Studies in Variation, Contacts and Change in English

Volume 18 – Exploring Recent Diachrony: Corpus Studies of Lexicogrammar and Language Practices in Late Modern English

Article Contents

1. Introduction

2. Get and its constructions

2.1 The get-passive

2.2 Get-constructions

3. The historical development

3.1 Earlier studies

3.2 Corpus data

4. Data on prescriptions

4.1 The material

4.2 Descriptively correct assessments of the get-passive

4.3 Proscription against get-constructions

5. The actual effect of proscription

6. Conclusion





Get, get-constructions and the get-passive in 19th-century English: Corpus analysis and prescriptive comments

Lieselotte Anderwald
University of Kiel


Prescriptivism as a factor influencing (at least written) language is regularly invoked when corpus results are unexpected, but is rarely substantiated. This paper presents one case study that raises interesting methodological and linguistic questions. The get-passive, together with have got, has seen considerable criticism by language mavens in the twentieth century (e.g. reported in Ballard 1939: 23–26; Mittins et al. 1970: 33–35). Since the rise of the get-passive is a Late Modern English phenomenon (Denison 1993, 1998; Hundt 2001; cf. also Gries and Hilpert 2012 for the twentieth century), it makes intuitive sense to link its stigmatization with prescriptive grammar writing, which was demonstrably at its height in the nineteenth century. And indeed, prescriptive grammars have been blamed for the comparatively slow rise of the get-passive over the course of the nineteenth century (Hundt 2001). However, this claim has not yet been supported by much evidence from the grammars themselves, and it is not clear whether the evidence from the twentieth century can be projected back in time. For this reason, this paper traces the rise of get over the nineteenth (and into the twentieth) century corpus-linguistically in COHA (Davies 2010–) and differentiates the individual constructions that get was (and is) used in. The shift in constructional use is then correlated with prescriptive comments in nineteenth-century grammar books, based on my Collection of Nineteenth-Century Grammars, which contains 258 grammar books from Britain and the US, and which allows us to investigate temporal correlations (surely a prerequisite for causal connections). As this paper shows, criticism of get and get-constructions does occur in the nineteenth century; however, the effect of the proscription against get seems to be minimal: it is small in scale, short in duration and quickly compensated by subsequent rises. If anything, the get-passive is less affected than get overall. The delay in the rise of the get-passive is thus highly unlikely to be due to nineteenth-century proscriptions.

1. Introduction

The get-passive in English (e.g. he got arrested) stands in interesting paradigmatic contrast to the more usual be-passive (e.g. he was arrested), and by various criteria is an instance of a non-canonical passive construction (in the sense of Alexiadou 2005, 2012; Alexiadou and Schäfer 2013). Compared to the passive with be, the get-passive is far less frequent in terms of text frequency, it is more colloquial, and it carries additional semantic information, typically connotations of adversity or benefit for the subject (she got promoted), and a notion of responsibility of the patient (in the sense of she got herself promoted), as regularly noted by descriptive grammars (cf. Quirk et al. 1985: 161–162; Huddleston and Pullum 2002: 1440–1443). Given these characteristics, it is not unexpected that its use was stigmatized at the beginning of the twentieth century (Leonard 1929; Ballard 1939: 23–26; Mittins et al. 1970: 33–35), and educational establishments tried to eradicate the get-passive (and other get-constructions) from their students’ language (e.g. Crumpton 1916). This paper investigates whether this prescriptive stance can be traced back to the nineteenth century – the century known first and foremost for its prescriptivism.

2. Get and its constructions

2.1 The get-passive

The term get-passive is an umbrella term for a number of constructions that contain get followed by a past participle – most usefully, these can be ordered on a scale from clear passives to constructions that only formally resemble passives, but where most syntactic and semantic tests for passives fail (as e.g. in Collins 1996). Thus, in (1) the agent is explicitly mentioned, got can be substituted by was with only little change in meaning, and an active equivalent is possible; (1) is thus an instance of a central get-passive. In (2), it is equally possible to explicate the cause of the experience (e.g. by the task), an active paraphrase is available, but a substitution of got by was changes the dynamic passive into a stative one and suggests a reading of frustrated as a participial adjective, and indeed the participle can be modified; (2) is thus less passive than (1). In (3), we have a typical instance of a reflexive get-passive where the agent cannot be explicated, since agent and patient are identical (roughly equivalent to we dressed ourselves); an active paraphrase is available (we dressed/we dressed ourselves), but not fully equivalent in meaning, and we are thus moving down the passive cline. In (4), drunk is adjectival in quality, there is no active equivalent, the agent cannot be specified in a by-phrase (*the girls got drunk by too much beer), and the subject is not the patient (compare the marginally possible the beer got drunk by the girls, an instance of a central get-passive). (4) by all these criteria can therefore not be called a passive any longer. (5), finally, contains constructions where the former participle has fossilized and become idiomatic, and there is no transparent relationship with a full verbal paradigm any longer; passive status is therefore also marginal.

(1) Central get-passive: He got arrested by the police.
(2) Psychological get-passive: I got frustrated.
(3) a. Reflexive get-passive: We got dressed.
b. Reciprocal get-passive: We got engaged.
(4) Adjectival get + past participle: The girls got drunk.
(5) Formulaic get + past participle a. He got stuck.
b. I will get rid of it.

As this scale already shows, the get-passive is thus not a homogeneous group. As Hundt (2001) argues, the rise of the get-passive is linked to the rise of get in a number of constructions more generally, and get-passives make up only a small part of grammaticalized uses of get. For this reason, we have to cast the net wider and investigate in which constructions get can (or could) be used overall.

2.2 Get-constructions

Get is probably one of the most polysemous verbs in English, and is found in a wide range of constructions, with either nominal or verbal elements following (on the polysemy of get cf. Downing 1996; Gronemeyer 1999). This bifurcation also seems to be related to the two central meanings of this word. With nouns following, get essentially indicates the ‘onset of possession’ (and is thus equivalent to obtain), as in (6a). Related to this by metaphorical extension, at least until the nineteenth century get could still be used in the sense of ‘acquiring knowledge’, as in (6b) – a meaning that survives in I don’t get it until today. From the ‘onset of possession’ meaning, get has lexicalized into constructions indicating ‘stative possession’ in (7), and have got today is equivalent to ‘have’ or ‘possess’. have got in its turn then in the nineteenth century grammaticalized into obligative have got to as in (8), in parallel to the grammaticalization path of have towards have to (Krug 2000), if slightly later.

(6) Get + NP: a. Of him oft ’twas said, That to get his daily bread, He made a vast deal more noise than work. (COHA 1814 FIC)
b. Fanny was quietly seated by the fire in the sitting room, getting her lesson for the next day. (COHA 1854 FIC)
(7) Have got + NP: I’ve only got one witness to prove my innocence, and that’s my wife. (COHA 1896 FIC)
(8) Have got to: if you say A in this matter, you’ve got to say B. (COHA 1853 FIC)

On the other hand, get can also convey a sense of ‘motion’, and this meaning is clearest in its combination with directional adverbs. Thus, in (9) get to is equivalent to arrive at. The sense of ‘motion’ is also inherent in many combinations of get with particles, as in get up, get in, get out, get away, get through, get over, etc. It can also be related to the inchoative sense of get, as in (10), where get is used as an equivalent of become. In addition, get has a causative use, as in (11), which cannot be directly linked with either ‘onset of possession’ or ‘motion’, where a direct object is followed by an infinitive or an adjective. Here, get is used in the meaning of cause, make. This construction (especially with reflexive pronouns and participial adjectives, as in (11c)) has traditionally been cited as the source construction for the get-passive (e.g. by Givón and Yang 1994), no doubt also because of the very similar semantics to the get-passive, but it is now accepted that this path of derivation does not fit the historical facts (Hundt 2001; Fleisher 2006). Instead, it was probably the inchoative use as in (10) that (through a reanalysis of participial adjectives as true participles) served as the source construction for the range of get-passive constructions in (1) to (5) that we can document today (cf. also (12)).

(9) Get to NP/Adv: he came to the wise conclusion that if he got to any settlement at all he would be content. (COHA 1870 FIC)
(10) Get + Adj: if she saw I was getting sad, she would begin some funny story (COHA 1853 FIC)
(11) Get + NP + VP: a. I want to get him to dine with me at Welckley’s (COHA 1880 FIC)
b. Get + NP + Adj: Get the saloon ready, Picard. (COHA 1821 FIC)
c. Get + reflexive pronoun + Adj/ppl: a mother who has placed her infant here and got herself employed as child’s nurse to the hospital (COHA 1838 NF)
(12) Get + past participle: there was Jim, telling a great yarn about the way you rode and how you got flung onto the gate (COHA 1854 FIC)

Although there are frequency shifts in the proportions of constructions over the course of the nineteenth century, and especially verbal constructions with get increase at the expense of more nominal constructions, all of them in fact undergo a massive rise in terms of text frequency, and contribute to the increase in use of get that we can observe in historical corpus materials, as the next section will show.

3. The historical development

3.1 Earlier studies

The most striking rise in frequency of (all) get-constructions has been documented to occur in the nineteenth century, as Figure 1 illustrates. Although get (in the use and senses relevant here originally a loan word from Old Norse, cf. OED: s.v. get, v.) has been present in English from Middle English times onwards, the corpus frequencies seem quite stable for much of the documented period until the second half of the eighteenth century, where we can observe the start of a typical S-curve of language change. The fastest rise seems to happen between 1800 and 1900, before the rise levels out towards the end of the twentieth century. Given this actual rise, it is perhaps not surprising that prescriptive grammarians of the time noticed this increase in frequency and reacted to it; as can perhaps be expected, this reaction was by and large negative.

GET-constructions and GET-passive in ARCHER (based on figures in Hundt 2001)

Figure 1. Get-constructions and get-passive in ARCHER (based on figures in Hundt 2001).

Figure 1 also indicates that the get-passive (defined widely as a form of get followed by a past participle) only accounts for a small percentage of all uses of get, and that it rises later than get overall. As Hundt (2001) suggests, this delay might plausibly be due to nineteenth-century proscriptions against the informal, marked get-passive specifically. However, for a more fine-grained analysis of what happens over the course of the nineteenth century we first of all need more detail in terms of corpus data to substantiate the different developments, and we also need actual evidence of prescriptivism during the nineteenth century to substantiate the claim of successful proscription against the get-passive.

3.2 Corpus data

For an in-depth investigation of get in American English, the huge Corpus of Historical American English COHA is the corpus of choice (Davies 2010–). COHA contains over 400 million words from four large text categories (FICTION, NEWSPAPERS, MAGAZINES, and NON-FICTION) spanning the time from 1810 to 2009. [1] The data is contained in temporal slices of just 10 years and thus allows for a much more detailed temporal analysis of phenomena of change than was possible before. Replicating the searches for get-constructions and the get-passive (defined again as any participle following a form of get) on this new resource, the essential development documented in ARCHER is confirmed, as Figure 2 shows.

