The passive. Is it really harder to understand than the active?

People who give advice on good writing often claim that sentences containing verbs in the passive form are harder for listeners and readers to process than sentences containing verbs in an active form. A paper I read recently provides evidence that these claims are overstated.

The paper is Processing unambiguous verbal passives in German, by Nino Grillo, Artemis Alexiadou, Berit Gehrke, Nils Hirsch, Caterina Paolazzi, Andrea Santi (2019), Journal of Linguistics, vol 55 Issue 3.

In this post, I cover:

  • adjectival passives and verbal passives
  • feautures of passive: English
  • features of passive: German
  • empirical test
  • findings
  • digression on Italian passives
  • conclusion

Adjectival passives and verbal passives

English has both verbal passives and adjectival passives:

  • verbal passive: the verb is a past participle, and is combined with a form of the passive auxiliary be. Example (1) shows an active sentence (1a) together with the corresponding long passive (1b, naming Mary as the agent who performed the act) and short passive (1c, not naming the agent).
    (1a) Mary broke the door.                        [active]
    (1b) The door was broken by Mary.        [long passive]
    (1c) The door was broken.                       [short passive]
  • adjectival passive. The past participle functions as an adjective, not as a verb. The adjective is linked to the subject of the sentence with a copula (linking verb): typically be. Example (2) is an adjectival passive.
    (2) The door was broken.                        [adjectival passive]

Although example (2) looks identical to example (1c), they are different. Example (1c) describes the event that led to the door becoming broken. In contrast, example (2) describes the state the door is in (broken), but does not describe the event of breaking the door.  

Features of passive: English

Features of the verbal passive (and of the adjectival passive) in English include the following:

  • English verbal passives of verbs referring to states (stative verbs) are often identical to an adjectival passive—and initially readers are often more likely to interpret them as adjectival than as verbal. Then, if readers encounter a reference to the agent (eg by the singer) later on, they seem to reanalyse the sentence as verbal. The time needed for re-analysis at that point makes it take longer to parse the clause than the corresponding active clause.
  • when made into a passive, stative verbs must be ‘coerced’ into an ‘inchoative’ meaning that refers not to a continuing state but to entry into that state. Coercion adds extra complexity, which seems to increase processing time.
  • in both passive and progressive actives, the verb follows an auxiliary (be). In contrast, the simple past active has no auxiliary. By showing that the next word is probably a verb, the auxiliary seems to help readers process the verb more quickly.
  • in verbal long passives, the preposition by precedes the noun referring to the agent. By showing readers that a noun phrase follows, the preposition seems to help readers process the noun more quickly.
  • the adjectival passive does not normally name the agent. But this is possible for some types of noun (bare nouns, weak definites, weak indefinites, generics).

Some of those features are important in the tests described later in this post.

Features of passive: German

In English, the auxiliary be accompanying a past participle in a verbal passive is the same as the copula be accompanying an adjectival passive. In contrast, the German verbal passive uses an auxiliary (werden ‘become’) differing from the copula (sein ‘be’) used with an adjectival passive. Thus, in German there is no early ambiguity between verbal passives and adjectival passives.

Examples (3, English) and (4, German) illustrate these differences. Example (3a) is a long passive and example (3b) is a corresponding active sentence.

(3a) The guitarist was rejected by the singer.       [long passive]
(3b) The singer rejected the guitarist.                    [active]

For German, example (4a) is a long passive and example (4b) is the corresponding active. German has no form like the English progressive.

(4a) Der Gitarrist wurde von der Sängerin zurückgewiesen.
[long passive]
The guitarist was by the.DAT singer rejected.
‘The guitarist was rejected by the singer.’

(4b) Die Sängerin hat den Gitarristen zurückgewiesen.
[active, present perfect]
The singer has the guitarist.ACC rejected.
‘The singer rejected the guitarist.’

abbreviations: DAT = dative. ACC = accusative
zurückgewiesen is the past participle of zurückweisen. The participle is formed by adding the prefix ge before the root, and changing the vowel of the root.

