As far as I can see, the structure of supplementary constructions like

Karen, being ill, was unable to go


John – her father – was unable to walk her down the aisle

or maybe

a washer-dryer


a paperweight-cum-ashtray

(i.e. compound nouns where the meaning is built up more by addition than modification)

don’t fit very neatly into either dependency or phrase structure theories, because they are not made up of a head plus dependents.

Given that these constructions basically function as appendages, you might say it's appropriate for the framework that describes them to be a theoretical appendage. Still, I was wondering whether there are other theoretical approaches out there that give more attention to this kind of construction, or integrate it better.

Also, is it fair to say that the more pragmatic the language, the greater the use of supplementation?

  • 2
    "Being ill" is tightly integrated into the clause, and hence does not qualify as a supplement. "Her father" is a supplementary appositive. Compounds are single words and thus can't be supplements. – BillJ Mar 20 '19 at 14:50
  • 2
    The easiest way to handle them is to treat them as intercalated utterances; certainly they have different pragmatics. – jlawler Mar 20 '19 at 15:01
  • @BillJ yes compounds are single words, but these examples are built up from smaller elements using suppletive logic rather than head-and-dependents logic. It seems EN is comfortable with suppletive relationship at clause level, and sometimes at word level, but resistant to it within clauses - it requires any suppletive elements to be very clearly marked. – user23078 Mar 22 '19 at 9:36
  • Against that background it's not surprising if linguistic theories are also oriented more towards head-and-dependents logic than suppletive logic - but this may just reflect an idiosyncrasy of EN, making this type of theory less than ideal for describing languages which are more flexible in terms of whether an element is parsed on a suppletive basis or on a head-and-dependents basis - and I was suggesting that this may be a feature of those languages that are often described as pragmatic. – user23078 Mar 22 '19 at 9:36

I agree with BillJ's first two comments: being ill in the first example and her father in the second are integrated well into their clause structures and hence do not qualify as supplements in the intended sense. They are, rather, post-dependents of the preceding noun -- on a dependency grammar analysis. I would analyze the first sentence as follows:

enter image description here

And the second sentence as follows:

enter image description here

In each of these structures, the subject noun takes a post-dependent, the post-dependent clearly modifying the preceding noun. The arrow dependency edge in the tree each time identifies the post-dependent modifying the subject noun as an adjunct; it is not a complement.

The third example is, I think, indeed a challenge to strict head-dependent analyses. Phrase structure has a means of addressing such cases, though. Phrase structures can produce exocentric structures, that is, structures that are headless. Hence I would attempt an analysis of a washer-dryer along the following lines, combining both dependency and phrase structure:

enter image description here

The horizontal connector above washer and dryer links the two words in such a manner that neither is head over the other. In this manner, an exocentric phrase structure is integrated into the greater dependency analysis. The analysis of the fifth example, i.e. paperweight-cum-ashtray could receive a similar analysis.

A related point concerns the analysis of coordinate structures. One can legitimately argue that coordinate structures are headless (or they have multiple heads). This suggests again that exocentric analyses are necessary. I have therefore argued that the following type of analysis is appropriate for coordinate structures:

enter image description here

This analysis links the nouns together without viewing the one as head over the other.

In sum, I agree with the underlying sentiment of the question, namely that strict head-dependent analyses are not always possible. My solution to the problem is to acknowledge exocentric phrase structures at times, and then to integrate these exocentric structures into the greater endocentric analyses of sentence (and word) structure.

  • I noticed that the first sentence is not quite the same as the one in the OP. In the original version (Karen, being ill, was unable to go), there is a link between being ill and unable to go that is lost in Karen, being ill, was the life of the party. It's that link, if anything, that makes being ill feel like an intercalation. Even so, I can see that this sentence is different from the others. – rchivers Mar 16 '20 at 13:55
  • While I'm at it, I think it's plausible that John and her father are not in a head/dep relationship in the second sentence. The difference that the head/dep analysis implies between her father, John, was unable to walk her down the aisle and John, her father, was unable to walk her down the aisle strikes me as illusory. I think the structure is more like [John/her father] was unable to walk her down the aisle - the subject is named twice, but the names don't modify each other so it doesn't really matter which comes first. That seems to be what the notation used for washer-dryer is showing. – rchivers Mar 16 '20 at 13:56
  • @rchivers. I have corrected the first tree. For me, the expressions being ill and her father are like non-restrictive relative clauses; they provide additional information about the preceding noun and should hence be subordinated to them. – Tim Osborne Mar 21 '20 at 5:20
  • @rchivers. The difference between John, her father, and her father John has in part to do with what is known and not known. In the first case, who John is is known and the speaker adds the additional information that 'John is her father'. In the second case, what is known is that she has a father and what is added is that 'her father's name is John'. I think both submit to an analysis in terms of heads and dependents. I also think, though, there must be a difference between narrow and broad apposition, although I do not know if that difference impacts the head-dependent analyses. – Tim Osborne Mar 21 '20 at 5:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy