ACL-2001 TUTORIALS
Paola Merlo and Gertjan van Noord

The ACL 2001 tutorials were scheduled for the Sunday between the
workshops and the main conference, with parallel tracks, both in the
morning and the afternoon (a number of excursions were planned on this
Sunday too).  Reimbursement for presenters was the same as last year
(US$500 per session plus $25 per registrant in the range 21-50 plus
$15 per registrant in excess of 50).

The call for tutorial proposals was sent out to email lists in late
November, calculating deadlines as for previous ACLs. The wording (but
not the substance) of the reimbursement was changed from previous
years' calls.

We solicited two tutorials on topics, which do not fall into the
typical ACL expertise.  Specifically, based on recent interest in
empirical corpus-based methods and evaluations, we solicited a
tutorial on experimental methodology and hypothesis testing
techniques. We invited Harald Baayen (MPI) because of positive direct
experiences as a tutorialist on these topics. We also solicited a
tutorial on bioinformatics. Sequence comparison and alignment are hot
topics in bioinformatics that are potentially relevant for NLP. We
asked Rolf Backofen, as an expert in bioinformatics (author of a
recent textbook on this subject) with a background in NLP.  The idea
of soliciting tutorials was quite successful, judging from the number
of registrants.

We received 11 submissions (including two solicited), which made
for a difficult selection process. We considered several options, and
selected five half-day tutorials, based on quality of submission,
appropriateness of topic, and diversity of expected audiences.
We considered having a one-day tutorial, but were
strongly discouraged from doing so.

Morning

T1. Statistical Analysis of Experimental Data for NLP     Harald Baayen
T2. Very Large Lexical Databases                          Pustejovsky and Hanks
T3. Recent Issues in Categorial Grammars                  Christian Retoré

Afternoon

T4. Bioinformatics					  Rolf Backhofen
T5. Automatic Summarization				  Mani and Maybury

The scheduling of the tutorial was motivated by expectations about
popularity of the topics. This setup gives rise to a statistical track
(1, 5), a large database track (2, 4), a formal languages track (3,
4), and an applied problem track (2,5).

Brief tutorial descriptions are on-line at the ACL 2001 web site, and
were incorporated into the conference registration brochure, the
conference proceedings companion volume and the CDROM. The fees for
tutorials were reduced from ACL 2000 (early registration: 650 FF; late
registration: 850 FF).  Each participant receives a printed
copy of the materials, containing the overhead transparencies and
a bibliography.

Registration numbers appear satisfactory. (Update on the 18th of June,
2001 is as follows T1: 45; T2: 49; T3: 6; T4: 36; T5: 40.)  As can be
seen, one tutorial attracted many fewer people than the others. This
was partly expected, due to the formal and specialised nature of the
topic. We thought it was very important to provide at least one
tutorial of this kind.

Based on our experience, we recommend:

1) Since there are strong valid reasons againsts one-day tutorials,
the call for tutorials should be reworded so as to solicit only
half-day tutorials.

2) Issues of copyright must be clarified.  It should be decided
whether the tutorial materials require a release of copyright. If so,
the presenters should be informed in advance.

3) Presenters were required to register to the main conference. We
recommend that this requirement is dropped for presenters who do not
participate in the main conference.

4) Tutorial materials are duplicated in black and white. This is
unfortunate since most presenters provide colored sheets, sometimes
with beautiful illustrations.  We recommend to investigate
the production of tutorial materials in full color.