Choosing between the active and the passive voice
The passive voice is a grammatical choice and not an error. It puts the thing acted on first and can leave the actor unstated, so it suits a sentence whose actor is unknown or beside the point, while the active suits one in which who acted matters. The choice also sets what a sentence emphasises. How much passive a medical text should carry, and whether any ceiling on it makes sense, is disputed.
What makes a verb passive
Voice describes the relation between a sentence's subject and its verb: in the active voice the subject acts, and in the passive the subject is acted on. In the active the agent comes before the verb, and most active sentences follow the order agent, verb, object; with a passive verb the agent comes after the verb. When the direct object of an active sentence becomes the subject, the verb is passive and the agent moves into a prepositional phrase or disappears (I ate the chocolate; The chocolate was eaten by me; The chocolate was eaten), and a pronoun moves from the objective to the nominative case (She loved him; He was loved by her). So Drug X reduced the inflammation in the lungs significantly is active, The inflammation in the lungs was reduced significantly is passive with no agent expressed, and The inflammation in the lungs was reduced significantly by Drug X is passive with the agent added.
English forms the passive with a past participle, usually but not always ending in -ed, as paid, kept, drunk and forgotten show, and an auxiliary that is always a form of to be, whose tense and aspect time the action (The chocolate is being eaten; The chocolate will have been eaten). Other languages have completely different ways of expressing the passive. The auxiliary need not be repeated if it applies to all the verbs in a series.
Constructions mistaken for the passive
Only transitive verbs have a passive, so He sleeps cannot be made passive; and if either the form of to be or the past participle is missing, the verb is not passive. The constructions some people might mistake for it — a modal (She has to eat), progressive tenses of intransitive verbs (He had been waiting; I am thinking), linking and mental verbs (They feel sick; He looks tired), a state-of-being verb (We became friends) and an expletive (It is raining; There is work to be done) — are all active.
An expletive construction is not a passive, although advice against the passive has mistaken one for it. Expletive comes from the Latin explere, to fill, and a syntactic expletive is a word meaningless in itself that performs a syntactic role, usually as a dummy subject: It is raining; There is no cause for alarm. Neither verb is passive, because there is a form of to be but no past participle. In a construction beginning there is or there are, the verb agrees in number with the complement (There is a rose in Spanish Harlem; There are no children here). Strunk and White's bad example in their discussion of the active voice, There were a great number of dead leaves lying on the ground, which they rewrite as Dead leaves covered the ground, is not a passive but a needlessly long sentence that begins with an expletive. Expletive constructions are grammatical, but they often mark an opportunity to omit needless words.
What the choice of voice changes
Active and passive constructions do not necessarily give interchangeable ways of saying the same thing, and because they do not create precisely the same statements, they must not be interchanged casually. In each of the pairs we predicted the pharmacokinetic properties of X and the pharmacokinetic properties of X were predicted, and Smith and colleagues have reported a clinical series… and a clinical series… has been reported by Smith and colleagues, the essential information is the same, but the active statement is about who did something and the passive about what had something done to it. The choice depends on whether the writer wants to focus on who or what acts, or on who or what is affected by the action.
Voice is also a way of choosing the emphasis, because what comes first in a sentence is usually, or should be, the most important information. The same finding can therefore be angled by the choice: in the Methods section of an abstract reporting that 474 patients completed the French version of a questionnaire, the active formulation stresses that there were 474 patients, the passive that the French version of the questionnaire was the one used.
When the passive suits
The passive says what was done to someone or something without saying who did it, and it is the more appropriate choice where the doer is unknown, or is obvious, unimportant or less important than the recipient of the action. In a methods section the specimens were washed puts the emphasis on the specimens, where it belongs, and we washed the specimens mentions a we that is unnecessary, since the researchers obviously washed them. Abdominal incisions were closed with absorbable sutures leaves out an agent that is obvious or unimportant, and The water supply to the hospital had been cut off one that is not known. The patient was injected with the test drug may be better than we injected the test drug into the patient, and in The patient was transported to the hospital by ambulance it does not matter who drove the vehicle. The American Medical Association's manual of style holds that the active voice is generally preferred if the agent is mentioned in the sentence and the passive preferable if the agent is unknown or irrelevant, as in He was shot in the abdomen and within 10 minutes was brought into the emergency department.
