by-topicalization
« previous post | next post »
Caught on NPR's Weekend Edition Saturday on 7 June, in a story by Wendy Kaufman on weightlifter Melanie Roach that had a section on the autism of Roach's son Drew:
(1:ex) Roach hopes that by talking about Drew’s autism, it will spur more research and assistance for families affected by it.
I've put the relevant clause in bold face. This clause has the form
(1:form) by VP1ing, it VP2
which is characterized by some composition teachers as "wordy". Some would criticize it as "vague" as well, because the anaphoric pronoun it has no noun antecedent (recall our earlier discussion of "vague" pronouns on Language Log, here, here, and here); instead, it refers to the action or event that VP1 denotes. In any case, using (1:form) to convey this meaning is non-standard.
The usual suggested fix is to compact the clause into a simple subject-predicate clause, of the form
(2:form) VP1ing VP2
(with VP1ing now a "nominal gerund", serving as subject; in (1), it's a predicate in a subjectless predicational adjunct). In the case at hand, the fix for the original would be:
(2:ex) Roach hopes that talking about Drew's autism will spur more research and assistance for families affected by it.
This is entirely standard, but it doesn't, I think, get the effect that people who use the construction in (1) are trying to get with it — which is to mark some discourse referent as topical in the discourse (in the by-adjunct) and then say something about it (in the main clause). That is, this "by-topicalization" construction explicitly separates "discourse topicality" and "sentence topic", while these two statuses are fused in the subject in (2).
There are other fixes for by-topicalization. One way is to change the subject in the main clause so as get a garden-variety subjectless predicational adjunct, of the form
(3:form) by VP1ing, NP VP2
(where the subject NP in the main clause supplies the discourse referent for the missing subject in the adjunct). For the Roach sentence:
(3:ex) Roach hopes that by talking about Drew’s autism, she will spur more research and assistance for families affected by it.
This is not semantically equivalent to (1:ex) and (2:ex); (3:ex) attributes the hoped-for spurring of research and assistance to Roach herself, while the others attribute it to Roach's activities. Granted, this is a subtle difference, and possibly not one that's important to the writer or speaker.
There are other possibilities. For instance, you can make the sentence explicitly biclausal, with the topicality marking and the main assertion divided between an initial subordinate clause and the main clause, respectively:
(4:ex) Roach hopes that when/if she talks about Drew’s autism, it/that will spur more research and assistance for families affected by it.
This makes the sentence longer, and it also introduces an instance of it or that that some would object to as "vague", so it's no great bargain.
Now, a bit of background. There are many types of subjectless predicational adjuncts. The ones we're looking at here have initial sentential adverbials of the form
Intro VPing
VPing is a VP with its head verb in the -ing form (which goes by a number of other names: present participle, gerund participle, Form P, PRP). Intro is an introductory element, which in traditional terminology is either a preposition or a subordinating conjunction (or a word that can serve as either of these):
(5) Preposition (P): In signing this letter, you agree to serve a five-year term. By asking that question, you betrayed your ignorance.
(6) Subordinating conjunction (Sub): When eating meat, you should use a fork. While working on their computers, many of the staff listened to music. While/Though not being fond of green peppers, Kim will eat them anyway.
(7) P/Sub: After eating lunch, you should brush your teeth. Before logging out, I always make sure that everything has been saved.
Each Intro contributes meaning to an adjunct (in some cases, more than one possible meaning). The use of by in by-topicalization, in particular, is a specialization of the causal or consequential use of by in examples like the one in (5).
The examples in (5)-(7) have entirely standard adjuncts in them; they are "default SPARs" (they are subjectless predicational adjuncts that require a discourse referent for the missing subject, and they pick this referent up from the subject of the main clause). But (1) is different. Its SPAR picks up its discourse referent (Melanie Roach) from preceding context, rather than from the subject of the main clause; it's a non-default SPAR (a.k.a. "dangling modifier"). The extra twist in (1) is the summative it as subject of the main clause.
I posted to ADS-L back in 2006 about an example of by-topicalization I'd heard on NPR, with by doing that, it … in it. That's when I discovered how hard it was to search for examples — the construction has a open VPing slot in it, with only the by and the it fixed — and how much work it was to weed out irrelevant hits (you have to look at each hit in its full context). At the time I searched on {"by doing that, it"} and got many thousands of hits, many of them SPARs. A large number of these were either default SPARs or non-default ones in which the subject it in the main clause was not summative (often it was a non-anaphoric "dummy it", a place-holder). Examples of each (with the SPAR bold-faced and the subject it italicized):
(8) By Knowledge & By Love is a sound and substantive book. Besides being in its own right an impressive and valuable contribution to Thomistic scholarship, by devoting attention to an important and timely aspect of St. Thomas’s moral theology, it performs the very important service of correcting serious misinterpretations of the Angelic Doctor’s thought. And by doing that it also contributes appreciably, on a broader scale, to the growing and much needed reaction to the kind of loose and wandering reasoning that in recent decades we have seen altogether too much of in moral theology. [the it refers to the book By Knowledge & By Love] (link)
(9) [documentary filmmaker Gini Reticker speaking] I feel the common bond of humanity is fascinating and so I was hell bent on making sure the women were able to tell their own story and they were portrayed in the way I saw them. I also felt that by doing that it is much easier to be inspired by them. [that is, ‘by my doing that it is much easier for people/viewers to be inspired by them’] (link)
Here are some by-topicalization cases I recently googled up:
"This is the first time that we dictated the tempo in the second half," [Detroit Pistons basketballer Tayshaun] Prince said. "By doing that, it really allowed us to get back in the game within the first three minutes." (link)
Bobbie said,
June 14, 2008 @ 2:56 pm
I'm sure you mean assistance, not assistant in your second example above
Stephen Jones said,
June 14, 2008 @ 4:22 pm
As you have said, in (8) 'it' is not a dummy subject as it appears to be in (9) and the original example, and your last three examples.