GET-constructions and GET-passive in COHA (text frequency per 100,000 words)

Figure 2. Get-constructions and get-passive in COHA (text frequency per 100,000 words).

Parallel to the data in ARCHER, we can observe a striking rise in get overall, which more than triples in text frequency from the beginning to the end of the nineteenth century. As in the data from ARCHER, get continues to rise until the 1940s (in fact, it increases by another 67% from 1900 to the 1940s), before levelling out towards the end of the century. This confirmation means that we can be quite confident that COHA essentially portrays the same development as ARCHER does. Figure 2 also shows that (again, as expected), the two main grammaticalized constructions involving get, i.e. have got and the get-passive, cannot account for the overall rise of get, although both also show significant rises. However, even taken together both their text frequencies make up only a very small part of get-constructions overall.

As already noted, get-constructions (and the get-passive) are highly register-sensitive. An analysis in terms of text types is therefore useful. Figure 3 splits the information contained in Figure 2 into the four text types contained in COHA; the dotted line repeats the overall averages from Figure 2. (Note that newspapers are only included in COHA from the 1860s onwards).

The text-type specific rise of GET in COHA (text frequency per 100,000 words)

Figure 3. The text-type specific rise of get in COHA (text frequency per 100,000 words).

Figure 3 shows clearly that the overall trend in the rise of get is influenced by the fact that all decades in COHA contain the same proportion of 50 per cent data from Fiction, and the shape of the averages curve (the dotted line in Fig. 3) thus essentially follows the figures for Fiction (for more detailed figures, cf. Table 2 in the Appendix). Given that get is found in particular in informal, colloquial speech, it comes as no surprise that Fiction (as the most informal text type of the four included) has by far the highest text frequency of get. What may be unexpected is the indication that this text-type sensitivity seems to have increased, rather than decreased, over much of the twentieth century. Much of the overall increase in the twentieth century is due to a continued increase in the use of get in Fiction, which reaches a high of over 360 occurrences per 100,000 words in the 1940s, followed by a striking increase in the text type of Magazines toward the 1950s. Newspapers, on the other hand, seem to have been rather reluctant users of get until the 1950s. Whereas overall, get levels out at a high level of 250 occurrences per 100,000 words, in Newspapers the use of get has increased only recently, to reach a text frequency of around 220 in the 1990s and 2000s.

If we now zoom in on the get-passive, the scale of Figure 4 already shows that only a small minority of instances of get are actually the get-passive: even today, the figure is only about 7% (get has a text frequency of 250, as we have seen, whereas the get-passive at the end of the twentieth century has a text frequency of about 17 on average; cf. Table 3 in the Appendix for more detailed figures). Although noticeable as a construction, then, the get-passive actually accounts for only a small part of all instances of get.

The rise of the GET-passive per text type in COHA (text frequency per 100,000 words)

Figure 4. The rise of the get-passive per text type in COHA (text frequency per 100,000 words).

In addition, the rise of the get-passive over the nineteenth century is much slower than the rise of get generally; the really dynamic increase in the get-passive does not start until the 1890s, and is thus a truly twentieth-century phenomenon. In fact, there is even a decrease in the nineteenth century from the 1870s towards the 1890s, and the get-passive only accelerates afterwards.

If we concentrate on the first attested verbs in the get-passive, Fleisher’s (Fleisher 2006) observation is confirmed that the get-passive starts out from formulaic constructions, the frequency of which remains relatively stable: especially idioms like get rid of are among the earliest attestations, as in (13). A second large group of semi-fixed expressions are constructions like get married, get acquainted, and get introduced, as in (14), and true get-passives only become more frequent than these formulaic instances after the 1850s. The most striking rise in central get-passives (those that are truly equivalent to be-passives) and its free combination with a large range of lexical verbs is thus also a late nineteenth century/early twentieth century development, as the examples in (15) illustrate.

(13) a. Bravo! my young gentleman! I know now a way of getting rid of you, when tired of your intolerable nonsense. (COHA 1814 FICTION)
b. I have seen something so like war to-day, that I can not get rid of the gloom it has thrown over my spirits. (COHA 1826 NON-FICTION)
(14) a. Alice was pale, dejected and miserable; her husband had got acquainted with a set of worthless beings who called themselves honest fellows (COHA 1828 FICTION)
b. She had agreed to run off with him to North-Carolina, and get married (COHA 1827 FICTION)
(15) a. I am used to all these ins and outs, and know about what I can do, even if I never happened to get caught just here before. (COHA 1892 FICTION)
b. The Spanish Archbishop of Manila is another gentleman who has made a fool of himself, but he is in the way of getting cured (COHA 1898 NEWS)
c. But whichever way he votes, his vote gets counted, and his will, whether it be feeble or sturdy, gets expressed. (COHA 1890 MAGAZINES)
d. Sir Robert Cary posted away, unsent, to King James of Scotland to inform him of the “accident,” and got made a baron of the realm for his ride. (COHA 1897 FICTION)
e. True, the trainmen get paid for their Sunday work, because they are paid by the trip or day. (COHA 1890 MAGAZINES)
f. If you seize him and chain him, he will answer your questions in order to get released. (COHA 1898 NON-FICTION)
g. [He] said that he had come out for a day’s fishing, had got separated from his companions, lost his way and was hungry and worn out. (COHA 1896 FICTION)
h. Yet existing habits, both in the United States and in England, make it probable that the majority [sc. of school leavers] will content themselves with getting taught in the evening after their day’s work is done. (COHA 1899 MAGAZINES)

Considering the delay (compared to get-constructions as a whole) and the prescriptions against get attested since the early twentieth century it is therefore a valid question to ask about the prescriptive influence in the nineteenth century on these developments.