The Materials and Methods section of an abstract, like the Results section, is where the passive is most appropriate. Written all in the active, such a section has to give every sentence a subject, which in medical writing is often a person, so it says that a clinical research assistant applied the questionnaires and the team statistician analysed the data. Who did those things is unnecessary information, and the passive makes the questionnaires and the correlations, the much more important information, the grammatical subjects of the sentences. The subject–active verb–object structure used in too many successive sentences makes a wooden and sometimes rather staccato text that is not comfortable to read. Where the person who did something is important, which is usually the exception, the active voice can be used, or the agent added with by in a passive sentence.
In general English, active structures are the norm, and they emphasise the agent by putting it in the subject position. Writing about science often focuses instead on who or what was in a particular state or had something done to it, so in parts of a scientific text a higher proportion of passive constructions is natural. Accounts of experiments, survey work or statistical calculations are naturally written mainly in the passive, because the reader should focus on the methods and materials rather than on who did what. In other sections of a typical paper, especially the introduction and the discussion, the writer usually needs to distinguish clearly between what the authors have thought and done and what other people have thought and done, and a mixture of constructions helps the reader decode the account reliably.
The heavy use of the passive in research writing probably results from a thematic, subject-focused pattern — the protein was isolated — rather than an agent-focused narrative one — we isolated the protein — because keeping the protein, or its equivalent, as the subject carries continuity from sentence to sentence, where we … we becomes monotonous and egotistical.
The passive also avoids assigning responsibility for the action and draws attention away from the agent, which can be used disingenuously, as in mistakes were made rather than we made mistakes, and thoughtfully, as in the prognosis of patients with this disease is generally poor rather than your prognosis is poor. A passive makes something less hostile: this bill has not been paid is softer than you have not paid this bill. It can also be used simply because it sounds better.
Orwell, who insisted "Never use the passive where you can use the active", used it in the same essay to powerful effect in his passage on pacification, transfer of population and elimination of unreliable elements. One writer holds that there it serves an important moral purpose: it focuses the reader's concern on the people who are suffering and reminds the reader that the moral significance of an act resides in its nature and expected consequences, not in the identity of the agent. The same writer adds that the device could serve an immoral purpose, as a smokescreen that lets an agent go on committing evil acts with impunity.
The case for the active
Active verbs make writing clearer, more natural and more personal, and passive verbs make it stuffy and impersonal. Passive constructions create an atmosphere of passivity, in which subjects have things done to them, while active structures create a sense of energy; cumulatively, a text with a majority of active verb structures usually gains in readability, because active structures are shorter and more direct, as in the concentration of lipofuscin increased against an increased concentration of lipofuscin was observed, and most sectors improved against improvement was seen in most sectors. Passive verbs need extra words to say the same thing.
The active voice emphasises the subject who performs the action, and the passive reverses that by emphasising the receiver, adding a form of to be, a past participle and possibly a by phrase naming the performer. A passive sentence that names the performer is wordier than the comparable active one, especially as sentences become more complex and demand more extensive revisions: the data that were collected by Smith were shown to suggest against Smith's data suggested. Because the active voice is clearer and usually more concise, it is the voice to choose unless there is a good reason to prefer the passive.
Few topics are as heatedly debated as active against passive writing, and many authors hold a strong view in favour of the tradition of passive writing. Proponents of the passive argue that the doers, such as the scientists, are not relevant and the outcome is what matters. One writer on scientific style answers that, appealing as that modesty is, it no longer complies with the scientific community's expectations, because peers now demand to know who carried out the work reported and require the transparency and clarity that the active voice gives. The active voice, that writer adds, is more than the personal pronouns I or we: it also means using active verbs in place of the ubiquitous verb to be.
The trial was conducted in 30 healthy subjects leaves open whether the authors themselves carried out the study; we conducted the trial in 30 healthy subjects settles that with the personal pronoun we; and the trial involved 30 healthy subjects, active through an active verb, wins for word economy and clarity while still not saying who carried out the study, which may become clear from the context if the authors generally use personal pronouns in the article.
Most grammarians, most readers and most authorities prefer the active voice.
Asked to compare four versions of part of a scientific article, members of the Biochemical Society voted for the most direct version, with verbs in the active voice, sentences of various lengths, and statements that are not too complex.
