Preposition, final.

It is a widely held belief that it is incorrect, unorthodox, or simply dispreferred to use sentences with prepositions in final positions in literary or written English. This is a misplaced belief. Geoff Pullum points out that it all starts in 1672 with an essay by John Dryden, titled "Defence of the epilogue", in which the author enumerates a series of examples of unorthodox writing in several contemporary authors' works. One of such mistakes was indeed to have a "Preposition in the end of the sentence". There are many criteria to establish that this is a made up canon. First of all clarity and conciseness. As the guys over at Language Log noticed, the On-Line New Yorker article search engine, when stumbling upon a search that returns no results, warns the user with the following message:

??I'm sorry I couldn't find that for which you were looking.

Compare this with the more natural,

I'm sorry I couldn't find what you were looking for.

As it should appear evident, the second form is more natural, clearer and more concise. If anything, to conform to a more ltiterary style, the contracted verbs could appear in their standard, non abbreviated form (I am / I could not).

The second issue here is that to look for with the meaning of to search or to seek is really not a generic verb supported by a prepositional complement. This would be the case of sharing a room with somebody, for example. In such cases, separating verbs from the prepositional complement is not nearly as bad to the ear as it is in the case of a phrasal verb, such as to look for. The sentence

That's not the man with whom I shared the room.

is much better than the above example. Still, That's not the man I shared a room with is just as clear, and slightly more concise (7 words against 8).

The interesting thing is that the English language - at least when compared with other languages, or cultures associated with other languages - has always had less of a prescriptivist character tied to it. Italy and France for example, still today have institutions dutybound to protect their languages and charged with the task of dictating their correct and standard usage. The OED - arguably the most authoritative source of information on history and usage of the English language - has always prided itself of its descriptivist approach to documenting the evolution and extent of the English idiom. It is nontheless curious that so many have fallen prey to the rants of a single, opinionated man. And HE didn't even have a blog as a pulpit.

Gee... you're sooo "so"!

Apparently the word "so" used to have by itself the meaning of gay, homosexual. The OED cites one of its early usages in this sense:

b. slang. Homosexual. Obs.

1937 in PARTRIDGE Dict. Slang. 1963 C. MACKENZIE Life & Times II. 254 ‘I've come to the conclusion,’ he told me, ‘that I'm not really “so” at all. I much prefer girls.’ At this date [sc. 1899] the cant word among homosexuals for their proclivities was ‘so’. That seems to have vanished completely from current cant. 1968 J. R. ACKERLEY My Father & Myself xvi. 192 A young ‘so’ man, picked up by Arthur in a Hyde Park urinal. 1973 Daily Tel. (Colour Suppl.) 23 Feb. 51/4 Wilde used to call him ‘the architect of the moon’. Rothenstein, Beerbohm,.. and Epstein were his more predictable friends, as he was not.. at all ‘gay’, as it is now called, or, as it was then called, ‘so’.

Language Hat readers entertain a discussion on the etymology of this use. The most inspiring hypotheses are a dutch derivation from the word "zo" and the more straightforward so = "that way", which contextually nudges to the salient meaning.

Copy-editors sued for [language] battery?

Last night I came across something very interesting. The folks over at Language Log have discussed at length the semantic effects of careless copy editing and of selective quotations. The example I came across – although not exactly falling under either one of these – seems to be just as striking. I assume it must be a common practice to edit out selected portions of judicial decisions in law textbooks, where the intent of the editor is that of eliminating information that interrupts the logical flow of the discourse (such as case or essay citations) and thus helping the student-reader to understand and analyze the legal reasoning appreciating its ramifications, without being sidetracked by pleonastic language or information intended for a completely different audience (as in fact, case citations would be). This seems to me a completely reasonable objective and thus a wholly legitimate practice – if, that is, it were to be applied sensibly.

Living with a fresh-out-of-the-first-week law student – and being the law and language geek that I am – I was welcomed last night at my return home with an interesting read. It was a 1994 decision of an Ohio appellate court, Leichtman v. WLW Jacor Communications, Inc., 634 N.E.2d 697 (Ohio Ct. App. 1994). As David asked me to read through the case and see if everything made sense to me, I found that I too was stumbling on exactly the same passage he was struggling with.

