taken to constitute a state (Vendler 1967, 108; Hirtle1967, 48; Hirtle 1975, 95) or at least “a durative situation with many of the proper-ties of a state” (Lyons 1977, 716, in Brinton 1987, 196).

Moreover, as Comrie (1976) points out, used to can be used to refer directly to astate, as opposed to a series of eventualities, as in his example, here (19). Accord-ingly, he concludes that what is common to all habitual expressions is that “they de-scribe a situation which is characteristic of an extended period of time, so . . . thatthe situation referred to is viewed not as an incidental property of the moment but,precisely, as a characteristic feature of a whole period” (27-8). This can involve ei-ther a protracted situation, as in (19), or, as in the usual habitual, a series of itera-tions of some type of eventuality. Thus, following Comrie, Jørgensen (1988, 348)says that “[used to] in contemporary English [expresses] repeated (or habitual) ac-tivities or continued states in the past,” and many definitions of habitual aspect—forexample, that of the Summer Institute of Linguistics (http://www.sil.org/linguistics/GlossaryOfLinguisticTerms/ WhatIsHabitualAspect.htm)—characterize it as “animperfective aspect that expresses the occurrence of an event or state as characteris-tic of a period of time.”

(19) The Temple of Diana used to stand at Ephesus.

But a habit is not a state, and the standard tests for stativity generally fail with ha-bitual expressions. States, for example, may be true at a point in time, as in “at noon,




Sue was in Rome.” But a habit cannot be true at a point in time (20a), only over aninterval of time (20b). More generally, states have the “subinterval property”(Bennett and Partee 1978). If a state holds of a certain interval of time, it holds ofany subinterval of that time whatsoever. Thus, if Sue was in Rome all last year, thenshe must have been in Rome all last June. But if Sue frequently visited Max lastyear, there is no guarantee that she frequently visited him in June or even visitedhim then at all. Nor, if she sometimes received gifts in the mail last year, is there aguarantee that she received any gifts in June.

(20) a. *At noon, Sue {used to eat/would eat/ate} bananas for lunch.b. {For years/in her youth}, Sue {used to eat/would eat/ate} bananas forlunch.

Stative clauses typically do not advance narrative time but do often function asbackground information contemporary with the foregrounded narrative line. Thus,in (21a), Sue stands up (stood up is an eventive expression) after John enters, but in(21b), her standing up (with the stative expression was standing up) occurs beforehe enters. If habitual sentences were stative, we would expect no narrative advance,but in fact, Sue’s standing up in both (21a) and (21c) occurs after John’s entrance.Of course, the when in (21c) can only mean whenever, and it is not the habit itselfthat follows his entrance but the single event of her standing. However, this receivesno explanation if habitual sentences are stative.

(21) a. When John entered, Sue stood up.b. When John entered, Sue was standing up.c. When John entered, Sue would stand up.

States are typically nonagentive, so it is generally not possible to have an imper-ative of a state (22), nor may the pseudo-cleft construction involve a stative predi-cate (23) (Dowty 1979, 55). Yet it is completely possible to have an agent in a habit-ual sentence (24) and likewise a habitual interpretation of a pseudo-cleft (25) or ofan imperative (26) sentence.

(22) *Be sleepy!(23) *What John did was be sleepy.(24) John {frequently stole/often steals/would frequently steal} cars.(25) He uses tried and tested methods, but what he does is to wave his magic

wand . . . (www.brianmac.demon.co.uk/rugby/strength.htm)(26) Be particularly wary anytime a “nice” person offers you an unusual deal. . . .

(http://www.december.com/simple/live/prepareurban.html)




States hold over intervals of time. Hence, they readily occur with temporaladverbials of duration (27). Used to (28) and would (29) do co-occur with such ex-pressions. Nonetheless, it is incorrect to say that the habit characterizes the intervalin the way that a state does. Although the habit itself holds over the interval, the in-dividual eventualities instantiating the habit do not fill out the period in question,which is why the subinterval property fails. Moreover, it is possible to define a pe-riod in terms of a state (when he was wealthy in (30)) but not of a habit. The whenclauses in (31) do not refer to the intervals during which the habits occurred; asshown by the plural times, these expressions refer instead to the individualepisodes.

(27) For years, Sue was quite content to let things slide.(28) For years she used to get up early in the morning and drive him to tourna-

ment sites all around the state of California. (http://www.rolemodel.net/tiger/tiger.htm)

(29) For years she would wake up at 2 or 3 am with all kinds of brilliant ideas thatshe would promptly forget by morning. (http://www.witi.com/growth/2004/environment.php)

(30) All his relatives and family who thought high of him when he waswealthy, now looked down upon him. (http://www.samratchana.net/about/baba/hisstory/en-us/default.asp)

(31) There were times when he would get on my nerves, and I’m sure therewere times when I ticked him off too. (http://www.eszlinger.com/family/fambenmemories.html)

To be sure, in examples of “stative” used to such as (19), the state holds of the en-tire interval of time in question. But Comrie (1976) is misled by such exampleswhen he concludes that “habitual” used to likewise involves a state characteristic ofthe interval in question. “Stative” used to (19) is indeed stative, but not by virtue ofbeing an expression of habituality, and we should neither be surprised that habitualwould cannot be used statively (the habit in (32) consists of a series of episodes, notstates as such, as indicated by the reference to “these occasions”) nor attempt to finda single definition that fits both the meaning of “stative” used to and “habitual” usedto (as well as habitual would). If habits are states, they are not states because theycan be expressed with used to. But neither do they function like ordinary states, forthey necessarily include gaps and hence fail the subinterval test.

(19) The Temple of Diana used to stand at Ephesus.(32) His solution was to move on, but every so often, he would be near her. On

these occasions, . . . (http://www.greyjedi.com/SWArchives/chronicles/archives-interim.html)




In examples such as those in (33), “habitual” used to and would hold over a cer-tain era, as shown by reference to “that time,” and to that extent, a habit may charac-terize an interval of time. But truly habitual expressions are not, ipso facto, stative.

(33) a. Why, all that time when he used to come breakfasting with Mr Pontifexmorning after morning, it took me to a perfect shadow the way he carriedon. (Samuel Butler, The Way of All Flesh, chap. 72)b. So I sit and think about other things, like Taaj’s smile, or of that timewhen he would laugh out loud everytime he saw himself laugh on video.(http://www.thebilns.com/index.php/weblog/2003/12/)

The Habitual as a Condition That No Longer Obtains

Traditionally, used to (34, 35) has been taken to imply, or even to mean, thatsome past condition no longer obtains; Schibsbye (1970, 88-9) extends this to ha-bitual would as well, and Tagliamonte and Lawrence (2000, 324) observe that“each of these forms [used to and would, as well as the preterit] is used . . . to de-scribe a situation that existed for a period of time but is no longer the case.” This issaid to be “particularly true with used to . . . as opposed to would . . ., where there isno implication of discontinuance” (Visser 1963-1973, 1413, in Tagliamonte andLawrence 2000, 331).

(34) I used to dance, but I don’t dance now.(35) It’s just not like it used to be at all. (Examples (34) and (35) from Taglia-

monte and Lawrence 2000, 331)

That a past habit is no longer the case is not part of the meaning of the expressionitself but a conversational implicature, as recognized by Comrie (1976, 29), whosays, “One can reasonably say, without self-contradiction, in answer to a questionwhether Bill used to be a member of a subversive organization: yes, he used to be amember of a subversive organization, and he still is. . . . Thus this cannot be an im-plication in the strict sense, since the putative implication can be cancelled by an ex-plicit denial of it,” and by Harrison (2002), who cites the example (36), in which thetag cancels the implicature. The other putative markers of past habituality likewisecan co-occur with “and still {does/is}” (37-38).

