Steven Clarke said:

>Does anyone know of studies investigating the effect of casing on
>readability of program source code?

The place to start is probably with studies of print readability in general.

The general finding IIRC is that ALL-UPPER-CASE has low legibility in terms of 
recognising individual words, almost certainly because the outline shapes of the words 
don't differ as much as lower-case. I don't recall what happened to with Title case or 
with camel case (where joinedUpWords are internally capitalised). They might not have 
invented camel case at that time.

I believe the first few letters exert more influence on recognition than the later 
ones. 

There was a lot of work on font legibility too. Not entirely conclusive, but I think 
I'm right in saying that the Road Reseach Lab in the UK designed its own font for 
motorway signs in the light of research, and that seems to be extrenely successful - 
one of the better examples of evidence-based practice. There may have been some 
research on those dreadful fonts which had no descenders, in vogue for technical 
reasons in the 70's.

I see no reason why programs should be different. 

It may also depend on whether the whole source code is being put into, say, uppercase, 
or only the keywords. Recognition is helped by high redundancy (in 
information-theoretic terms), and redundancy is increased when the ensemble of 
candidates is small or when some items are more probable than others. Programs have 
high redundancy for keywords like 'for' but rather low redundancy for general 
identifiers. So you could get away with keywords in capitals much more successfully 
than having the whole text in capitals.

NB, different types of source code have different degrees of redundancy and different 
syntactic constraints. 

I'm afraid I can't pull out a reference. The legibility literature is quite old, which 
makes it harder. I think Patricia Wright did some, at the tail end of the period, and 
she might be a starting point for searching the data bases. See who she cites. Also 
look for a guy called (Derek?) Twyman, a typographer, I think at Reading Uni or 
possibly the OU, I think he probably cites the relevant studies.

If there have been studies with program code, it will have been perhaps in the 
mid-80s: look at the ESP series, but that period produced a few papers from people new 
to experimental research with rather casual experimental designs or with less 
thoughtful interpretations than might be hoped for, so don't ingest the conclusions 
too uncritically.

Thomas



27 Allerton Park, Leeds LS7 4ND
+44-(0)-113-226-6687
http://homepage.ntlworld.com/greenery/
 
----------------------------------------------------------------------
PPIG Discuss List ([EMAIL PROTECTED])
Discuss admin: http://limitlessmail.net/mailman/listinfo/discuss
Announce admin: http://limitlessmail.net/mailman/listinfo/announce
PPIG Discuss archive: http://www.mail-archive.com/discuss%40ppig.org/

Reply via email to