1. Introduction
English control verbs take infinitival complements that include the particle to:
(1)
a. She tried to read one book each week.
b. She promised to read one book each week.
The verb use has a special application which qualifies it as a type of control verb, albeit with aspectual semantics and certain peculiar syntactic characteristics that make it comparable to a modal verb:
(2) She used to read one book each week.
The negated versions of the sentences in (1) and (2) are of course as follows:
(3)
a. She didn't try to read one book each week.
b. She didn't promise to read one book each week.
c. She didn't use to read one book each week.
What is striking is that instead of (3c) one very frequently encounters the following:
(4) She didn't used to read one book each week.
Note that because the [d] in “used to” is assimilated and not heard in actual speech, it is necessary to base this study on written data. This is actually advantageous. Written data provide a certainty and a formality that are often lacking in the spoken language. When speakers write something even though they do not hear it, one can be much more certain that it reflects their linguistic competence and is not a performance issue.
While sentences like (4) are striking, it is equally striking that many speakers who accept (4) consider (3c) to be ungrammatical. I have presented sentences (3c) and (4) to a number of educated native speakers of American and British English whose education ranges from a bachelor's degree to a doctorate, and only one of them, a Scotsman over sixty years of age with a doctoral degree, categorically rejected (4). Nearly all the others found (3c) strange and preferred (4). One informant, an American English teacher with a master's degree, preferred (3c) but said that she found both sentences strange and that she would only say, “She never used to read one book each week”, thereby avoiding the dilemma.
But the reader does not have to draw any conclusions from my little survey. The Internet contains millions of examples of both didn't use to and didn't used to, and an Internet search for “didn't used to” yields countless websites in which it is debated which construction is correct. Opinions differ greatly. Roger Woodham, who answers questions about grammar on the Learning English page of the website of the BBC World Service, writes that didn't use to and didn't used to are both correct.Footnote 1 My favourite quote, from Kenneth's ESL Blog, is the following:Footnote 2
“Did he used to” or “Did he use to”? Interesting question. As a native British English speaker I would argue that “used to” is correct and “use to” is wrong. But people say both, and a BBC website says both are okay. So it's probably not worth losing any sleep over!
These are but a few of the many comments and discussions that one can easily find. It is a fact that many, maybe even most, English speakers, regardless of whether they are British or American and regardless of their educational level, write “didn't used to”. And yet, these speakers would never say the following:
(5)
a. *She didn't tried to read one book each week.
b. *She didn't promised to read one book each week.
There is evidence that “didn't used to” has had to fight for recognition. The following quote, borrowed from Tagliamonte & Lawrence (2000), is both revealing and amusing:
Though such forms as ‘Did you used to live in Whitechapel?’ and ‘I didn't used to work at Billingsgate’ are much used by illiterate persons it is highly improbable that they will ever come to be employed by speakers of education and refinement. (T. O. Lees, 1931, English ‘English’, Tokyo, Osaka, p. 96)
In Women in Love, written in (1921), David H. Lawrence uses the expression “didn't used to” to reflect the relative lack of education of a servant woman who is talking about how happy the servants were whenever one of the spoiled Crich children was punished:
And didn't we used to be thankful when one of them caught it. They were the torment of your life.
The way this subject is handled in grammar books provides additional evidence that didn't used to has evolved from an unacknowledged, non-standard expression to one that has achieved much wider acceptance. If one consults reference books such as Foley & Hall (Reference Foley and Hall2004), the Longman Dictionary of Contemporary English (1995), Fowler (Reference Fowler1965), Leech et al. (Reference Leech, Cruickshank and Ivanič2012) and Dirven et al. (Reference Dirven, Pütz and Jäger1989) one finds that didn't use to is the prescribed form and there is no mention of didn't used to. Quirk et al. (Reference Quirk, Greenbaum, Leech and Svartvik1985) mention both variants but state that didn't used to is often considered non-standard. However, other reference books, particularly more recently published ones such as Swan (Reference Swan2005), the Collins Cobuild English Grammar (2011), the Oxford English Grammar (1996) and the Cambridge Grammar of English (2006) are very much reflective of current usage. They treat didn't used to not as incorrect but as a variant of didn't use to. Christophersen & Sandved (Reference Christophersen and Sandved1969) do not condemn didn't used to but imply that it is less frequent than didn't use to. This may have been true in 1969, but it would be difficult to make the same claim today. Table 1 shows the results of an Internet search conducted on 30 March 2013. Whereas an Internet search such as this one can certainly not be relied upon for any kind of statistical precision, it does serve to demonstrate that didn't used to co-exists with didn't use to and is frequently the preferred form.
