Department of Computer Science,
National University of Ireland, Maynooth,

John Harpur's Home page

(a.k.a John Harper's Home page)

Previous Staff Page Staff Index Home Next Staff Page

These pages are dying on their feet and a new site will be available shortly based on CSS files. Apologies for the flat file structure here, but initially it seemed the most HCI friendly structure for users (and browsers). Segmented linear stream of information. Easy to search. Easy to model cognitively (viz. remember and recall). Anyway this will change for November 2003. I am experimenting with colour at the moment - hypothesis being that names or short items you want users to remember, should be in deep colours and hues of the same. Not Earth shattering.

There are still a few glitches in the pages below. Some of the links to code may be broken. Can't do anything about that now as I am trying to accelaerate awy from this layout. Please don't plagirise the contents of this site. Give credit where credit is due.


Workshop on Constraints in Discourse

Venue: NUI Maynooth

Dates: 7th - 9th July


The goal of this workshop is to provide a forum for presenting recent research on constraints in discourse. The target areas include the recognition of discourse structure as well as the interpretation and generation of discourse in a broad variety of domains. The workshop offers a forum for researchers from diverse formal approaches, including but not limited to:

  • Rhetorical Structure Theory (RST)
  • Segmented Discourse Representation Theory (SDRT)
  • Tree Adjoining Grammars
  • Head-driven Phrase Structure Grammar (HPSG)
  • The QUD Modell
  • Plan Based Reasoning
  • Abductive Reasoning
  • Gricean Pragmatics
  • Speech Act Theory

Invited Speakers

  • Barbara Kaup, TU Berlin
  • Alex Lascarides, U Edinburgh
  • Ivan Sag, Stanford U
  • Candy Sidner, MERL

Program Committee

  • Anton Benz, U of Southern Denmark
  • Markus Egg, U Groningen
  • John Harpur, NUIM
  • Barbara Kaup, TU Berlin
  • Peter Kuhnlein, U Bielefeld
  • Alex Lascarides, U Edinburgh
  • Gisela Redeker, U Groningen
  • Ivan Sag, Stanford U
  • Candy Sidner, MERL


Harpur, J, Lawlor, M and Fitzgerald, M (February 2006). Succeeding with interventions for Asperger syndrome adolescents: a guide to communication and social interaction therapy. London and Philadelphia: Jessica Kingsley Publishers

Harpur, J, Lawlor, M and Fitzgerald, M (2004). Succeeding in college with Asperger syndrome. Jessica Kingsley Publishers: London and New York. You can order either through Amazon, your local bookstore or here is a downloadable order form.

The book ( pages 270) will be available from mid-November 2003.

We have another book in progress looking at how interactive video can be used to teach social skills. All going well that should appear Autumn 2004. Some of this work is based on material we presented at the IMFAR 2002 (International Meeting for Autism Research), in Orlando, FL.

Our website is undergoing a face lift. Due to lack of broadband in Ireland (yes there are people living outside Dublin) we pulled some of the Flash content from the site. A thinner more bandwidth friendly site will be available in November.

Described by the international Asperger syndrome expert, Dr Tony Attwood, as likely to 'become the book of first choice' in this area. Read his full review in Child and Mental Health, November 2005 - Vol. 10 Issue 4, p 214.

Research Grants:

The past year has been productive on the research grant front - information I sometimes forget to stick in grant applications because it seems so obvious to me (bit of frontal lobe difficulty there, Jerry).

Department of Rural, Community and Gaelteacht Affairs funded an exploration in establishing materials for an online support group for adolescents with Asperger syndrome, parents and professionals. The project helped generate an enormous amount of media materials including videos. October '02 to September '03.

Partnering with the Access Office in Maynooth here, the HEA Disability fund helped us make a small series of themed social competence videos for 3rd level students with Asperger syndrome. November 02 to November 03.

Between the two projects we have produced approximately 60 short videos. Most of these have been trialled with small number of subjects with AS. We will have these available on DVD by the end of October. These materials are copyrighted ©.

