The horse (that) raced past the barn fell

[ná'oolkiłí]

In the psycholinguistics portion of my introductory linguistics course, the sentence "the horse raced passed the barn fell" was given as an example of a so-called garden path sentence. Supposedly its syntax tricks you into parsing it ungrammatically at first, but nonetheless has a grammatical reading. Other garden paths where omitted relativizers are (partially) the culprits, like "the man whistling tunes pianos" or "the cotton clothing is made of grows in Mississippi", were all eventually sensical for us, but no one could wrestle a grammatical reading out of this one (besides one where "barn fell" is some sort of compound noun).

Is this sentence grammatical in anyone's dialect? If not, why can't the "that" able to be omited, and what are other situations where you can't? I'd also be interesting in hearing about the limits of omitting relativizers or the existence of garden path sentences in other languages.

[Jipí]

Prosody is what makes this less gardenpath-y.

[finlay]

I can only grammaticalise it as 'the horse that was raced past the barn fell'... the sense 'the horse that raced past the barn fell' is completely nonsensical and ungrammatical for that sentence. Raced would have to be a passive adjective or something. I think there may be something in the fact that it would be the subject of the relative clause that means you have to insert a non-zero complementizer. or something like that.

Frankly, this is a perfect example of why I discontinued syntax and its ilk; they don't use real data but idealised hyperreal constructed sentences that they can shoehorn into a mathematical model. It's not so interesting for me when they say 'well this sentence disproves X' and I'm just like 'but nobody in their right mind would ever say something like that and I find it ungrammatical anyway' and then they're like 'well for the purposes of this course it's grammatical' 'fffffff-'.

[linguoboy]

It's grammatical for me, but I could be influenced by hearing it used so often. I'm trying to come up with parallel sentences with the same syntax, but it's tough to find another verb used this way that's also more transitive than intransitive. (I thought of "the man elected three times lost", but elected is more object-oriented with singular nouns than subject-oriented.) Maybe "the man called for assistance died"?

[finlay]

It's grammatical for me, but I could be influenced by hearing it used so often. I'm trying to come up with parallel sentences with the same syntax, but it's tough to find another verb used this way that's also more transitive than intransitive. (I thought of "the man elected three times lost", but elected is more object-oriented with singular nouns than subject-oriented.) Maybe "the man called for assistance died"?

You see, the only way I can grammaticalise the final sentence is if the man who was called for assistance died... so somebody cries for help and the person who comes to help him dies... which is a bit odd but conceivable really.

[HandsomeRob]

Add a comma:
The horse raced past, the barn fell.

[Ossicone]

I think I was conditioned to punch someone every time I hear that sentence.

[ná'oolkiłí]

I can only grammaticalise it as 'the horse that was raced past the barn fell'... the sense 'the horse that raced past the barn fell' is completely nonsensical and ungrammatical for that sentence. Raced would have to be a passive adjective or something. I think there may be something in the fact that it would be the subject of the relative clause that means you have to insert a non-zero complementizer. or something like that.

That passive interpretation was discussed in class, but our professor was adamant the active sense was the intended one (then again our professor was a bit of an idiot). Now that you remind me of it I can grammaticalize it that way too, but only after some concentration; IMD I think there's something very unnatural to drop the "that was" in relative passives (but I suppose that's the point of the garden path!).

Frankly, this is a perfect example of why I discontinued syntax and its ilk; they don't use real data but idealised hyperreal constructed sentences that they can shoehorn into a mathematical model. It's not so interesting for me when they say 'well this sentence disproves X' and I'm just like 'but nobody in their right mind would ever say something like that and I find it ungrammatical anyway' and then they're like 'well for the purposes of this course it's grammatical' 'fffffff-'.

I'm only one semester in and I'm getting this sense too =(

It's grammatical for me, but I could be influenced by hearing it used so often. I'm trying to come up with parallel sentences with the same syntax, but it's tough to find another verb used this way that's also more transitive than intransitive. (I thought of "the man elected three times lost", but elected is more object-oriented with singular nouns than subject-oriented.) Maybe "the man called for assistance died"?