4. Data on prescriptions

4.1 The material

In order to base an assessment of the (potential) influence of prescriptive grammar writing on a sounder footing than mere anecdotal evidence, I have been collecting nineteenth-century grammar books over the past few years. The Collection of Nineteenth-century Grammars (CNG) now contains 258 grammars published in Britain and North America between 1800 and 1900 (in their first edition). The books were collected from Google Books between 2007 and 2011. My collection contains only full texts that were then manually analysed and annotated. In this way, the corollary databases now allow me to collect all comments on a linguistic phenomenon and compare them across authors, across regions, compare them to other phenomena, and trace developments over time.

4.2 Descriptively correct assessments of the get-passive

Overall, as for many other phenomena (Anderwald 2016), British grammars are less critical (less prescriptive) than American ones of get, and from the 1850s onwards, some descriptively correct comments can be found. Thus, the anonymous English Grammar (1853) notes that “get is now in good use as an auxiliary in passive forms – To get dressed” (Anonymous 1853: 27); James Gow 1892[1878] recommends using get in particular in the place of uninflected be with the progressive: “The expression ‘be being’ is practically never used. In forming passives we generally say ‘be getting’, as ‘If I be getting tired’, etc.” (Gow 1892: 106; although it has to be noted here that get tired is not an instance of a passive, but is clearly adjectival). [2] Frederick White in 1882 even has recourse to Greek and Latin to describe the semantic difference between the get-passive and the be-passive:

Get […] is also an animated sort of copula, differing from be much as gignomai from eimi, and fio from sum. Thus, “The boy was whipt,” gives the dry fact merely; “The boy got whipt,” is a sort of picture […] Thus get is of the same class as prove, become, turn out, grow; but is the nearest of all them to a simple copula, in that it is used with the perfect participle. (White 1882: 168)

Also in American grammars, some authors note the (relatively new) function of get. Thus, W. Dwight Whitney in 1877 states: “Phrases of nearly the same meaning with the ordinary passive ones are made also with the verbs become and get: thus, he became frightened; he has got beaten; but it is not usual to reckon them as passive.” (Whitney 1877: 129)

Towards the end of the century, Samuel Ramsay similarly notes that “More subtle and liable to escape notice is the formation of a kind of middle voice by using get as an auxiliary: ’I got up,’ ‘He got tired,’ ‘They got married,’ ‘He got elected’.” (Ramsey 1892: 475), curiously pre-empting the present-day generative analysis of the get-passive as, indeed, a kind of middle voice. In sum, however, a linguistically adequate description of the get-passive is a minority stance: out of 42 grammars which mention get, only 6 are, if not positive, at least neutral – this is a proportion of only 14%, or one grammar in seven. Nevertheless, it is an important insight that, rare as they might be, comments on the get-passive are generally not negative; the strong proscriptive stance against the get-passive that is reported for the twentieth century can thus not be traced back to the nineteenth century. The picture is quite different if we look at comments on get more generally, as the next section will show.

4.3 Proscription against get-constructions

In contrast to the (still rare) construction of the get-passive, get more generally is usually assessed negatively, as Figures 5 and 6 illustrate. [3]

Mentioning and criticizing GET-constructions in British grammars

Figure 5. Mentioning and criticizing get-constructions in British grammars.

Mentioning and criticizing GET-constructions in American grammars

Figure 6. Mentioning and criticizing get-constructions in American grammars.

Especially in British grammars, a critical stance towards get in its many constructions is an undercurrent of nineteenth-century grammar writing, at least up to the middle of the century, as Figure 5 shows. In fact, it continues an eighteenth-century strand of grammar writing, not only in the stance towards get, but also in the examples that are regularly cited (for get and for some other examples, cf. Anderwald 2015). What is criticized in particular are idioms like get rid of or get into a scrape that are designated “low and provincial” or “vulgar”, and the lexicalized have got, which is said to be “improper”, “ungrammatical”, “superfluous”, “unnecessary” or a “solecism” throughout the century. Occasionally, an “overuse” of get is deplored more generally (e.g. by Booth 1837: 236; Beard 1854: 180). However, overall it has to be said that none of the uses of get escapes criticism; thus, even the inchoative use (get wet) is occasionally said to be “not authorized by grammar” (Kelke 1885: 112).

The proscriptive stance of American grammars, by contrast, is stronger; as noted above, negative comments on get are also found almost twice as frequently as in British grammars. The range of constructions that are cited and the actual examples, however, are strikingly similar to the British examples, and clearly continue a tradition of grammar writing (that includes usage advice) that is common to both Britain and North America. Again, the construction criticized the most is lexicalized have got. In addition, the further grammaticalized obligational have got to comes into the focus of criticism especially in American grammar writing. In fact, have got to is first mentioned in the 1820s, actually predating anecdotal evidence (Krug 1998: 61–62) and the OED (OED: s.v. get v. I.24.): according to Joseph Hull’s appendix of (American) ‘Vulgarisms’, “Have got to go for I must go, is a common errour” (Hull 1828 [1827]: Appendix p.5). [4] For the other constructions, the close link between British and American examples is at least partly explained by (repeated and successive) plagiarism. Thus, the American grammarian Peter Bullions (1845 [1834]: 184) copies passages on the proscription of get directly from his British colleague William Allan from twenty years before (Allen 1824 [1813]), and Bullions himself is then copied by his compatriot Simon Kerl some thirty years later (Kerl 1868 [1861]). In this manner, in particular the specific assessment of get into a scrape as “low and provincial” for Britain in the 1810s is revived, generalized, and becomes part of the grammatical canon of American grammar writing, notwithstanding the fact that the assessment of provincial must have been curious for American readers.

As in Britain, practically none of the uses of get escapes criticism, thus we find proscription of full verb uses of get, especially in the ‘motion’ sense (thus Kirkham corrects I got there to I arrived in his section on ‘Provincialisms’, Kirkham 1834 [1824]: 206), or of causative get (e.g. John Goldsbury cites she has got her sails furled in his examples of ‘False Syntax’, Goldsbury 1842: 53), but also criticism of phrasal verbs with get: for example Brantley York (York 1862 [1854]: 160) corrects I got up to I rose, and like him, many other authors preferred Latinate or Romance-based (monomorphemic) verbs to Germanic phrasal verbs with get. [5] Inchoative get is also criticized; thus William Balch in 1839 claims that get ready employs “unnecessary words […] which exceedingly injure the style of composition” (Balch 1841 [1839]: 159) – although the reader is left to wonder what an alternative construction could possibly be, and which of the two words (get or ready?) is deemed to be unnecessary here. However, it is the grammaticalized constructions have got and have got to that continue to be criticized until the end of the century, and thus bear the brunt of the proscription against get. [6]

5. The actual effect of proscription

For American English, we are now in the fortunate position to be able to compare our detailed corpus-linguistic data with  data of the proscription against get that is almost as detailed over the course of the nineteenth century. Figure 7 repeats the text-type specific rise of get in COHA for the nineteenth century, and adds (on a separate scale) the percentage of grammars in the CNG (per decade) that are critical of get. For a more realistic picture, these figures should probably be higher, because we saw that most grammars did not even mention get. This means for readers that if they did encounter a comment on get, this was almost always negative. Calculating negative comments on get in relation to all grammars in the CNG thus presumably underestimates their prescriptive impact, since negative evidence (not encountering get in a grammar) would not have served to counteract the proscription against get; Figure 7 is thus a rather conservative estimate.

Did proscription against GET have any effect on the rise of GET in COHA?

Figure 7. Did proscription against get have any effect on the rise of get in COHA?

As Figure 7 clearly shows, there were three distinct peaks in the proscription against get in the CNG. Criticism was strongest in the 1830s, where half of all grammars contained negative comments on get, but it also flared up again in the 1860s and the 1880s, where just under a third of grammars contained critical remarks. If we compare this to the overall rise of get (in the dotted line), we can see that indeed ten years later in each case, the use of get declines, or the increase falls off markedly. This is due mainly to the shape of the curve for Fiction – not surprisingly, since this text type makes up around 50 per cent of the corpus in each decade, and thus dominates the total figures. In Fiction, there is a distinct slump in the rise of get from the 1830s to the 1840s, another visible slow-down from the 1860s to the 1870s, and a noticeable decline from the 1880s to the 1890s (which, as Figure 3 has shown, is then more than made up by the subsequent rise of get over the first half of the twentieth century). Less conclusively, another case can perhaps be made for a slow-down of the rise of get in Non-Fiction texts from the 1830s to the 1840s, where the curve again visibly flattens.

By contrast, the proscriptive impact on get-passive is not as clear, as Figure 8 details.

Did proscription against GET have any effect on the rise of the GET-passive in COHA?

Figure 8. Did proscription against get have any effect on the rise of the get-passive in COHA?

This is partly due to the fact that the get-passive has a much lower text frequency overall, and with absolute occurrences under 100 (per decade) frequency curves typically become more volatile, and subject to random fluctuations (or rather, individual texts get a disproportionate influence on the averages). Absolute frequencies of the get-passive are below 100 for Fiction up to (and including) the 1820s, for Magazines up to the 1860, and for Newspapers and Non-Fiction texts for all of the nineteenth century. (This is indicated in Figure 8 by thinner lines; cf. Table 3 in the Appendix for more details.) My analysis of the effect of proscription therefore concentrates on the most robust figures, those from Fiction. With all due caution, there seems to be a corresponding dip from the 1830s to the 1840s in the rise of the get-passive in Fiction, and thus a slow-down in the average figures, after massive (general) proscription against get in the 1830. There is another decline of the get-passive clearly visible in Fiction texts from the 1850s to 1860s (again reflected as a slow-down in the averages) that cannot be related to proscription. As in Figure 7 for get, we observe an overall decline towards the end of the nineteenth century, before the get-passive then really takes off in the twentieth century, and starts its trajectory to notoriety. This cautious interpretation suggests at most a small, temporary influence of the general proscription against get on the rise of the get-passive specifically. In the wake of general proscription against get, all get-constructions (including the passive) seem to be avoided temporarily. This proscriptive effect does not seem to be longer-lasting, however, and it is comparatively small in comparison to the overall dynamic of the rise of get and the get-passive.

