Well, that's a tough one. Normally, dislocations (topicalised material) are distinct from focalisations because the latter but not the former leads to a resumption in the clause, but not every language has clitic or morpheme resumption. Also, topicalisations in some languages have a variable position (left or right periphery should be fine, as in the French/Italian translations of (Paul,) he's just a dork, (Paul)), but Focus is generally in a fixed position. Topicalisations of the dislocation type can be found in subordinates, but not generally Focus (I very much doubt that my book, Paul read it; ?(?)I very much doubt that MY BOOK Paul read, not yours). Finally, Focus would be expected to be contrastive, as per your examples, but it is not always the case in some languages (Focus Fronting in French in "100 euros il me faudrait" (100 euros I would need) isn't necessarily contrastive).

Levinsohn has a different category for when there is resumption. He calls it "left dislocation" which he says is "a type of sentence in which one of the constituents appears in the initial position and its canonical position is filled by a pronoun or a full lexical noun phrase with the same reference." Actually, that definition is taken from Crystal, same with topicalization above.

Actually, I have found Levinsohn's statement about distinguishing points of departure from preposed focal constituents to be helpful. He said, "a propisitional topic or point of departure must either be already established in the text or be easily related to material that is already established, whereas focal information either has NOT been established in the tezt or needs to be RE-established." In the case of (2) above, the contrasted leg is related as "my other leg" which is "easily related to material that is already established." So, in (2) it makes sense why it would be topicalization (which is considered a point of departure) as opposed to a preposed focal constituent. In (1) and (3) the preposed element is new information. Thanks!

From a Role and Reference Grammar [RRG] perspective , the representation of topicalised constituents and dislocated (detached) ones is structurally different .The former lies in the Precore slot (PrCS), a clause internal position , and the latter in the detached (left or right) position (L/RDP), a clause external position. Topicalised and focused units take the PrCS because they are said to be within the focal range while detached /dislocated units are beyond the scope of focus.

In certain languages, the intonation pattern (type of stressed on the word/phrase under consideration, and intonation patern and stress on the rest of the sentence) can help to differentiate the topicalized word/phrase from the sentence-initial focus.

A lookout from a functional grammar , in particular, SFL provides some insights; as your approach is structurally based.

According to Lambrecht (1994) a topic must satisfy the following conditions:

1. The topic constituent must be referential.

2. It must be focus of attention. That is, it must be currently under discussion.

3. The predicate must be about the topic and more importantly, provide relevant information.

4. The referent of the topic must be part of the discourse register. To put it differently, the topic referent must be among the referents common between the speaker and the listener at the time of the utterance.

Two examples with minimal context:

1a. What did the children do next?

b. The children went to school.

(stress on school)

2a. Who went to school?

b. The children went to school.

(stress on children)

In 1b, the children as the subject possesses all the prerequisites; therefore, it is the topic, and the sentence has a topic_comment reading.

In 2b, the subject (children) satisfies non of the above_mentioned conditions, hence serving as focus, the non_ recoverable information at the time of utterance. The sentence is an identificational one. It identifies the children as focus, as an element equalling X in the presupposition " X went to school".