Table 1: The number of occurrences of “didn't use to” vs. “didn't used to” in declarative sentences
Negation is of course not the only syntactic operation in which did and use co-occur. Interrogation, which also involves do-insertion, also results in this phenomenon. The following examples are from discussions on the Internet and show that didn't used to and didn't use to co-exist in questions:
(6)
a. Did you use to live in London?
b. Was Edward always a vampire or did he used to be human?
An Internet search of interrogative sentences on 30 March 2013 yielded the results in Table 2.
Table 2: The number of occurrences of “didn't use to” vs. “didn't used to” in questions
Needless to say, if a sentence is both interrogative and negated, did and used also co-occur:
(7)
a. Didn't you used to go to the Hacienda? Footnote 3
b. Didn't he used to be Ben Affleck? Footnote 4
The results of another Internet search on 31 March 2013 can be seen in Table 3.
Table 3: The number of occurrences of “didn't use to” vs. “didn't used to” in negative questions
I have three aims in writing this article: to show that the co-occurrence of did and used is a fairly recent development; to show that reanalysis is the probable cause; and, most importantly, to offer an explanation for why the reanalysis has taken (is taking) place. The explanation is actually that the reanalysis in question is the continuation of a process that started hundreds of years ago. It is well-known that use (as a kind of control verb) began as a full lexical verb and gradually took on modal-like characteristics. What I will claim is that the evolution of use did not stop at modalization but has continued to adverbialization. The article is organized as follows:
In Section 2 I present historical data from three different corpora and from some relevant articles and the Oxford English Dictionary to show the development of the use to construction. It will be shown that originally, beginning in the 14th century, use as a modal-control verb behaved syntactically like a lexical verb. In Section 3 I provide a summary of how, syntactically, use began to show more and more modal-like characteristics while retaining some of its lexical verb qualities, with the end result that its status became very uncertain in the minds of English speakers and it became susceptible to reanalysis. Section 4 presents the hypothesis that reanalysis is the best explanation for the development of didn't used to and discusses the factors that probably played a role in the reanalysis process. Section 5 is a brief summary.
2. The history of use to
The verb use is an old lexical verb of Norman French origin that means “to make use of”. However, use also functions as a kind of modal that selects infinitival complements with to and is semantically aspectual. The semantics of used to has been discussed in the literature by Comrie (Reference Comrie1976), Hantson (Reference Hantson2005), Tagliamonte & Lawrence (2000) and others and is typically described as former habituality, or habituality that no longer holds. In this article, to distinguish the two functions of use I will apply the terms lexical use and modal use.
In Present Day English, modal use is defective. It only occurs in the simple past tense, never in the present or in participial or gerundive form. It cannot be selected by modals and auxiliaries. This makes it very much like a modal verb. Things were not always so. From the Middle English period well into the Early Modern English era modal use was able to occur in the past, present and perfect tenses, it was able to occur with a modal or auxiliary, it had a gerundive form, and it could even be passivized. According to Nagle (Reference Nagle1985), who is citing the 1933 edition of the Oxford English Dictionary, the first documented occurrence of modal use was in 1303:
(8) For ryche men vse comunly Sweryn grete othys grysly.
for rich men use commonly swear great oaths grisly.
‘For rich men commonly are in the habit of swearing great, grisly oaths.’