Yeah, I tried coming up with similar sentences, but the voice and valency interplay really really seems to be what is crucial. Race is a very peculiar verb in this respect. I think "The horse raced fell" could only be said to have very poetic grammaticality; "The horse raced in the derby fell" is much better (though I still get a hint that it should be "the horse that raced in (like ushered in, I guess) the derby fell").
Hmm, "the man called for assistance died" looks like a pretty good parallel, but there's something strange about it. Maybe passivizing light verbs is iffy in general? What would the active form of "the man was called for assistance" be?

Add a comma:
The horse raced past, the barn fell.

Yeah, that's one possible reading, but other garden paths are made sensible by realizing there are supposed to be relative clauses.

I think I was conditioned to punch someone every time I hear that sentence.

Haha, I have a similar reaction =P

[Rory]

Garden Path sentences that I use in my course include:
The old man the boats
The author wrote the novel was likely to be a best-seller
While the woman dressed the baby threw up

I really dislike the horse sentence, partly because I find it so hard to parse, and so do most students. An easier sentence, but with the same structure, is:
The woman driven to the hospital fainted

Most garden path sentences are prosodically disambiguable. (Is that a word? It is now.) Indeed, even so-called "globally ambiguous" sentences can be solved through prosody (e.g. "the man saw the girl with the telescope").

[Nortaneous]

While the woman dressed the baby threw up

I don't get this one. How is it a garden path?

[ná'oolkiłí]

The baby could be seen as the patient of "dressed".
I can't get a garden path out of The woman driven to the hospital fainted, though =S

[zompist]

I can't get a garden path out of The woman driven to the hospital fainted, though =S

It's not one; it was given as an example of the exact same sentence structure as the horse sentence.

[Radius Solis]

It's grammatical for me, but I could be influenced by hearing it used so often. I'm trying to come up with parallel sentences with the same syntax, but it's tough to find another verb used this way that's also more transitive than intransitive. (I thought of "the man elected three times lost", but elected is more object-oriented with singular nouns than subject-oriented.) Maybe "the man called for assistance died"?

For one thing, the garden-pathing relies on the past tense and past participle being the same for regular verbs:

1a) The rock thrown into the water splashed.
1b) *The rock threw into the water splashed.

It also relies on confusing whether the subject is an agent or a patient of the following VP, which depends partly on animacy; rocks are inanimate and cannot be expected to do any throwing themselves, so let's change that:

2a) The man thrown into the water splashed.
2b) *The man threw into the water splashed.

And we might find animals to be ideal subjects, because they are animate enough to be easily interpreted as agents while still often being patients:

3a) The dog thrown into the water died.
3b) *The dog threw into the water died.

So now let's go back to a verb that doesn't have a separate past participle:

4) The dog walked around the block died.

Perfect! Another garden-path. Although the verb is not what you were looking for in terms of being more often transitive than intransitive, I'm not sure why that matters so much, so long as it can go either way. We can even do a nice garden path with a more strongly intransitive verb like "pray":

5) Our father prayed silently calms me down.

(That should be read as: "Our Father", (when) prayed silently, calms me down.")


NINJA: About five more posts were made since I began typing this.

[Thomas Winwood]

I can make it work if I add an implied second clause, something like the horse raced past the barn fell (but the horse raced over the hill kept going).

[Rory]

While the woman dressed the baby threw up

I don't get this one. How is it a garden path?

[While the woman dressed the baby] [threw up]

You stumble at the "threw up", as it doesn't make any sense. You quickly resolve it, though:

[While the woman dressed] [the baby threw up]

[zompist]

Frankly, this is a perfect example of why I discontinued syntax and its ilk; they don't use real data but idealised hyperreal constructed sentences that they can shoehorn into a mathematical model. It's not so interesting for me when they say 'well this sentence disproves X' and I'm just like 'but nobody in their right mind would ever say something like that and I find it ungrammatical anyway' and then they're like 'well for the purposes of this course it's grammatical' 'fffffff-'.

I'd say this is a failure of your professors rather than of syntax. Radius and Rory have just shown that the structure of the horse sentence is perfectly grammatical.

It's been noticed for years that grammaticality judgments can depend heavily on context. People don't evaluate single sentences in a vacuum, after all; they interpret conversations and texts. Do you have any problem with the paragraph below?

We placed microwave emitters of varying strength in various farm buildings and tested whether they had any impact on the horses as they galloped by. The horse raced past the silos showed no effect at all. The horse raced past the farmhouse stumbled. The horse raced past the barn fell.