6. Conclusion

My study has shown that it is now possible to correlate detailed corpus data with similarly detailed data on prescriptive attitudes by nineteenth-century grammar writers. Because of the wide variety of sources (and stances in those sources) it is indispensable to investigate a large number of grammars, rather than pick out anecdotally the one that fits one’s preconceptions. Only in this way can we judge which pronouncements were ‘mainstream’ in the community of grammar writers, which linguistic features were commented on (comparatively) much or little, and whether proscriptions were particularly harsh, or mild, for the times. In this way, this paper has shown that there is no evidence in the data from the CNG that proscription against the get-passive was in any way stronger, more pronounced or more protracted than against get more generally. On the contrary, we have seen that the get-passive was not specifically singled out for criticism. In fact, it attracted less proscription than get overall. However, as Chapman has pointed out (Chapman 2012), prescription generally seems to act like a “bulldozer” (razing everything in its path, whether the intended object or not). In this sense, we saw that proscription against get (rare as it was overall) may also have had a (small) corollary effect on the get-passive. However, it is no longer tenable to claim that the rise of the get-passive was delayed (vis-à-vis the rise of other get-constructions) because of specific nineteenth-century proscriptions against it. Even the more general proscription against get could be shown to be effective only briefly (typically for a time-span of one decade), on a small scale (halting or slowing down the dynamic increase for a time), and temporarily (the subsequent rises in all cases more than compensate the preceding slow-down). Overall, however, instead of generalizing from anecdotal evidence, or speculation instead of evidence, it is clear that we need many more investigations of this type to evaluate more generally the effectiveness of prescriptive grammar writing. For get, the effect seems to have been noticeable, if small. It remains to be seen if detailed correlations of corpus data and comments in grammars (or other prescriptive sources) can corroborate this picture.


[1] Although it has to be said that the distribution of these text types is skewed. In all decades, Fiction texts make up ca. 50% of the material, Newspapers are only included from the 1860s onwards, and earlier decades contain less material than later ones (cf. Table 1 in the Appendix for details). Despite this skewing, COHA is still one of the biggest diachronic corpora that are freely available, and a valuable resource for this reason. [Go back up]

[2] One of Gow’s motives may have been avoiding the progressive passive, especially in non-finite constructions. On the much hated progressive passive, cf. Anderwald (2014). [Go back up]

[3] Although it also has to be said that overall, comments on get are rather infrequent: only 15 British grammars (or 11% of the CNG) mention get, and only 27 American ones do (22%). Although this is twice as many as the British grammars, it still means that almost 4 in 5 American grammars do not comment on get. Compared to the many other phenomena I looked at (Anderwald 2016: chapter 8), this makes get the linguistic feature with the lowest frequency of comments. [Go back up]

[4] It is interesting that this (quite short) list of ‘vulgarisms’ observed by Hull himself is expanded (by him) by items taken from John Pickering’s Americanisms (Pickering 1816); Hull mentions that Pickering’s “treatise” was out of print at the time (Hull 1828[1827]: Appendix p.3). However, Pickering does not mention either get, have got or have got to in his 206 pages. [Go back up]

[5] I noted above that full verb get today has two distinct main senses (‘obtain’ and ‘motion’). Increasingly, only the first of these is allowed by prescriptive grammarians over the course of the nineteenth century, which would explain proscription of the ‘motion’ sense. This tendency is in accordance with the widespread belief that one word can only have one ‘true’ meaning (what Arnold Zwicky has called the principle of ‘One Right Way’, cf. Zwicky 2009), and with the Myth of the Original Meaning (Watts 2011). Historically, this is not tenable (cf. OED: s.v. get v. again), but historical facts do not seem to enter into this debate much. [Go back up]

[6] The rarity of comments on the get-passive cannot have been due to its text frequency alone; the progressive passive is comparable in terms of text frequency, but one of the most hated (and most frequently commented on) constructions of nineteenth-century grammar writing (cf. Anderwald 2014). [Go back up]


COHA = The Corpus of Historical American English: 400 million words, 1810–2009. 2010–. Compiled by Mark Davies. Available at http://corpus.byu.edu/coha/


Alexiadou, Artemis. 2005. “A note on non-canonical passives: The case of the get-passive”. Organizing Grammar: Studies in Honor of Henk van Riemsdijk, ed. by Hans Broekhuis, Norbert Corver, Riny Huybregts, Ursula Kleinhenz & Jan Koster, 13–21. Berlin & New York: Mouton de Gruyter.

Alexiadou, Artemis. 2012. “Noncanonical passives revisited: Parameters of nonactive Voice”. Linguistics 50: 1079–1110. doi:10.1515/ling-2012-0036

Alexiadou, Artemis & Florian Schäfer, eds. 2013. Non-Canonical Passives. Amsterdam & Philadelphia: John Benjamins.

Allen, William. 1824. An English Grammar; with Exercises, Notes, and Questions. Third edition, revised and improved edition. London: G. and W.B. Whittaker. [First published 1813].

Anderwald, Lieselotte. 2014. “‘Pained the eye and stunned the ear’: Language ideology and the progressive passive in the nineteenth century”. Contact, Variation, and Change in the History of English, ed. by Simone Pfenninger et al., 113–136. Amsterdam & Philadelphia: John Benjamins.

Anderwald, Lieselotte. 2015. “The 19th-century perspective on 18th-century grammar writing”. Anglistentag 2014 Hannover, ed. by Jana Gohrisch & Rainer Emig, 35–47. Trier: Wissenschaftlicher Verlag.