(36) Erik used to be a member of the Volapük League, and he still is.(37) He would (and still does!) tell any one he can get to stand still long enough

how GREAT this dog is! (http://www.bigdawgsboarding.com/_wsn/page3.html)

(38) My father always said it and still does, to fill in time. (http://www.phrases.org.uk/bulletin_board/37/messages/471.html)




In the case of used to, this raises an interesting question, for used to presupposesa contrast between a past state of affairs and the present one. How used to can coex-ist with the implicature-canceling tag is discussed in the section “Used to.”

Generic Aspect

In addition to habitual aspect, it is generally assumed that there is a generic as-pect. Harrison (2002) says, “The generic aspect occurs in broad, general statementssuch as ‘squirrels live in trees.’” Copley (2004) refers to “a reading of will with ge-neric aspect” and Ferrari-Bridgers (2004) to “Habitual Generic Aspect mor-phemes.” But while Dahl (1985, 99-100) and Cutrer (1994, 194) both refer to ge-neric sentences, Cutrer notes that “in terms of tense, the grammar in English doesnot make a distinction between” generic and habitual sentences, and Dahl notesthat “it seems to be rare for a language to have an overt and unequivocal TMA[tense-mood-aspect] marking of a sentence as being generic. The most frequentcase is for generic sentences to be expressed with the most unmarked TMA cate-gory, as in English, where the Simple Past is used.”

Generic sentences concern recurrent eventualities but differ from habitual sen-tences in that we can neither individuate (39, 40) nor quantify (41) the occurrencesin the series of situations referred to, in the way that we can in the case of habitualsentences (42, 43), because the generic, unlike the habitual, does not concern aseries of actual eventualities.

(39) !Each time, the dinosaur {hunted/would hunt} for meat.2

(40) !On each occasion, the dinosaur {hunted/would hunt} for meat.(41) !Many times, the dinosaur {hunted/would hunt} for meat.(42) On each occasion, Lincoln{fired/would fire} his head general.(43) Many times, Lincoln{fired/would fire} his head general.

Habitual sentences are predicational; they predicate acts or states of individuals:“individual entities (or groups of entities) are assigned habitual properties” (Cutrer1994, 146). Van Geenhoven (2001, section 3.2) assigns example (44) three read-ings: habitual, ‘Mary handles the mail from Antarctica regularly/once in a while’;prescriptive, ‘Mary is supposed to handle the mail from Antarctica’; and descrip-tive, ‘if mail arrives from Antarctica, Mary will handle it.’ Only on the habitualreading does the sentence ascribe to Mary actual acts of mail handling. In this re-gard, Brinton (1987, 206) contrasts a pair of instructive examples from Jespersen([1932] 1961, 192), respectively generic (45) and habitual (46). What (45) does isassign a certain property (that of drinking) to the set of properties that belong to theconcept of “smoker,” while (46) generalizes over the individuals who belong to thatclass and predicates of those acts that they are typically in the habit of performing.




(44) Mary handles the mail from Antarctica.(45) Smokers always drink. = All smokers are drinkers.3

(46) Smokers are always drinking. = Smokers are constantly so occupied.

An expression such as smokers is ambiguous and can refer either to members ofa class or set taken individually (the members of the class act or acted separately) orcollectively, as a group (the members of the subject class do or did something to-gether). Accordingly, (47), in its habitual reading, can mean either that a group con-sisting of several men would come, or that a single man would come, on each occa-sion. Example (47) also has a semelfactive reading, referring to one occasion, onwhich several men came as a group.

(47) Men came to the door and asked after Max.

The purported difference in meaning between habitual and generic aspect is infact a difference in interpretation, forced by the interpretation of such sentence ele-ments as the subject and object. Cutrer (1994, 146) comes close to this positionwhen she writes,

Habitual expressions are handled in the same way as generics. . . . In habitualspaces, particular properties are assigned to a specific entity or set of entities.In contrast to generic spaces, the habitual spaces set up for the interpretationof examples (4.4) are more specific, since they have particular entities ratherthan roles.(4.4) a. John sees a therapist.

b. On Tuesday, John sees a therapist.

Thus, we call “generic” a statement that does not predicate a property or proper-ties of specific individuals or groups of individuals, but rather assigns a property orproperties to the set of properties defining a class or set. This distinction is not un-like that between sentences predicating stage-level and individual-level propertiesof an individual. A stage-level property such as being tired is predicated of its sub-ject over a certain interval of time and hence is impermanent, while an individual-level property such as being male tends to permanence over the whole lifetime orexistence of the entity in question. The reason for this is that a stage-level propertyis contingent, while an individual-level property is in some sense essential andforms part of the identity of the entity. To be the Eiffel Tower is to be a tower inmuch the way that to be a smoker is to smoke. Thus, we may speak of the “general-ity” or “globality” of an example such as (48), which takes as its reference time thewhole of the period in which ancient Rome existed and generalizes over the wholeof that period in much the way that a generic sentence generalizes over a whole




class, as opposed to examples like those in (49), which take particular times—whether explicitly stated, as in (49a), or implicitly given by the context, as in(49b)—as reference times.

(48) Ancient Rome was the largest city in the then known world. (http//www.historylearningsite.co.uk/ancient_rome.htm)

(49) a. In 1871, Chicago was truly a boom town. (http://www.prairieghosts.com/great_fire.html)b. London’s population had continued to grow and many lived in squalorand poverty. The only way people had to get rid of rubbish was to throw itout into the streets. . . . As a result, London was filthy. (http://www.historylearningsite.co.uk/plague_of_1665.htm)

Preterites like was in (48) are “indefinite,” and those like was in (49) are “defi-nite.” The difference is in whether the interpretation is global (over the whole of acertain period of time, such as that in which ancient Rome existed) or limited to acertain temporal perspective (such as that of 1871). In the same way, the differencebetween habitual and generic lies in whether the interpretation is global over an ab-stract class of individuals or is limited to a particular instantiation, or particularinstantiations, of the class. Example (44, repeated below) is “generic” when Maryis taken globally, to refer to Mary abstractly; it is “habitual” when it refers to a seriesof her stages associated with a series of actual eventualities. Which is the appropri-ate reading depends on the context and many other pragmatic factors.

(44) Mary handles the mail from Antarctica.

Thus, there are no distinct habitual and generic meanings in English as such andhence no distinction of habitual and generic aspects in English.

Used to4

Used to is not a past tense. Like the present perfect, it is a present tense. With itspast-tense morphology and present-tense semantics, used to is rather like thepreterito-present modals—might, must, and so on. Just as these modals use peri-phrastic constructions (e.g., might have) to express past meaning, used to has as itspast-tense counterpart (at least in British usage) the pluperfect had used to (50)(Quirk et al. 1985, in Jørgensen 1988, 348).

(50) His hide was less shiny than it had used to be. (George Orwell, AnimalFarm, chap. 9; in Jørgensen 1988, 349)




The distribution of used to, as described by Tagliamonte and Lawrence (2000),is similar to that of the present perfect. Both are what Benveniste (1959, 70; 1966,241) calls tenses of discours (“discourse”), as opposed to tenses of histoire(“story”), and what Weinrich (1973, 23) similarly calls tenses of “commentary,” asopposed to those of “narrative.”5

Tenses of discours are deictic and relate the time of the eventuality to that of thespeech act, so they do not require a contextually defined reference time and canreadily appear in absolute position (e.g., in a title) or in the first sentence of a text ordiscourse (51, 52).