The Health Research Board awarded us an interdisciplianry grant, October 03 to October 05, to investigate pragmatic assessment techniques in adolescents with Asperger syndrome. The outputs from the project will include a set of methodological recommendations for both computer-based (a) social skills programmes and (b) pragmatic competence assessment. The project has three facets. It can be viewed as a mental health project, an assitive technology project and finally as an attempt to scientifically ground a certain type of computer based coaching in a specific human domain.

The foundation for much of our recent progress was laid by Intel (Ireland) through the sponsorship of high end computers and ongoing support. It would have been impossible to move the project forward without Intel support.

...await the new pages in November for better organisation of information...


Awarded the Applied Ergonomics Award 1999 for contributions to the field: cf paper Harper & Fuller et al. (1998) below. The award is jointly sponsored by the Ergonomics Society, J.Applied Ergonomics and Elsevier Science. See The Irish Times, 2nd August 1999 for an article on the award and the research.

Current Research Briefs

Research Sponsors

Intel (Ireland) and Microsoft (Ireland)

Emotional Intelligence and Asperger's Syndrome:

This is multidisciplinary project involving two child psychiatrists, a speech and language therapist, an education programme evaluator and two computer scientists. We are designing a prototype multimodal system to aid adolescents with Asperger's Syndrome (AS) acquire emotional intelligence (EI).

Really? What's that? Been spending to long in California...I hear you mutter.

Well, EI is simply the capacity to monitor and differentiate between the behaviour and emotions of oneself and others, and to use this information to forge effective social communication. In traditional philosophy and linguistics this is referred to as pragmatic understanding. Normal children tend to pick it up and use it pretty handily. However not everyone has EI. The self monitoring component which is crucial to making sense of feedback may not be operated by all children equally (if at all). For example, it is kind of useful when you go out on a date to be able to read the signals correctly. The definitive lay person's guide to understanding EI is Daniel Goleman's Emotional Intelligence, Bloomsbury Press 1996, but other works on self-efficacy (especially by Bandura) provide interesting viewpoints.

Key characteristics of Asperger is the combining of lack of empathy with poor comprehension of nonverbal communication. AS children have a seriously diminished capacity to put themselves "in another's shoes". It is generally accepted that AS children occupy the high performance end of an autistic spectrum. Many have above average IQ and good verbal skills. Funnily enough many end up in maths and computer science. You'll look twice at your colleagues having read this piece. While there are at least three competing theories attempting to account for Autism and its derivatives, the one that emphasizes impaired 'theory of mind' has gathered a substantial amount of experimental support. Many of the competing accounts and experiments are reported in the Journal of Child Psychology and Psychiatry (cf. papers by Baron-Cohen, Frith, Boucher, Swettenham etc.). Baron-Cohen's essay on autism and the theory of mind hypothesis is entitled Mindblindness and published by MIT Press, 1995. An excellent focused account of AS is given by Tony Attwood, a clinical psychologist, in Asperger's Syndrome, published by Jessica Kingsley Publishers, 1998. Attwood references interesting work by Carol Gray on the use of social stories to help AS children rehearse behaviour. Carol Gray has a book coming out next year on her techniques by the same publisher.

This is where we pick up the thread computationally. Could we make computerized versions of the social stories? Having read Ros Picard's Affective Computing, (MIT Press, 1998), it didn't seem far fetched to consider designing a system that would help children pick up cues which are so important to EI. Our work is based on a nonlinear model of social stories which can be played out on the computer screen. AS adolescents are offered an adaptive scenario based on determinations of their levels of EI proficiency and self monitoring skills. Apart from the obvious therapeutic benefits, it is an interesting exploration in digital media manipulation. In addition we plan to have the adolescents synthesize those voice features which they believe will convey relevant non verbal information (I'd like to reference Janet Cahn's work at MIT on speech synthesis here). This is a tricky step for these children as it involves the sequencing of skills that most of us take for granted. We expect the outputs of the project to be generally applicable to modeling and treating Social Communication Disorders. We were examining ways of integrating biosensory information into the feedback mechanism. This work is quite 'conjectural' and we won't be progressing it any further in the immediate future. As an experimental exercise, connected with measuring skin conductance, John Ghent has built a prototype Galvanic Skin Response sensor using a lot of initiative. Our own in-house lie detector. There are other 'gadgets' but we don't have a budget for them and the clinicians are sceptical about their value in this setting.