A thorough review of Kroch & Taylor (Reference Kroch and Taylor2000), the Penn-Helsinki Parsed Corpus of Middle English, Release 2 (PPCME2), provided relevant data. Some example sentences follow. Note that in the first example the expression use to is in the perfect tense, that is, in past participle form:
(9) “Sir,” seyde sir Kay the Stywarde, “if ye go now unto youre mete ye shall breke youre olde custom of youre courte, for ye have nat used on thys day to sytte at your mete or that ye have sene some adventure.” Footnote 5
The following example from Chaucer's Canterbury Tales, cited in Nagle (Reference Nagle1985), shows that modal use even had a gerundive form:Footnote 6
(10) A theef he was for sothe, of corn and mele … and usuant to stele. A thief he was forsooth of grain and meal ... and using to steal
The following example from the OED shows that modal use could even be passivized, like a true lexical verb:Footnote 7
(11) It was in old times vsed ... for men to shaue themselues. (1621 R. Montagu Diatribæ Hist. Tithes 531)
Despite the fact that use to was present in Middle English, one cannot expect to find examples of didn't use to or didn't used to for the simple reason that do-insertion in negated and interrogative sentences had not yet entered the language. A search for didn't used to must therefore begin in Early Modern English. The corpus that I have used is Kroch, Santorini & Delfs (Reference Kroch, Santorini and Delfs2004), the Penn-Helsinki Parsed Corpus of Early Modern English (PPCEME). What is interesting is that even in the Early Modern English period modal use still behaved like a full lexical verb (was not restricted to simple past tense form). The following examples show modal use in the perfect tense, the present tense and in combination with modals.
(12) For if you will haue a tree beare more fruite then it hath vsed to do; it is not any thing you can do to the boughs ...Footnote 8
(13) I'll creepe vp into the chimney. There they alwaies vse to discharge their Birding-peeces.Footnote 9
(14)
a. To whom a physycyon cam to gyue hym councell and seyd that he must vse to ete metis that be light of dygestyon as small byrdys.Footnote 10
b. … by mine aduise, the childe shall vse to speake no latine.Footnote 11
The following example shows that use to could even be used in the passive voice and present tense with the meaning of habitualness. Note that it differs from the passive in (11) because it does not involve impersonal it:
(15) But leauing to speake any further of these Circles, because they are not vsed to be described in Spheares but onely in Astrolabes, I will now treate of the foure lesser Circles ...Footnote 12
The modal use could also be negated without do-insertion:
(16) “I vse not to gest,” quoth this parson, “when I speake so earnestly.” Footnote 13
(17) Forsooth for I wyll not, I vse not to kisse men.Footnote 14
Nonetheless, as the following sentence shows, at some point in Early Modern English do-insertion for negation was possible, but didn't used to had not yet appeared:
(18) And so, he that vsed to teache, did not commonlie vse to beate, but remitted that ouer to an other mans charge.Footnote 15
And at the time when do-insertion came into use for negation, it was still possible to employ use to in the present tense:
(19) Look you, though in strictness, unless the Party be dead, we do not use to admit of any such Evidence.Footnote 16
The following example shows that in 1685 do-insertion had entered the language for interrogative sentences but reanalysis of used to had probably not yet taken place:
(20) Did your Lady use to sup below stairs or above? Footnote 17
And the following example shows that when do was used for emphasis in Early Modern English, use could accompany it, but in infinitival form, not tensed:
(21) I believe thou dost use to bake on Sundays, dost thou not? Footnote 18
The data viewed so far show that use to existed in Middle and Early Modern English and behaved like a lexical verb, including being affected by do-insertion. There is, however, no evidence that in Early Modern English did or didn't and used to could co-occur, which means that we need to focus on Modern English. The Corpus of Historical American English includes examples from written American English from 1810 to 2009. A search for occurrences of didn't used to and didn't use to by decade yielded the results shown in Table 4.