The case deals with an anti-smoking activist who sued a radio co-host of a show where he was a guest, alleging battery, following the intentional and provocative act of said co-host who had lit up a cigar and blown smoke across the activist’s face. The trial court held that blowing smoke onto somebody could not be considered battery and dismissed the claim. The appellate court reverses and remands the case for further proceedings, thus legitimizing the claim of battery. It so does on grounds that are set forth in the following passage, with the editing shown as in Torts & Compensation, 5th Ed., Dan B. Dobbs and Paul T Hayden, Thomson West, pp. 44:

In determining if a person is liable for a battery, the Supreme Court has adopted the rule that "[c]ontact which is offensive to a reasonable sense of personal dignity is offensive contact." [...] It has defined "offensive" to mean "disagreeable or nauseating or painful because of outrage to taste and sensibilities or affronting insultingness." [...] Furthermore, tobacco smoke, as "particulate matter," has the physical properties capable of making contact. R.C. 3704.01(B) and 5709.20(A); Ohio Adm.Code 3745- 17.

As alleged in Leichtman's complaint, when Furman intentionally blew cigar smoke in Leichtman's face, under Ohio common law, he committed a battery. No matter how trivial the incident, a battery is actionable, even if damages are only one dollar. [...] The rationale is explained by Roscoe Pound in his essay "Liability": "[I]n civilized society men must be able to assume that others will do them no intentional injury—that others will commit no intentioned aggressions upon them." Pound, An Introduction to the Philosophy of Law (1922) 169. [...] We do not [...] adopt or lend credence to the theory of a "smoker's battery," which imposes liability if there is substantial certainty that exhaled smoke will predictably contact a nonsmoker. [...]

The incriminated passage here is

We do not […] adopt or lend credence to the theory of a "smoker's battery," which imposes liability if there is substantial certainty that exhaled smoke will predictably contact a nonsmoker.

The first time around, this passage seemed incoherent to both of us. The passage strikes the reader with immediate and apparent contrastive semantics to its preceding context and yet something in it was eliciting a different reading, namely that of the recap or conclusion to the rationale exaposed in the paragraph. On semantic grounds this seemed completely incoherent. The court had just terminated stating that the act in question – indeed exhaling smoke which will predictably contact a non-smoker – was indeed a battery. Furthermore, the facts in the case seem to apply to what one would at first describe as “smoker’s battery”. Now, it’ll be apparent to all those who are familiar with the legal ramifications of terms such as substantial certainty[1] that this passage is actually preventing a generalization of the rationale applied here from being made to the general case of substantially certain second-hand smoke. And here is my point exactly. While the passage as edited requires some additional level of domain-specific knowledge to construct the correct contrastive discourse/rhetorical relation with the previous text, if the text had been included in full, such knowledge wouldn’t have been necessary at all (thus considerably facilitiating the job of the poor first year student!). This is substantiated by the fact that as soon as I could read the full text (thanks google!) of the decision, the coherence of the passage was restored and I was actually able to explain what it was that the court was addressing in writing that paragraph. And now, the mysterious removed text, the language that holds the key to the correct interpretation, will be revealed!

We do not, however, adopt or lend credence to the theory of a "smoker's battery," which imposes liability if there is substantial certainty that exhaled smoke will predictably contact a nonsmoker.

In some theories of discourse analysis, linguistic features hold a key role in guiding the discourse syntactic interpretation assigned to natural language texts (or speech, for that matter). Absent the however bit from the previous paragraph, everything else linguistically suggests a structure incoherent with the semantic content of the text. It is not clear to me what are the exact linguistic conditions that made such structure salient, as this is one the current foci of my personal research, but in retrospective I seemed to have instinctively chosen a structure where the final passage is in some way recapping the argument, and providing the final word on how the rationale in toto was to be construed. Arguably - and one's prosodic production in an out-loud first instinctive reading of this passage could indeed provide some useful insight in settling the matter - I unknowinlgy and subconciously assumed that the missing illocutionary operator was something along the lines of therefore rather than however - which of course would have made no sense, thus causing the puzzling over the case in the first place.

When we analyze the unedited text, the presence of the strong discourse marker however suggests and makes salient a wholly different structure. The conclusive caveat put forth by the court is contrasted with what has been said thus far, as to limiting the scope of applicability of the first part of the argument. What we have here - it seems - is a quite complicated and insidious case of garden path discourse: a semantic incongruence is requiring a correction on a discourse structure made salient by linguistic factors (or more likely by statistical distributions conditioned on linguistic information and genre constraints). What is quite interesting is that the mere reinsertion of the original discourse operator seems to prevent the GP effect, probably affecting the salient statistics.

