LINGUIST List 12.2811

Fri Nov 9 2001

Sum: Future Markers in English

Editor for this issue: Marie Klopfenstein <>


  1. Benedikt Szmrecsanyi, questionnaire on future markers in English

Message 1: questionnaire on future markers in English

Date: Fri, 9 Nov 2001 19:50:11 +0100
From: Benedikt Szmrecsanyi <>
Subject: questionnaire on future markers in English

Hello all,

as you might remember, a short while ago I asked the list to participate in
an online questionnaire which dealt with the expression of futurity in
English. Here is the summary which I promised to deliver. The questionnaire
is still online at , so you can go back and
check it out for reference purposes.

I should perhaps mention that the better part of my study relied on
empirical findings stemming from a corpus based approach, i.e.
on the computerized analysis of ca. 10 million words of spoken English (both
British and American).
With the questionnaire, I set out to replicate my earlier, corpus-based
findings through elicitation tests and acceptability judgements.

In all, the questionnaire yielded a total of 274 respondents, 196 of which
were native speakers of English. For methodological purposes, I had to
exclude all non-native speakers' data from analysis in the end. 57% of
respondents were male, 43% female, and the mean age of respondents was 36
Respondents were rather well-educated individuals, the vast majority of
which holds academic diplomas. Also, for obvious reasons, the questionnaire
oversampled respondents from North America, which cannot be helped given
that so many people are actually living there :-).

With questions I.1 and I.5, I set out to test whether Haegeman's (1989) and
my own claim that BE GOING TO (i.e., both variants, "gonna" and full "going
to") and WILL/SHALL (i.e., the variants "will", "'ll", "won't", and "shall")
are truth-conditionally equivalent is correct. Indeed, this claim was borne
out by the data, with the vast majority of respondents rating I.1 and I.5 as
unacceptable. From this, it can be inferred by virtue of logic and of formal
semantics that the two paradigms are semantically interchangeable, while, of
course, nothing is thus said about their pragmatics. A question that dealt
with pragmatics was II.5, where Hall (1970) suggested that "will" would be
preferable to "be going to" because of the implicatures it carries.
Respondents conformed with Hall's (1970) claim. They did not, however, seem
to agree to Binnick's (1971) assertion that future markers would be
interchangeable in II.1; something - possibly pragmatic factors - in II.1
induced respondents to prefer "will" over "going to" in that specific
sentence. In summary, because material like "I am going to marry, but I will
not" was rated nonsensical by a very vast majority of respondents, it is
clear - at least to me - that both ways of expressing futurity mean -
semantically - the same thing.

 Results from the section of the questionnaire dealing with "'ll not"
revealed that while respondents judged sentences containing this
construction as decently acceptable (meaning that "'ll not" is clearly not
ungrammatical), they would not frequently choose "'ll not" when being able
to choose from among other negated future markers (a minor exception to this
finding was II.7, where "'ll not" was selected by relatively many
respondents). Moreover, I could show that those respondents who opted for "'
ll not" were disproportionally often from Scotland and also England, while
US respondents appeared to avoid "'ll not". As to other negated future
markers, there is evidence in the data that "will not" was chosen rather
infrequently by respondents too. While "won't" was clearly the preferred
negated future marker for British respondents, US respondents actually
tended to slightly prefer "not gonna" over "won't" (this became especially
evident in question II.6). These findings, too, were in line with the
findings from the corpus linguistic part of this study.

 The part of the questionnaire that addressed conditional clauses (or
if-clauses) as well bolstered confidence into the corpus-based findings.
Contrary to what Binnick (1971) claimed, "I'll get back to Tom, if you
really want me to" was preferred by respondents over "I'm going to get back
to Tom, if you really want me to". As, however, suggested by Binnick (1971),
"I'll talk to Sam if you ask me to" was more popular among respondents than
"I'm going to talk to Sam if you ask me to". This is in accordance with my
earlier finding that WILL/SHALL markers are very frequent in
main clause slots, and especially conditional apodoses. With regard to
protases, I could confirm Comrie's (1985) assertion that BE GOING TO,
under certain circumstances, is preferable to WILL/SHALL in conditional
subclauses. Based also on other material on conditional sentences in the
questionnaire (i.e., III.1 and III.4), there was, then, a clear tendency for
respondents to prefer WILL/SHALL in conditional main clauses and to prefer
BE GOING TO in conditional subclauses. These preferences, again, conform
with the findings that stemmed from the corpus-based approach to future

 In the two questions in the questionnaire that addressed interrogative
sentences, respondents clearly preferred WILL/SHALL in main clauses of
interrogative sentences (II.6) and BE GOING TO in interrogative subclauses
(III.1). This is what was expected given what I knew from analyzing corpus

Results with regard to COMPL clauses could only partially corroborate
corpus-based findings (in II.7, respondents conformed with the research
hypothesis, while - by preferring "'ll" - they did not in III.2). The same
muddiness must be stated concerning respondents' preferences concerning
cause clauses (especially III.9) and time clauses, where respondents'
preferences were as expected in III.6, but contrary to my expectations in
III.8. As to relative clauses, respondents preferred WILL/SHALL by a wide
margin in III.3, although I would have expected them to prefer BE GOING TO

Occasionally, the data that were returned by the questionnaire were
stratified according to educational level; in one case, I detected a pattern
of gender stratification. In many cases, unsurprisingly, there were
statistically significant regional differences. In sum, US respondents opted
(a) for BE GOING TO markers more often than did British respondents, and
(b), they opted for the contracted / cliticized variant forms more often
than did British respondents, which is especially true for "gonna". Subject
to the limits of the data source, both findings clearly conform with
findings that stem from the corpus-based approach, although the differences
among respondents, especially concerning "gonna", are surprisingly marked.
However, elicitation tests are about how respondents *think* they talk;
regional differences between respondents, then, might be exaggerated in that
while Americans commonly pride themselves of their informality, British
respondents might think they talk less colloquially and informally than they
actually do. Relatedly, British respondents might simply be more aware of
prescriptivist traditions than are US respondents.

Should you want to more detailled information, please feel free to go to , where I uploaded the
entire chapter of my thesis that dealt with the questionnaire and its
results (20 pages). Again, I want to express my thanks to all those who

Benedikt Szmrecsanyi
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue