For the paper described in the previous post, I ended up deleting a section which described an investigation into whether student use of capital letters and full stops could be used as a proxy for writing in sentences and paragraphs. We looked at this because it is a time-consuming and labour-intensive task to classify student responses as being ‘a phrase’, ‘a sentence’, ‘a paragraph’ etc. – but spotting capital letters and full stops is easier (and can be automated!).
I removed this section from the paper because the findings were somewhat inconclusive, but I was nevertheless surprised how many responses finished with a full stop and especially by the large number that started with a capital letter. See the table below, for a range of questions in a range of different uses (sometimes summative and sometimes not).
| Question | Number of responses (and percentage of total) that started with a capital letter | Number of responses (and percentage of total) that finished with a full stop |
| A-good-ideaAYRF
S154 10J |
1678 (60.9%)
622 (60.0%) |
1118 (40.6%)
433 (41.8%) |
| Oil-on-waterS154 10J | 500 (53.9%) | 294 (31.7%) |
| MetamorphicSXR103 10E | 297 (41.6%) | 166 (23.2%) |
| SedimentarySXR103 10E | 317 (39.9%) | 178 (22.4%) |
| SandstoneS104 10B | 954 (58.2%) | 684 (41.7%) |
| Electric-forceS104 10B | 673 (56.7%) | 445 (37.5%) |
Answers that were paragraphs were found to be very likely to start with a capital letter and end with a full stop; answers that were written in note form or as phrases were less likely to start with a capital letter and end with a full stop. Answers in the form of sentences were somewhere in between.
The other very interesting thing was that capital letters and full stops were both [sometimes significantly] associated with correct rather than incorrect responses.