Anderwald, Lieselotte. 2016. Language between Description and Prescription: Verb Categories in Nineteenth-Century Grammars of English. Oxford: Oxford University Press.

Anonymous. 1853. English Grammar and Composition. London & Edinburgh: William and Robert Chambers.

Balch, William S. 1841. A Grammar of the English Language; Explained According to the Principles of Truth and Common Sense, and Adapted to the Capacities of All Who Think. Designed for the Use of Schools, Academies, and Private Learners. Fourth edition. Boston: B.B. Mussey. [First published 1839].

Ballard, Philip Boswood. 1939. Teaching and Testing English. London: University of London Press.

Beard, John R. 1854. Cassell’s Lessons in English; Containing a Practical Grammar, Adapted for the Use of the Self-Educating Student. London: John Cassell.

Booth, David. 1837. The Principles of English Grammar. London: Charles Knight and Co.

Bullions, Peter. 1845. The Principles of English Grammar; Comprising the Substance of the Most Approved English Grammars Extant. With Copious Exercises in Parsing and Syntax, for the Use of Academies and Common Schools. A New Edition, Revised and Corrected; with an Appendix of Various and Useful Matter. Thirteenth edition. New York: Pratt, Woodford & Co. [First published 1834].

Chapman, Don. 2012. “Enforcing or effacing useful distinctions?: Infer vs. imply”. Paper presented at ICEHL 17, August 2012, Zurich, Switzerland.

Collins, Peter. 1996. “Get-passives in English”. World Englishes 15: 43–56. doi:10.1111/j.1467-971X.1996.tb00091.x

Crumpton, Claudia E. 1916. “Better Speech Week at Montevallo”. The English Journal 5: 569–570. doi:10.2307/800989

Denison, David. 1993. English Historical Syntax: Verbal Constructions. London & New York: Longman.

Denison, David. 1998. “Syntax”. The Cambridge History of the English Language. Vol. IV: 1776–1997, ed. by Suzanne Romaine, 92–329. Cambridge: Cambridge University Press.

Downing, Angela. 1996. “The semantics of get-passives”. Functional Descriptions: Theory in Practice, ed. by Ruqaiya Hasan, Carmel Cloran & David Butt, 179–205. Amsterdam & Philadelphia: John Benjamins.

Fleisher, Nicholas. 2006. “The origin of passive get”. English Language and Linguistics 10: 225–252. doi:10.1017/S1360674306001912

Givón, Talmy & Lynne Yang. 1994. “The rise of the English get-passive”. Voice: Form and Function, ed. by Barbara Fox & Paul J. Hopper, 119–149. Amsterdam & Philadelphia: John Benjamins.

Goldsbury, John. 1842. A Sequel to the Common School Grammar; Containing, in Addition to Other Materials and Illustrations, Notes and Critical Remarks on the Philosophy of the English Language; and Explaining Some of its Most Difficult Idiomatic Phrases. Designed for the Use of the First Class in Common Schools. Boston: James Munroe & Company; New York: Collins, Brother & Co.; Philadelphia: Thomas, Cowperthwait & Co.; and Baltimore: Cushing & Brothers.

Gow, James. 1892. A Method of English for Secondary Schools. Part I: Grammar Chiefly. London & New York: Macmillan and Co. [First published 1878].

Gries, Stefan Th. & Martin Hilpert. 2012. “Variability-based Neighbor Clustering: A bottom-up approach to periodization in historical linguistics”. The Oxford Handbook of the History of English, ed. by Terttu Nevalainen & Elizabeth Closs Traugott, 134–144. Oxford: Oxford University Press.

Gronemeyer, Claire. 1999. “On deriving complex polysemy: The grammaticalization of get”. English Language and Linguistics 3: 1–39. https://www.cambridge.org/core/journals/english-language-and-linguistics/issue/5172D3936B5024BB896959C1EB9AC0DD

Huddleston, Rodney D. & Geoffrey K. Pullum. 2002. The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press.

Hull, Joseph Hervey. 1828. English Grammar by Lectures; Comprehending the Principles and Rules of Syntactical Parsing, on a New and Highly Improved System; Intended as a Text Book for Students; Containing Exercises in Syntax, Rules for Parsing by Transposition, Critical Notes, and a Lecture on Rhetoric. Fourth edition. Boston: Printed by Lincoln & Edmands, for the author. [First published 1827].

Hundt, Marianne. 2001. “What corpora can tell us about the grammaticalisation of voice in get-constructions”. Studies in Language 25: 49–88. doi:10.1075/sl.25.1.03hun

Kelke, William Henry Hastings. 1885. An Epitome of English Grammar for the Use of Students. Adapted to the London Matriculation Course and Similar Examinations. London: Kegan Paul, Trench & Co.

Kerl, Simon. 1868. A Comprehensive Grammar of the English Language. For the Use of Schools. New York: Ivison, Phinney, Blakeman & Co., and Chicago: S.C. Griggs & Co. [First published 1861].

Kirkham, Samuel 1834. English Grammar in Familiar Lectures, Accompanied by a Compendium; Embracing a New Systematic Order of Parsing, a New System of Punctuation, Exercises in False Syntax, and a System of Philosophical Grammar in Notes: to which are Added an Appendix, and a Key to the Exercises. Designed for the Use of Schools and Private Learners. Thirty-Sixth edition, enlarged and improved edition. Rochester, NY: Marshall & Dean. [First published 1824].