At present we are setting out our stall and looking for funding. I have a document describing the computational objectives in greater detail, and if you are interested in collaborative research or offering support I am willing to make it available to you.

Related Work: A final year student, Alana Conroy, is working with me on the provision of a tool for assessing bullies 'theory of mind'. Recent research indicates that bullies have a much more sophisticated understanding of their victims' psychological states than previously assumed.

Mental and User Modeling formation in Wireless Computing:

Following his successful final year project on the use of Wireless Application Protocol (WAP) enabled devices to access and share medical information, Aidan McHugh, is working with me as a postgraduate student. The objective of the research is to assess how scaling of applications to PDA sized devices affects mental and user model formation.

The Irish Times 14th May 2001, page 8, carried a substantial article on the earlier application under the title "Got an injury? WAP it up."

Despite the seeming ubiquity of wireless computing devices very little research has focused on questions about usability transference from one scale of application functionality to another. Micro browser initiatives, and web searching through 'drilling' down through cropped web pages to find information has crept up on us with very little reaction from HCI researchers - for the past three years CHI Conferences has had no more than a smattering of papers on these developments.

A key property which we would expect to affect the 'take up' of PDA based applications is their prediction of, and control for pragmatic overshoot by the user. The notion of pragmatic overshoot arises from Natural Language Processing work in the eighties's into question structure, especially the degree to which it revealed user assumptions about the problem domain. "Can you tell me which department sells both buckets and spades, please?" implies that the speaker (user) believes one department has both types of tool. Overshoot occurs due to users having reasonable but unsupported presuppositions about a system; (what we might call over-assuming). We are interested in the connection between overshoot, mental modeling and usability transference in scaled applications.

Currently we are working on determining how users perform relevant feature extraction from PDA scaled version of a workstation package, and whether there is any way of flagging overshoot so as to minimise user frustration. Underpinning some to the work are certain ideas I have about the presence and role of cognitive dissonance in application migration environments.

Discourse Analysis with Children: Looking at attachment patterns: .

It's a fact, but when you look at most of the work in natural language processing and automated reasoning, we begin always with the finished product. i.e. adult competence in these areas. Very little work has been done on computationally modeling the child's view of the world as mediated through language. The smaller vocabulary of children, their use of shorter sentences, and pretty bald grasp of nonverbal communication makes their discourse an interesting candidate for computational modeling. While one can assemble good arguments for dealing with adults only on the grounds of finished skills, in reality adults had to pass through childhood. Perhaps, studying children's discourse, especially how they report attachment patterns will help us gain an insight into adult discourse structures that has eluded us in the main.

Currently I am looking into this area with a Child Psychiatrist using both a theory of discourse analysis specifically aimed at extracting attachment information from adults, and Mann and Thompson's rhetorical structure theory (RST) as a basis for contrastive analysis. We hope to have transcripts of children;s dialogues available for use before Xmas. We begin recording in December.

Related work: A final year student, Rowan Nairn, is currently working on an implementation of Kamp and Reyle's Discourse Representation Theory, which will prove a useful foundation for the above work.

Computer Adaptive Testing:

Secondary research stream. In line with my philosophy of socially responsible computing, Computer Adaptive Testing (CAT), is a prime example of a potential High Brow High Impact development. The traditional model of education has probably best suited both the well off and the bright. Joe Average has had a tougher time however. Computer Adaptive Testing doesn't promise Joe Average more marks but it does offer a handle on a finer means of understanding Joe's performance. CATs operate on the basis that each question is selected by an algorithm which compares Joe's performance against that of the statistically average student during the exam. If Joe is doing well, the questions get harder. If Joe is slipping the questions get easier. Both Joes can emerge with a sense of satisfaction after the exam. In fact they may have answered the same number of correct questions, but because difficult questions attract more marks, their scores will be different. Adaptive testing is the future, especially in commercial training situations where companies want to quantify take up of information from training courses using more fine grained tests than currently available. Both elearning and lifelong education initiatives provide strong grounds for investment in national CAT infrastructures.

We are still in funding limbo on this one however.

Older Research Trends



Previous Staff Page Staff Index Home Next Staff Page