Ideally we'd use sentences drawn from actual conversation (or, with caveats, actual written text)— and some linguists, such as Geoffrey Sampson, insist on doing so.

What's more tricky is that judgments can vary by dialect/idiolect, and may be graded (that is, there are levels of unacceptability). Of course it's awkward for the professor if the point he's trying to make depends on a judgment not shared by some of the students.... that's one reason that any point in syntax should ideally be supported by several arguments!

It's also rather unfair to talk about syntacticians shoehorning the data into their models. No one enjoys breaking syntactic models like another syntactician... hell, if you can come up with a good sentence that breaks a model it'll be named after you, like Bach-Peters sentences.

[Radius Solis]

It's been noticed for years that grammaticality judgments can depend heavily on context. People don't evaluate single sentences in a vacuum, after all; they interpret conversations and texts. Do you have any problem with the paragraph below?

We placed microwave emitters of varying strength in various farm buildings and tested whether they had any impact on the horses as they galloped by. The horse raced past the silos showed no effect at all. The horse raced past the farmhouse stumbled. The horse raced past the barn fell.

Oh, interesting.... since I know you're quite right about judgements depending on context, I was expecting to have no problem at all with whatever the text was. And then I read it. And despite expecting to see one of the very beasts we've been talking about and just came up with several of myself, I still stumbled when I came to the first one here. Swapping in "silos" for "barn" was all it took to trip me!

The human brain moves in mysterious ways.

[zompist]

Yeah, it's interesting why we have such problems at all. The "garden path" idea is really a theory: we are building a parse tree and if the next item can't be slotted in, we balk. Chomsky has always been notoriously uninterested in how his mental grammars are actually implemented, and so far as I know he has no explanation for the variable performances we see here.

But Tomasello has one. He and some other cognitivists throw out the parse trees, more or less, and concentrate on individual constructions. From their point of view, we don't yacc-and-lex our way through a sentence building a parse tree; we compare the sentence to constructions we know. As knowledge varies by person, is contextual, and can be firmer or softer, this nicely explains the sorts of trouble people have with this construction.

And from that point of view, if you read it enough times you'll learn the construction, and ultimately get impatient with undergrads who can't get it.

[Bob Johnson]

In a vacuum:

The horse that raced past the barn fell.
The man who is whistling *pause and prosody reset* tunes pianos.
The cotton *prosody jump* clothing is made of *pause* grows in Mississippi.
The old *pause and prosody reset* man the boats.
The author wrote *speech quotes (they're like fingerquotes except different)* the novel was likely to be a best-seller.
While the woman dressed *speech comma* the baby threw up.

I don't know what to call those prosody effects exactly. zompist's paragraph makes the sentence clearer, but it still sounds weird.

I agree with the idea that we should use actual conversational or written language, rather than contrived examples, no matter how reasonable they may seem. Rubin (1998) explains quite well that many of the example sentences given in Japanese textbooks are, in fact, quite odd; for example, Zō wa hana ga nagai (p. 43; 象は鼻が長い). This could be simply explained as "As for elephants, their noses are long;" but in actual conversation, this would only come up as part of a list describing various animals and their identifying traits: "Giraffes have long necks, lemurs have big eyes, ... and as for elephants, well, they have long noses." (p. 44) Normal speech would drop either the topic or subject phrase, or perhaps both -- "Oh, look, an elephant!" "Wow, what a nose!" "Yeah, it's long!"

I also agree with the hypothesis that we compare word-strings to familiar constructions; this could, perhaps, explain the ongoing loss of grammatical subject/object contrast in English pronouns. For example, "Him and me are going to the movies" -- one wouldn't say "Me are going", but "Him and me" gets chunked into a noun phrase without case checking since it's a familiar word-string. (I also think this particular example is castigated out of proportion to its actual usage rate.)

Rubin, Jay. (1998) Making Sense of Japanese. Kodansha International. ISBN 4-7700-2802-4

[ná'oolkiłí]

We placed microwave emitters of varying strength in various farm buildings and tested whether they had any impact on the horses as they galloped by. The horse raced past the silos showed no effect at all. The horse raced past the farmhouse stumbled. The horse raced past the barn fell.

