This week Babel’s Dawn has been looking at an article titled “The Simpler Syntax Hypothesis” that appeared in the September issue of Trends in Cognitive Science (abstract here). It points towards solutions for several issues of interest on this blog, notably the evolution of syntax (discussed yesterday) and certain technical problems connected with unspoken meanings.
There is an old lyric sung by Andy Williams and Julie Andrews, “Not a word need be spoken in our language of love.” Very romantic, but obviously something of a problem in theories of language that say words are the pilots of attention.
Formal syntax seeks to show the structure of sentences so completely that the full meaning of the sentence is laid bare. The challenge for syntacticians is that many sentences do not actually express the fully understood meaning. Monday’s post (here) looked at a sentence (“Ozzie tried not to drink”) that will seem transparent to any reader, but provides challenges to formal syntax, challenges that Noam Chomsky first resolved by proposing a deep structure with hidden, or unspoken, elements.
The article by Peter. W. Culicover and Ray Jackendoff examines many other types of sentences that are readily understood by humans and yet do not include the logical meaning in the words used. The most fully discussed cases of such omissions are in Bare Argument Ellipsis in which much of the formal meaning has been “stripped” from a sentence. Such ellipsis, if used as an initial sentence, can only cause confusion. If I come up to you and say, “Yeah, scotch,” you may be a little puzzled, but if you say, “Ozzie says that Harriet’s been drinking,” and I say, “Yeah, scotch,” the exchange makes sense. It has a context.
Formal syntax can interpret this last sentence as “Yeah, [Harriet’s been drinking] scotch.” The portion in brackets has been omitted in actual speech but was stated in exactly this way in the preceding remark. The hidden meaning was found by grafting a complete syntactic structure from the first sentence onto the second. Since the interpretation works fine, formal syntax is looking good.
The authors then go on to cite a few well-known kinds of sentences where the tactic does not work so well. Here’s an exchange, for example, where taking a preceding syntactic structure will not work:
“John met a guy who speaks a very unusual language.”
We could all expand this to mean “Which [unusual] language [does the guy who John met speak]?” That expansion, however, is nowhere in the original sentence. If we used the formal tactic of grabbing a syntactic structure from the first sentence, we would get an expansion like “Which language [did] [John meet a guy who speaks?” and that does not seem very helpful.
There can also be sentences that intervene between the stripped ellipsis and its source. The authors give this lovely example:
It seems we stood and talked like this before. We looked at each other in the same way then. But I can’t remember where or when.
If you can settle down long enough to get the Rodgers and Hart tune from your head, you can probably work out the ellipsis: But I can’t remember where or when [we stood and talked like this before]. The formal tactic of looking to the previous sentence, however, would give you either
But I can’t remember where or when [we looked at each other] or
But I can’t remember where or when [in the same way then].
Neither of those readings is going to get you a C in your English class.
There can even be ellipsis where nothing relevant has ever be said: — What are you looking for? —Those (pointing to scissors). The speaker says Those instead of That because scissors are idiomatically plural. There is no possible analysis of the first sentence that will uncover this explanation.
The authors also mention other common forms of expression in which formal syntax requires the insertion of a hidden meaning that is nowhere stated in the formal context:
Metonymy: The use of a word or phrase to stand in place of another word or phrase, e.g., “The ham sandwich over there wants more coffee” (The [person] over there [who ordered a ham sandwich]…); “Chomsky is next to Plato up there” ([The book by] Chomsky is next to [the book by] Plato…).
Sound for motion: The use of a sound to identify a motion, e.g. “The trolley rattled around the corner.” (The trolley [went] around the corner, [rattling]. Personally, I don’t think that interpretation is correct. If you understand speech in terms of attention and perception rather than scientific accuracy, there is nothing paradoxical in the sentence as it stands. The speaker is merely choosing to draw attention to the sound heard rather than the motion observed.)
Beneficiary dative construction: The use of a single word to serve as the understood indirect object and also as the dative (recipient) object. Datives are not normally specified in English syntax, but Latin speakers will recall that in a sentence like, “Give the ball to Mary,” requires a special dative case marker for Mary. e.g., “Build Mary a house.” (Build a house [for] Mary [and give Mary the house].)
None of these meanings can be explained in formal syntax without recourse to hidden (unspoken) elements, and sometimes they cannot be explained at all. Such hidden elements are difficult to reconcile with a linguistics that sees speakers piloting the listeners’ attention. Tomorrow we will discuss explanations that can be reconciled with that view.