Table 4: Occurrences of “used to” vs. “use to” in the Corpus of Historical American English by decade
In this corpus didn't used to doesn't appear until after 1850 whereas didn't use to was around before that. Until around 1940 didn't use clearly dominated. After that the trend started to reverse itself and in the 1980s didn't used to became quite dominant. Incidentally, whereas Table 4 suggests that didn't used to appeared in the middle of the 19th century in the United States, the OED places the first occurrence in England in the 18th century. The following sentence is from a dialogue and is spoken by a “glassman”, a person who makes or sells glass products (non-nobility, non-gentry):
(22) ‘Dad’, (said the glassman. pulling out his pocket-handkerchief) ‘I didn't used to be so melch-hearted.’ (1782 E. Blower George Bateman II. 111)
In this section we have seen that use in its modal function behaved like a lexical verb. However, probably because of its aspectual semantics it gradually took on a number of modal-like characteristics. This will be the subject of the next section.
3. The incomplete modalization of use, and its consequent uncertain status
It is not surprising that the verb use, even though it had the aspectual semantics of a modal, behaved like a full lexical verb, since it was derived from the lexical verb use meaning “to make use of”. At some point, however, use in its modal function began to be treated, logically, as a modal. The history of use to is thus linked to the history of modals in general. For a detailed history of English modals, the reader is directed to Fischer (Reference Fischer and Hart2003) and Lightfoot (Reference Lightfoot, Anderson and Jones1974). For the purposes of this article it is important to know that once the Middle English period had begun, modals began to undergo grammaticalization, which included the loss of infinitival and participial forms and the loss of the ability to combine with other modals and auxiliaries. Past tense forms of modals also lost their past semantics. This led to the appearance of the so-called quasi-modals like ought to and have to. The expression use to, also sometimes categorized as a quasi-modal, went through many of the changes that modals went through. These changes began during the Early Modern English period and will be discussed below.
Modal use was probably pre-destined to become reanalysed as a modal because it has one characteristic that differentiates it from other lexical verbs that take a to-infinitive: used to only serves to add habitual modality to a verb. It does not assign a thematic role to its subject like a control verb. In the following sentences, one could not maintain that use assigns a thematic role any more than one could claim that the habituality modal would does:
(23)
a. John used to get up early every day.
b. John would get up early every day.
Let's now look at the changes that use underwent as a result of its status as a quasi-modal. As we have already pointed out, it can only be used in the past tense.
(24)
a. Mary used to play football every Saturday.
b. *Mary uses to play football every Saturday.
It also has no gerundive form and it cannot combine with true modals or auxiliaries:
(25)
a. *John was using to visit his mother on Sundays.
b. *Mary could use to run faster than her brother.
c. *Before Peter studied law he had used to be a farmer.
Further evidence of the modal-status of use is that even in the 20th century there are examples of its being used without do-support. The following examples are from Jørgensen (Reference Jørgensen1988):
(26)
a. The Mistress usedn't to sleep well at night. (from Dumb Witness (Poirot Loses a Client) by Agatha Christie, 1937)
b. ‘This the only strip in the vicinity?’ – He nodded again. – ‘They used not to have a strip at all.’ (from The Rainbow and the Rose by Nevil Shute, Norway, 1958)
c. He used not to sweat like that. (from You Only Live Twice by Ian Fleming, 1964)
These sentences are of course for most speakers antiquated at best and would normally be constructed with do-support, but they illustrate that use has retained modal-like characteristics up until very recent times. On the other hand, it has not completely surrendered its status as a lexical verb. The most obvious similarities between use and a normal control verb are that they both take infinitival complements and they both require do-support in certain environments:
(27)
a. You promised to be on time, didn't you?
b. She really tried to impress us, didn't she?
c. He used to live in England, didn't he?
There are other signs that modal use has been slow to forfeit its lexical verb qualities. The following sentences are from rather recent times. In the (a) sentence, from Nagle (Reference Nagle1985), modal use appears in the present tense. In the (b) sentence, from Jørgensen (Reference Jørgensen1988) we see modal use as a past participle, in the pluperfect:
(28)
a. All this time, of course, they went on talking agreeably, as people of birth use, about the Queen's temper and the prime Minister's gout. (from Orlando by Virginia Woolf, 1928)
b. His hide was less shiny than it had used to be. (from Animal Farm by George Orwell, 1962)
One can see that the modal semantics of use has been (and is) in conflict with the fact that it was derived from a still existent lexical verb. The result is that it behaves sometimes like a lexical verb and sometimes like a modal. This dual status causes confusion in the minds of speakers and renders use to more susceptible to reanalysis. This is the subject of the next section.
4. Reanalysis as the explanation
I begin this section with a definition of reanalysis. I will follow the definition in Langacker (Reference Langacker and Li1977), who says that reanalysis is “change in the structure of an expression or class of expressions that does not involve any immediate or intrinsic modification of its surface manifestation.”Footnote 19 Brinton & Traugott (Reference Brinton and Traugott2005) expound on this definition by adding that reanalysis entails a change in constituency, a change in category labels, and boundary loss. We will see that all of these definitions and criteria are very applicable in the case of used to.
What we saw in Section 3 is that use underwent a kind of reanalysis or grammaticalization, from lexical verb to modal or quasi-modal, while never surrendering all its lexical verb qualities. I think that it might be useful at this point to compare this with what happened in the case of the perfect auxiliary have. When the lexical verb have meaning “to own or possess” was grammaticalized as the perfect auxiliary, it did not disappear as a lexical verb. It simply took on an additional function. When do-insertion became obligatory for lexical verbs in negated and interrogative sentences, have was (and still is, in some cases) exempt, due to the fact that it is both a lexical verb and an auxiliary. Sentences like the following are fairly common:
(29)
a. Have you the ability to make this project a success?
b. I haven't a clue.
The modal verb use shows comparable behaviour. Because it has undergone a kind of reanalysis or grammaticalization, a sentence like (26a) (The Mistress usedn't to sleep well at night) is rare but not totally impossible. However, since use has never lost its status as a full lexical verb, The Mistress didn't use to sleep well at night would be more normal. The grammaticalization of use was thus not complete. When used as a modal it forfeits its ability to appear in any tense but the past, its ability to combine with modals and auxiliaries, etc. but because of its dual status as a lexical verb it normally requires do-support in questions and negation.
The question that we will now address is why didn't use to has become didn't used to. I will argue that reanalysis is the best explanation. The cornerstone of my argument is the fact that an English verbal projection cannot contain more than one tensed or finite verb. Consequently, in the phrase didn't used to either didn't or used must be something other than a verb. Since the clitic negation marker n't is only attached to verbs, it is clearly didn't that is the finite verb, which means that used must be something else. It is very plausible to construe this as reanalysis. This is strongly supported by the fact that many people who say “didn't used to” not only would never say “didn't use to”, they find it ungrammatical, although they would never say “didn't promised” or “didn't tried”. For these speakers, the word used in the phrase didn't used to does not have the status of verb. Since it is not a verb, noun, adjective or preposition, it is an adverb. Because it combines the semantics of the adverbs usually and formerly, it could be considered an adverb of time and frequency. I will now present some factors that I think might have played a role in bringing this reanalysis about.
The most significant factor is the uncertain status of modal use, a hybrid that vacillates between being a lexical verb and being a modal. If something is neither fish nor fowl there is some possibility that it will be categorized as something else eventually. Another factor is that in the history of English the particle to has been involved in reanalysis and grammaticalization before. In Los (Reference Los2005), Lightfoot (Reference Lightfoot1979) and Nagle (Reference Nagle1985) there is discussion of how to was originally a preposition appearing not only before nouns but before nominalized verbs. At some point in the later Middle English period, the combination of to plus a nominalized verb was reanalysed as VP (to plus verb) so that to was, in addition to being a preposition appearing before nominals, also a non-finite tense marker. Nagle suggests that this dual status of to created opacity that opened the door to reanalysis.
Another important factor is that modals do not have an infinitive form. Since use has so many modal qualities and can be said to have undergone grammaticalization, it is understandable that use in its modal function would seem strange in infinitival form. In the following sentence, modal use is in fact in infinitival form, and it is not unreasonable to assume that many speakers reject the sentence simply because they cannot accept a modal as an infinitive.
