Language Log

More passive tense

As Geoff Pullum has just noted, on NPR's Morning Edition of 5/11/07, Steve Inskeep, interviewing U.S. Army Gen. Dan McNeill, pursued a question about current U.S. policy in Afghanistan:


Ok, a mistake.  But WHICH mistake is it: passive tense for passive voice (a mistake in grammatical terminology, as Geoff thinks), or passive tense for past tense (a mistake in word retrieval)?


Inskeep goes on:


The accent on the past tense form was (and the absence of anything that looks like a passive) suggests that Inskeep was aiming for past tense and pulled out passive instead of past -- an easy mistake to make, since the words are phonologically similar and also (both being grammatical terms) semantically similar.  I am myself given to saying verb for vowel, and vice versa; I'm not confused about the concepts, just retrieving the wrong v-initial technical term of linguistics.

A few other examples of doubly motivated retrieval errors:


Other retrieval errors are mostly phonological in motivation (pollution for pollination, of flowering plants by bees) or mostly semantic ("this particular noun [consonant] cluster").  Some are mostly phonological in motivation, but also have probable triggers in the context: my "a preposition [presentation] of brutal masculinity", about gay porn, but not long after a discussion of stranded prepositions; a friend's "a biological orgasm [organism] that reproduces by...", an error undoubtedly facilitated by planning for the word reproduces.

Such errors suggest that the "mental lexicon" is organized according to both phonological and semantic properties (as well as syntactic properties: retrieval errors almost always preserve syntactic category) and that retrieval is sensitive to the context in which language is produced, in that errors can be facilitated by what's "in the mind" of the speaker or writer.

But wait!  There's more.  This is not our first brush with passive tense here at Language Log Plaza.  If you google on {"passive tense"}, you'll get a distressingly large number of hits -- in the tens of thousands -- almost none of them (if any) for word retrieval errors; instead, there are a few for lists of the tense forms of the passive voice in one language or another (more on this below), but most of the hits are, alas, for errors in grammatical terminology (passive tense for passive voice).  And as the top-ranked hit you'll probably get Geoff Pullum's earlier plaint about an occurrence of this error in The Economist, where they really should know better.  Soon you'll get to Mark Liberman's response to someone asking for advice on "how to avoid passive tense".

Particularly distressing is the fact that so many of the errors in grammatical terminology you google up are, omigod, on pages giving advice to writers or students.  A few examples:


At this point you suspect that an awful lot of people are using tense to mean something like 'verb form'.  For the English passive even this isn't quite right, since the passive isn't a verb form but a syntactic construction (most commonly composed of a form of the verb BE and a VP with its head verb in the past participial form).  But let that pass.  There are bigger things to worry about. 

Look at English-Zone's "Active and passive tenses chart".  English-Zone (which describes itself as "the BEST English-Learner's site on the 'Net!") talks about active and passive VOICE, so there's one problem avoided.  The chart begins with the "simple present" and the "simple past", for which the "forms" of the passive given are, respectively:


This is already fairly silly, since it treats these forms as fresh things to memorize.  But really all the learner needs to know is (a) that (more or less in English-Zone's terms) the passive is composed of head verb BE + past participle, and (b) that the present forms of BE are am/is/are, the past forms was/were.  Only the first of these is new information for the learner.

The chart then continues in this vein, giving the "present and past continuous (progressive)", "present and past perfect", and "future" forms of the passive.  These forms are entirely systematic, being constructed on the basis of fact (a) and the principles governing the corresponding forms of the active.  English-Zone has managed to turn a pretty simple system into a massive pile of unrelated formulas.

Put that aside, and note that this is supposed to be a chart of TENSE FORMS, where the tenses include: simple present, simple past, present progressive, past progressive, present perfect, past perfect, and future.  They're all "tenses", presumably because their associated meanings have some temporal component.  This is not some idiosyncrasy on English-Zone's part; such a use of tense is all over the pedagogical literature for English and dozens (if not hundreds) of other languages, and it's not unknown among language professionals:


Some treatments of English go on from the list above to the future progressive, future perfect, present perfect progressive, past perfect progressive, and future perfect progressive (in the passive: will have been being given!).

Linguists factor the grammatical categories into tense (for English, present, past, sometimes future, but see below), having to do, in many of its occurrences, with the location of situations in time, and aspect (for English, unmarked vs. perfect and progressive, with the possibility for these latter two to co-occur), having to do, in many of its occurrences, with the internal organization of situations over time.  There are then twelve possible combinations, enumerated in the preceding paragraph.

Given this extended sense of tense in many quarters, it's probably no surprise that some people have extended it to cover voice as well as tense and aspect. 

There are two different extensions of careful technical vocabulary here: tense is extended to cover all sorts of verbal categories (often realized by morphology on the verb); and form is extended to cover multi-word combinations -- periphrastic expressions -- as well as single words.  These extensions are in principle independent of one another. 

