This semester I'm taking a pretty cool semantics class. It deals with a lot of very interesting complicated subjects that only linguists really worry about, like binding theory, and discourse sensitive referentiality. To do this we use logic, and logic is basically math. Now the question that comes up a lot with my phonologist friends who are like, "uh, yeah, we don't really care about that. We care about sound change." is, "why do you think math is actually useful, in any way, to model meaning?"
What editing has taught me about linguistics is that sentences are incredibly, incredibly perverse. Sentences really are a lot like mathematical formulas. A sentences is a complex combination of variables (like pronouns) and functions (like verbs). And when you hear a sentence, or read one, you fill in all this real-world knowledge, like, where you are, who you're talking about ,what words usually mean, what words possibly mean, how nouns and verbs relate to each other, etc, etc. And if the sentence is well constructed, all these crazy calculations take you exactly where the author wants you to go. You get a meaning.
But there is always some variability in this process. When we learn math we learn things like order of operations and how to combine fractions and the meaning of e and the meaning of ^2. And these are complicated irritating things that we are all universally taught, hopefully clearly, and have to memorize and consciously apply repeatedly before they become at all easy. But they're consistent.
When we're processing language, when we're looking at a sentence: a language formula, we use a method of solving it that we learned basically by accident. We learned it by virtue of growing up as speakers in a speaking environment. We learned it because our parents spoke to us and expected us to understand. If our parents had locked us in a closet and never spoken to us, we would not have any ability to solve this formula. But we do. However, everyone puts language together differently. People who speak the same language don't have vastly different grammars. They have generally overlapping vocabularies, and basic sentence structures, but not everything is going to be the same. Luckily, we can adapt and learn new things as we meet new people.
The other problem is that written text and spoken speech are not precisely the same. Speech is learned naturally, has immediate corrections and interactions. Written texts are supposed to be interpretable by a simple decoding of signs to sounds. But writing isn't speech. Writing is a way of modeling speech, it models not only the words, but the context in which the words are said. It models reality. And we learn this, also, with practice. We read, are read to, have parents or teachers who help and explain, until we get used to it. But speech and writing are not the same. With speech, you can ask and say, "hey, did you get that," and your interlocutor can say, "uh no, I didn't get that at all," and you can say, "okay, I'll try again." With writing, you get one chance to get it right.
Now this was a long digression into language acquisition, and reading acquisition, but I did have a point here. What I'm trying to say, is that language is a way of modeling the world. It's not as formal as math, due to its method of acquisition and development, but it is just as precise, and that is the relevant thing about sentences.
Think of a sentence as a formula, not some abstract sort of change of state functional programming sort of formula, but as an engineering calculation. Formulas, you want them to be simple. You want there to be direct correspondence between your measurements and your numbers (compositionality, lets call it.), you don't want anything extra. But that doesn't mean you only look at the big important numbers. You need the little fiddly ones too. You need friction, and gravity, and air pressure, and wind speed. You might need weather, and temperature, and rate of decomposition of your materials, depending on what you're trying to build.
Sentences are exactly the same. Okay, sure, you need the core meaning. You need the participants. You need the verb. Then you need all the fiddly stuff, like the context: tense, voice, aspect, mood. And you need to know, that every tiny piece of this formula that you're spitting out as if it were easy, as if it were natural, is important. Every choice you make, by inclusion or omission, will become part of the formula, and will change the result, maybe just slightly. But slightly in engineering is a bridge that can survive a windy day and a bridge that falls apart. Slightly in writing, okay, it's not life or death, but it can be life or death to the power of your prose to keep a reader's attention.
So, when you're editing, don't be fooled by the ease at which words spill from your lips or your fingertips. Humans are amazing at online processing. With writing, every word counts, and sometimes it counts more than you could ever imagine.