(51) Well, we have lived to see the end of civilization in England. (T. H. White,The Age of Scandal, chap. 1)

(52) Gordon Brittan never used to think animals had minds. (http://www.montana.edu/wwwpb/univ/animinds.html)

Like the present perfect, the used to construction shows a preference for highlysalient categories such as first person and animate subjects (Wallace 1982, 212-3),particularly with nonstative predicates. For example, a Google search found ap-proximately 43,000 Web pages for I used to eat and 405,000 for I have eaten. Foryou used to eat, only 4,330 pages were found, and for you have eaten, 72,000. Simi-larly, (s)he used to eat found 10,200 pages, and (s)he has eaten found 42,700. Wemay compare the 1,440,000 pages for I ate, 271,000 for you ate, and 678,000 for(s)he ate. Another search found approximately 11,500,000 Web pages for (s)he hasbeen and almost twice as many (22,600,000) for it has been. But the 729,000 pagesfor it has done are less than half the 1,910,000 pages for (s)he has done. Given thevarious contingencies that affect what appears on Web pages, and in the absence ofa statistical analysis, these numbers can only be taken as qualitatively and roughlyindicative, but nonetheless they are revealing. The preponderance of the first personis not as great in the case of the preterite as in those of the tenses of discours, whichare likely to occur in contexts that put a premium on deicticness and proximalityand hence high salience.

Insofar as the perfect and used to both speak to the current state of affairs, itwould be, and is, relatively unusual for either to be negated. A Google search found3,300,000 Web pages with I have made or I’ve made but only 311,000 with either Ihave not made or I haven’t made. Versions of I have been appeared on some 40 mil-lion pages; the negated versions appeared on less than 3 million. I have lived oc-curred on over a million pages but the negated versions on just over 45,000.Tagliamonte and Lawrence (2000, 338) observe that though negated used to doesoccur (53), in their data, it “is virtually nonexistent,” which corroborates Denison’s(1993, 323) observation that its negation is avoided.




(53) That didn’t use to go down really well, did it? (Tagliamonte and Lawrence2000, 338)

This avoidance has been attributed to uncertainty regarding the form of the ne-gated verb (Denison 1993, 323 and Jørgensen 1988, in Tagliamonte and Lawrence2000, 337). To be sure, there is considerable variation and confusion over the“proper” negation. A Web-based survey (at http://www.livejournal.com/users/pne/408823.html?mode=reply) evoked the forms (in descending order of occurrence)didn’t use to, didn’t used to, used not to, used to not, and use to not, along with oth-ers, though no examples of either usedn’t to or usen’t to. (The former seems to oc-cur on the Web only in discussions of grammar and is so rare that (54a) must be ahypercorrection, especially in light of (54b)). However, negated used to is commonenough: a Google search for didn’t use to found 24,700 pages. Even a search for therelatively rare usen’t to found 867. That the present perfect, which does not sharethis morphological confusion, exhibits a similar avoidance of negation suggeststhat the relative lack of negation with used to is not to be accounted for onmorphological grounds.

(54) a. The Mistress usedn’t to sleep well at night, . . . (Agatha Christie, PoiretLoses a Client)b. You usen’t to be like that. (Agatha Christie as Mary Westmacott, Giant’sBread)

Nor is used to a marker of habituality, for, like the present perfect, it concerns thepresent state of affairs resultant from past actions and not a recurrent series of even-tualities. Hence, both constructions exhibit present relevance (55) and require re-peatability (56), and both disallow definite adverbials of the past (57), though theyallow indefinite ones (58).

(55) a. {Bill Clinton/*Abe Lincoln} has been president.b. {Bill Clinton/*Abe Lincoln} used to be president.

(56) a. Woody Allen has directed {*Annie Hall/a film a year/important films}.b. Woody Allen used to direct {*Annie Hall/a film a year/important films}.

(57) a. *I’ve lived in York in 1914.b. *I used to live in York in 1914.

(58) a. . . . my greens have sometimes frozen up on me. (http://forums.egullet.org/lofiversion/index.php/t45442.html)b. The pipes on the farm used to freeze sometimes. . . . (http://forums.nervousness.org/ lofiversion/index.php/t943-5600.html)




The used to construction functions as a kind of present perfect, but instead oflinking a past event with the present state of affairs by dint of placing it in the pres-ent era (i.e., in an “extended ‘now’ ”), the used to construction does just the reverse,divorcing the past situation from the present era, which results precisely from an al-teration of the previous state of affairs. As shown by example (36), the implicaturethat the situation no longer holds in the present is cancelable. That this is possible isdue to the fact that while normally the present era contrasts with the immediatelypreceding one precisely in regard to the situation predicated by the sentence (i.e.,the situation in question defines the preceding era and hence its contrast with thepresent), in those cases in which the implicature is cancelable, the current era is notdefined in terms of the predicated situation itself but of something else. The differ-ence between the past and present eras in (36) cannot be Erik’s being or not being amember of the Volapük League. When the defining characteristic is thus not thatpredicated by the sentence, it may be rendered explicit, as by when at Cambridge in(59), and it is presupposed that the past era in question has ended: in (59), that theauthor is no longer at Cambridge.

(36) Erik used to be a member of the Volapük League, and he still is.(59) When at Cambridge I used to practise throwing up my gun to my shoulder

before a looking-glass to see that I threw it up straight. (The Autobiographyof Charles Darwin)

Since the point of the used to construction is not to report a habit in the past butrather to contrast the present state of affairs with those obtaining in the immediatepast, the requirement for contrast always trumps the implicature that the past situa-tion no longer obtains, which is why (60) is not as odd as (61), in which the onlypossible contrast involves the predication asserted in the sentence. The eras in (60)contrast not in whether the United States has occasional disputes with its allies butinstead in whether George Bush père is president.

(60) When George Bush père was president, the U.S. used to have occasionaldisputes with its allies.

(61) The U.S. used to have occasional disputes with its allies.

If used to is neither a past tense nor a marker of habituality, it is not surprisingthat it shows significant differences in its use and distribution from both would andthe simple past tense, as well as significant similarities to the present perfect tensein both regards.




The Simple Tenses

The simple tenses are markers neither of habituality nor of genericity—these arenot meanings but contextual interpretations due to elements of context or cotext.

The preterite has a nonhabitual default interpretation (Bybee et al. 1994, 153 inTagliamonte and Lawrence 2000, 340)6 and “requires some temporal and/or con-textual specification for [a habitual] reading” (Quirk and Greenbaum 1972, 41-2 inTagliamonte and Lawrence 2000, 340). Thus, always renders (62) explicitly habit-ual, while (63) remains ambiguous between a habitual reading, referring to a seriesof actions, and a semelfactive reading, referring to one. (Examples (62) and (63) arefrom Tagliamonte and Lawrence 2000, 340.) The present can have a semelfactiveinterpretation only in special uses such as the narrative present (which includes thehistorical present, as in Napoleon dies in exile), descriptive present (a caption,Washington crosses the Delaware, or a stage direction, Hamlet exits), or the “in-stantaneous” present (he scores!), and in such uses, the predicate is normallyeventive, and the semelfactive is generally the preferred reading.

(62) He always wrote with a special pen.(63) @He wrote with a special pen.

