'second oldest' - well, actually it's the oldest because it's just traditional school grammar, as in 19th-century Reed and Kellogg diagramming (https://en.wikipedia.org/wiki/Sentence_diagram). Tesniere's main contribution was to unify the analysis by taking the verb as root, instead of treating the top verb and its subject as equals. See attached notes from Percival (1999).

Why? I don't see why this is more of a problem for DG than for PSG.

Precisely as suggested in Word Grammar. As I say, there's not only one version of DG!

I think everyone in DG would agree that syntax and semantics are separate.

How about this ?: Robin's goal was to take the same time to do A, B and C; and at last she did it: she did A, did B and did C in exactly 30 seconds each.

Not really 'phrase structure' because the larger unit need not be a phrase.

Precisely - and that's probably why T didn't adopt your analysis!

But Tesniere was happy with junction as a special relation, so why are you revising his theory?

I think it's misleading to present Gross and Osborne's analysis, as it's neither typical nor particularly successful. Nobody else distinguishes heads and governors in that way, and G&O don't show how 'head' translates into linearization. Nor do they explain how their principle excludes examples such as "great with difficulty", where "with" qualifies as the head of "great difficulty". And their 'g' notation implies the underlying dependency even though their diagram doesn't show it. In contrast, there are well-developed analyses in other theories, including my own Word Grammar analysis which I've published in several places, including a whole chapter of discussion in Hudson 1990, which you mention in your footnote 13.

This (fn 9) is standard Word Grammar, e.g. Hudson 1990:167, Hudson 2007:190.

This is like saying that dependency grammar doesn't focus on word classes. No 'dependency grammar' discusses nothing but dependencies, just as no PS-based grammar discusses nothing but part-whole relations. A 'dependency grammar' is one that uses dependencies, and the fact is, as you've already said, that different dependency grammars make different assumptions e.g. on linearization. Since everyone agrees that linearization is important, it's just a question of how to handle it, so it must be part of any dependency grammar, whether implicit or explicit. E.g. in Word Grammar (2015) linearization is handled by 'landmark' relations which are separate from dependency relations.

What you can say is that dependency structure, as such, is independent of linearization. Just as with the ID/LP split in GPSG.

And Hudson 1990 etc

You're suddenly assuming a very particular version of DG. DG is just a tradition, not a theory; it's no more unified than PS-based theories. As soon as you discuss details such as passives, you have to say which version you're discussing. The analysis in Fig 11.7 is very different from the one in Word Grammar.

But every dependency-based theory includes a separate semantic structure which dependencies are mapped onto. It's quite easy, especially since the asymmetry of dependency is motivated by meaning, but as you say, the mapping is sensitive to word order.

Not really. Projectivity doesn't invoke a separate phrase node, so the dependent and head aren't part of a larger node, as required by constituency.

You do need them if you recognise mutual dependency!

Strictly speaking, the part of speech information is included as well, though sometimes it's omitted just to simplify diagrams. This diagramming system is only used in Word Grammark, I think.

You should also mention Karlsson's ependency-based Constraint Grammar which as fully functioning parsers for several languages. It's described at ttp://beta.visl.sdu.dk/constraint_grammar.html and the languages and parsers are listed here: http://beta.visl.sdu.dk/constraint_grammar_languages.html . This online system has also been used in school teaching.

Hudson 1989 certainly doesn't deserve a mention - it was just a mickey-mouse parser with a tiny vocabulary.

Lafferty, John D., Daniel D. Sleator, and Dennis Grinberg. “A Robust Parsing Algorithm for Link Grammars,” 1995. Sleator, Daniel D., and David Temperley. “Proceedings of the Third International Workshop on Parsing Technologies.” In Parsing English with a Link Grammar, 277–92. Tilburg, 1993

Hudson followed Anderson: Anderson, John. The Grammar of Case: Towards a Localistic Theory. . Cambridge: Cambridge University Press, 1971.