Frankly I'm glad the construction is not standard English. I'm only sorry that it's English at all.
Arnold Zwicky said,
June 14, 2008 @ 5:24 pm
To Bobbie: yes, of course, the most ordinary of typos, now corrected. It would be more useful to send these minor corrections to the original poster, rather than making them the topic of comments on the substance of a posting, which then invites further comment.
Lance said,
June 14, 2008 @ 6:29 pm
There are a few interesting restrictions on by-topicalization that one could add. First, I think it can only be used on a subject:
(1a) I'm trying to encourage people to start [talking about Fight Club] (by doing so myself).
(1b) *By talking about Fight Club, I'm trying to encourage people to start it.
(2a) I'm trying to make people more open to [supporting gay marriage].
(2) *By supporting gay marriage, I'm trying to make people more open to it.
This may be related to the other fact, which I'm having trouble formulating, but roughly: the gerund in the by-phrase must be subjectless, but it must have the force of a gerund with a (non-PRO) subject. That is, "Roach hopes that by talking about Drew’s autism, it will spur etc." means, in fact, that she hopes that her talking about Drew's autism will spur etc. But you can't in fact say
(3) *Roach hopes that by her talking about Drew's autism, it will spur…
Similarly,
(4) *By us dictating the tempo, it really allowed us…
(5) *By them forcing books on their children, it…
(6) *By me asking the question, it keeps me probing.
even though the paraphrases are "Us dictating the tempo allowed us…", "Me asking the question keeps…" and so forth.
Or so it seems to me, at any rate.
Aristotle Pagaltzis said,
June 14, 2008 @ 6:57 pm
I am surprised that you did not mention switching VP_2 to passive voice:
“Roach hopes that by talking about Drew's autism, more research and assistance for families affected by it will be spurred.”
This is semantically equivalent to 1:ex and 2:ex. However, end weight is not respected.
Rick S said,
June 14, 2008 @ 8:09 pm
I think I see another fix for by-topicalization: Convert the main clause to a passive construction. In all four cases (the original and the last three examples) this yields a grammatical sentence in standard English (I think):
"Roach hopes that by talking about Drew's autism, more research and assistance for families affected by it will be spurred."
"By doing that, we were really allowed to get back in the game within the first three minutes."
"By doing that, they are made to feel more different."
"By asking the question, I am kept probing."
Could it be that by-topicalization is merely a hypercorrection of Avoid Passive, or at least originated that way?
john riemann soong said,
June 15, 2008 @ 12:28 am
"**By supporting gay marriage, I'm trying to make people more open to it."
This doesn't actually seem to me that non-standard — it just is slightly semantically different. (Open to gay marriage as opposed to open to supporting it — and the difference is subtle.)
Phil said,
June 15, 2008 @ 2:08 am
Lance's example 2 is completely acceptable for me (Brit, if it's relevant).
Example 1b could be acceptable with appropriate stress (for example, if "talking about" and "start" were stressed, it would be acceptable, except the "it" would refer to "Fight Club", not "talking about Fight Club".)
Actually, 1b seems to be unacceptable in the meaning Lance wants because it would be odd to use "it" after start to refer to a VP. For the meaning in 1a, you could have: "By talking about Fight Club myself, I'm trying to encourage other people to start."
Arnold Zwicky said,
June 15, 2008 @ 11:44 am
To several commenters: it's important that "by"-topicalization is not merely the form in (1:form), but this form *in combination with* a particular interpretation. (This is true of constructions in general: they are not just forms, but form-meaning pairings.) My examples (8) and (9) are there, in fact, as instances of things that have the same form as "by"-topicalization but are *not* instances of it; (8) is a default SPAR with anaphoric "it" as subject in the main clause, (9) a non-default SPAR with dummy "it" as subject in the main clause. Searching on word strings pulls up huge numbers of both types, and they are irrelevant to "by"-topicalization.
Lance's (2) is, like (8), a default SPAR, and it's entirely grammatical (for all speakers of English, so far as I know) on the interpretation in which the discourse referent for the missing subject is supplied by the subject, "I", in the main clause. But that's not "by"-topicalization.
To Aristotle Pagaltzis and Rick S.: to start with, I didn't claim to be surveying all the possible alternatives to "by"-topicalization. (For some discussion of the way I react to comments like yours, see my posting "This blogging life: Not doing, failing to do, neglecting to do" at
http://itre.cis.upenn.edu/~myl/languagelog/archives/005201.html ) In fact, I deliberately chose not to discuss the passive versions, because they introduce another layer of complexity, one that's not directly relevant to "by"-topicalization.
To start with, the passive versions are mostly awkward; for the Roach example, the agentless passive "spurred" (in "Roach hopes that by talking about Drew's autism, more research and assistance for families affected by it will be spurred") is especially awkward, to my ear, anyway.
More important, some of the passive versions have non-default SPARs. For sticklers, these are all "dangling modifiers" and so are verboten. So, in the passive version of the Roach example, the discourse referent for the missing subject of "talking about Drew's autism" is supplied, not by the subject of the main clause to which the SPAR is adjoined (as sticklers would insist it must be), but by "Roach" in the preceding context. Most people probably wouldn't see this as a problem, and I myself would judge this particular example to be an entirely acceptable non-default SPAR.
Sometimes the conversion to passive produces a default SPAR: "By doing that, we were really allowed to get back in the game within the first three minutes"; "By asking the question, I am kept probing". These are clunky, but the modifiers aren't "dangling".
Other times the conversion has unfortunate results: "By doing that, they are made to feel more different." The intention is that it's the parents who do that (i.e., force books on their children) and that it's the children who are made to feel more different, but it's very hard to avoid the understanding that it's the children who do the forcing, even though that doesn't make any sense. Even in context, we want the subject of the main clause to supply the referent for the missing subject for "doing that". Dangler time.
I have two more examples of "by"-topicalization in my files (but not in my posting), and both turn into nasty danglers by conversion to the passive. So at least some of the time, the passive version is a bad choice.
Aristotle Pagaltzis said,
June 15, 2008 @ 1:18 pm
Certainly. I did not in fact claim that the Roach example was a good demonstration of a conversion to passive voice – it’s awkward for a whole range of reasons. However, all of them seem avoidable with editorial liberties I did not want to take (eg. substituting “encouraged” for “spurred,” using “affected families” to reduce weight, or one of several other possible fixes).
I thought the passive version particularly noteworthy because the sole problem with the by-topicalisation construct (insofar as there is one) lies with the subject of VP_2. Since passive voice obviates the need for an explicit subject, it would seem that when the effect of discourse topicality/sentence topic separation is desired, recasting in passive voice is the best standard strategy – notwithstanding the fact that it doesn’t always work.
Tearlach said,
June 16, 2008 @ 3:43 am
The original question was wether or not it was vague. It is not. Everyone who speaks English knows exactly what both "its" in the original sentance refers to. And as the original poster finally got around to saying, the construction in question shows a high level of complexity, again, being vague. That means information was conveyed and information was recieved. That makes it top shelf English. Who cares if it conforms to the arbitrary level of "standard?"
Tearlach said,
June 16, 2008 @ 6:21 am
"the construction in question shows a high level of complexity, again, being vague. "
I meant …WITHOUT being vague.
Sorry about that.
outeast said,
June 16, 2008 @ 7:43 am
How many people do suffer from Drew's autism, then?
Tearlach said,
June 17, 2008 @ 1:53 am
We apparently don't know. It was too vague.
Jason Orendorff said,
June 18, 2008 @ 6:37 pm
Whew. I have a lot of responses to this one.
1. I disagree with this interpretation of (1). I think the word "by" there kinda evokes action, suggesting something at least a bit more (3)ish than (2). Apart from that, (1) and (2) have the same meaning to my ear.
2. Many of the examples come from unrehearsed speech. My first thought was that this turn of phrase indicates a mid-sentence mental change of direction. You start saying something without thinking about how the sentence is going to turn out. You get to the point where you've said "by doing that", heading for "by doing that, we really…", but then you decide what you actually want to say, and you (subconsciously) recast what you've said so far as an NP by throwing in an "it".
I might be wrong about that, though; maybe "By doing that, it really allowed us to get back in the game…" sounds OK to Tayshaun Prince.
3. This is the first time Language Log has described a bit of nonstandard English that I find flat-out repulsive, so it was a nice opportunity for introspection. The buzzer sounded in my mind as soon as I got to the word "it" in (1). The urge to rationalize was strong. :-) But of course, I have to admit there's nothing actually barbaric about this turn of phrase.
The post is about something specific, but the sentence struck me as an instance of a much broader category of errors where you just plug together the parts of a sentence wrong. I'm still not convinced "by-topicalization" really exists as such. Is it just my bias? Maybe now I'll start hearing it everywhere. :-P
4. In the last example, "By asking the question, it keeps me probing.", the "by" form has a poetic advantage. Judging from the rest of the quote, I'm guessing the author knew it, and I bet this consideration weighed more heavily than the fine semantic distinction you draw between (1) and (2).