[Sorry… the images in this post are all fouled up. I would put forth a lot of effort to fix it, but I’ve got a lot on my plate this week. Speaking of which, come to the Berkeley Linguistics Society meeting this weekend and you’ll hear an exciting talk on environment prototypicality from me, live and in the flesh!]

This post carries with it an important disclaimer. I do not like Yanni’s music. Not a bit. Nor would I want to read his book. However, I am in the habit of going to thrift stores and rummaging through their books, and that’s where our story begins. The other night at the thrift store, my apartmentmate found a copy of Yanni in Words and showed me the back cover, which Amazon has been so kind as to digitize. What it says is:

“What are the chances that a poor kid from the seaside town of Kalamata, Greece, who doesn’t read music but taught himself to play piano at eight, who doesn’t dance, doesn’t sing, doesn’t write lyrics, doesn’t conform to any particular musical style, is fiercely independent, and doesn’t want to play the show business ‘game’ — what are the chances that this kid will ever succeed, much less become a composer and performer revered in every corner of the globe?”

This is an oddly phrased question for a couple of reasons. The first is the odd lack of parallelism. Everyone from schoolchildren to undergraduates to businessfolk are exhorted to maintain parallelism in writing and speech:

(1) I gave gifts to Pikachu, to Diglett, and to Ponyta.
(2) I gave gifts to Pikachu, Diglett, and Ponyta.
(3) (*) I gave gifts to Pikachu, to Diglett, and Ponyta.

(1) & (2) are both considered fine, but (3) is considered bad. Linguistically, (3) uses the conjunction and to join two prepositional phrases and a noun phrase, and that’s just not the way and works. It’s supposed to join like elements. Also, the omission of to in (3) makes it sound like one of the thoughts you’re making this sentence out of is I gave gifts Ponyta. I wouldn’t go so far as to say (3) is ungrammatical, but I definitely prefer (1) & (2) to (3).

In the Yanni sentence, you’ve got two different non-parallelisms. The first is that who is used for the first two relative clauses, but not for subsequent ones:

… who doesn’t read music but taught himself to play piano at eight, who doesn’t dance, doesn’t sing, doesn’t write lyrics …

This gives the sentence a structure like:

Yanni 1

There’s not anything inherently wrong with this; each new description (for instance, that Yanni doesn’t sing) can be added either as a new relative clause (with its own who) or as a new conjunct within the current relative clause (without its own who). So to get this structure, you just make who doesn’t dance a new relative clause, conjoined to the relative clause who doesn’t read music…, and all subsequent descriptors are conjoined to the verb phrase doesn’t dance within the second relative clause — not the relative clause itself. But this ends up leading to a new problem; there is no conjunction to join the two relative clauses:
Yanni 2

There is a single and in this question, and it is within the second relative clause, so this is an ill-formed sentence. So that’s the first bit of non-parallel weirdness to this sentence.

The second bit is that the negations are not parallel. The relative clauses are all negated but one. You’re going along: doesn’t, doesn’t, doesn’t, is, doesn’t … whoa! Again, this does not make the sentence ungrammatical, but it does make it harder to comprehend (at least for me — I actually had to stop and think for a second when I got the non-negative verb phrase). It’s something that makes a sentence less acceptable, although not necessarily less grammatical.

The last weird thing about this sentence that I want to discuss is its use of resumption. (This is also why my apartmentmate brought it to my attention in the first place.) The question has a huge(!) noun phrase in it, almost 50 words long (a poor kid … business ‘game’). This NP is longer than most sentences — I think it’s longer than any I am using in this post — and that means it is entirely possible for the reader to forget what the beginning of the sentence was by the time they get to the end of the NP. Try it yourself. Can you remember the structure of the beginning of the question was by the time you reach the end of the NP? I had only a foggy memory of it in my head. So to help the reader out, Yanni actually restarts the question with an anaphor replacing the giant noun phrase. The 49 words used at first are reduced to two (this kid), so that the opening structure can be retained in the reader’s mind. It’s something you do when you talk, but it’s just not allowed in written English.

Which I think is a shame.

This is that this is how people actually speak, and resumption is used judiciously in other languages to help the reader out. French, for example, uses resumptive pronouns when you have some long subject that might confound the reader [please, please excuse how dusty my French is]:

Cet homme avec les chapeaux, avec les yeux bleus, avec les chats et les chiens, il est tres gros.
(with which I intend to mean: That man with the hats, with the blue eyes, with the cats and dogs, he is very fat.)

The long noun phrase is introduced at the beginning, recalled with il, and the sentence uses this resumptive il as the subject. That’s just not how it’s done in English. You only get to say the subject once, so you can’t go on and on and then try to simplify it to a pronoun. You’d say “That man with the hats and all that jazz is very fat.” Or rather, you’d write that. But in saying it to myself as I wrote it, I snuck in a he and said “That man blah blah blah, he’s very fat.” It’s especially common in my head when you’re pointing someone out: “That man over there, he’s very fat” or “That man over there, isn’t he fat?” In speech you assume that the listener can’t go back and look at what you said, so to be safe, you’d better repeat it to make sure the idea will come out right, even if that means stepping on the toes of a few grammatical rules. In writing, though, you assume that you should generally stick to grammatical norms and that the reader can always go back if their memory gets hazy.

This is why Yanni’s question is odd. It sounds relatively fine to me, because each grammatical norm it violates is a sort of justifiable. The first one, that he’s got one too few ands or one too many whos, is commonplace in speech, where you can’t go back and check what you’ve been conjoining your phrases to and you don’t necessarily know at the outset how many conjuncts you’ll end up having. It’s just weird to see in writing, because you do have the chance to go back and check the sentence over and instill some parallelism to it. The second problem, the non-parallel negation, is also not so bad because it’s hard to make a negative phrase that means the same as is independent. And the third problem, the resumption, is alright by me because it makes the question easier to comprehend.

That said, the question is still quite awkward to me — but I think that’s the point. It uses the grammar of speech instead of the grammar of writing, probably to subtly establish that this book is intended to read like a conversation or narration, not as a dusty old biography. But it also goes to show that speech and writing really have different grammars, and what sounds fine in speech can send up some red flags when written down.

Summary: The grammar of speech and the grammar of writing are markedly different. What would sound fine spoken can look pretty odd written. A lot of that’s because speech and writing have different constraints; you can go back and re-read writing, but you can’t re-hear speech. As such, the two modalities care about different things: speech cares about clarity, and writing cares about formalism. When you sacrifice clarity in speech or formalism in writing, things seem a little odd. Maybe they shouldn’t.