The presence or absence of adverbials makes no appreciable difference withused to or would, but there is naturally a strong tendency to use adverbials with thepreterite (Tagliamonte and Lawrence 2000, 341), simply because otherwise itmight not clearly receive either a semelfactive (64, 65) or habitual (66-68) reading.(Examples (66) to (68) are from Tagliamonte and Lawrence 2000, 341.)

(64) One time I went up to the top of Sir W. Batten’s house. . . . (Diary of SamuelPepys, Feb. 21, 1662)

(65) Later that morning, they played catch and ran laps in a line without passingone another. . . . (http://www.nldline.com/social_camp.htm)

(66) He changed it every day. (Hugh Walpole, The Secret City, chap. 10)(67) Used it quite a lot when I learned, and now when I write HTML. (http://

project5.freezope.org/oldsite/links/links.html)(68) Often they were punished for doing this. . . . (H. B. Stowe, Uncle Tom’s

Cabin, chap. 8)

The simple tenses quantify cumulatively, unlike would and used to, which quan-tify distributively. Thus, adverbials of quantity apply to the total number of timesthe episode or event recurs over the total period in question, whereas with wouldand used to in its “habitual” use, a quantifier typically applies to each episode.Compare the sentences in (69). Similarly, quantification of the subject or the objecthas different effects with the three expressions. With used to or would, a plural sub-




ject and/or object has only a habitual and, hence, a group reading, in which numer-ous episodes each involve a group of entities (70a), but the simple tenses (70b) haveeither a semelfactive reading (in which a group of men come to the door on oneoccasion) or quantify cumulatively, so that men come to the door on several occa-sions—either as a group or singly. Compare (71), the habitual reading of whichleaves open the possibility that on any given occasion more than one, but not neces-sarily five, men come to the door—for example, on one occasion, two come; on an-other, three.

(69) a. John {left/leaves} several times. (There were/are several episodes of hisleaving.)b. John would leave several times. (In each episode, he left several times.)c. John used to leave several times. (In each episode, he left several times.)

(70) a. Several men {would/used to} come to the door.b. Several men {came/come} to the door.

(71) Five men {came/come} to the door.

Because they quantify cumulatively, the simple tenses—for example, the preterite—refer to the interval as a whole, while, quantifying distributively, used to andwould refer to each member of the series. Thus, the habitual preterite can only an-chor (serve as antecedent to) an expression that takes the whole of the interval as itsreference time, while would and “habitual” used to can anchor an expression takingthe time of one of the episodes as its reference time. In (72), he would borrow refersto each time he went fishing and so can readily take as antecedent used to, but in(73), he borrowed is odd because it seems to refer to a single act of borrowing, whilethe context would seem to call for multiple acts, as in (74). Similarly, the ambiguityin (74) proceeds from the fact that the when-clause could be either semelfactive(when = one time when) or habitual. But the ambiguity of (75) is also caused by theambiguity of the second clause, though the semelfactive is the preferred reading.We have more to say about anchoring in the section “Will and Would” below.

(72) He used to go fishing. He would borrow tackle from his brother. (=eachtime)

(73) @He used to go fishing. ?He borrowed tackle from his brother.(74) @When he was younger, he went fishing. He would borrow tackle from his

brother. (=each time)(75) When he was younger, he went fishing. @He borrowed tackle from his

brother. (the semelfactive is the preferred reading)

The habitual and/or generic readings of the simple tenses are interpretations incontext, due to elements of context and cotext, and a number of different factorsplay a role in determining the reading.




The first is Aktionsart. With an eventive predicate, as in (76), the preferred, de-fault interpretation of the past tense and the special uses of the present cited above isa definite, semelfactive one, that is, as a single event at a specific temporal nexus.

(76) a. John {turned/turns} blue.b. Susan {went/goes} home.c. The dam {broke/breaks}.

An activity or process expression acts like a stative when unbounded, so that thepreferred reading of (77a) is ‘she habitually traveled’ (and [77b] characterizes a se-ries of travels); and for (78), ‘she never traveled’or ‘she was in the habit of not trav-eling;’it acts like an eventive one when bounded, and hence the preferred reading of(79) is ‘on some occasion, she traveled for three days’ (cf. (80)).

(77) a. She traveled.b. Julius Caesar travels not in a chariot but a horse-drawn Edsel automobile.(http://www.sfbrain.co.ik/misc/anime.html)

(78) She didn’t travel.(79) She traveled for three days.(80) When Caesar travels to Utica with his army. . . . (http://www.musicweb_

international.com/classrev/2005/Aug05/ferrandni_catone_OC901.htm)

In the preterite, an unbounded process has a generic (81) or habitual (82) defaultreading, and some explicit indicator (as in (83)) is required for a semelfactive inter-pretation. A bounded process without an explicit adverbial, however, is normallyinterpreted as semelfactive (84), and some indicator (as in (85)) is required for ahabitual reading.

(81) Annette Kellerman loved to swim. She swam for fun. . . . (http://www.edhelper.com/ReadingComprehension_33_106.html) (generic)

(82) But she couldn’t wear them when she swam. . . . (http://umanitoba.fitdv.com/new/articles/article.html?artid=58) (habitual)

(83) She swam, then met with some small boats which conveyed her to theLucrine lake. . . . (Tacitus, The Annals, Book XIV, 1-16) (semelfactive)

(84) She swam for an hour, then got tired and began looking for pretty seashellsalong the beach. (http://illusionsworld.tripod.com/story/warmagic.htm)

(85) She swam for an hour every morning, and always ate. . . . (http://sfgate.com/cgi-bin/article.cgi?file=/ chronicle/archive/1997/08/28/DD23604.DTL)

With states, it matters whether the state is a “temporary,” stage-level predicate,or a relatively “permanent,” “essential,” individual-level predicate. With a stage-level predicate (86a), the most normal interpretation is definite; the state obtains at




the reference time, as in (86b), where she was tired is true of a time on the day shewas arrested. As Partee (1973, 602f.) points out, negating a definite past (as in (87))simply indicates that the predication failed to hold at the time in question, while ne-gation of an indefinite (88) means that the predication never held. The aspects havestage-level readings, hence definite readings (89).

(86) a. She {was/is} tired.b. On that famous day when she was arrested, it would have been much easierfor Rosa to give up her seat. . . . But Rosa was tired. (http://www.socialstudies.com/c/@ca5ECqLPQRruU/Pages/rosaparks.html)

(87) a. John didn’t turn blue.b. Rosa wasn’t tired.

(88) Pompeii wasn’t large.(89) a. Susan {was(n’t)/is(n’t)} reading.

b. Susan {had(n’t)/has(n’t)} gone to the store.

With an individual-stage stative predicate, the preferred reading of the past tense(90) and of one of the special uses of the present (91) is indefinite, simply predicat-ing a property as a global characteristic, without specifying any particular time atwhich it holds, other than the lifetime of a living subject or period during which aninanimate subject exists.

(90) Lincoln was tall.(91) . . . Hamlet is male and Ophelia is female. (http://www.123helpme.com/

assets/4506.html)

We have been considering the use of the preterite as an absolute tense, but it alsois used as a relative tense, a present-in-the-past, and when so used, the default read-ing is definite, as with a stative expression such as tall (92); cf. the context in (93).An individual-stage stative predicate also has an absolute, indefinite reading, as in(94), which does not mean that the man was tall when she spoke but instead that theman who came to her door was a tall man. A stage-level predicate lacks an indefi-nite reading, however, and (95) cannot mean that he was globally tired in the past,only that he was tired at that time, as in (96).