The context doesn't help me either— i still think "raced" is an intransitive verb of motion and don't visualize any rider. If as they galloped by is replaced with as we spurred them on (coaxing out the reading "The horse [that was] raced [by us] ~ [that was caused to] race...") or perhaps as the men on ATVs urged them to compete with them in a gallop ("The horse [the men on ATVs] raced..."), however, I get a fairly clear meaning.

But Tomasello has one. He and some other cognitivists throw out the parse trees, more or less, and concentrate on individual constructions. From their point of view, we don't yacc-and-lex our way through a sentence building a parse tree; we compare the sentence to constructions we know. As knowledge varies by person, is contextual, and can be firmer or softer, this nicely explains the sorts of trouble people have with this construction.

And from that point of view, if you read it enough times you'll learn the construction, and ultimately get impatient with undergrads who can't get it.

Haha, that is very true! The construction is becoming less unintuitive to me.

[finlay]

Frankly, this is a perfect example of why I discontinued syntax and its ilk; they don't use real data but idealised hyperreal constructed sentences that they can shoehorn into a mathematical model. It's not so interesting for me when they say 'well this sentence disproves X' and I'm just like 'but nobody in their right mind would ever say something like that and I find it ungrammatical anyway' and then they're like 'well for the purposes of this course it's grammatical' 'fffffff-'.

I'd say this is a failure of your professors rather than of syntax. Radius and Rory have just shown that the structure of the horse sentence is perfectly grammatical.

I did have a series of crap lecturers. I dunno. It's your call; I do understand this sentence now that it's been made clear to me that I was right in thinking that it was the passive construction (and I've heard it plenty of times before), so it was more the fact that the OP put '(that)' in the title rather than '(that was)', and by proxy the fault of his lecturer for saying that it should be 'that'. But I had plenty of examples of this kind of lecturing come up, and it's probable that it was something to do with undergrad linguistics where they do try and shoehorn things into out-of-date but simpler theories – lies-to-children and all that – because they just need to teach us the bare minimum so that we all have some knowledge of it and those who want to continue with it can. And for me it's a fine line between wanting to know the next stage of the theory and just getting pissed off at what they're teaching us which just seems obviously wrong. And liking phonetics better anyway.

But no matter whether it's because of the lecturers or the subject itself, that's why I couldn't continue with it.

[zompist]

I was taught by a professor deeply involved in the syntax wars of the '70s, so we didn't just learn syntax, but how to argue about syntax. One of the neat things about syntax at that time, at least, was that it was a young science, still discovering its facts— you could discover new things in the course of a class discussion or writing a paper. Try that with physics!

[Radius Solis]

And for me it's a fine line between wanting to know the next stage of the theory and just getting pissed off at what they're teaching us which just seems obviously wrong.

You just perfectly described my own reaction to the introductory syntax texts I've read: an unstable balance between wanting to know what the real theory is all about and being upset with all the lies-to-children that I can see through. Unstable, because at times one thread of that dominates and at other times the other does.

I'm having a similar reaction to the Optimality Theory book I'm presently reading. Much of it is absolutely terrible, and much of it is excellent. It stands head and shoulders higher than generative grammar in scope, elegance, and explanatory power... while simultaneously relying on a ludicrous core model of speech production. Plus I have yet to run into one falsifiable sentence. So I don't know whether to love it or run away as hard as I can, and knowing me, I'll end up doing a bit of both.

[Jashan]

Similar to another poster, I can grammaticalize The horse raced past the barn fell only if it's intended as a passive: The horse [that someone] raced past the barn, fell.

There's no way that I know of in any 'standard' variety of English allows for an active interpretation of raced in this context.

What I think your professor is thinking, is the dropped 'that', as in The horse (that/who) raced past the barn, fell. But I don't think that you can drop it in this instance. Even if you eliminate the adverbial/prepositional phrase, it doesn't sound right: *The horse raced fell. You simply have to have the "that" explicitly: The horse that raced, fell.

Why this particular construction demands that, I don't know.

[Radius Solis]

Why this particular construction demands that, I don't know.

Ambiguity avoidance. Normally we leave a gap in relative clauses where the relativized noun goes:
a) the man that I saw 0
b) the man that 0 saw me

So when you drop the relativizer you get:
c) the man I saw 0
d) *the man 0 saw me

Compare to the independent clause:
e) The man saw me.

Since (d) and (e) are identical (discounting prosody), English bans one of them, presumably because the benefit in terms of confusion avoidance is strong enough to motivate doing so.