(30) She didn't use to like broccoli.
The rejection of use as an infinitive may well have encouraged the use of used to instead.
Still another factor, closely related to the last one, has to do with an observation made by Jørgensen (Reference Jørgensen1988). He says that the phenomenon “didn't used to” might have arisen from the fact that when used appears in its quasi-modal tensed form the past ending is always present, and this gives speakers the subconscious feeling that the past ending is a “permanent fixture”. This would prevent one from saying “use” when using the verb in its modal sense. This argument by Jørgensen is not implausible.
Phonology could also have played a role. For many speakers the pronunciation of used to when use is a modal verb deviates from the pronunciation of used to when used is employed as a transitive verb. Note the pronunciation of the sibilant in the following examples:
(31)
a. This is the camera that Mary used to take pictures.
b. This is the place where Mary used to take pictures.
For many speakers of Standard British and American English, the pronunciation of used is [juzd] in (a) but [jus] in (b). Contrast this with a sentence containing another control verb, such as Mary refused to take pictures. In this sentence, the [z] and [d] phonemes in refused are not devoiced and assimilated to the following [t]. The merging of the [z] and [d] phonemes in used with the [t] in to may very well have facilitated the reanalysis of used and to as a constituent.
It has actually been suggested to me that the used to/use to distinction is a simple spelling problem. Nothing could be further from the truth. A spelling problem exists when a speaker doesn't hear a particular letter pronounced and spells accordingly. People may spell “climb” <clime> because they don't hear a [b] or “often” <offen> because they don't hear a [t]. In the case of “used to”, people don't hear or pronounce the [d] on “used” but insist that it is there. That is not an issue of spelling but of grammar.
Another factor, mentioned before, is the fact that modal use cannot be used in the present tense:
(32) *She doesn't use to go home on time every day.
It is not difficult to imagine that a speaker might think that if this sentence is impossible, its past tense equivalent should also be impossible:
(33) She didn't use to go home on time every day.
In other words, if used to, which is in the past tense, cannot also appear in the present tense the way any normal verb can, then it must not be a verb.
Another important factor was probably analogy. One cannot help but think of the expression to be used to. According to Lightfoot (Reference Lightfoot1979) and Nagle (Reference Nagle1985), to be used to first appeared in the late 1500s, when modals were separating from main verbs and quasi-modals started to appear. Nagle argues that it spread as a result of analogy with expressions such as be wont to and be accustomed to. It most likely originated as an adjectival past participle. I say this because one of the meanings of transitive use was “to accustom”. Note the following examples from the OED:Footnote 20
(34)
a. Some moderate skill in it will use a man to reason closely. (1688 T. Shadwell Squire of Alsatia ii. i. 29)
b. It is not surprising that the seal ... should use her little ones to live under water. (1783, J. O. Justamond tr. G. T. F. Raynal, Philos. Hist. Europeans in Indies (new ed.) VII. 91)
Although use to came into existence at least two hundred years before be used to, since the latter expression invariably involves the adjectival past participle used, it is certainly possible that the invariability of used in this expression had an effect on creating Jørgensen's above-mentioned “permanent fixture” in didn't used to. This idea becomes, I think, all the more plausible when one considers the fact that to be used to could originally take to-infinitives as complements, the same as modal use. This similarity between the two constructions would seem to make use to much more susceptible to change through analogy. Note the following additional examples from the OED:Footnote 21
(35)
a. He hath bene vs'd Euer to conquer. (1616 Shakespeare Coriolanus)
b. I'm not used to be used in this manner! (1796 F. Burney Camilla IV. viii. vii. 329)
I would also like to say that I think it is significant that the reanalysis of used to as an adverb began with negated sentences, interrogative sentences, and negated interrogative sentences, all of which involve do-insertion, subject-auxiliary inversion, or both. These are more complex structures, syntactically speaking, and this added complexity combined with the already uncertain status of modal use may well have contributed to the reanalysis.