(92) Sue knew that John was tall.(93) She kept her eyes on the hall, her other senses working together to form a pic-

ture for her. She already knew that the intruder was male. Now she knew thathe was tall. . . . (http://members.tripod.com/reality_dysfunction/pitchblack/games8.htm)

(94) KATU spoke with one woman who said a white man in his mid 40’s came toher door. And she said that he was tall and slim, dressed well, and was verypolite. (http:// www.katu.com/news/story.asp?ID=57037)




(95) Sue knew that John was tired.(96) The jerk stopped throwing punches at her. She knew that he was tired as

much as she was. (http:// tannim.anifics.com/Fushigi_Mizu/fm2.txt)

Eventive sentences do not have relative tense uses unless there is a special ad-junct that forces an indefinite, habitual interpretation. A search on the Internet forexamples such as she knew that John turned blue found none in millions of Webpages. But adding often found the example (97). However, adding such an adjunctforces an indefinite reading of stage-level predicates as well, as in (98). We summa-rize these relationships in Table 1.

(97) Mulder would have teased her forever if he knew that she often turned inwardto find where he was, if he was all right. (http://vickiemoseley.freeservers.com/follies2.html)

(98) She knew that he was often up all night. (http://www.geocities.com/osakabe_yoshio/ Haruki/Books/Honey-Pie-E.html)

The quantification of subjects, objects, and other adjuncts plays a role also, in-teracting in various ways with elements of context and cotext. In the preterite, a sin-gular subject with a singular object takes as its default reading, in the absence of anexplicit or implicit adverbial indicating plural occasions, a semelfactive reading(99a), whereas a plural object (99b) or subject (usually with plural object) (99c)renders preferred a habitual reading. But even a singular subject and singular ob-ject, and even in the absence of a frequentative adverbial, can give rise to a habitualreading in an appropriate context (e.g., the last sentence in (99d)).

(99) a. . . . Tim wrote a novel set in his fictional village. (http://www.bloc-online.com/about writing/writing-fiction/TimPears.htm)b. . . . Henryk Sienkiewicz wrote several novels set in the medieval conflictsbetween Poles and the Teutonic Knights. (http://www.answers.com/topic/historical-novel)c. The authors were important individuals who were famous writers; theywrote novels with social messages. . . . (http://www.odessa.edu/dept/english/dlane/eng1302/Pages/Background.html)


Table 1Predicate Type and Tense Interpretation

Absolute Relative

Eventive John turned blue.(definite, semelfactive)

Stative—individual level John was tall. (indefinite) (Sue knew that) John was tall. (definite)Stative—stage level John was tired. (definite) (Sue knew that) John was tired. (definite)



d. . . . Lora . . . occasionally interfered and took her part when she was veryunjustly accused, but no one seemed really to care for her, and she often feltsad and lonely. Mr. Dinsmore, though her own grandfather, treated her withentire neglect. . . . (Martha Finley, Elsie Dinsmore, chap. 2)

The interpretation of a simple-tensed sentence as habitual depends not only onthe presence, either in the sentence or the context or both, of plural or frequentativeelements but also on a number of pragmatic factors. For example, in a polygamoussociety, a sentence such as (100) could receive a semelfactive, group reading it isotherwise likely to lack.

(100) John {married/marries} two women named Sue.

The habitual interpretation of both simple tenses thus depends on a number ofsemantic and pragmatic factors and occurs only in special contexts. Similarly, thegeneric reading of the simple tenses is restricted and depends on a number of fac-tors, the most important of which is the presence of a generic subject or object. Asentence with a singular, definite subject and singular definite object (101) can re-ceive a habitual reading in the appropriate context but cannot receive a genericreading. What is required is a generic subject (102) or object (103) or some elementof context or cotext (such as any) that licenses a generic reading (104). As in thecase of habitual interpretations, pragmatics may enter into a generic interpretation.Generic sentences usually involve individual-level properties, which are the kind ofproperties that may be assigned to the set of properties characterizing the class inquestion (105a); they do not usually involve stage-level properties, and a sentencesuch as those in (105b) is at best not as natural as those in (105a). Adverbs offrequency (105c) render such examples habitual and hence natural.

(101) John {seeks/sought} a new client.(102) Roman generals {love/loved} a parade.(103) John {drives/drove} trucks.(104) They intercept any HTTP or HTTPS requests from the user’s web

browser, go and retrieve the page themselves, and then hand it back to theuser. (http://www.upenn.edu/computing/web-security/websec-proxy.html)

(105) a. {Dinosaurs were/Crocodiles are} {big/swift/ferocious}.b. ?*{Dinosaurs were/Crocodiles are} {tired/furious/hungry}.c. {Dinosaurs were/Crocodiles are} {frequently/often/never}{tired/furious/hungry}.

When a nominal expression may receive either a generic or group reading, theinterpretation depends on the nature of the property. A stage-level predicate forcesa nongeneric reading (106a), while an individual-level predicate forces a generic




reading (the second cats in 106b). The second cats in (106a) could be replaced bythe cat (106c), but that in (106a) cannot (106d). It is only the cotext that revealswhich reading is intended.

(106) a. That year was really bad for rats and cats were everywhere with such anabundant food supply. (http://www.doc.govt.nz/Conservation/002~Animal-Pests/001~Control-Methods/1080-in-Action/017~Stewart-Island-Sentinel.asp)b. Cats were popular because the ancient Egyptians believed that cats hadprotective qualities. (http://www.google.ca/search?q=%22Cats+were+popular%22&hl=en&lr=&ie=UTF-8&start=10&sa=N)c. Cats were popular because the ancient Egyptians believed that the cathad protective qualities.d. *That year was really bad for rats and the cat was everywhere with suchan abundant food supply.

Genericity, like habituality, is not a meaning of the simple tenses; it is an inter-pretation dependent on a number of semantic and pragmatic factors.

Will and Would

In contrast to the simple tenses, will and its past tense form, would, are markersof habituality, though not of genericity. Habituality is, however, only one of a num-ber of uses of this modal verb. That these uses are so disparate and allow for ambi-guity even in narrowly restricted contexts suggests that the modal is polysemousand that habituality is a different meaning from, say, futurity, as opposed to merelya contextual interpretation. How many different meanings the verb has is not withinthe scope of the present article to consider, but certainly it has, inter alia, the follow-ing uses: future/conditional (future-in-the-past), supposition (that will be John),7

volition or intention (if you will turn now to page 21),8 and habituality (107, 108).

(107) Patch is very affectionate. She would prefer to be by your side all day. Shewill jump up and head butt you to get your attention. (http://www.hsfn.org/cats.htm)

(108) She would follow me every place. I couldn’t turn without her being there.If I sat in a chair to watch television, she would jump up on and curl up inmy lap. (http://www.in-memory-of-pets.com/aboutus.asp)

Would differs from the simple past in that unlike the latter, it does not require anyspecial adjuncts to be interpreted as habitual; it is inherently habitual, although con-ditional readings are naturally possible, out of context, for examples such as (109).As we have noted, the default reading, out of context, of an example like (110) is




nonhabitual (i.e., semelfactive). Since individual-level predicates do not readilyform series of episodes, would, unlike the simple past (111a), is not readily usedwith such predicates (111b) (except in the conditional, i.e., future-in-the-past, read-ing), but would is common with stage-level predicates such as aware (112). Will/would also has generic readings (113, 114).

