Viewing language data in CSV files
Note: the
information below was current as of November 2011, when MAT 1.3
was released. It's possible that Excel's import features have
improved since then. We provide this information as representative
of the problems you may encounter.
The MATScore and the MATReport tools both produce CSV files which
contain snippets of your input document. Viewing these CSV files
is a bit complicated, and deserves some attention.
The short answer is: use
OpenOffice rather than Excel.
Excel 2007 CSV import has some very unpleasant features which
will compromise your ability to view the data cleanly.
- If the file extension of your CSV file is .csv, when you open
the file normally (either by double-clicking or selecting "Open"
from the main menu), Excel will not offer you an import wizard,
which will cause Excel to try to digest and interpret dates. So
if your annotation happens to span a date, Excel will recognize
it as such and process it. Changing the column format to Text
after you import is useless, because Excel has already discarded
the original data. To avoid this, create a new workbook, then
select the "From Text" option in the "Data" tab. This option is
available only on Windows; Mac Excel doesn't allow you to do
this at all with a .csv file, as far as we can tell.
- If the file extension of your CSV file is .txt, Excel will
offer you the import wizard (although now you can't open the
file with a double-click). The import wizard allows you to
select the column delimiter (comma), and also allows you to
change the column format for the columns you select. You should change the
columns which contain actual text to Text. However, the import
wizard screws up newlines in column data; even if they're
delimited with double-quotes, the import wizard treats them as
separate entries. So if there are any newlines in the spans of
text displayed, this strategy won't work.
Another issue is character encoding. All the CSV documents
created by the MAT tools are encoded in UTF-8. In order to view
this data correctly on Excel 2007, you must use the import wizard.
Again, this option is only available on Windows.
Because there's no consistent way of viewing the data in its
clean form, Excel isn't an appropriate tool, especially on the
Mac.
Fortunately, OpenOffice 3 does
do the right thing. You'll be offered an import wizard when you open
a .csv file. Select the column delimiter (comma), and make sure to
change the column format to Text for each column which contains
spans of text.