Krug, Manfred. 1998. “String frequency: A cognitive motivating factor in coalescence, language processing and linguistic change”. Journal of English Linguistics 26: 286–320. doi:10.1177/007542429802600402

Krug, Manfred G. 2000. Emerging English Modals: A Corpus-based Study of Grammaticalization. Berlin & New York: Mouton de Gruyter.

Leonard, Sterling Andrus. 1929. The Doctrine of Correctness in English Usage 1700–1800. Madison, WI: University of Wisconsin. [Reissued 1962].

Mittins, William Henry, Mary Salu, Mary Edminson & Sheila Coyne. 1970. Attitudes to English Usage. London: Oxford University Press.

OED. 2011–. Oxford English Dictionary online. Oxford: Oxford University Press. http://www.oed.com.

Pickering, John. 1816. A Vocabulary, or Collection of Words and Phrases which have been Supposed to be Peculiar to the United States of America. To which is Prefixed an Essay on the Present State of the English Language in the United States. Originally Published in the Memoirs of the American Academy of Arts and Sciences; and now Republished with Corrections and Additions. Boston, MA: Cummings & Hilliard.

Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A Comprehensive Grammar of the English Language. Harlow: Longman.

Ramsey, Samuel. 1892. The English Language and English Grammar. An Historical Study of the Sources, Development, and Analogies of the Language and of the Principles Governing its Usages. Illustrated by Copious Examples from Writers of all Periods. New York & London: G.P. Putnam’s Sons.

Watts, Richard. 2011. Language Myths and the History of English. Oxford etc.: Oxford University Press.

White, Frederick Averne. 1882. English Grammar. London: Kegan Paul, Trench & Co.

Whitney, William Dwight. 1877. Essentials of English Grammar. For the Use of Schools. Boston: Ginn and Heath.

York, Brantley. 1862. An Analytical, Illustrative, and Constructive Grammar of the English Language. Accompanied by Several Original Diagrams, Exhibiting an Occular Illustration of some of the most Difficult Principles of the Science of Language; Also, an Extensive Glossary of the Derivation of the Principal Scientific Terms Used in this Work, in Two Parts, for the Use of Every One who may Wish to Adopt it. Third edition. Raleigh: W.L. Pomercy. [First published 1854].

Zwicky, Arnold. 2009. “One Right Way”. http://arnoldzwicky.wordpress.com/2009/06/28/one-right-way/.


No. words Fiction No. words Magazines No. words Newspapers No. words Non-Fiction
1810 641,164 88,316 451,542
1820 3,751,204 1,714,789 1,461,012
1830 7,590,350 3,145,575 3,038,062
1840 8,850,886 3,554,534 3,641,434
1850 9,094,346 4,220,558 3,178,922
1860 9,450,562 4,437,941 262,198 2,974,401
1870 10,291,968 4,452,192 1,030,560 2,835,440
1880 11,215,065 4,481,568 1,355,456 3,820,766
1890 11,212,219 4,679,486 1,383,948 3,907,730
1900 12,029,439 5,062,650 1,433,576 4,015,567
1910 11,935,701 5,694,710 1,489,942 3,534,899
1920 12,539,681 5,841,678 3,552,699 3,698,353
1930 11,876,996 5,910,095 3,545,527 3,080,629
1940 11,946,743 5,644,216 3,497,509 3,056,010
1950 11,986,437 5,796,823 3,522,545 3,092,375
1960 11,578,880 5,803,276 3,404,244 3,141,582
1970 11,626,911 5,755,537 3,383,924 3,002,933
1980 12,152,603 5,804,320 4,113,254 3,108,775
1990 13,272,162 7,440,305 4,060,570 3,104,303
2000 14,590,078 7,678,830 4,088,704 3,121,839

Table 1. No. of words per text type in COHA (based on http://corpus.byu.edu/coha/ > help/see texts and registers).

M-co get Fiction M-co get Magazines M-co get Newspapers M-co get Non-Fiction average M-co get
1810 57 43 6 36
1820 43 13 24 33
1830 80 18 36 61
1840 82 21 37 58
1850 113 23 47 79
1860 137 42 70 40 94
1870 145 59 69 72 109
1880 163 26 71 50 114
1890 154 67 96 43 111
1900 208 64 125 48 142
1910 290 84 111 92 196
1920 282 107 102 53 184
1930 353 161 76 47 228
1940 366 196 94 58 248
1950 352 201 89 74 243
1960 350 181 103 60 237
1970 351 168 121 75 239
1980 317 159 157 69 224
1990 344 187 216 74 254
2000 313 220 224 110 255

Table 2. The rise of get in COHA (text frequency of occurrences per 100,000 words, rounded).

occ. in Fiction occ. in Magazines occ. in Newspapers occ. in Non-Fiction
1810 13 1 4
1820 78 9 11
1830 266 34 68
1840 257 55 62
1850 425 51 85
1860 430 126 4 62
1870 629 128 34 87
1880 658 107 59 105
1890 626 139 38 63
1900 940 134 38 90
1910 1173 186 29 128
1920 1109 244 120 90
1930 1567 350 66 66
1940 1826 424 80 84
1950 1926 497 91 96
1960 1901 568 133 101
1970 2134 536 186 146
1980 1921 581 370 135
1990 2584 1027 614 138
2000 2751 1275 701 263

Table 3. The rise of the get-passive per text types in COHA, absolute numbers.

University of Helsinki