CSV ignores lineterminator

[Jeffrey Barish]

With

input_data = ['word1\tword2;word3\tword4;',
'word5\tword6;word7\tword8;']

and

delimiter = '\t'
lineterminator = ';'

shouldn't csv.reader(input_data, dialect='mydialect') return

['word1', 'word2']

as the first row?  I find that it doesn't matter how I set
lineterminator, csv always terminates at the end of the line returned
by the iterable object passed as its first argument (input_data, in
this case).  I must be missing something basic here.

I may be confused about the interaction between what iterable object
defines as the next row and what csv.reader defines as the next row. 
The documentation for csv says that the line returned by csv.reader can
span multiple input lines.  I assume that this would happen if the
lineterminator is defined as something other than \n, in which case
csv.reader would keep pulling in lines until it finds the
lineterminator.  This is not the behavior I am observing.

I'm using Python 2.3 on Linux 2.4.23.
-- 
Jeffrey Barish