Fear and Loathing of the English Passive

Geoffrey K. Pullum

School of Philosophy, Psychology and Language Sciences, University of Edinburgh
Dugald Stewart Building, 3 Charles Street, Edinburgh EH8 9AD


This is a browsable version of the text of an article that appeared in Language and Communication 102 (vol. 26, no. 2, June 2010), 34-44.
Please refer to the printed version when citing or quoting.

Abstract
Writing advisers have been condemning the English passive since the early 20th century. I provide an informal but comprehensive syntactic description of passive clauses in English, and then exhibit numerous published examples of incompetent criticism in which critics reveal that they cannot tell passives from actives. Some seem to confuse the grammatical concept with a rhetorical one involving inadequate attribution of agency or responsibility, but not all examples are thus explained. The specific stylistic charges leveled against the passive are entirely baseless. The evidence demonstrates an extraordinary level of grammatical ignorance among educated English language critics.

1  Introduction

The references to passive constructions in the vast body of work on English grammar, usage, style, and writing are unremittingly negative. Passives, we are told firmly, over and over again, should be shunned. A diverse assortment of unpleasant maladies will afflict your work, it is claimed, if you use passives: your writing will become weak, dull, vague, cowardly, bureaucratic, and dishonest.

Issuing such warnings is very much a 20th-century practice. As Brock Haussamen points out (1997: 54), 19th-century writers on grammar and usage explained the structure and function of passives without any negative spin. But early in the 20th century we start to find minatory statements about it, over and over again:

(1) a. "Do not use the passive voice when such use makes a statement clumsy and wordy... Do not, by using the passive voice, leave the agent of the verb vaguely indicated, when the agent should be clearly identified." [Edwin Woolley, Handbook of Composition, 1907, p. 20]
  b. "Use the active voice... The active voice is usually more direct and vigorous than the passive..." [William Strunk, The Elements of Style, 1918]
  c. "As a rule, avoid the passive voice" [Foerster and Steadman, Writing and Thinking, 1931]
  d. "Never use the passive where you can use the active" [George Orwell,`Politics and the English language', 1946]

More recently, writing pundits emboldened by Orwell's charges about official evasiveness issue their warnings in wilder terms. Sheridan Baker appaently sees the passive as a kind of lethal drug fed to us by scientists and bureaucrats:

The passive voice liquidates and buries the active individual, along with most of the awful truth. Our massed, scientific, and bureaucratic society is so addicted to it that you must constantly alert yourself against its drowsy, impersonal pomp. [Baker 1985, p. 121, quoted by Haussamen 1997: 53]
And Sherry Roberts, in materials for her seminars on business writing, allows a strangely ill-chosen metaphor to run away with her completely:
A sentence written in passive voice is the shifty desperado who tries to win the gunfight by shooting the sheriff in the back, stealing his horse, and sneaking out of town. [Sherry Roberts, 11 Ways to Improve Your Writing and Your Business, on the Web here.]

The writing gurus who say such things, while insistent that novice writers should avoid passives, pay very little attention to defining or describing what it is that should be avoided. We shall see in section 3 that there is rampant confusion about what `passive' means linguistically. But first, in section 2, I will provide a brief and informal descriptive review of the relevant grammatical facts, because the available grammars for educational use and popular consumption come nowhere close to being either accurate or comprehensive.

2  Describing English passive clauses

The descriptive failure of English grammars to describe the passive adequately is partly due to a centuries-old tradition of talking about English grammar as if English were typologically similar to Latin, the international language of higher learning at the time when grammatical study of English began. Latin had morphologically distinguishable infinitives, gerunds, gerundives, present participles, subjunctives, future tense forms, and passives, all with their particular endings on the verb. English has none of this, but is traditionally described as if it did.

It is not verbs that exhibit passive voice in English, but larger units. The underlined verb in a passive clause like She has been questioned by the detectives is exactly the same as the one underlined in the active clause The detectives have questioned her. Passive clauses are marked by the use of devices (participial inflections and accompanying verbs) that have other uses too. To construct an adequately general description of passive constructions we need to focus not on the verb but on larger units, specifically the verb phrase (VP) and the clause.

2.1 Passives with be

There are at least two obvious ways of making a direct claim about the universal admiration for Anne's scholarship within the department: the active transitive in (2a) and the passive counterpart in (2b).

(2) a. Everyone in the department admires Anne's scholarship.
  b. Anne's scholarship is admired by everyone in the department.

The sentence in (2b) illustrates the most familiar and neutral kind of passive clause. The main clause verb is a form of the verb be, and it has a past-participial VP complement. The direct object noun phrase (NP) of (2a) appears as the subject of (2b), and the subject of (2a) appears within the VP of (2b) as the complement in a preposition phrase (PP).

Thus in a sense two NPs exchange syntactic positions: the subject of (2a) appears as an internal complement within the VP in the (2b), and vice versa. Yet the core meanings are the same, in the sense that if (2a) is true, then so is (2b), and conversely.

(The semantic equivalence claim must be carefully hedged, of course: as linguists have long stressed, when negation, quantifiers, or generics enter the picture, actives may not be synonymous with their passive counterparts. Many Americans did not support Obama is true, but Obama was not supported by many Americans is not. Beavers build dams is a true generalization, whereas Dams are built by beavers is not.)

The usage literature talks about the roles in semantic terms that seem quite peculiar, frequently claiming that ``the subject receives the action expressed by the verb'' in a passive. This strange locution ``receives the action'' is found solely in the thousands of mutually plagiarizing bad descriptions of the passive construction. The whole notion of ``action'' is irrelevant here anyway: no action is expressed or implied by passive clauses like the ones in (3).

(3) a. Not much is known by biologists about the coelocanth.
  b. It was alleged by the committee that the dean was incompetent.
  c. That capitalism works is no longer doubted by anyone.

In none of these does the subject denote anything that could conceivably be said to ``receive'' or be targeted by an action: in (3b) it is a dummy pronoun and in (3c) it is not even an NP, but rather a clause.

2.1.1 Short be-passives

The by-phrase in a passive (which CGEL calls an internalized complement, because it is internal to the VP but would have been external to it if expressed as the subject of an active clause) is usually omissible, so some passives contain no a counterpart of the NP that would have been the subject in the related active. When a passive clause has a by-phrase, as in (4a), I will call it a long passive. The kind that does not, as in (4b), will be called a short passive.

(4) a. The president's authority has been much diminished by recent events in Washington.
  b. The president's authority has been much diminished.

A short passive does not have an exact active counterpart; (4b) entails that some unspecified person or circumstance is responsible for the diminution of the president's authority, but says nothing about who or what it might be.

2.1.2 Other kinds of passive clause

The variety of passive constructions beyond the simple be passive is hardly touched on in pedagogical grammars or books on usage and style, but I think it is nonetheless worth surveying the full range of passive constructions here.

2.2 Prepositional passives

Passive clauses frequently have a transitive verb that lacks its direct or indirect object (for example, admired in (2b), or diminished in (4), or given in He was given a medal); but there is another possibility: the NP complement in a PP complement of the verb may be missing. Thus, for example, (5a) has the passive counterpart (5b), where the (normally obligatory) complement of at is missing.

(5) a. His friends laughed at him.
  b. He was laughed at by his friends.
  c. He was laughed at.

Such prepositional passives may be long, like (5b), or short, like (5c).

2.2.2 Bare passive clauses

It is possible for a passive clause to contain just a subject and the past participle of a verb. CGEL calls such clauses bare passive clauses. Having no tensed verb, they cannot generally occur as full sentences, but they do occur as adjuncts, as seen in the underlined parts of the Wall Street Journal corpus examples in (6).

(6) a. That said, however, Korea is Korea, not the Philippines.
  b. One of its ads shows a washed-out manager, arms folded, sitting in a corner.

One context in which bare passive clauses occur on their own is newspaper headlines:

(7) a. France accused of running vast data surveillance scheme to keep tabs on population [The Independent, 5 July 2013]
  b. 28 injured in accidental detonation [Los Angeles Times, 5 July 2013]
  c. Chinese swimmers driven away by smelly green algae [BBC News website, 5 July 2013]

And occasionally we find what might be analysed as a bare passive used as an independent clause:

(8) A cybersecurity expert duped by an email scam! Amazing.

2.2.3 Embedded passives

Passive clauses also occur embedded in active clauses. Various transitive verbs with causative, inchoative, or perception meanings take subjectless bare passive complements (that is, in effect, past-participial VPs). In (9a) we see an active complement clause (investigate the case) but the roughly synonymous (9b) has a passive counterpart (underlined).

(9) a. The government had the police investigate the case.
  b. The government had the case investigated by the police.

These are roughly synonymous in the sense that if one is true, so is the other, though they differ in focus or viewpoint: (9a) strikes us as a statement about the instructions that the government gave to the police, while (9b) seems more like a statement about the government's action on the case.

Further examples of this sort, slightly adapted from attested ones, are given in (10), with the passive VP complements underlined.

(10) a. Western Canada has found itself treated by Ottawa as a colony.
  b. I had the suit made by my tailor in Rome.
  c. David Swan, a geneticist, got himself transferred within the CDC.
  d. Manufacturers saw themselves pushed to the brink of going out of business.

Subjectless passives with be can also occur as complements in active clauses. Under the now widely accepted assumption that the auxiliaries of English are complement-taking verbs, sentences like those in (11) illustrate this possibility.

(11) a. Marie has been photographed by a journalist.
  b. Marie will be photographed by a journalist.
  c. Marie could have been photographed by a journalist.

2.2.4 Adjectival passives

The term `adjectival passive' is often applied (perhaps not very felicitously) to active clauses with predicative adjective phrases in which adjective derives from the past participle of a verb and has a passive-like meaning. There is frequently an ambiguity between be passives and adjectival ones. For example, The door was locked is ambiguous: as a be passive it says that at a particular time someone took the action of locking the door, and as an adjectival passive it says that during some past time period the door was in its locked state. Since the complement in this kind of clause is an adjective phrase, verbs other than be can be used (The door seemed locked, as far as I could tell), and so can adjectives derived with the negative prefix un- (The island was uninhabited by humans).

2.2.5 Get passives

The intransitive verb get, which is not an auxiliary, has developed a special grammaticized use in marking an additional type of passive. The active in (12a) has two different corresponding passives, (12b) and (12c).

(12) a. A journalist photographed Marie.
  b. Marie was photographed by a journalist.
  c. Marie got photographed by a journalist.

There are two minor semantic or pragmatic differences between (12b) and (12c). The first is that the get passive is somewhat more informal in style than the be passive. The second is that the get passive carries a weak tendency toward implying that the event described was either some kind of misfortune or some kind of benefit for Marie, and that she may to some extent have brought the situation upon herself or arranged it. These get passives from the Wall Street Journal illustrate:

(13) a. ... more hardcore bears than the ones who got spooked in January and February ...
  b. A man got clobbered with a steel chair during a secret meeting of local contractors last October.
  c. ... he got hooked on skeet shooting ...
  d. He even got appointed to a coveted civic post.

The writer of (13a) is implying that the pessimistic investors let themselves succumb to spooking; the man in (13b) was certainly unlucky to get clobbered and may perhaps have brought it upon himself; the man in (13c) permitted himself to get hooked or was unfortunate enough to get drawn in; and we can assume that the man in (13d) eagerly sought his coveted civic post. Thus the get passive sounds natural in each case. By contrast, it would be distinctly odd to write anything like this:

(14) ??The Nobel Prize in Physics got awarded to Peter Higgs in 2013.

The prize does not bring the awarding upon itself, or receive either benefit or misfortune by having the Nobel committee award it, so a get passive sounds strikingly inappropriate here.