Extended sense of tense.  I am sorry to report that you can find references to infinitive, conditional, subjunctive, negative, causative, permissive, inceptive, plural, imperative -- Mark Liberman complained here about this one three years ago -- and interrogative tenses in one language or another.  No doubt there are many more verbal categories that have been labeled "tenses", but I gave up my search in sorrow after finding these.

What we have here is a terminological morass.  The way out is to distinguish different types of categories; these are customarily given labels that suggest something about the semantics associated with each type: tense, aspect, voice, mood, mode/modality, polarity, finiteness, evidential status, and the like. 

These types can be relevant for some languages and not for others.  For any particular language, within each type we distinguish various forms belonging to  that type, and those too are customarily given labels that suggest something about the associated semantics: present vs. past tense, for example. 

All of this labeling is problematic: the "same" type in different languages will cover rather different territory, and that's also true of the "same" form in different languages; and within a single language, almost all forms are multifunctional (associated with a variety of meanings), so that choosing a label for any particular form is a somewhat arbitrary process.  For these reasons, for some time I've been advocating assigning arbitrary labels whenever we need to be absolutely clear about how a language works; see my posting on the "subjunctive" in English for some development of this proposal.

But for informal discussion of English (or any other language), traditional labels will usually do.  Still, they should be used accurately.  "Passive tense", "infinitive tense", and the like are not accurate uses.

Extended sense of form.  There are two motives for extending the term form to multi-word combinations: semantic parallels between periphrastic and inflectional realization in one language (for instance, between periphrastic future will see and inflectional present see(s) and past saw, and between periphrastic to see and inflectional seeing); and periphrastic realization in one language corresponding to roughly equivalent inflectional realization in another (for instance, the English passive construction vs. the inflectional passive of many other languages, among them Latin).

Both of these moves are unwise, because they take us down a slippery slope.  If we treat the expression of future time via the modal will ("I will be your assistant") as a tense, then why not say the same for the futurate quasi-modal be going to ("I'm going to be your assistant"), the future-plan modal be to ("I am to be your assistant"), and the inceptive future quasi-modal be about to ("I am about to be your assistant")?  And the future-in-past modal would ("I would soon see why the idea was problematic").  And we pick up at least two more past tenses, expressed via the quasi-modal used to ("I used to be your assistant") and the modal would ("When I was a child, I would always help set the table").

Somewhat more subtly, if we're picking out tenses on the basis of meaning, why don't we say that "I leave at noon tomorrow" and "I am leaving at noon tomorrow" illustrate two more future tenses (rather than saying that they are futurate uses of the present tense)?  And why don't we say that English has several more tenses -- for instance, a gnomic tense, for (putatively) universal truths ("Ice melts at 32 degrees F."), a narrative tense, as in "A panda walks into a bar,...", and still others?  The usual way people talk about these phenomena is as "uses of the present tense", and that's basically right.  But how do we avoid calling these things different "tenses" simply because they have different kinds of temporal reference?

In fact, calling the progressives and perfects of English "tenses" follows from this identification of "tense X" with "meaning Y": unmarked aspect in "I see you" (a state description) describes something happening at the moment, but so does progressive aspect in "I am jumping over an anthill" (an event description); unmarked aspect in "I will go to Hawaii" describes a future event, but so does perfect aspect in "By the time you arrive, I will have gone to Hawaii" (a past in the future).

The multiplication of verbal categories will continue if we analyze English on the basis of the inflectional categories of other languages, where voice, mood (declarative, imperative, interrogative, subjunctive, conditional, etc.), mode/modality (expressing necessity, obligation, permission, possibility, likelihood, intention, and much more), polarity (postive vs. negative), finiteness, evidential status, and various other meaning ranges can be expressed via inflectional morphology.  Going down this road gives English at least one "verb form" for each of its modal verbs, plus more for the constructions with have to, want to, ('ve) got to, (had) better, etc.; two passive "forms" (for the be and get passives); a whole pile of moods; and much more.

This is craziness.  The way out that I favor is to split the description of the data into two parts: an inventory of the inflectional morphology of the language and an inventory of its syntactic constructions.  The forms in the inflectional inventory are "bits of stuff" that can be used in the constructions ("grammatical words" are another kind of stuff, and there are still other kinds).   Each construction is a set of conditions on the composition of expressions -- think of the conditions, taken together, as a recipe for putting a construction together -- associated with a meaning for the construction as a whole. 

Such an approach gets us down to two verb forms for English that can be referred to, informally, as tenses; call them Form:R and Form:T.  They can be used in constructions with present-time meaning and past-time meaning, respectively -- this is, in some sense, their customary use -- but they can also be used in constructions with other meanings, and the temporal-location meanings can be expressed by other means. 

This isn't the place to develop an analysis of English along these lines -- I've probably strained your patience already -- but it should be enough to show that we don't have to slide down that slippery slope.

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at May 25, 2007 04:59 PM