Features of the German passive include the following:

  • German verbal passives of stative verbs can refer to states without being coerced into an inchoative meaning.
  • by using different auxiliaries (werden vs sein), German passives avoid early ambiguity between verbal and adjectival passives. These auxiliaries appear early in the sentence. (examples (4a) and (4b))
  • as the passive examples (3a) and (4a) show, the German word order differs from the English word order: the past participle (zurückgewiesen) appears at the end of the sentence, after the prepositional phrase referring to the agent (von der Sängerin). This is the same position as in the present perfect (German example (4b)).
  • in verbal long passives, the preposition von (‘by’) precedes the noun referring to the agent. By showing readers that a noun phrase follows, the preposition seems to help readers process the noun more quickly.
  • although spatial or temporal event modifiers may intervene before the verb in the active (after the direct object) or in the passive (after the prepositional phrase headed by von), these modifiers are less likely to occur in the passive.
    If present, those modifiers delay the appearance of the verb. That delay makes it more likely that the next word will be the verb. Thus, after a von-phrase, readers may be able to predict that the next word will be a verb (past participle) more confidently than they can after a direct object. If they predict it more confidently, it seems they process the verb more quickly.

Some of those features are important in the tests described later in this post.

Empirical test

After summarising previous research, the paper discusses 4 empirical tests: 2 tests on English verbs (reported just by summarising previous work by some of the authors); and the other 2 tests on German verbs (reported in detail). Each test asked volunteers to read several sets of sentences, with each set containing 2 or 3 sentences.

Tests: English

  • Test 1 examined English verbs expressing an event (eventive verbs). This test consisted of sets containing 3 English sentences along the lines of examples (5a)-(5c).
    Example (5a) is the same as example (3a).
    Example (5b) is the same as example (3b), but with the roles (and, hence, order) of the nouns guitarist and singer swapped. That swap minimises the differences between examples (5a) and (5b).
    Example (5c) is the present progressive form of the active. The progressive uses the same auxiliary (be) as the passive, but combined with a different participle (present instead of past).
  • Test 2 examined English stative verbs. The test sets contained 2 sentences along the same lines as examples (5a) and (5b). There was no equivalent to example (5c) because stative verbs have no progressive form.

(5a) The guitarist was rejected by the singer.  [long passive]
(5b) The guitarist rejected the singer.              [active, simple past]
(5c) The guitarist was rejecting the singer.      [active, present progressive]

Tests: German

  • The other tests examined German eventive verbs (test 3) and stative verbs (test 4), using test sets along the lines of examples (6a) and (6b), similar to examples (4a) and (4b). Example (6b) swaps the nouns Gitarrist and Sängerin for the same reason as example (5b).
    Example (6b) uses the present perfect (rather than the simple past)—partly because the German perfect often corresponds to English simple past, but mainly because the German perfect contains an auxiliary and so using this form means that the passive and active forms both contain an auxiliary.

(6a) Der Gitarrist wurde von der Sängerin zurückgewiesen.          [long passive]
The guitarist was by the.DAT singer rejected.
‘The guitarist was rejected by the singer.’

(6b) Der Gitarrist hat die Sängerin zurückgewiesen.                        [active, present perfect]
The guitarist has the singer.ACC rejected.
‘The guitarist rejected the singer.’

The sentences actually used in the tests included some additional modifying material (adjectives, additional subordinate clauses). For simplicity and to highlight the different sentence patterns more clearly, examples (5) and (6) exclude those modifiers.

Findings

The tests investigated both:

  • comprehension—including whether participants correctly identified the agent performing the action and the person undergoing the action.
  • reading time—the time taken by the participant at each word to read that word.  

Comprehension

Participants understood passives just as well as actives. This was the case for both eventive and stative verbs in German and also for eventive verbs in English. But for stative verbs in English, participants understood actives better than passives.

The authors conclude that:

  • passive verbs are inherently no more difficult to understand than active verbs.
  • for English (but not for German), people are more likely to misunderstand stative verbs in the passive than in the active. This is the case even though the presence of the auxiliary and of by speeded up the initial stages of parsing for passives.

Misunderstanding passives of English stative verbs

The authors suggest 2 reasons why people are more likely to misunderstand English stative verbs in the passive:

  • in forming a passive, English stative verbs must be ‘coerced’ into an ‘inchoative’ meaning that refers not to a continuing state but to entry into that state. That coercion adds extra complexity, and also increases processing time.
  • verbal passives of English stative verbs are often interpreted initially as adjectival. The reader then reanalyses them on encountering the reference to the agent (eg by the singer). Reanalysis makes processing more complex, and increases the risk of mis-understanding. As noted below, this effect also influences reading times at the time when reanalysis occurs.   

Neither of these 2 reasons exists in German. German verbal passives of stative verbs:

  • refer to states without being coerced into an inchoative meaning; and
  • avoid early ambiguity between verbal and adjectival passives by using different auxiliaries (werden vs sein). These appear early in the sentence (examples (4a) and (4b)).