In short, there are – it’s true – numerous situations when the removal of information-less pieces of language such as discourse cues and markers does not alter the preferred structure that the language interpreter is building as she goes along, because of the fact that linguistic [and epistemological] evidence is redundant and complementary. On the other hand, there can be instances when the contributions given by such devices are essential for a speedy and effortless correct interpretation.

Furthermore, on the subject of copy editing, it seems to me that it should be a desirable exercise in humility to show deference for the careful way in which language is often crafted by the courts in their decisions for the specific purpose of requiring the minimal amount of guesswork in interpreting their findings; which in turn implies putting more care in editing language which has the potential of carrying significant rhetorical force.

NB: A reminder for the jurists that wander these places: this entry’s purpose is to show how there’s something funny going on here about the linguistic properties of this particular case of editing. It furthermore involves issues in discourse interpretation that interest me greatly. It seems quite clear to me that those who are particularly knowledgeable about tort-law (which a first year law-student after one week of class is probably not) would not necessarily be puzzled by the passage at hand. 20/20 hindsight will also play the same trick; in fact I can’t seem to feel as easily befuddled by the language as much as I did the first time I read it. Nonetheless both David and I puzzled over it for several minutes. I consider this worth of mentioning because in my view it legitimizes some of the belief I hold about discourse processing and the importance of linguistic evidence (as opposed to semantic and knowledge-intensive) for decisions about discourse grammaticality.


[1] Or who are familiar enough with tort law to actually recognize the phrase “smoker’s battery” from the legal literature as charged with specific meaning and associated with well-defined contexts. I most certainly was not. Only the next day after some research it appeared clear to me that one could have used this knowedlge as a signficant clue and therefore not be confused at all. The fact that the editors most likely possessed such knowledge and were therefore convinced of the clearness and coherence of the edited text would seem to contribute to this thesis.

To have or not to have....

Over at Language Log, Arnold Zwicky describes some seriously non standard use of have (meaning to own) and have (meaning to carry, to have with you) as concise and effective at disambiguating the intended sense:

They were supposed to do a half-hour of silent reading and write about it, but only a handful brought books. The rest... were allowed to write an essay on why it's important to bring your book. "If I write, 'I ain't got it; that's why I don't got it,' is that worth points?" asked one of three boys who taunted the young teacher the entire two hours.

Zwicky dismisses the own/have with me option as slightly off, semantically - since the student in this specific example would not be owning the book but just have it assigned it to him for the duration of the school term. Even accepting this fact, there are at least two options - one of which sounds perfectly fine to me, the other which I believe to be slightly non-standard in American English. If anybody cared to comment and give me their native-speaker intuition on either one of these....

The first one - which in my intuition sounds standard and evokes the right semantic framing - would have been to use "If i write, 'I don't have one; that's why I don't have it,". In this case the usage of an existential quantifier versus the definite anaphoric reference to the book evokes the correct semantics, at least for me. Does it, for anybody else?

The second one is even better, at least from my intuition, although I fear it may not be standard in American English. Any native British English speaker cares to comment on whether it is ok for them? So, how about if the language used had been "If I write, 'I haven't got it; that's why I don't have it,"? Is it still ambiguous? Is it non-standard English? Both questions are no's for me but the jury's still out. Of course, "If I write 'I haven't got one; that's why I don't have it," seems to be the best choice of all. Clear, succinct, unambiguous, and quite standard.

Garden Path discourse

I've sometimes mentioned this particular example with friends and colleagues because I always thought it was particularly interesting. And kind of funny too. Anyway, if you've ever taken a Muni car in San Francisco and are a little observant, you may have seen a sign on the side of one door that stated - I go by memory so if somebody sees it and can correct me, the better. Especially for spacing and line breaking, I don't recall exactly - something like this:

PLEASE HOLD ON
SUDDEN STOPS
NECESSARY

Now this wouldn't be particularly funny, or interesting if it weren't for the fact that both I and a few others that noticed the sign experienced a variation of the well known garden-path phenomenon. Garden Path sentences are sentences that - because of the statistically unlikely usage of certain words in combination - initially lead the reader to build the wrong interpretation of the sentence, and only going back and rereading the sentence taking the less-likely local interpretation (the garden path, indeed) can the sentence be assigned a coherent meaning. A famous example is "Fat people eat accumulates". Now, mostly this phenomenon tends to occur in processing written language because prosodic information can give a great deal of clues about which path is actually the more likely to yield the correct interpretation (making statistical distributions actually dependent on more than just morpho-syntactic characteristics of the words/sentence).

The example from the SF Muni cars is interesting because the text is actually a small discourse fragments. The correct interpretation is naturally "Please hold on / Sudden stops [are] necessary." It is interesting that my first experience with this sign - and reputedly, that of a few other people whose reactions I could observe when first confronted with the sign - was something to the effect of "Please hold [on sudden stops]PP. / ???Necessary" at which point the interpretation had necessarily gone bad and needed that I went back and through the garden, so to speak.

Now, the interesting fact is that while new-line information is initially consistent with the sentence breaking (a CF after hold on should lead us to believe that there is a sentence break there) I in fact seemed to have ignored this fact and interpreted the remaining part as direct continuation of the first line. In my intuition two factors are probably affecting this, one there is probably statistical evidence that hold on is less likely to be interpreted as a phrasal-verb construction than it is to be used in its literal meaning, with a PP argument. A curious side comment is that a statistical evaluation in the language interpreter's mind is most likely contextualized by her surroundings. Hold on in its phrasal verb construction (with on as a particle) is more likely to mean "wait a second" than it is to mean "hold on something"(*) In the context of a riding a streetcar - which I was - I was quick to (correctly) assume the sense of grabbing for the word hold, in which case the PP argument was more likely. This notwithstanding the clear clue of a new line after the particle on. Admittedly, the genre constraints given by the fact that it was a sign - and therefore that new lines are more spatially than semantically constrained - must have certainly played a significant role in all this...

If anybody can find any other example of garden-path discourses, make sure to post them here. I'm sure they're out there.

(*) I've tried to quickly verify this with google but was vastly impeded in my attempt by some recurrent Green Day lyrics....

Why "The Harmonic Mind"?

I am finally finding the time and the topic for my first post. First a word of caution for anybody who has found themselves reading this blog. I intend to use this space for several different purposes. First and foremost I want this to be about language. Language and its aspects affect and account for so much of what defines who I am, what I do, and what I like, that in very recent times I have felt the need for a space where I can write about the things I am interested in, the ones I don't quite understand, the ones that puzzle me, or that inspire me to more thought. There are a few other blogs out there - some very influential ones (I'll be building a blogroll soon) - that fulfill in some shape or form the same purpose. They do so, though, for their respective authors. I enjoy reading them and I will continue to do so, but on the other hand I wanted to have a space of my own where I can initiate similar discussions, or continue some of the ones that have been spun off elsewhere.

On the other hand, I hope with time to make this a place of exchange with others who share some of the same curiosities, passions and interests about the topics I will write about. This is more of a wish than a real objective for this effort.

Finally I intend this to be a tool for myself, for my own development and learning. Although I'm working full-time as a researcher in computational linguistics, I am putting a lot of thought and effort in completing my academic journey. I have finally somewhat made up my mind about my dissertation topic and I expect that there will be traces of the progress I will make, in this blog as well. From time to time, I will reflect on the issues that more closely relate to my research and by that time, hopefully enough people will be reading this that I will actually be able to benefit from feedback, and other people's insights and opinions about such topics.

As a final cap to this first post, I will spend just a few words on the title of this post, and not so incidentally of course, the title of this blog. As I believe it often happens with any name, or title, or brand that one finds particularly appropriate or that just fits right whatever communicative intent one wants to exercise, this name too, has been used before. Curiously enough, I didn't know, until I came up with it and bothered to google it. Incidentally though, its previous usage fits the frame and the purpose it will serve here. "The Harmonic Mind" seems to have been the title of a series of talks first, and of a book in more recent times, by Paul Smolensky. It refers to the parallelism and profound links between the connectionist view of cognitive processes and language processing and the optimality-theoretic framework of harmonic selection invented by Smolensky and Prince. I had chosen it because it appropriately describes at least one of the general themes I will try to touch which closely connects to my thesis research - harmonic selection in discourse-level language processing - and because it sums the wish I have for this forum, a harmonious convergence of my ramblings and questions, and the precious exchange of ideas with whoever will stop by and be interested enough to read any further.

Coming Soon