Medical writing is moving toward more use of the active voice, especially in reporting original research, in phrases such as "We aimed to assess the efficacy and safety of…" in The Lancet Infectious Diseases in 2011, "We report, at a median follow-up of 5.3 years, the efficacy of…" in the New England Journal of Medicine in 2016, and "We undertook a systematic review and meta-analysis…" in JAMA in 2016. One editor puts the changeover to the active voice down to the lack of single-author papers, since it is much more comfortable to write "we" than to write "I". The same editor favours a good balance between active and passive voice, because in a subtle way it reflects the balanced attitude to the subject that should be apparent throughout the document.
The same writer on scientific style calls it good advice to apply the active voice wherever possible and to limit the passive to statements that are more natural in the passive than the active. A sentence such as he was rushed into hospital by helicopter after an accident in the mountains could be rephrased in the active, especially by the nurse or pilot involved, but there is little sense in making a truly passive situation grammatically active. The rule usually applied in science writing states that no more than 30% of all verbs in an article should be passive, and in a standard research paper writers tend to use most of the allowed passive verbs in the methods section, where active sentences involving we are sometimes stilted if used throughout. The best guide, that writer concludes, is common sense, as long as active verbs are preferred in most situations and the active is used in most sentences.
The passive and comprehension
No study shows that the passive by itself reduces comprehension, and some authors' recommendation to avoid it at all costs is not supported by the research: the passive is understood just as well as the active. It is usually as easy to understand as the active and sometimes even easier. A 1979 psycholinguistic study of how people understood jury instructions established that the confusion associated with legal documents was the result of the interaction of three grammatical constructions: the passive voice, nominalisations, and negatives with qualifiers. What reduces comprehension is the interaction between the passive and nominalisations, verbs turned into nouns. A passive combined with one or more nominalisations increases not only the length of a sentence but also its grammatical complexity, and grammatical complexity reduces comprehension.
Blanket rules on voice
One editor warns against any recommendation that reduces the issue to a simple statement such as make more use of the active voice, avoid the passive voice, or even do not mix the active and passive voices. The same editor gives The Elements of Style as an example of such advice, useless for medical writers and editors, and especially bad because the five examples it gives of the passive voice are not in the passive voice. Its recommendations are not intended for people writing the range of texts the life-science field requires, and they have been indiscriminately reproduced over the past fifty years in recommendations and style guides from all sorts of other sources.
Orwell's "Politics and the English Language" and Strunk and White's The Elements of Style both urge writers to avoid the passive, yet Orwell's essay shows by example that it is sometimes appropriate, and some of the sentences Strunk and White criticise in their discussion of the active voice are not passive at all; neither the passive nor an expletive construction is bad grammar, nor always bad writing.
Whether to use the active or the passive is not just a high-level consideration, and no blanket advice on when to use one or the other is possible, because different sections of one document, such as a publication or a study report, may require different approaches. Where it is unimportant whether the reader knows who performed an action, which is usually the case when reporting results in a publication or investigations in a case report, the passive is usually the best choice. To bring immediacy and directness to a controversial discussion in an editorial, where who said what is important, a writer will probably use the active.
For objective scientific and medical writing, one editor declines to support Orwell's rule "Never use the passive where you can use the active". The passive is much maligned in style guides, most of which were not written specifically for scientific texts, and the source of the advice writers say they have been given to use more active voice is often unclear, often turning out to be a native speaker vaguely recalling a rule. That advice may be sound for a novel or for scientific writing intended to have popular appeal, but in an objective piece of scientific and medical writing the passive has an important place, and a reasonable mix of active and passive voices is best.
Newcomers to scientific writing are often told during their education that scientific texts should be written in third-person, impersonal constructions, in the past tense and in the passive voice. One adviser on writing calls that harmful advice, suspects it is responsible for much ponderous, uncomfortably encoded scientific text, and holds that it is entirely natural to mix the full range of structures and forms available in English.
How much passive medical articles carry
What a count of medical articles found
Excessive use of the passive voice is a common criticism of medical writing, yet until a 2009 study there were no published data on how often medical articles use it. That study counted the share of sentences containing at least one passive construction in 90 articles published in 2006 in JAMA, the New England Journal of Medicine and The Lancet — ten opinion papers, ten review articles and ten original research reports from each — against 30 front-page articles from The Wall Street Journal published in 2007.