I will continue this discussion by adding data that provide evidence that used to has been reanalysed as an adverb. What is interesting and important is that some of the examples involve neither negation nor interrogation. The following sentences from the Midwestern United States are from Nagle (Reference Nagle1985). Nagle even refers to (36a) as “preposed adverbial used to”. (He does so without elaboration or discussion, but he was clearly thinking in a manner consistent with the hypothesis being presented in this article.)
(36)
a. Used to he called home every week.
b. Mary used to didn't play tennis.
These sentences are at best highly marginal for most speakers, but they do exist. Consider the following additional sentences:
(37)
a. Musicians used to came up to me after I had written a review.Footnote 22
b. That was two years ago, when she used to came down to the corner of Chrystie bare-headed to meet you after supper.Footnote 23
c. We used to took a vacation every summer but we stopped.Footnote 24
Given the placement of used to in these examples, it seems very reasonable to conclude that used to has been reanalysed as an adverb.
I would now like to return to the definition of reanalysis presented at the beginning of this section to show that it applies to used to. Langaker's criterion is that a reanalysed syntactic element shows no superficial sign of having changed. This is certainly the case with used to. Consider the following two sentences:
(38)
a. Catholics used to eat fish on Fridays.
b. Catholics didn't used to eat meat on Fridays.
In the (a) example used is the finite verb in the sentence. In (b), the finite verb is did and the complement of did is the VP eat meat on Fridays. The finite verb also bears the clitic negation marker, and the entire verbal projection didn't eat meat on Fridays is modified by the aspectual adverb used to. The phrase used to looks and sounds the same in both sentences. Recall that Brinton and Traugott state that reanalysis will involve a change in constituency, a change in category labels and boundary loss. These criteria are all met in the case of used to. Constituency has definitely been realigned. In (38a), to is the non-finite tense marker and forms a constituent with eat. In (38b), to forms a constituent with used, that constituent being an adverb. Category labels also change, from verb to adverb. And given that used and to have combined to form an adverb, one can also speak of boundary loss, which is enhanced by the coalescence of phonemes.
5. Summary
This article has been about the reanalysis of the quasi-modal used to as an adverb. The reanalysis is the continuation of a process that began centuries ago when use was reanalysed/grammaticalized as a modal. That is, when the verb use meaning “to make use of” took on the additional function of aspectual modal early in the 14th century, it continued to behave syntactically as a lexical verb. More than two hundred years later, English modals began to undergo reduction resulting from grammaticalization. Probably because of its modal semantics, use was associated with modals and began to be treated more and more like a modal. And yet, most probably because of its having been derived from the still existent lexical verb use, modal use led a “double life”, with many speakers treating it not as a modal but as an anomalous lexical verb. The result of this series of events is that up until very recent times there has been inconsistency among speakers in the way they employ modal use, and this has created uncertainty concerning the actual category of the verb. This made it a good candidate for reanalysis.
It appears that the reanalysis began during the first part of the 19th century in the United States and a bit earlier in England. While the phenomenon began as a non-standard expression it has gradually become more and more accepted. This is attested by the countless discussions and debates being held on various websites and by the fact that the coexistence of didn't used to and didn't use to is discussed in renowned books on English grammar, both American and British.
It is probably not a coincidence that the reanalysis process began with negated and interrogative sentences. These kinds of sentences are more complex, involving subject-auxiliary inversion or do-insertion, and this may have enhanced the confusion surrounding the categorial status of use. In any case, if used to as an adverb first appeared in questions and negated sentences, there is evidence that its use is spreading to other kinds of sentences.
The overwhelming evidence that reanalysis is the most plausible explanation for this phenomenon is that in Standard English a clause may contain only one finite verb.
ROBERT CIRILLO completed his bachelor's and master's in linguistics at Indiana University in the 1970s. He interrupted his doctoral studies to pursue a business career in Europe. After working for over twenty years in Germany and the Netherlands, he returned to the field of linguistics, completing his doctorate at the University of Amsterdam in 2009. Since then he has remained active in the field by participating in conferences, publishing articles and teaching generative grammar at the University of Amsterdam. Email: ciril000@planet.nl