(109) Sue would swim across the pool.(110) Sue swam across the pool.(111) a. John was tall.

b. !John would be tall.(112) . . . from time to time he would be aware of occasions when the sound had

changed for no apparent reason—he just knew that the sound had changedwith no conventional explanation as to why it had. (http://www.belt.demon.co.uk/bh.html)

(113) Once this happens the tiger will hunt for a slower prey, humans. (http://www.milmdl.k12.hi.us/nmd/Malama/M3/Wild%20Animals/Tigers/Tigers.html)

(114) The sender would dictate his or her letter (say it out loud) to the scribe, andthe scribe would write what he was told. (http://educ.queensu.ca/~fmc/march2005/signedandsealed.html)

When the simple tenses cannot express habituality, will/would can, and whenthe simple tenses are ambiguous, will/would is unambiguously habitual because itlacks semelfactive readings, both with eventive predicates (contrast 115b with115a) and stative predicates (116b versus 116a). Whereas the absolute eventivepast has as its preferred interpretation reference to a single event, would renders itexplicit that a series of events is in question. Thus, in examples such as (117) and(118), the use of would is clearly intended to render unambiguous that a habit is re-ferred to, and the event in question is part of a series and not just a single event. If wewere to delete would in these examples (as in 119 and 120, respectively), asemelfactive reading would be not only possible but the preferred reading.

(115) a. John turned blue.b. John would turn blue.

(116) a. John was too tired to run.b. John would be too tired to run.

(117) On landing in the evening, Good would at once set to work . . . to build a lit-tle “scherm,” or small enclosure. . . . (H. Rider Haggard, Allan Quater-main, chap. 2)

(118) So now when the huge loaf came in on a gigantic trencher . . . and, the tourof the table once made, seemed to have melted away, Elias and Catherinewould look at one another and say, “Who is to find bread for them all whenwe are gone?” (Charles Reade, The Cloister and the Hearth, chap. 2)




(119) On landing in the evening, Good at once set to work . . . to build a little“scherm” . . . and to light a fire.

(120) So now when the huge loaf came in on a gigantic trencher . . . Elias andCatherine looked at one another and said, “Who is to find bread for themall when we are gone?”

As we have seen, the simple past may be rendered, and indeed usually is ren-dered, habitual by quantifying a subject (by not all in (121a) or simple plurality in(122a)) or object (by any in (123a)) or specifying either the frequency of occur-rence (from time to time in (121a), every now and then in (124a), every week and of-ten in (125a)) or the period over which the series of repeated occurrences took place(never in (122a), as a child in (126a)).

(121) a. Not all his visits were to the sick; from time to time he made surprise vis-its to clandestine dances. (http://www.maristoz.edu.au/champagnat/glances/page5.html)

(122) a. My teachers said I’d never amount to anything. (http://www.copycoach.com/sampo.htm)

(123) a. When I made any remark to him on Geology, he never rested until hesaw the whole case clearly, and often made me see it more clearly than Ihad done before. (The Autobiography of Charles Darwin)

(124) a. . . . they had prizefights every now and then, and cockfights and evendogfights. (Upton Sinclair, The Jungle)

(125) a. So every week they received reports as to what was going on, and oftenthey knew things before the members of the union knew them. (UptonSinclair, The Jungle)

(126) a. . . . as a child, I went to Church, Sunday school, the Church choir. . . .(http://www.users.globalnet.co.uk/~edges/online/issue31/p5.htm)

Without such adjuncts or quantifying terms, the past is ambiguous, as in (127),which could mean either (1) that on one occasion, the narrator made a remark andthat on that occasion, the person he spoke to didn’t rest until he saw (on that occa-sion) the whole case clearly, or (2) that habitually the narrator made remarks andthat on each occasion the person didn’t rest until (on that particular one of the vari-ous occasions) he saw the case clearly.

(127) When I made a remark to him on Geology, he didn’t rest until he saw thewhole case clearly.

In those cases in which the simple past is ambiguous, would can disambiguatethe past, and where the past tense lacks a habitual reading, would can supply thewant. Example (128), unlike (127), is unambiguously habitual. John was tired isdefinite and refers to a single nexus, but John would be tired refers to a series of epi-




sodes of tiredness. Thus, sometimes we can account for the use of would as simply adevice for marking a habitual interpretation when the simple past itself cannot do soor forcing such an interpretation when the past tense is ambiguous.

(128) When I made a remark to him on Geology, he wouldn’t rest until he sawthe whole case clearly.

We have been considering cases in which would makes up for the ambiguity ofthe simple past tense. But in the (a) sentences of (121) to (126), the simple past tenseis explicitly rendered habitual (and neither semelfactive nor generic) by various ad-juncts and quantifying expressions. Hence, the replacement of the simple past bywould in the corresponding (b) sentences cannot be explained by the need to unam-biguously mark habituality. How, then, do we account for the use of would in suchcases of an unambiguously habitual simple past?

(121) b. Not all his visits would be to the sick; from time to time he would makesurprise visits to clandestine dances.

(122) b. My teachers would say I’d never amount to anything.(123) b. When I would make any remark to him on Geology, he would never rest

until he saw the whole case clearly, and would often make me see it moreclearly than I had done before.

(124) b. . . . they would have prizefights every now and then, and cockfights andeven dogfights.

(125) b. So every week they would receive reports as to what was going on, andoften they would know things before the members of the union knewthem.

(126) b. . . . as a child, I would go to Church, Sunday school, the Church choir. . . .

The key to answering these questions is the concept of distributivity. Brinton(1987, 203) states that the habitual “views a situation as iterated or distributive overa period of time.” She also comments, by way of explaining the incompatibility ofhabituality with the progressive, that “habits, although dynamic, are never continu-ous in the frame considered, but distributive” (207). She says no more regardingdistributivity or the distributive nature of habitual aspect, but her comment cor-rectly places distributivity at the heart of habituality. Similarly, Van Geenhoven(2001, section 5) characterizes frequency adverbs such as often as “bindingsubevent times by distributing them over the overall event time.”

The simple tenses are not habitual. They do not represent a series of events as ahabit because they quantify over events cumulatively, while habitual expressionssuch as the modal will/would quantify distributively: the past tense speaks of allevents within the frame, while would speaks of each event. Thus, even in the habit-ual, as opposed to the generic, interpretation, the simple tenses characterize the pe-




riod of the frame as a whole, while the modal will distributes individual events overthe period of the frame.

Because generic (129), serial (130), and individual-stage stative (131) sentencesconcern a period of time, they cannot take as their anchor a moment in time (129a-131a) but must instead anchor at an interval of time (129b-131b). Habitual would,too, characterizing as it does an entire period of time, cannot take as its anchor a mo-ment (132a) but must anchor at an interval (132b). In a sentence such as (133), thewhen clause is interpreted as a whenever clause, precisely because a when clause,interpreted literally, refers to a moment.

(129) a. !When Suzanne entered, lions ate meat.b. Before they became vegetarians, lions ate meat.

(130) a. !When Suzanne entered, Bill commuted to work by train.b. Throughout 2003, Bill commuted to work by train.

(131) a. !When Suzanne entered, Bill was American-born.b. All his life, Bill was American-born.

(132) a. !When Suzanne entered, Max would stand up.b. Every time Suzanne entered, Max would stand up.

(133) When a man got to the mast-head, he would come slowly down again toget something which he had forgotten. . . . (Richard Henry Dana, TwoYears before the Mast, chap. 12)

It is striking, then, that would can co-occur with each time (134). This expres-sion, unlike every time in (132b), appears to refer not to a set of times (and henceimplicitly to an interval) but instead to but a single time. Yet would can take as itsreference or frame time the time denoted by a phrase headed by it.

(134) Each time Suzanne entered, Max would stand up.

