Info-str library and ICONS-KEY constraints in clauses

I am experimenting with the Inormation Structure library currently, creating basic grammars which place focus and contrastive focus in various positions in the sentence. The library creates phrase structure rules such as this one:


The problem is the grammar does not load, complaining about an invalid constraint. Turns out the invalid constraint is the ICONS-KEY on the head daughter (meaning, if I remove that constraint, the grammar loads).

I wonder how I should fix it (assuming this is a bug and not an invalid choices file).

Here’s the choices; I did not include any information structure markers, perhaps that is problematic?


Assuming these choices are a valid set, how should I proceed?

I am not sure what the problem is, as there is no clash between the type and any of its parents that would be obvious to me. Furthermore, removing the parents and just saying the parent is simply sign does not help. Is this something about the feature geometry, like the ICONS-KEY somehow ends up not appropriate in the type?

Dear Olga,

In the type hiearchy of info-str, “topic” and “focus” are presumed to be mutually exclusive. Thus, an element cannot be assigned “topic” and “contrastive-focus” at the same time.
Thus, the following two choices are contradictory to each other.


When I created the regression tests with the pseudo languages, as far as I remember, I thought a situation simialr to this (topic: first and focus: clause-initital). I think the related regression test is “infostr-foc-initial-topic-first”.

At that time, probably I implemented only “non-contrastive focus” with this regression test. There is a clue as to how to fix this problem in the regression test.

I will follow up this QnA.


Thank you for the answer, Sanghoun!

To clarify: I see that, in the “infostr-foc-initial-topic-first” psudolanguage, the combination is:

word-order = svo
topic-first = on
focus-pos = clause-initial

The two test sentences for this language, if translated into English, are:

Kim chases the dog
The dog, Kim chases

the first sentence is licensed by the head-subject rule and has underspecified focus marking and an empty ICONS list.

The second sentence is licensed by the head-filler rule and has one item on the ICONS list, relating to the dog which is said to be focus-or-topic.

I have a question about information structure marking here.The MKG feature for the second sentence is fc-only, FC +, TP -. What does it mean? I read chapter 9 of Song 2014 but I am still not sure I understand the relationship between MKG and ICONS and how a combination such as in this pseudolanguage should be interpreted.

(To anchor this discussion a bit, what I am actually doing right now is working on a Russian grammar for the wh-questions library, where I have free word order, clause-final focus and clause-initial contrastive focus (according to the literature, anyway). So, what I need is to decide on a set of information structure choices that I will be using and then make sure that I am getting what is expected from the customization system, before I start thinking of any changes).

If so, I think the value of ICONS-KEY might be better to be “contrast-or-topic”.

The MKG structure primarily aims to constrain the scrambling constructions in Japanese and Korean. You can see Section 10.3 of my book “Modeling information structure in a cross-linguistic perspective”. The book is the upgraded version of my dissertation. You can download it from the following.

Regarding the wh-questions, I think you can ignore the MKG structure, but let me think of how to structure the preposed wh-words with or without MKG […].


1 Like