Fwd: usage ridicule

Victor Steinbok aardvark66 at GMAIL.COM
Thu Mar 29 04:57:15 UTC 2012

I'd like to believe that, but I can't. Let me explain.

Instead of a single pair "a historical,an historical", let's consider a
double pair "a historical,an historical,a historic,an historic"

If you take a close look, you will find the picture far more
complicated. The only interesting thing that occurs in the 1940s is that
the "a" variants both overtake the "an" variants. But that's where the
similarities between the two pairs end. For "historic", both variants
rise fairly consistently from roughly 1840 to 1920. Through almost 1950,
both variants remain nearly constant and about equal. In fact "a
historic" experiences a mild decline, then starts a recovery, then
rising fairly steadily from the mid-1950s. The "an" version experiences
a mild decline from 1950 forward.

If the issue was a simple change in dominance from British publications
to American ones, the two pairs of graphs would have similar
corresponding periods of divergence, increase and decline between the
parallel graphs. They do not. In fact, aside from some possibly random
local fluctuations, there is a clear increasing trend in both versions
of "historical" throughout the 19th century. "An historical" peaks near
1890, "a historical" peaks near 1900. Both suffer a decline through
roughly 1920, a brief period of stability, then a gradual increase in
frequency for "a historical". The other version does not suffer a
decline until at least the 1980s.

Aside from the fact that the graphs are in no way similar, there is the
problem that restricting to only British English duplicates most of the
picture, but not quite. Until the WWII years, "a historical" and "an
historical" behave similarly. Both start a mild decline in the 1890s,
but the decline is much shorter than the overall graph, continuing
increasing precipitously from the late 1950s.


An it is in /British English/ that the frequency of "a historical"
starts increasing in the late 1930. There is no corresponding decline
/overall/ in "an historical" between 1900 and 2000--the frequency
remains essentially the same. The other pair of graphs is similar to the
general one, except for a rather significant bump to "an historic" from
1930 to 1950, remaining nearly constant thereafter.

It's in the American English version that things become more clear.

The graphs are again very similar, except for the fact that the two
"historical" versions are a bit closer to each other from the 1810s to
the 1950s largely because the "a" version is more frequent and the "an"
version is less frequent than the British. But the point remains the
same--the decline in "an historical" starts a bit earlier (1880s
perhaps) and the recovery of "a historical" is a bit sharper from 1930s,
than for the general English combination. And there is not 1930s-40s
"bump". Otherwise, the shape is very similar.

So far from being an American influence on the general trend, it was a
similar pattern from both sides. In addition, both sets of data are
heavily corrupted. The British list includes US publications reprinted
in the UK and vice versa. So the simultaneous trends are not surprising.
The main issue, however, is the eventual sharp increase in the "a"
variants, not the substitution of one for the other. It's important to
note that these are relative frequencies of occurrence, not absolute
numbers. So the conclusions are very different.

The combined frequencies don't help much. Comparing "a humble, an
humble", however, gives a very different picture.


Compare all that to the use of "humble". This one has a general
precipitous decline from the 1830s. The rates differ somewhat along the
way, but the decline does not stop, except in the early 1840s and the 1920s.


Separating into the "a" and "an" variants gives a very different
perspective from the "historical" and "historic" graphs.


The "an" variant dominates the graph through the 1860s. From that point,
the "an" variant suffers a precipitous drop, virtually disappearing at
the modern end, while the "a" version also declines, although much
slower. From about 1900, the overall graph is dominated by the "a" variant.

The trick is that the British graph is virtually identical, although the
frequency is a bit higher across the board. The American graph is also
similar, except that the crossover occurs a bit later--it's in the 1850s
for British and in the mid-1860s for American. Other than that, the
trends are basically similar.

Whatever the explanation might have been for "historical" it completely
fails for "humble". It looks like the frequency is much more affected by
the individual words than by which continent they occur on. This is the
danger of drawing conclusions from a single data set without running any
controls. Any "generalizations" drawn in this manner immediately break
down as soon as we look at any other data.


On 3/28/2012 2:00 PM, Joel S. Berson wrote:
> ...
> But someone there asked Google for an Ngram for "a" vs. "an
> historical".  It's interesting what happened about 1940.  Quoting him
> without his permission:
>> Here's a graph plotting the relative popularity of "a historical"
>> and "an historical" from 1800 to 2000:
>> http://books.google.com/ngrams/graph?content=a+historical%2Can+historical&year_start=1800&year_end=2000&corpus=0&smoothing=3
>> The explosion of "a historical" around the middle of the last
>> century probably has to do with America's increasing dominance of
>> the publishing world.  Still, limiting it to British English shows
>> "a historical" is still the more common form.
> Joel

The American Dialect Society - http://www.americandialect.org

More information about the Ads-l mailing list