Notice that although (10a), repeated here as (15), has the verb get, it is not a get passive.

(15) David Swan, a geneticist, got himself transferred within the CDC.

The get clause here is an active clause with a transitive verb that has a direct object and a subjectless passive clause as complement. Get has causative force: (15) it does not just vaguely imply that Swan might have done something resulting in his being transferred; it entails that Swan took some active part in obtaining his transfer.

2.2.6 Concealed passives

Finally, there are also certain passive clauses, referred to in CGEL as concealed passives, that have a gerund-participle rather than a past participle as head:

(16) a. This rug badly needs washing.
  b. The situation needs looking into by experts.

In some dialects (especially Scotland, Northern Ireland, western Pennsylvania, eastern Ohio, northern West Virginia, central Indiana), need also takes a past-participial bare passive complement, so that This needs washed is grammatical.

2.3 Summary

What I have stressed so far is that contrary to popular belief, passives do not always contain be and do not always contain a past participle. They also do not always obscure the role or responsibility of the doer. They may or may not have a subject (the passive clause in any monument defaced by vandals does not), and they may or may not have a by-phrase (The president has been assassinated does not). Sometimes they specify the agent of an action very clearly (as in It was thrown at them by hooligans), and sometimes not (as in It was thrown at them); sometimes they specify the undergoer (as in A surfer was attacked by a shark) and sometimes not (as in Being attacked by a shark is no fun). Often (as in (3) there is no action whatsoever, rendering the strange phrase ``receives the action'' inappropriate.

As we shall see in section 3, the usage advice literature, and the comments about usage and style by writers and critics in general, show an extraordinary level of ignorance of simple facts of this kind. But first I need to cover an additional aspect of passives: their sensitivity to discourse conditions.

2.4 Information-structure constraints on passives

Almost universally overlooked in both the pedagogical grammar literature and transformational-generative syntactic discussions are two important information structure constraints governing the felicitous use of passive clauses in discourse.

2.4.1 The state-affecting condition on prepositional passives

The first one applies solely to prepositional passives. There are two slightly different kinds of prepositional passives, which I will call the idiomatic and locative types (see Huddleston and Pullum et al. 2002, Ch 16, sec. 10). In the idiomatic type, the preposition is specified by the construction itself, as in toyed with, taken advantage of, or looked up to by everyone, where the particular prepositions cannot be replaced by others. But in the locative type the head preposition just has its ordinary locative meaning, as in My hat was sat on by one of the guests or The shop was broken into during the night. In prepositional passives of the locative type, the VP has to denote either a salient and significant property of the entity it is predicated of, very often determiningn a physical state change in that entity. Compare these two examples:

(17) a.  It may surprise you to learn that this chair was once sat in by Sir Winston Churchill.
  b. #It may surprise you to learn that Swindon has been walked in by Sir Winston Churchill.

It is arguably a significant historical property for a chair to have had the great British prime minister sit in it; but no one could think it a significant property of a Wiltshire town that Churchill once walked in its streets, and that makes (17b) anomalous.

Again, note the contrast between (18a) and (18b):

(18) a.  This bed has been slept in.
  b. #The monastery has been slept in.

It changes significant properties of a bed to sleep in it; but it doesn't change any important property of a monastery to have had people sleep within its walls.

2.4.2 The new-information condition on by-phrases

The second constraint is more important and more general. It applies solely to long passives (whether prepositional or not). We can state it informally as (19):

(19)  The denotation of the passive complement NP in a passive clause must denote something at least as new in the discourse as the subject.

The force of the constraint is illustrated by the infelicitous character of case (c), but not case (b), in (20).

(20) a. Have you heard the news about YouTube? It was bought by Google. [acceptable because the subject is old and passive complement is new]
  b. Have you heard the news about Google? #YouTube was bought by it/them. [unacceptable because the subject (YouTube) is newer in the discourse than the by-phrase]

The reference to newness in the information-structure constraint is relative: it is possible for a passive complement to embody old information, but only if it is not older than what the subject contributes. This is illustrated in (21).

(21) a. A man walks into a pub leading a bear on a leash. #A beer and a pot of honey are ordered by him...
[unacceptable because the subject is newer than passive complement NP]
  b. A man walks into a bar leading a bear on a leash, and says to the barman: "Give me a beer and a pot of honey." So a beer and a pot of honey are handed to him by the barman...
[acceptable because subject and passive complement NP are both discourse-old]

This information-packaging constraint is a real, important, and fully general part of the way long passives work. It is remarkable that, as we shall see immediately, hardly any attempts at describing the passive make any mention of it.

3  Critical incompetence

Keeping the information-structure constraint (19) in mind, take a look at the beginning of the section headed `Use the active voice' in William Strunk's original 1918 edition of America's most famous book on how to write, The Elements of Style:

The active voice is usually more direct and vigorous than the passive:

   I shall always remember my first visit to Boston.

This is much better than

   My first visit to Boston will always be remembered by me.

The latter sentence is less direct, less bold, and less concise.

Directness, boldness, and concision are not even relevant here: Strunk's disrecommended example is a passive that flagrantly violates the general information-structure constraint (19), and cannot be used in any normal kind of context, because the NP in the by-phrase denotes the utterer, and the utterer is discourse-old by default (because there is guaranteed to be an utterer in every discourse).

The sort of factor that can override this and make by me a perfectly acceptable by-phrase is an emphasized contrast between the utterer and someone else:

(22)  The picture you've been admiring wasn't painted by Picasso at all; it was painted by me!

If Strunk wanted to explain why passives are dispreferred relative to their active counterparts, he should at least have started with a passive that stood some chance of being sensibly used. To support a prejudice against the passive we would need grounds for claiming that even (22) is unacceptable.

But as Strunk proceeds with his explanation of why every writer should favor the active voice, his exposition drifts off the topic of passives, asserting that "Many a tame sentence of description or exposition can be made lively and emphatic by substituting a transitive in the active voice for some such perfunctory expression as there is, or could be heard."