Reading time

For eventive verbs:

  • in English, reading times at the verb for the passive were significantly shorter than for simple (non-progressive) actives, but did not differ significantly from reading times for progressive actives. This seems to be because an auxiliary (be) precedes both passive and progressive actives, but there is no auxiliary in the simple past active. By showing that the next word is probably a verb, the presence of the auxiliary seems to help readers process the verb more quickly.
  • in German, reading times at the verb were shorter for stative verbs, although there is no statistically significant difference in reading times at that point for eventive verbs. I discuss this point below.  
  • in German, reading times within the 2nd noun phrase in the sentence were shorter for the passive (where der Sängerin ‘the singer’ follows von [‘by’] in example (6a)) than for the present perfect (where die Sängerin ‘singer’ is a direct object in the accusative case in example (6b)). The authors suggest that the preceding von helps readers predict more confidently that a noun phrase follows than they can when the preceding item is a direct object; thus, they process the noun phrase more quickly.

In the case of stative verbs:

  • in English, reading times at the verb were shorter for the passive than for the active—as was also the case for eventive verbs.
  • in English, reading times within the initial regions of the 2nd noun phrase were shorter for the passive than for the active—this was also the case for eventive verbs. When readers encounter the preposition by, this primes them to expect a noun phrase next. That expectation helps them process the noun phrase more quickly.
  • in English, reading times for the passive in a later region of the 2nd noun phrase were longer. It seems this was because of the factors already mentioned above for comprehension: (1) coercion of English statives into an ‘inchoative’ meaning; and (2) reanalysis from adjectival passive to verbal passive after (although not instantly after) encountering the agent.
  • in German, reading times were shorter within the 2nd noun phrase, as was also the case for eventive verbs. This seems to because the preceding von helps readers predict more confidently that a noun phrase follows.
  • in German, reading times were also shorter at the verb. I discuss this point below.

Shorter reading times for verbs in German passives?

As noted above, reading times at the verb were shorter for German stative verbs, although there is no statistically significant difference in reading times at that point for eventive verbs.The authors suggest that this difference between findings for stative and eventive verbs may arise from the test having too little statistical power.

There seem to be 2 competing factors at play here:

  • As examples (6a) and (6b) above show, the German verb (past participle) itself occurs later in the sentence in the same place in both active and passive. Because that location is the same in both cases, I wouldn’t expect that location to lead to a difference in reading times.
  • as noted above, spatial or temporal event modifiers are less likely to intervene before the verb in the passive (after the prepositional phrase headed by von) than in the active (after the direct object). Thus, the authors suggest, the presence of von allows readers to expect more confidently that the verb will follow, and so process the verb more quickly.
    (No such difference can arise in English because the verb (rejected in example (5)) precedes (rather than follows) both the agent (by the singer in the passive) and the direct object (the singer in the active)).

Overall reading times

The authors tested local reading times at each individual word in the test sentence. For example, they tested whether reading times were faster:

  • for verbs if they follow an auxiliary; and
  • for nouns if they follow the prepositions by or von.

As far as I could tell, they did not test whether reading times were also faster:

  • for the sequence auxiliary+verb than for a verb that has no preceding auxiliary.
  • for the sequence by / von + noun, than for a noun without that preceding preposition.

If such a test were to be carried out, I’m not sure what implications its results would have.  

Digression on Italian passives

The authors suggest that it might be worthwhile to test Italian passives because they display some properties like English passives and other properties like German passives:

  • Like English but unlike German, Italian uses a single verb meaning ‘be’ (essere) both as a copula with adjectival passives and as the auxiliary for most verbal passives.
  • Like German, Italian has an auxiliary (venire ‘come’) that can be used in verbal passives instead of essere (‘be’).
    But, like English and unlike German werden-passives, Italian venire-passives force stative verbs into an inchoative interpretation (referring to a change of state).
    Venire-passives are not available for all stative verbs.

Conclusion

In summary, readers take longer to process some English clauses containing verbal passives of stative verbs. But fact that the clauses are passive is not the reason for the slower processing. The causes of slower processing seem to be that:

  • readers mis-interpret those clauses initially as containing adjectival passives and then need to back track to re-analyse them.
  • readers need to coerce those stative verbs into a meaning that denotes a change in state.

There are both good reasons to use the passive and good reasons not to use it. From the evidence discussed in this post, the good reasons for not using the passive do not seem to include more difficult comprehension or slower reading.

For more on the passive, please see https://languagemiscellany.com/tag/passive/

Leave a comment

Your email address will not be published. Required fields are marked *