• 21 per cent of a Twitter poll preferred ?with regards to over with regard to.
  • In a Twitter corpus the regards versions happen more often than the regard ones
  • With regard to is three times more frequent in academic journals than in general language, whereas in regard to has a very similar frequency in both.
  • In a general corpus, with/in regard to occur with greater than expected frequency in formal and technical writing.

This blog is the continuation of an earlier one on the same topic. As mentioned in that previous blog, I look at language corpora to give me an objective basis for what I write. Below I show the results of searching in two Oxford University Press corpora initially, one of general language launched in February 2014 which I have often consulted and one of academic journals from June 2015.

What tweeps think

Before I get on to the figures it’s worth mentioning a couple of things. First, a poll on Twitter produced some interesting reactions. Here’s the question and the results.

Which reaction most closely describes yours to ‘with regards to’ for ‘with regard to’?

48% Totes unacceptable

22% Acceptable but worse

09% What’s the problem?

21% Better >with regard to

It was interesting that 21 per cent preferred ?with regards to. As for the 9% ‘What’s the problem’, I framed the question badly because it could mean either a) I don’t understand what the issue is, which is what I meant (Call yourself a lexicographer! Ed.) or b) the respondent doesn’t have a problem with either. I have just fired myself as a questionnaire writer.

Seriously, though, if nearly a quarter of respondents are happy with ?with regards to and, as Table 4 below suggests, nearly 15 per cent of examples are of it, it becomes problematic to talk about ‘correct’ and ‘incorrect’ versions, despite what editors (see previous blog) say. I have therefore adopted the terms ‘accepted’ and ‘stigmatized’, even though you could argue that this merely pushes the elephant of authority and correctness into the next room.

Now for the corpora

The two major ones are enormous:

Table 1: Corpora size

Numbers of: Gen. Lang Corpus Acad. Journals Twitter corpus
‘Documents’ 64,049 467, 954 1,689,217
Sentences 112,453,774 64,135,538 ditto
Words 2,145,689,388 1,367,119,159 17,764,948

My first research question, if you want to put it that way, was: how often are ‘in regard to’ and ‘with regard to’ used in academic writing compared to general writing?

And my second question was: how often do the stigmatized versions ‘?in/with regards to’ occur in academic writing compared to general writing?

However, once I started looking at those two sources, it struck me that I should also look at the Twitter corpus and add it to any comparisons. (See the table above for figures showing its much smaller size.)

And that led to the third question: what does the data show about the style or register of language in which the two phrases are used?

As regards or regarding question 1, then, as one might expect, with regard to occurs almost three times as often per million words*(PMW) in the academic journals corpus as in the general one while on Twitter it is negligible. In regard to is much less frequent than with regard to at 3.57 PMW in general corpus  and 3.69 in academic journals. I have no convincing explanation for this very similar frequency across the two corpora but wonder if it suggests that it is less stylistically marked in people’s minds than with regard to. Any suggestions for an explanation are welcome.

Table 2: Relative frequency of standard phrases in three corpora

  Gen. Lang Corpus Acad. Journals Twitter corpus
Total PMW Total PMW Total PMW
with regard to 26,925 10.71 46,868 28.20 5 0.22
in regard to 8,969 3.57 6,138 3.69 2 0.09

*Frequency per million words is a useful way of putting words in rank order. For example, to put the figures for with/in regard to in context, happy, a common word, occurs 114.52 PMW in the general corpus but only 8.88 in the Academic Journals.

The next table shows data relating to question 2 (How often do the stigmatized versions ‘in/with regards to’ occur in academic writing compared to general writing.)

Table 3: Relative frequency of ‘regards’ phrases in three corpora

  Gen. Lang Corpus Acad. Journals Twitter corpus
Total PMW Total PMW Total PMW
?with regards to 3,772 1.50 3,621 2.18 11 0.49
?in regards to 3,810 1.52 1,088 0.65 15 0.66

This data shows, obviously, a much lower frequency PMW for the stigmatized versions in both the general language corpus and academic journals, but a higher one in the Twitter corpus, where there are in any case more stigmatized than accepted versions.

Another way of looking at the data is to calculate the relative percentage contribution of the accepted and the stigmatized versions in each corpus. The left column for each corpus shows the absolute numbers, the right the percentage contributed by each form to the combined totals.

Table 4: Percentages of accepted:stigmatized versions across three corpora

  Gen. Lang Corpus Acad. Journals Twitter corpus
# of examples % Total % Total %
with regard to 26,925 85.65 46,868 92.82 5 31.25
?with regards to 3,772 14.35 3,621 07.18 11 68.75
TOTAL BOTH 30,697 100 50,489 100 16 100
in regard to 8,969 70.18 6,138 84.94 2 11.77
?in regards to 3,810 29.82 1,088 15.06 15 88.23
TOTAL BOTH 12,779 100 7,226 100 17 100

The percentage of accepted version versus the stigmatized version is higher in the academic journals corpus than in the general corpus. In contrast, in the Twitter corpus, the accepted versions are a minority, and, for in regard to, a rather small one.

As regards question 3, the general corpus data is stratified according to register into five categories:  standard, formal, technical, informal, non-standard. There is also a sizeable category ‘0’, meaning ‘unknown’.

Though one could perhaps predict this, it is still worth pointing out that both phrases occur with a much higher than predicted frequency in the formal and technical registers. Conversely, the altered or mistaken versions occur with somewhat greater than predicted frequency in the informal register, and with very much less than expected frequency in technical and formal registers. The next two tables show the absolute numbers per register and the actual vs expected percentage.

Table 5: Distribution of accepted forms by register

Register Gen. Lang Corpus
with regard to in regard to
Number % Number %
standard 7,768 52.10 2,746 55.30
formal 7,652 163.90 2,200 141.40
technical 5,727 206.70 1,363 147.70
informal 1,521 40.80 576 46.40
non-standard 176 15.90


35 9.50


unknown 4,081 93.80 2,049 141.40

Table 6: Distribution of ‘regards’ versions by register

Register Gen. Lang Corpus
with regards to in regards to
Number % Number %
standard 1,766 84.50 1,631 77.30
formal 494 75.50 385 58.30
technical 260 89.60 211 53.80
informal 608 116.50 828 157.00
non-standard 98 63.30 206 131.70
unknown 546 89.60 549 89.20


Finally, in the earlier blog I suggested that in/with regard to could sometimes be replaced by about but that often that would not work. A reader kindly suggested that regarding could do the job, in addition to e.g. as regards, with respect to, in reference to, etc. I agree that it could, but think that while it is shorter than with/in regard to, it is still noticeably more formal and starchy than, say, a simple about, over and other single-word prepositions.

Rather more frequent than with/in regard in general corpus, it occurs with greater than expected frequency in the formal register and could often be replaced by something simpler, as in the following example:

When we disagreed with each other regarding whether to include a study, we discussed the inclusion criteria until we reached a decision.

about? over?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.