The confused discussion is only amplified and worsened in the revisions made in later editions of The Elements of Style by E. B. White, where small changes to the examples are made. The most recent version of the book (Strunk & White 2000, p. 19) contains the following tableau, where the sentences on the left are supposed to be bad, and the ones on the right are proposed as suitable corrections or replacements:

There were a great number of dead leaves lying on the ground.     Dead leaves covered the ground.
At dawn, the crowing of a rooster could be heard.     The cock's crow came with dawn.
The reason he left college was that his health became impaired.     Failing health compelled him to leave college.
It was not long before she was very sorry that she had said what she had.     She soon repented her words.

The original theme of replacing passive clauses by active transitives has now disappeared completely. The main difference between the examples in the two columns is that the favored ones on the right are significantly shorter than the allegedly bad ones. But there is not a single case in Strunk & White's tabular display of a passive in the left column corresponding to an active transitive on the right. In the second pair of examples, the right-hand version has an intransitive verb; and none of the others has a passive on the left (impaired, for example, is not a passive VP—become cannot take passive clause complements, as we see from *Kennedy became shot by Oswald).

But Strunk & White's confused discussion was only the beginning. Today, mistaken charges of using the passive voice are commonplace. In (23) – (34) I present a collection of quotations from published sources illustrating the depth of the misdiagnosis problem and the extraordinary diversity of constructions to which the grammatical term `passive' has been wrongly attached.

(23) Tim Levell in a BBC blog post (here) about how the BBC strives to be sensitive when covering stories about upsetting events such as school shootings:

We use passive constructions (`Five girls have died', not `The man went in and shot five girls')...
But Five girls have died is not a passive clause.

(24) Steve Inskeep interviewing U.S. Army General Dan McNeill (NATO commander, Afghanistan) on National Public Radio (USA) in 2007 (see this Language Log post):

Let's put it in the passive tense: there was a ceasefire agreement in Southern Afghanistan with some members of the Taliban at one time.
But the passive is not a tense, and there was a ceasefire agreement is not a passive clause.

(25) From a blog discussion of Paul JJ Payack's claims about the 2008 vice presidential debates in the USA (see this Language Log post):

Though most of Obama's verbs were in the active voice, 11 percent of the sentences were in the passive voice, a dependable method of deflecting responsibility, Payack said. He cited Obama's `There will be setbacks and false starts' as an example. `He's spreading the responsibility around,' Payack said. `He didn't say, ``I will have setbacks. I will be wrong. I will make mistakes.'' He used the passive voice for those types of constructions.'
But There will be setbacks and false starts is an existential clause, not a passive. No use of the passive voice is cited.

(26) From a post by Tom Maguire on the blog JustOneMinute, discussing the arrest affidavit for George Zimmerman after he shot the unarmed black teenager Trayvon Martin (see this Language Log post):

I especially like the passive voice at the critical plot point: `a struggle ensued'. Those pesky struggles, ensuing like that! One might have thought the prosecution would at least argue that Zimmerman initiated the struggle, in addition to the verbal confrontation.
But A struggle ensued is not passive. Clauses with verbs of event occurrence (continue, ensue, go on, happen, occur, transpire, ...) are active clauses.

(27) A 2003 media analysis report by the `Honest Reporting' group alleges that Reuters shows anti-Israel bias in its headlines, using passive voice to mask Palestinian responsibility (see this Language Log post). Three examples of this biased use of passives are cited in the report. These are the first two:

  1. `New West Bank Shooting Mars Truce' (July 1) Palestinian not named as perpetrator; Israeli not named as victim; shooting described in passive voice.
  2. `Bus Blows Up in Central Jerusalem' (June 11) Palestinian not named as perpetrator; Israelis not named as victims; described in passive voice.
But neither example contains a passive clause.

(28) From `Weapons of denial,' an article by Marie Murray in The Irish Times, 31 July 2009:

`The passive voice is especially useful where apologies are required: personal apologies for what people have done personally. Because instead of having to say, ``I'm sorry'', the passive voice allows a culprit to say ``It is regrettable''. Instead of saying ``I made a mistake'' the abstract term ``mistakes happened'' can be evoked.'
But neither it is regrettable nor mistakes happened is a passive.

(29) From an article by Jon Fine in Business Week about a plugin for the Firefox browser called `SpinSpotter,' which will allegedly identify rhetorical devices as passives:

passive voice
(example: a story says `bombs land' without stating which party is responsible for them)...
But Bombs land is an active clause.

(30) From an article in the Washington Post by Charles Krauthammer of Time, described by the Financial Times in 2006 as the most influential commentator in America, discussing President Obama's gentle treatment of Muslims:

On religious tolerance, he gently referenced the Christians of Lebanon and Egypt, then lamented that the `divisions between Sunni and Shia have led to tragic violence' (note the use of the passive voice).
But Divisions between Sunni and Shia have led to tragic violence is not in the passive voice. (31) From an article by law professor Richard A. Epstein in the Hoover Institution journal Defining Ideas (30 July 2013, here):
Instead of suggesting policies to reduce the impact of government on production, Obama reverts into a lament for the lost middle class. He notes that our economic engine has, over time, ``began [sic] to stall'': ... But his careful use of the passive voice—the engine ``began to stall''—conceals that he has no explanation of the ebbs and fall of the overall system.
But began to stall is not a passive construction. (Note: At some time during the day of publication the text of Epstein's article was silently amended: by about noon Pacific time the reference to passive voice had been removed. I thank Linda Seebach for sending me a transcript of the relevant passage the way it was when she read it earlier that day.)

(32) From an article by Nancy Franklin in The New Yorker commenting on fraudster Bernie Madoff's having said about his Ponzi scheme, `I believed it would end soon':

...he betrayed no sense of how absurd it was to use the passive voice in regard to his scheme, as if it were a spell of bad weather that had descended on him.
But (although Madoff does indeed appear to speak as if the possibility of his criminal misdeeds ending were independent of his volition) It will end soon is not in the passive voice.

(33) From a UK government publication on literacy, the Year 7 Sentence Level Bank issued in April 2001 by the UK Department for Education and Employment as part of the UK National Literacy Strategy, exemplifying the passive voice:

But The butler was dead is not a passive. An earlier version had The butler was murdered, which is a passive, but no one noticed that the change had ruined the point of the example (see Cajkler 2004).

(34) From a blog by a writing teacher at Boston University complaining about undergraduates' excuses: (see here):

The plane stub was accompanied by a breezy email: `thought I would be able to return by tonight, but it does not appear that will be in the cards.' I found the switch to the passive in the latter clause of that sentence to be extremely irritating, as if he were freeing himself from blame. If you're going to be absent, just own up to it like a responsible human being.
But It does not appear that will be in the cards is not a passive.

This list provides clear evidence that the state of the general public's education regarding the notion `passive voice' is nothing short of disastrous, even when we consider people like professional writers, journalists, and authors of usage guides. However, people have suggested to me several times that the sheer number of failures to apply the term `passive' the way grammarians do indicates there must be something wrong with my assumptions: perhaps I am being too prescriptive, and in reality there is a vernacular sense of the term that simply isn't the technical term in use among grammarians. In other words, perhaps I am in the position of someone who, ludicrously, objects to the claim that the World Wide Web was responsible for a transformation in book retailing on the grounds that the term `transformation' actually refers to a mapping from phrase markers to phrase markers.

And there is a rather vague discourse-semantic property that one might say is shared by all the quotations I have given so far: in the examples cited, the critics seem to think that in some sense agency has been being obscured or attenuated, or some admission of responsibility has been evaded.

But this does not cover come anywhere near covering the full range of the evidence. It could perhaps apply to most of the examples given so far; but now consider the cases in (35) – (44):

(35) The commenter `Ivenson', on the Wonkette website, talking about Sarah Palin (see here):

Gave up when she said `you guys are wanting to dissect the past.' Awful passive voice manner of speech ...
But there is no `passive voice manner of speech', and you guys are wanting to dissect the past is in the active voice, and neither the allegation expressed not the reference of the subject (you guys) could possibly be any clearer.

(36) Brad DeLong, a well-known blogger, lists uses of the passive in an English translation of a passage by Wolfgang Mommsen (see his post here and my discussion in this Language Log post):

the misfortune that befell Germany and Europe...
the Reich had to face a superior coalition...
the war turned out to be...
the ... situation that isolated Germany...
It was above all the bloody reckoning...
But not one of these five examples illustrates the use of passive voice. And DeLong's suggestions for "alternatives that focus more attention on agency" are often radical rewritings. For example, in his post he proposes replacing `the Reich had to face a superior coalition' by `the German government struck first and struck at Belgium and France, creating the coalition of Russia, France, and Britain it then fought'; but the replacement is almost triple the length, and has a subject (the German government) that is referentially the same as the original (the Reich), so nothing about passive voice seems conceivably relevant here.

(37) From the group blog of The American Prospect in July 2007 (see this Language Log post):

First Read, a reliable purveyor of Beltway conventional wisdom, tries out the passive voice: `As for the media, we've allowed this story over race [to] bury one of the more consequential weeks of Obama's presidency thus far (the financial reform legislation becoming law, Senate passage of the jobless benefits, and Kagan clearing the Senate Judiciary Committee). Whether it's Sherrod, Gates, or Jeremiah Wright, the topic of race pushes the media's buttons like no other issue.'
But there are no passives whatever among the five occurrences of nonauxiliary verbs in the quotation (allow, bury, become, clear, and push), and I see no obscuring or attenuating of agency or responsibility.

(38) From a letter by Dave Bruce of Hoboken addressed to the public editor of the New York Times about a correction to a Maureen Dowd column (see this Language Log post):

The passive-voice note that she `failed to attribute a paragraph' seems to play down what actually occurred.
But Maureen Dowd's column failed to attribute a paragraph is in the active voice, not the passive. Dowd is clearly specified as the responsible party for the failure.

(39) From The Aspen Handbook for Legal Writers by Deborah E. Bouchoux:

Most sentences that include dangling modifiers are written in the passive voice. Changing to active voice corrects the dangling modifier because an actor or subject is identified in the phrase that begins the sentence.
Example:
When a boy, my father changed careers (passive voice).
When I was a boy, my father changed careers (active voice, actor identified in modifying phrase).
But When a boy is an elliptical phrase containing a predicative complement NP without a target of predication. It is not a passive. The complaint here is that if a verbless clause consisting of just a predicative NP were used, and the the hearer or reader assumed that the matrix clause subject provided the implicit subject thereof, the sentence would have been mistakenly interpreted (see Huddleston & Pullum 2005: 206–209). But the possibility of an incorrect on-the-fly resolution of an ambiguity has nothing whatever to do with the deliberate concealment of agency.

(40) From the BBC News Style Guide:

Compare these examples. The first is in the passive, the second active:
But the former is not a passive, and no clear agency or responsibility issue arises (in both versions the youths threw the stones, and in neither version is the instigator of the riots named or implied).

(41) From the Canadian Press Stylebook, the standard usage guide for all working journalists in Canada, under `Common faults':

Active vs. passive
Think of active verbs as power words—words that drive your sentences, keep the reader's attention and move her briskly along.
Not: The economy experienced a quick revival.
But: The economy revived quickly.
Not: At first light there was no sign of the ship.
But: The ship vanished in the night.
But neither The economy experienced a quick revival nor At first light there was no sign of the ship contains a passive clause. (Notice that in the second pair, making the ship the subject of the intransitive verb vanish seems no different from making bombs the subject of the intransitive verb land in (29), yet The ship vanished in the night is being cited as the preferred case, not exemplifying the passive.)