There is more to observe regarding each time. Each time can serve either as anadverbial or as a subordinating conjunction. When it is a conjunction, it does nottake as its complement a clause in would but does take one in a simple tense (135).But when it is an adverbial, the reverse is true—it does not take a clause containingthe simple past, but instead would (136) (or, as in (137), will).

(135) a. She had always liked thrift stores and each time she went into one, ideaswould start to form. (http://www.secondhandsavvy.com/ez5/ez5_3.html)b. . . . she is writing Atlantic series history each time she drives the #24 Ar-gent Mortgage/Toyota/Swift. (http://www.toyotaatlantic.com/Feature.asp?ID=598)

(136) Schools for gardening, astronomy, cooking, mathematics, tailoring andfor the martial arts, but from each of these came a note, “I can teach hernothing if she does not want to learn.” And each time she would reply, “It




is really too difficult.” (http://www.hindu.com/thehindu/yw/2002/10/ 26/stories/2002102600160300.htm)

(137) Now I put the water in a bowl and each time she will drink about half bowlof water especially after a walk. (http://www.puppy.com.my/forum/Dog_and_Puppies_Talk_C3/Behaviour_F15/My_pug_make_the _floor_wet_after_he_drink_water_P6392/)

We may ascribe this contrast to the different functions of each time in the twostructures. The subordinating conjunction serves to individuate the members of aset of events. Example (138a) refers to a series, a set, of events; (138b) to each oneof them. But while the simple tenses refer to a set of events and hence can provideinput into the semantic operation of individuation, will and would do not refer, asthey seem to do, to sets of events but instead already refer to each one, so no furtherindividuation is possible. On the other hand, the adverbial each time refers to one ofa series of times and hence is compatible with an expression such as will, referringto one of a series of events, but not with an expression, such as a simple tense, refer-ring to the series of events as a whole. As we would expect, while used to is quiteuncommon with the adverbial each time, examples like (139) do occur. On the otherhand, used to does not occur with the conjunction.

(138) a. I always left at noon.b. I would always leave at noon.

(139) They used to come twice or three times a week and each time they used totake from our detainees 30-40 to unknown destinations. . . . (http://www.crohdi.org/english/hostages/abbas.asp)

The period of time designated by used to can serve as an anchor for would, as in(140). For that matter, so can that denoted by would itself (141). In fact, just aboutany interval of time will serve to anchor would. In (142), these intervals are ren-dered explicit by the phrases I have italicized here, but in (143), they are given bythe context or are otherwise merely implicit. Thus, the period in question in (143a)is defined (by the preceding sentence) as “Pauline Breedlove’s younger years”; in(143b), it is defined as the period in which she suffered old age and deterioration.

(140) When I used to fly with my parents, I would always tell the security guards“No guns or bullets!” (http://iusedtobelieve.com/the_law/police/)

(141) When I would write to music, I would often produce a great deal. (http://www.outminds.com/outspoken/outspoken_column.cfm?cid=434)

(142) a. When she was in Mumbai, she would often go to the Shanti Durga tem-ple in Goa. (http://www.hinduonnet.com/thehindu/mp/2003/09/29/stories/2003092901610400.htm)




b. Upon returning home she would collect the letters and use them to writethe memoirs of her travels. (http://sunshinedubois.greatnow.com/wrklsr.html)c. As a child, Judy would often creep up into the attic of the family home toplay. She would dress up in all sorts of homemade costumes and make-upand star in her own feature films. (http://www.yourlibrary.ws/childrens_webpage/j-author22000.html)

(143) a. Pauline Breedlove’s younger years are described. It explains how shewould often go to the movies, and because of this eventually became fasci-nated with Hollywood ideals of beauty. (http://www.bookrags.com/notes/be/SUM.htm)b. I know that the indignities of old age and deterioration were almostmore than she could bear, and how she would often go without rather thanask people to do things for her. (http://www.benedira.com/)

Where there is an expression in the sentence that would force a habitual readingof the past tense, the past can be substituted for would in sentences anchored onused to (144a), would (144b), or an interval otherwise designated (144c).

(144) a. When I used to fly with my parents, I always told the security guards“No guns or bullets!”b. When I would write to music, I often produced a great deal.c. As a child, Judy often crept up into the attic of the family home to play.

But if we examine sentences such as those in (144), we notice that a subtle shiftin meaning has taken place from the corresponding would sentences. Example(145) connects each telling to each event of flying. The sense is that each time thenarrator flew, he told the guards “No guns or bullets!” But (144a) merely says thatduring the period in which the narrator flew with his parents, on every occasion hespoke to the guards thus, and the connection between the events of flying and theevents of saying “No guns or bullets!” is indirect, mediated by the fact that alwaysrefers not to all times within that period but only to those occasions on which thenarrator flew with his parents.

(145) When I used to fly with my parents, I would always tell the security guards“No guns or bullets!”

Similarly, (146) says that it was often the case that an occasion of writing to mu-sic was an occasion of producing a great deal. But (144b) says something slightlydifferent—namely, that occasions of producing a great deal were fairly denseamong the occasions of writing to music. Of course, this implies that it was oftenthe case that an occasion of writing to music was an occasion of producing a greatdeal, that is, (144b) implies what (146) says, but it is not what (144b) says.




(146) When I would write to music, I would often produce a great deal.

It is interesting to observe that “stative” used to and “habitual” used to serve asanchors in rather different ways. “Habitual” used to regularly serves to anchorwould, as in the examples (144a) and (145) above. It can also anchor another used to(147).

(147) a. Because when he used to drink, he used to be very violent towards me.(http://www.datec.net.pg/~phowley/oldst1.html)b. When he used to travel to his customers, farmers he used to exchangeproducts with, he used to rehearse the order of prayers for the high holi-days. (http://www.zabludow.com/chosenpagesbaaltefilah.html)

On the other hand, “stative” used to tends to anchor the simple past tense, eitherwith an eventive predicate (i.e., the past tense event occurred in the interval denotedby the used to phrase, as in (148a)) or a stative one (the state persisted over the inter-val denoted by the used to phrase, as in (148b)).

(148) a. . . . when he used to be a wrestler, he faced the Bossman once in a battleroyal in Cordele, Ga. back in 1992 and was eliminated by him. (http://www.raytraylor.net/Wrestling/ Interviews/1_6_02/1_6_02.shtml)b. When he used to be straight, I thought he was adorable. (http://www.geocities.com/cbaglad/malecast.html)

When a “stative” used to anchors the past (e.g., (148)), it is possible to replaceused to with the past tense, with no essential change in meaning (149). “Stative”used to, the past tense with a stative, and expressions such as when I was a child allsimply denote an interval that can serve as anchor for any clause requiring an inter-val as anchor.

(149) . . . when he was a wrestler, he faced the Bossman once in a battle royal inCordele, Ga. back in 1992 and was eliminated by him.

But “habitual” used to is like would, and as such, when it serves as anchor for anexpression with would or another used to, we understand the two to coordinate, sothat there is a link between the members of each pair of events. In (147a), for exam-ple, each episode of drinking involves an episode of violence toward the narrator.Example (150), substituting the simple past for used to, lacks this kind of linking.

(147) a. Because when he used to drink, he used to be very violent towards me.(150) Because when he drank, he used to be very violent towards me.




Thus there is something about the meanings of would and “nonstative” used tothat differs from the meanings of “stative” used to and the past tense, that interactsdifferentially with the meaning of the past tense, and that affects how the past tenseanchors to the various types of expressions—used to, would, the past tense itself,and other expressions.

