Zero vs. "that" relatives--when linguistic evidence and popular belief conflict

Herb Stahlke hfwstahlke at GMAIL.COM
Tue Jan 20 05:17:05 UTC 2009

Let me apologize first for the length of this posting.

A variant on this thread has come up on the ATEG list (NCTE's Assembly
for the Teaching of English Grammar), and it has raised a question for
me that I don't know how to answer.  On morphosyntactic grounds, the
"that" of "that" relatives is clearly a subordinating conjunction and
not a pronoun.  The best arguments for this can be found in
Jespersen's Modern English Grammar on Historical Principles and in
Huddleston&Pullum, who provide a few arguments that Jespersen missed.
However, a number of ATEG listers, who are largely school and college
English teachers with a few linguists thrown in, are insisting that
the sense of relative "that" is pronominal.  They present no evidence
beyond strong feeling, rather like the conviction that develops around
an eggcorn.  I know of know way of testing or demonstrating the
validity of this sense, but I know of one other grammatical case where
a similar conflict between morphosyntactic evidence and popular
conviction seems to be leaning in the favor of popular conviction.
This is the status of what is popularly identified as a subject
pronoun in Yoruba.  I've taken the liberty of including a portion of
my latest posting on the matter.

I'm not particularly interested in the linguistic problem of
determining what the form is, which is, in both cases, pretty
straightforward, but rather in possibility of a change taking place
that sets aside all the structural evidence in favor of popular


>From the ATEG posting:

On the other hand, Edith raises the interesting question of whether
"that" could be changing its function from subordinator to pronoun.  I
certainly don't reject that possibility.  Such grammatical change is
not unusual, and more radical cases abound.  My problem with the
claim, though, is again an absence of evidence beyond, present company
excepted, naïve assumptions about grammar.  A long time ago, 1973, I
think, I published a paper that included an internal reconstruction of
the Yoruba preverbal morphemes, including the subject pronouns. This
is a fairly complex problem, and internal reconstruction is a
historical linguistic methodology for extrapolating earlier stages of
a language from synchronic alternations and irregularities.  On
historical grounds, what every grammar and every Yoruba teacher I had
called a third person singular pronoun was nothing of the sort.
Historically I could explain every phonological and morphological
property of the form, and none of it had any historical source in the
pronoun systems.  Rather, the third singular in the present
affirmative indicative was a zero form, just as it was in all the
other subject pronoun sets in the language, and there were different
paradigms depending on tense, modality, and negation.  In spite of a
total lack of morphosyntactic evidence that the form was a third
singular subject pronoun and in the face of overwhelming evidence to
the contrary, grammar writers, Yoruba language teachers, and speakers
of the language who also spoke English insisted that the word was a
pronoun and did in fact translate as English he/she/it (the language
is totally without gender marking).  On the basis of that, I can
accept that for modern speakers that form has changed from what it was
historically, a combination of two different morphemes neither of them
pronominal, to a subject pronoun.

Edith suggests that something similar has happened in English with
relative that, but in the English case the grammars are far from in
agreement, and the best of them, Jespersen and Huddleston&Pullum,
reject the idea.  Grammar teachers are rather more in agreement, but I
suspect that's because of what they've been taught, which tends not to
come from the best sources.  So the change may be in progress, but
there's no way of detecting it yet.

On Mon, Dec 29, 2008 at 1:53 PM, Benjamin Zimmer
<bgzimmer at> wrote:
> ---------------------- Information from the mail header -----------------------
> Sender:       American Dialect Society <ADS-L at LISTSERV.UGA.EDU>
> Poster:       Benjamin Zimmer <bgzimmer at BABEL.LING.UPENN.EDU>
> Subject:      Re: Zero vs. "that" relatives
> -------------------------------------------------------------------------------
> On Mon, Dec 29, 2008 at 1:23 PM, Arnold Zwicky <zwicky at> wrote:
>> > but the levels of "that" were low in the 1920s-1940s (lower than in
>> > the 2000s, in fact).  is there any reason to think that prescriptive
>> > pressure increased in the 1950s-1970s?
>> >
>> > here it would be nice to have data from a source other than Time, to
>> > find out whether the change was the result of changing editorial
>> > practices at the magazine.
> My feelings exactly. It might be hard to extrapolate the Time data to
> journalistic usage more generally, since the magazine has been so
> stylistically idiosyncratic over the years. Most notorious is their
> "inverted syntax," which was finally phased out in 2007:
> --Ben Zimmer
> ------------------------------------------------------------
> The American Dialect Society -

The American Dialect Society -

More information about the Ads-l mailing list