(42) From the section on infinitives at the grammarist.com site (see this page):

Active voice and passive voice
In the active voice, to is usually dropped from the infinitive. For example, in this sentence, the to is dropped from the infinitive to stay: He should stay in Reno, where he belongs. If the same thought were expressed in the passive voice, the to would remain: I urge him to stay in cable, where he belongs.
But there is no passive clause in I urge him to stay in cable, where he belongs; and no issue of agency or responsibility is in play.

(43) From the style guide of a large corporation (which the informant preferred not to name: see this Language Log post):

Use active voice rather than passive voice.
Active voice is easier to read.
Instead of `we have decided,'
write `we decided.'
Instead of `we will be implementing a program,'
write `we are implementing.'
But neither have decided nor will be implementing is a passive. The writer here seems to have confused the notion `passive' with the property of containing an auxiliary verb.

(44) From Doostang, a corporate job-related advice site, giving examples of "the passive voice" that you should expunge from your resumé:

Indicators of the passive voice:
But not one of the four examples cited is a passive, and in every case the implicit subject of the cited phrase is clear (it is the author of the resumé). Doostang claims Responsible for management of three direct reports is worse than Managed 3 direct reports because the latter `is a shorter, more direct mode of writing and adds impact', so passive voice is apparently being confused with verbosity.

Once we take cases of this sort seen in (35) – (44) into account, we see that there is simply no unified meaning for the term `passive' that could possibly encompass all of the data. Some writers seem to think that `passive' means `contains an auxiliary verb' or `contains a nominalization'; others are even more mysterious. Consider the entire array of constructions we have now seen mistaken for passives:

No folk rhetorical property could yoke together this diverse array of constructions. What is going on is that people are simply tossing the term `passive' around when they want to cast aspersions on pieces of writing that, for some ineffable reason, they don't care for. They see a turn of phrase that strikes them as weak in some way, or lacks some sort of crispness or brightness that they cannot pin down, and they call it `passive' without further thought. And such is the state of knowledge about grammar among the reading public that they get away with it.

The most likely potential source for this fuzzy notion of passivity is nicely exemplified in a passage found by Arnold Zwicky (on Language Log, in this post). A handbook of style published nearly 80 years ago (Jensen et al. 1935) stacks on top of the familiar negatively prejudiced view of the passive voice a whole new set of charges about something called "the passive style". Zwicky quotes from the second edition (1941: 437–438):

Another kind of wordiness, the most pernicious kind of all, comes partly from laziness, partly from fear. This we may call the `passive style' as distinguished from the `active.' It is full of cumbrous qualifications: `in general it may be said that,' `under ordinary circumstances it will be found,' `it is probably safe to say that.' It has long and unnecessary transitions: `Now that we have seen how the machine functions, let us take a view of its advantages to social progress.' Worst of all, the writer of the passive style converts his verbs into abstract nouns and uses passive verbs and verbs of being. He thus robs his writing of its greatest strength: action. He takes good honest verbs like separate, develop, bewilder, make, and steals their life away by turning them into the abstract nouns separation, development, bewilderment, manufacture, or, much worse, the making of. With his verbal ideas thus abstracted, the writer of the passive style must cast about for other verbs to fill his sentences. First he looks for verbs of being. To say that a thing is or seems or becomes is almost never as good as to say it does something. He who robs his thoughts of action robs them of half their life, for life is action and readers like to think in terms of action. Especially is this evident in another characteristic of the passive style, the use of verbs in the passive voice. A passive verb shows action in reverse. It represents a subject not as doing something but as being done to. Hence it too makes meaning static. That is the great defect of the passive style. It pictures for a reader life in the abstract, life without action: still-life.
Here the notion of passive voice in the grammatical sense, which gets a brief mention, has been rolled together in a confused (and astonishingly wordy) way with all sorts of other factors: verbosity, laziness, cowardice, circumlocution, weakness, abstractness, inactivity, and lifelessness.

But of course one could imagine prose being written in this style without containing any passives at all. The style might better have been described as ponderous.

Allegations about passives

The kind of vague but highly prejudicial writing about style seen in the above quotation from Jensen et al. and elsewhere (e.g., from Strunk) has contributed to what is now an epidemic level of confusion among educated Americans. They are presented with a variety of different allegations about the passive voice (or the `passive style'—the two are not clearly distinguished) and what is so bad about it. Yet those allegations are hardly ever seriously supported.

It is surely not too much to ask that those who claim that the passive is bad should have some definition of the notion `passive' in mind, and that their examples of passives should be passives according to that definition. We have seen that it is very common for critics who complain about the passive to be entirely unable to meet these conditions. But we could also reasonably ask that they should justify their claims about what is wrong with the examples. If using a passive construction is a hallmark of bad writing style, we need to know what is supposed to be bad about it—what justifies the allegation of stylistic badness.

But in fact that condition is not met either. The claims about the alleged faults of passive clauses are never justified. Passives are variously alleged to be

All the allegations are unsupported, and to some extent clearly false.

Dull and static?

The article in Wikipedia on President John F. Kennedy has a section headed `Assassination' which begins thus:

President Kennedy was assassinated in Dallas, Texas, at 12:30 pm Central Standard Time on November 22, 1963...
One could hardly imagine a way of making a statement introducing that shocking event that would be more vivid. And with Kennedy as the topic of the entire article, having his name as the subject like this is obviously the best stylistic option.

Sneaky or evasive?

Sherry Roberts' astonishing metaphorical diatribe against the passive was quoted earlier. Here it is with the preceding context, and the following examples:

A sentence written in the active voice is the straight-shooting sheriff who faces the gunslinger proudly and fearlessly. It is honest, straightforward; you know where you stand...

A sentence written in passive voice is the shifty desperado who tries to win the gunfight by shooting the sheriff in the back, stealing his horse, and sneaking out of town. [Here, in section 7, `Be Active'.]

Active: The committee will review all applications in early April.

Passive: In early April, all applications will be reviewed by the committee.