Median passive-voice frequency was similar across the medical journals and across the types of medical article and much higher than in the newspaper: 20–26% across the journals and 21–24% across the article types, against 3%, with p-values below 0.0001. The only difference between medical journals to reach significance was between JAMA and the New England Journal of Medicine, 20% against 26%, and the differences between the three types of medical article were not significant. More than 20% of the sentences in the medical articles carried a passive construction, and percentages above 30% were not uncommon; the authors take these results to validate what writing experts have said for many years about excessive use of the passive in medical writing. The range within each medical journal and each type of article was very large and the lowest value in all of them was under 10%. And there was no meaningful correlation in the medical articles between passive-voice frequency and the use of first-person pronouns.
Within original research reports the Methods section carried a median passive-voice frequency of 44%, more than twice any other section, against 16% for the abstract, 15% for the introduction, 11% for the results and 15% for the discussion. Only the comparisons involving Methods reached significance — abstract against Methods, introduction against Methods, and Methods against results and discussion, all below p = 0.0001 — while every other pairing among the sections was not significantly different at p above 0.5. Against the newspaper, the abstract, introduction, Methods and discussion each differed significantly, while the results section did not. The Methods section was also the most variable, with a standard deviation of 20 percentage points against 9 to 16 for the other sections.
Far more passive sentences left the actor out than named it. The study separated two forms — "doer mentioned", as in data were collected by a member of our research team from patients with pneumonia, and "doer not mentioned", as in data were collected from patients with pneumonia — and found the second the most common form everywhere: 88% of passive sentences in JAMA, 85% in the New England Journal of Medicine and 91% in The Lancet, against 79% in the newspaper, 83% to 93% across the types of medical article, and between 86% and 100% across the sections of an original research report.
The median frequency of active sentences containing I or we was 2% to 4% in the medical journals against 0% in the newspaper, but the ranges were wide, up to 34% in the medical articles against a maximum of 4% in the newspaper, and within original research reports the median ran from 3% in the results to 12% in the introduction. The widest range was in the Methods section, from 0% to 70%. The scatter of the two variables shows no meaningful relationship, and its low p-value is misleading: excluding the few articles whose first-person frequency exceeds 15% raises it to 0.9.
A word processor's grammar checker will report a passive-voice percentage far more easily than a manual count. Against the manual count over the same 90 medical articles, the grammar checker in Microsoft Office Word 2003 had a median difference of only 3 percentage points, overcounting slightly, but the differences ran from 25 points under to 19 points over, with a standard deviation of 9, so the program made major errors in a small percentage of articles, and no feature of an article or its formatting explained them. The manual method is time-consuming but manageable and brings considerable value to the finished product.
Why writers use the passive
The debate over the role of the passive voice has two sides. One argues that medical writers should minimise it because it is less clear, less forceful and more verbose than active alternatives; the other that it should be the predominant style of most medical writing because it reflects the objectivity of the scientific method, avoids using a first-person pronoun in a way that would be impolite, and conforms to the style with which medical peers are most familiar. No publication guideline mentions goals or limits for the passive, and some of the most prestigious references are worded in a way that may encourage authors to use it whenever it is acceptable: the AMA manual of style says that "Authors should use the active voice, except in instances in which the actor is unknown or the interest focuses on what is acted on."
The study's authors found no survey or consensus statement addressing why authors of medical journal articles use the passive so often, and writing experts have suggested five reasons: to reflect objectivity, to avoid first-person pronouns, to appear scholarly and sophisticated, to avoid responsibility, and to conform to established writing style. The study's authors found no publication that recommends the passive specifically to demonstrate objectivity, and the experts who take a position criticise that argument as fundamentally incorrect and an obstacle to clear writing, Lester King holding that objectivity in science is in large part a myth and urging its devotees to apply themselves to clear expression rather than to indefensible dogma. Some journals prohibit first-person pronouns in the abstract, but no publication guideline requires the passive where an active construction would be grammatically appropriate, and many writing experts are passionate about changing the convention of avoiding active sentences with a personal pronoun, one urging young scientists not to be afraid to name the agent of an action, even when it is I or we. Another theory holds that the habit may come from the erroneous idea that first-person pronouns are somehow impolite, whereas one clinician-author holds that using the active, first-person voice is a way of respectfully taking ownership of one's words. To writers who use the passive to appear scholarly and sophisticated, the study's authors endorse the advice that formal, unassertive language does not make a work scholarly or scientific; it just makes it lifeless and vague. The writing on the urge to avoid responsibility that the study reviewed points to the psychological urge to avoid going out on a limb as a reason for writing in the passive. The same clinician-author suggests that the passive may simply seem easier, that it may let writers take a little less responsibility for their thoughts, and wonders whether writers are more likely to use the active when confident of positive results and the passive when confidence is lacking or the results are equivocal.
