Unfortunately, I'm not giving a presentation at the «Participants Workshop of the summer school on Generative and Transformational Techniques in Software Engineering (GTTSE 2005).» The competition was a quite tight: there were 32 extended abstracts and only 14 time slots. Well, anyway, at least I again have got a valuable feedback.
My extended abstract: “Using DSLs on top of Scheme VM.”
DSLs are important instruments in lifting the level of abstraction in software engineering. The most important obstacle for the introduction of DSLs is the lack of convenient development infrastructure for them; an issue that your work addresses.
You make some claims that are not so easy to substantiate. For instance: "Languages are usually designed better than casual code". Actually, your main example, XPath, could be argued to be poorly designed, lacking for example, abstraction constructs, proper typing.
You do not mention related work. Many meta-tools exist for DSL design and implementation. Several authors have written about relative advantages and disadvantages of DSL versus (combinator) libraries and embedded languages, modular languages, extensible languages, user-defined syntax, etc. It would be interesting to see how you position your work in this respect.
The objectives and approach of your work remain a bit unclear. This is partly due to the lack of linking it to existing work. Also, you speak about so many embeddings, mappings, transformations, implementations, etc. that the reader gets lost. A clear overview, perhaps with a picture, of your overal architecture seems to be needed.
"Scheme VM" -> "a Scheme VM" or "Scheme VMs"
"Software development process" -> "The software development process"
"only than" -> "only then"
"best suiting" -> "best suited"
"it's" -> "it is"
The remark "There is no general advice. Each case should be handled personally" seems to be out of place, or at least cryptic.
"the future research" -> "future research"