Authorial and Text Style
Task B - Comparing sentence lengths
Now we have some initial impressions, let's do a little bit of elementary
statistical work to begin to characterise the styles of the different
writers. As styles have to do with overall tendencies in writing, quantitative
analysis will often be revealing (though we should remember that as the
passages we are looking at are small - in order to keep your workload
down - the statistical work we do will be limited and illustrative rather
than conclusive).
For each passage, count the number of words in each
sentence and record them by filling in the relevant cells in the table
below. How does what you find correlate with your initial impressions
of the styles of the different passages? Compare your conclusions with
ours below. You can find a link to the passages under 'Useful Links' on
the left hand menu.
Note that, for comparative purposes, we have given you a sentence length
average 'norm' for prose writing, calculated in the 1970s by a Swedish
linguist called Ellegård. His "norm" was based on a 1
million words corpus of 20th century American English writing called the
Brown Corpus (the corpus was collected at Brown University in 1964).
A Statistical Comparison of the Passages
our analysis
Our Statistical Comparison of the passages
| Category |
Austen |
Steinbeck |
Lawrence |
Ellegård Norm |
|
Words
|
168
|
92
|
149
|
|
|
Sentences
|
4
|
5
|
11
|
|
|
S-Length average
|
42
|
18.4
|
13.5
|
17.8
|
Don't worry if you did not get exactly the same answers as us (but
if they are a long way out, check your maths!). It turns out that even
counting words in passages is not as simple as it seems, because there
are different ways to define what a word is!
For example, you have to decide whether to count 'good-looking' and
brother-in-law' in the Austin passage as one word each or as two and
three words respectively, and whether to count 'wasn't' in the Lawrence
passage as one word or two. What matters is that you use the same understanding
of what a word is when counting each passage.
Below is a comparative graph of the passage averages compared with
the Ellegård norm:
|