Notice that she unthinkingly uses a passive (the bare passive adjunct written in passive voice) while making the above statements And note also that it would be truly eccentric to suggest that the second example's account of what will happen to the applications in April is sneaky. The administrative detail of who will undertake the examination of the applications (the committee) has been made fully explicit. The prejudice expressed in Roberts' absurdly metaphorical characterization is utterly unsupported by her exemplification. She unthinkingly repeats and embroiders the stock criticism of the short passive (that it leaves the agent unexpressed), but exemplifies with a long passive to which the stock criticism is inapplicable, failing to notice the inconsistency.

Certainly, one of the objections to other people's prose that the objectors often seem to have in mind is some kind of avoidance of attribution of agency, or vagueness about responsibility. But the belief that the passive necessarily embodies such qualities is transparently false. Omission of the passive complement (not simply use of the passive) permits agents to be left unspecified, but there are plenty of other ways, nominalization being just one of those. And long passives not only give details of the agent, they are ideal for laying special emphasis on the agent.

But the stock criticism is of course nonsense even as applied to short passives. They are not the slightest bit sneaky or evasive when the identity of the agent is either thoroughly irrelevant or entirely unknowable. None of these examples (in which the verbs of the passive clauses are underlined) would be improved by struggling to rephrase them in the active:

(45) a. When the patient was first diagnosed with cancer her symptoms were minor.
  b. Perhaps the mysterious mound was constructed as a memorial.
  c. Since metallic sodium reacts violently with water it is usually shipped in oil-filled canisters.
  d. The strange object found in the crater was apparently made from a material unknown to terrestrial science.

In (45a–d) and innumerable other such examples, specifying an agent is neither necessary nor even advisable.

Feeble rather than bold?

Decrying passives as feeble, ineffective, and weak rather than bold, muscular, and strong seems to presuppose two things that need a defense. The first is that there is something weak about, say, The World Trade Center has been attacked by terrorists as compared to Terrorists have attacked the World Trade Center. The second is that the right style for every kind of writing is bold, muscular, and strong (rather than, say, subtle, delicate, and restrained). These two claims are quite independent, but neither gets any vestige of support in most of the usage literature.

To address the first point we might ask whether the passive avoided in tough, muscular prose. Take a look at a classic example of tough-guy writing style: Lee Child's first Jack Reacher novel Killing Floor, which won the prestigious Anthony Award. It is narrated in the first person by Jack Reacher, a tough and ruthless former military policeman who frequently deals with bad guys by breaking knees or necks. Brutally short sentences are the norm in Lee Child's prose. Yet the book opens with this sentence:

(46) I was arrested in Eno's diner.

There is no sense in which this is feebler than an active would have been. Indeed, filling in an agent as the arresting authority would actually have spoiled the exposition: in the first scene of the novel Reacher is taken into custody suddenly and without warning, and has no idea what is going on, who has ordered his arrest, or what the motive might be.

Moving to the second point, is it always right to go for bold, muscular, tough writing rather than subtle, delicate, restrained, evocative, or descriptive writing? The writing advisers seem to have no doubt in their minds that every sentence is supposed to come at you like a punch in the gut from Jack Reacher's fist, on top of assuming unjustifiably that passive clauses can't deliver a gut punch. Both assumptions seem completely indefensible.

Avoided in the best writing?

It matters very little whom you may decide to regard as an author exemplifying fine writing. Although there has apparently been a long decline in the frequency of use of be passives in the USA (see Mark Liberman's Language Log post "American passivity" here), authors who avoid the use of the passive voice entirely do not appear to exist. It would certainly be ridiculous to suggest that the passive is absent from the work of admired authors in the early 20th century, when Strunk was teaching at Cornell and Orwell was a boy.

Take H. G. Wells, the inventor of modern science fiction. He opens his classic book The War of the Worlds thus (and I underline the verbs of the passive clauses):

No one would have believed in the last years of the nineteenth century that this world was being watched keenly and closely by intelligences greater than man's and yet as mortal as his own; that as men busied themselves about their various concerns they were scrutinised and studied, perhaps almost as narrowly as a man with a microscope might scrutinise the transient creatures that swarm and multiply in a drop of water.
There are 6 occurrences of potentially transitive verbs: believe, watch, busy, scrutinize (twice), and study. 50% of them are in passive clauses.

I picked Wells at random from books available to me in e-text form, but one could do something similar with almost any literary work, as the reader may verify. And it makes little difference if you decide to look at prose written by the advisers on usage themselves. Consider the beginning of E. B. White's introduction to his revision of The Elements of Style (Strunk & White 2000). I underline the passive VPs:

At the close of the first World War, when I was a student at Cornell, I took a course called English 8. My professor was William Strunk Jr. A textbook required for the course was a slim volume called The Elements of Style, whose author was the professor himself. The year was 1919. The book was known on the campus in those days as `the little book,' with the stress on the word `little.' It had been privately printed by the author.
Six transitive verb forms (underlined) appear here: took, called, required, called, known, and printed. Five of them are in the passive. That's over 83 percent.

Or take George Orwell, the man who said `Never use the passive where you can use the active'. His celebrated essay `A Hanging' begins thus:

It was in Burma, a sodden morning of the rains. A sickly light, like yellow tinfoil, was slanting over the high walls into the jail yard. We were waiting outside the condemned cells, a row of sheds fronted with double bars, like small animal cages. Each cell measured about ten feet by ten and was quite bare within except for a plank bed and a pot of drinking water. In some of them brown silent men were squatting at the inner bars, with their blankets draped round them. These were the condemned men, due to be hanged within the next week or two.

   One prisoner had been brought out of his cell. He was a Hindu, a puny wisp of a man, with a shaven head and vague liquid eyes.

The verbs that are capable of being used transitively here are drape, condemn, hang, and bring. (The first occurrence of condemned is best treated as an adjective, and the same is true of shaven.) 100% of the occurrences of these transitive verbs are in the passive voice.