The factor is that of quantification. The stative use of used to, like that of the pasttense, simply quantifies cumulatively over the whole of the period, while thenonstative use, like would, quantifies distributively. The used to in (145, repeatedbelow) speaks to each occasion of flying, so that it is not the totality of the period inwhich the narrator flew with his parents that each occasion of speaking to theguards anchors to. But the past tense can only anchor to the period as a whole.Hence, the link, the logical relationship, between each event of flying and each actof telling cannot be directly expressed by the past-tense version (144a).

(144) a. When I used to fly with my parents, I always told the security guards“No guns or bullets!”

(145) When I used to fly with my parents, I would always tell the security guards“No guns or bullets!”

Conclusion

Traditionally, used to, would, and the simple tenses are considered markers ofhabituality and genericity in English. Habits have been said to be conditions consti-tuting a state, and past habits no longer obtain. However, expressions of habitualityin English fail most tests for stativity, and expressions of past habituality do not nec-essarily express conditions that no longer apply.

Used to is neither a past tense nor a marker of habituality. It is a kind of presenttense, a deictic tense of discours. Its distribution and usage are similar to those ofthe present perfect, and it functions as a kind of anti-present-perfect, serving to di-vorce a past state or series of events from the present state of affairs. In its presentsemantics and past morphology, it is similar to the preterito-present modals, and inBritish English, at least, it has a past-tense counterpart (had used to).

Nor are the simple tenses markers of habituality or genericity. These are possiblecontextual interpretations of the simple tenses (and not only the simple tenses), butthese readings come not from the meanings of the tenses but from elements ofcotext or context. The simple tenses quantify cumulatively, not distributively, andcan neither anchor, nor link, individual members of a series in the way that expres-sions with truly habitual uses do.

The modal will (past tense would) has among its various meanings that ofhabituality, though not genericity, the generic reading (as in boys will be boys) be-ing, as in the case of the simple tenses, a contextual interpretation. Will can receive a




habitual reading out of context and in the absence of any special adjuncts, and itquantifies distributively, not cumulatively.

Notes

1. Pages on the World Wide Web are highly volatile. However, a search of theWeb should find comparable examples to those on pages that have disappeared.

2. The exclamation point indicates that the sentence is grammatical but with adifferent meaning than that assumed, in this case, a habitual reading referring to aspecific dinosaur.

3. The example sentence is ambiguous, but the gloss makes clear the intendedsense.

4. A fuller exposition of the argument and discussion summarized in this sec-tion may be found in Binnick (forthcoming).

5. Weinrich specifically distinguishes (1964) the besprochene Welt (1973,“monde commenté”—‘commented world’ or ‘world of commentary’) and theerzählte Welt (“monde raconté”—‘narrated world’ or ‘world of narration’).

6. Bybee, Perkins, and Pagliuca (1994) say that the default is perfective. This iscontroversial, however, and in the present context, the point is simply that habitual-ity is not the default reading.

7. Suppositional would seems never to be used as the past-tense counterpart ofwill but always as a less direct present-tense modal.

8. Would seems not to share this meaning.

References

Bennett, Michael, and Barbara Hall Partee. 1978. Toward the Logic of Tense andAspect in English. Bloomington: Indiana University Linguistics Club.

Benveniste, Emile. 1959. Les rélations de temps dans le verbe français. Bulletin dela société linguistique de Paris 54:69-82.

⎯⎯⎯. 1966. Problèmes de linguistique générale. Paris: Gallimard.Binnick, Robert I. Forthcoming. Used to and Habitual Aspect in English. Style.Brinton, Laurel J. 1987. The Aspectual Nature of States and Habits. Folia

Linguistica 21:195-214.⎯⎯⎯. 1988. The Development of English Aspectual Systems. Cambridge, UK:

Cambridge University Press.Bybee, Joan L., Revere D. Perkins, and William Pagliuca. 1994. The Evolution of

Grammar: Tense, Aspect, and Modality in the Languages of the World. Chicago:University of Chicago Press.

Comrie, Bernard. 1976. Aspect: An Introduction to the Study of Verbal Aspect andRelated Problems. Cambridge, UK: Cambridge University Press.

⎯⎯⎯. 1985. Tense. Cambridge, UK: Cambridge University Press.




Copley, Bridget. 2004. Aspect and Scope in Future Conditionals. http://semanticsarchive.net/Archive/GQ3ZDY5Z/copley.aspect.scope.future.conditionals.5.27.04.pdf

Cutrer, L. Michelle. 1994. Time and Tense in Narrative and in Everyday Language.PhD diss., University of California, San Diego.

Dahl, Östen. 1985. Tense and Aspect Systems. Oxford, UK: Blackwell.Denison, David. 1993. English Historical Syntax. Harlow, UK: Longman.Dowty, David R. 1979. Word Meaning and Montague Grammar. Dordrecht, The

Netherlands: Reidel.Ferrari-Bridgers, Franca. 2004. [V-N] Compound Nouns Formation in Italian. Pa-

per presented at 20th Scandinavian Conference of Linguistics, January 7-9, Hel-sinki, Finland. http://www.ling.helsinki.fi/kielitiede/20scl/abstracts.shtml.

Freed, Alice F. 1979. The Semantics of English Aspectual Complementation.Dordrecht, The Netherlands: Reidel.

Harrison, Rick. 2002. Verb Aspect. http://www.rick.harrison.net/langlab/aspect.html

Hirtle, W. H. 1967. The Simple and Progressive Forms: An Analytical Approach.Québec: Les presses de l’université Laval.

⎯⎯⎯. 1975. Time, Aspect, and the Verb. Québec: Les presses de l’universitéLaval.

Jespersen, Otto. [1932] 1961. A Modern English Grammar on Historical Princi-ples: Part IV. Syntax (Third Volume). Reprint, London: Allen & Unwin.

Jørgensen, Erik. 1988. Used to (+ infinitive). English Studies 69:348-54.Leech, Geoffrey N. 1987. Meaning and the English Verb. London: Longman.Lyons, John. 1977. Semantics. Cambridge, UK: Cambridge University Press.Partee, Barbara Hall. 1973. Some Structural Analogies between Tenses and Pro-

nouns. Journal of Philosophy 70: 601-9.Schibsbye, Knud. 1970. A Modern English Grammar. Oxford, UK: Oxford Uni-

versity Press.Tagliamonte, Sali, and Helen Lawrence. 2000. “I Used to Dance, but I Don’t Dance

Now”: The Habitual Past in English. Journal of English Linguistics 28:324-53.Van Geenhoven, Veerle. 2001. Aspect, Pluractionality, and Adverbial Quantifica-

tion. Paper presented at Perspectives on Aspect, December 12-14, Utrecht, TheNetherlands. http://www-uilots.let.uu.nl/conferences/Perspectives_on_Aspect/Proceedings/geenhoven.pdf

Vendler, Zeno. 1967. Linguistics in Philosophy. Ithaca, NY: Cornell UniversityPress.

Wallace, Stephen. 1982. Figure and Ground: The Interrelationships of LinguisticCategories. In Tense-Aspect: Between Semantics and Pragmatics, edited byPaul J. Hopper, 201-23. Amsterdam: Benjamins.

Weinrich, Harald. 1964. Tempus: besprochene und erzählte Welt. Stuttgart: W.Kohlhammer.

⎯⎯⎯. 1973. Le temps: le récit et le commentaire. Paris: Éditions du Seuil.




Robert I. Binnick is Professor of Linguistics at the University of Toronto. Hisresearch interests include the semantics and pragmatics of tense and relatedareas.




Documents

Habitual Aspect in English