Conformity to established style is the reason the study's authors accept: frequent use of the passive is a habit writers acquire by mimicking the style of their journals and learning from reviewer feedback, without understanding the alternatives. Some suspect that heavy reliance on the passive is more a matter of tradition than a formal requirement, and it has been called a fashion the medical community is in the process of rejecting. Young researchers often report that their professors have changed their vigorous sentences back into the passive because that is the way science is written, and the study's authors had a journal reviewer recommend revising a manuscript to use the passive as much as possible because it would make the paper sound better. Another author reports a journal editor accepting a paper on condition that every statement be changed to the passive, and the printed version lacked the punch of the original and took more words to say the same thing.
A 10% ceiling and its critics
The study's authors take the width of the range within each journal and type of article to suggest that writing with a high passive-voice frequency is a style of choice rather than a requirement for publication. Their data, they hold, suggest that 10% is a reasonable upper limit for all types of medical article, because multiple articles in every analysis met it.
The study's authors make four recommendations:
- Replace "doer mentioned" passive constructions with the active.
- Limit the passive to the Methods section of an original research report, where describing who did an action distracts the reader from what was done methodologically.
- Use passive-voice frequency in the whole paper as an endpoint for evaluating the quality of the writing.
- Medical journal editors should make a passive-voice frequency of 10% or less a publication requirement for all types of article.
Every resource on medical writing the authors cite gives situations where the passive is appropriate, chiefly when the action is more important than the doer, but writers concerned about using it too often need more detailed instructions. The real question, on their reading, is not whether authors use the passive too often but why prestigious journals routinely publish articles that would be shorter, clearer and easier to read if revised to minimise passive constructions. Several veteran editors attribute it to a change in the role academic publications play and to editors' workload, holding that many medical articles seem to be written to be published and cited but not to be read. Editors say they are overwhelmed by the volume of articles in bad English, and many of them write editorials urging simpler English or run training sessions, but nothing changes.
A response to that study accepts that the passive is often undesirable and overused in scientific publications, but argues that its conclusions are not supported by decades of more informed research. It holds that comparing the percentage of passive sentences in three medical journals with the same percentage in a business newspaper is meaningless, because the purpose and context of journals and newspapers differ greatly, even if their readership does not.
The response calls the recommendation of a 10% ceiling unsupported by any research and probably counterproductive: it agrees that the passive has a place in scientific publications, but holds that identifying an evidence-based maximum percentage of passive sentences is impossible. Against using passive-voice frequency as an endpoint for evaluating the quality of writing, it argues that this assumes quality is defined by that frequency, which it is not: the quality of writing is traditionally evaluated by how well readers understand, recall, find and use information, not by the countable characteristics of the text. It adds that arbitrarily specifying textual characteristics, such as sentence length or the average number of polysyllabic words per sentence, has never been successful and is often counterproductive.
Voice misagreement
However common the passive, some voice misagreements impede immediate comprehension. The commonest, particularly among writers whose first language is not English, is an inverted relation between subject and direct object, and such writers also commonly put transitive phrasal verbs such as consists of into the passive. Why some verbs are ungrammatical in the passive seems to have no explanation yet.
A revision changes the voice, from active to passive or from passive to active, or inverts subject and direct object to correct an active-voice misagreement, as in these cases:
- In a Methods section, interviews focused on genetic, medical, and family history gives an active phrasal verb to an abstract, inanimate subject — a personification, the classic example being this paper discusses — and is revised into the passive: interviews were focused on genetic, medical, and family history.
- Also in Methods, the system was consisted of six main features puts what is probably a transitive phrasal verb into a passive that is ungrammatical; the active the system consisted of or the system comprised is correct.
- In Results, the concentration of DHA was increased from the maternal to foetal liver makes it appear that the authors were responsible for the increase, because was increased can be read as a passive verb, denoting an external action, or as a linking verb with an adjectival participle, denoting an observation. The intransitive DHA concentration increased removes the ambiguity, and the existential, somewhat wordy there was an increased DHA concentration shifts from a narrative to a descriptive pattern befitting an observation in Results.
- In an Introduction's hypothesis, β-catenin may involve early liver development sets up an illogical relation between subject and direct object; either the passive β-catenin may be involved in early liver development or inverting subject and object, early liver development may involve β-catenin, repairs it.