This is not a statistical quirk. Orwell's writing features significantly more passives than typical prose. By one count, on average in typical prose about 13% of the transitive verbs are in the passive, whereas in Orwell's essay `Politics and the English language' it is 20% (Merriam-Webster's Dictionary of English Usage, 1994, p. 720, citing Bryant 1962). My own counts, adhering strictly to the definitions of Huddleston & Pullum et al. (2002), are somewhat higher (probably because I include passive participles used as modifiers rather than complements, which for some reason many grammarians miss), but the ratio is unchanged: by my count, about 17% of the transitive verbs in random prose are likely to be passive, while a careful count of the whole of Orwell's essay shows that 26% are passive. By either count, then, Orwell uses more than one and a half times as many passives as typical writers.

In some cases usage commentators make little jokes by deliberately using the passive in the same sentence where they condemn it (as when Brad DeLong says of Mommsen's alleged passives, `Alternatives that focus more attention on agency are almost immediately thought of by one', and perhaps also when a Microsoft Word template for papers in the American Psychological Association (available here) says `The active voice, rather than passive voice, should be used in your writing'); but there is no sign that White or Orwell were joking. It appears that they simply had no idea of how often the passive occurred in their own writing, and never thought to run a check on themselves, or even on the paragraph in which they denigrated the passive.

The conclusion has to be that they would not dream of taking their own assertions about the passive seriously. Nor should we, I submit.

Conclusions

The topic of this paper is not so much a construction as a strange cultural trend emerging in the 20th century among language mavens, writing tutors, and usage advisers. Unneeded warnings against sentences that have nothing wrong with them are handed out by people who actually don't know how to identify instances of what they are warning against, and the people they aim to educate or intimidate don't know enough grammar to reject the nonsense they are offered. The blind warning the blind about a nonexistent danger.

How and why has this all happened? Oversimplification and overkill by well-meaning advisers may have a lot to do with it. Arnold Zwicky has noted on Language Log (here) that usage advisers tend to go for overkill, adopting a kind of zero-tolerance principle: that if students or novice writers do something too much, or if doing it sometimes gets them in trouble, they should be told not to do it at all.

It is right and good, of course, to instruct students and novice writers in how they might improve their writing. But handing them simplistic prescriptions and prohibitions is not doing them any favors. `Avoid the passive' is typical of such virtually useless advice.

The claims about why you should avoid passives—the allegations about why they are bad—are all bogus, and the interesting point (the discourse condition) is always missed. The advice is often supplied by advice-givers who don't respect their own counsel—though they are unaware of that that because they are commonly hopeless at distinguishing passives from actives. But the recipients of the advice can't identify passives either, so they are powerless to spot the blunders of their teachers. Even if they managed to follow the advice rigorously (which they can hardly do if it is not clear to them what a passive is), it would usually not improve their writing one whit. It would certainly make them write less like great writers of the past—and more like a little child.

The standard teaching about shunning the passive should be abandoned entirely. But that does not mean abandoning the teaching of grammar. College students should certainly be taught enough grammar to permit them to recognize passives in what they write; and it might be reasonable to teach the discourse condition on passives stated in (19) explicitly. Intelligent discussion of how non-canonical clauses fit into discourses just might have beneficial effects on the work of novice writers (note the interesting and perceptive account by Joseph Williams of how using passives can assist in ensuring a coherent flow of subject choices throughout a paragraph: Williams 1990: 54–55). There is at least some reason to hope so.

One thing is certain: such teaching could hardly have worse results than the policies in place now, which have given us usage critics, writing tutors, and even style guide authors who have no idea what they are warning against when they hand out the standard warnings against the perennially hated passive voice.

Acknowledgments I owe a major debt to the many Language Log readers over the years—literally too many to list here or even remember—who sent me references or links to erroneous criticisms alleging use of the passive voice. You know who you are, and I thank you en masse; I could never have found all those cases without you, and before you started sending them I never realized how many there are. I have been assisted in my study of this subject by correspondence and conversation with a number of colleagues, notably Rodney Huddleston, Richard Hudson, Mark Liberman, Geoffrey Nunberg, and Arnold Zwicky. Rodney Huddleston was particularly generous with his time in advising me about how to frame the descriptive section. The version published here is less technical and (I hope) much more useful than it would have been, and if it is still not perfect, that is certainly not his fault.

Bibliography

Baker, Sheridan. 1985. The Practical Stylist. New York: Harper and Row, 6th ed.

Bryant, Margaret M. 1962. Modern American Usage. New York: Funk and Wagnalls.

Cajkler, W. 2004. How a dead butler was killed: The way English national strategies maim grammatical parts. Language and Education 18(1):1–16.

Foerster, Norman and J. M. Steadman. 1931. Writing and Thinking. Boston: Houghton, Mifflin.

Gilman, E. Ward, ed. 1994. Merriam-Webster's Dictionary of English Usage. Springfield, CT: Merriam-Webster.

Haussamen, Brock. 1997. Revising the Rules: Traditional Grammar and Modern Linguistics. Dubuque, IA: Kendall/Hunt, 2nd ed.

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

Huddleston, Rodney and Geoffrey K. Pullum. 2005. A Student's Introduction to English Grammar. Cambridge: Cambridge University Press.

Jensen, Dana Olaf, Robert Morell Schmitz, and Henry Francis Thoma. 1935. Modern Composition and Rhetoric. Boston: Houghton Mifflin. Quotation and page reference from 2nd edition, 1941.

Orwell, George. 1946. Politics and the English language. Horizon 76 (April 1946), 252–264.

Strunk, Jr., William. 1918. The Elements of Style. Privately printed in Geneva, NY, at the Press of W. F. Humphrey. Text online here.

Strunk, Jr., William and E. B. White. 2000. The Elements of Style. New York: Allyn and Bacon, 4th ed.

Williams, Joseph M. 1990. Style: Toward Clarity and Grace. Chicago: University of Chicago Press.

Woolley, Edwin. 1907. Handbook of Composition. Boston: D. C. Heath.