fraction | Gyre&Gimble

Everyone knows that computer programs have bugs. In fact, languages have bugs, too, although we don't usually call them that.

Bugs in English

Right

Q: "Should I turn left at the next corner?" A: "Right". Probably most Americans who drive now know this bug. The answer could mean "yes" or "turn right". So we have to stop and think how to answer this question. That makes it a bug.

Too, two

Comment: " We will take Route 30". Answer: "We will take Route 30 too". This bug is probably responsible for the survival of the word "also".

Note that unlike the case of "right", this is a bug only of spoken English.

Subject and predicate

In Comma rule found dysfunctional, I wrote about the problem that in formal English writing there is no way to indicate where the subject ends and the predicate begins. This causes a problem reading complicated sentences with many clauses such as academic writing often uses. Of course, one way around this is to write short, simple sentences! (That sounds like the subject of a future blog…)

Bugs in the symbolic language of math

Fractions

In both Excel and Mathematica, "1/2*3" means 3/2. Now, I would think "1/2a" means "1/(2a)", but younger mathematicians are taught PEMDAS (see Purplemath), which says that division and multiplication have the same precedence and operations are evaluated from left to right.

If in Mathematica you define a function f[a_] := 1/2a, f[3] evaluates to 3/2, so Mathematica (and most other computer languages) agree with PEMDAS. (Note: When you write 1/2a in a Mathematica notebook, it automatically puts a space between the 2 and the a, and space in Mathematica means times, so it does warn you.)

Nevertheless, my ancient education would lead me to write (1/2)a for that meaning. This means I must learn to write 1/(2a) for the other meaning instead of 1/2a.

Questions:

Did the language really change or was I always "doing it wrong"? I would like to hear from other ancient mathematicians. (But I don't know very many who would read blogs or Purplemath.)
Should such a phenomenon be called a bug?

Repeated exponentiation

In Excel, "2^2^3" means $(2^2)^3$, in other words, 64. In Mathematica, it means $2^{(2^3)}=2^8=256$. My impression is that most mathematicians expect it to mean $2^{(2^3)}$.

References: This post in Walking Randomly, my post Mathematical Usage, Wikipedia's article.

Exponentiation on functions is ambiguous

If $f:\mathbb{R}\to\mathbb{R}$ is a function, $f^2(x)$ can mean either $f(f(x))$ or $f(x)f(x)$, and both usages are common. You should tell your students about this because no one is ever going to make one of the usages go away.

A far worse catastrophe is the fact that in calculus books, $\sin^2x=(\sin\,x)(\sin\,x)$ but $\sin^{-1}x=\text{arcsin}\,x$. I betcha (lived in Minnesota four years now) we could succeed with a campaign to convince calc book publishers to always write $(\sin\,x)^2$ and $\arcsin\,x$.

Bugs in the Mathematical Dialect of English

The mathematical dialect of English is what I call Mathematical English in the abstractmath website. It is a different language from the symbolic language, which is not a dialect of English.

I have written about the problems with Mathematical English in a ridiculous number of places. (See references in The Handbook of Mathematical Discourse). It is normal for a dialect of a language to use words and grammatical structures that in the original language mean different things. (See Dialects below).

Words with different meanings

A set is a group in standard English, but not in math English.
The number 2+3i is a real number in standard English, but not in math English.
And so on.

Use of adjectives and prefixes

A "noncommutative ring" has commutative addition.
A "semigroup" has a fully defined binary operation.

If, then

The bug that grabs math newbies by the throat and won't let go is the meaning of "If P, then Q".

"If a number is divisible by 4, then it is even" in math dialect means a number not divisible by 4 might be even anyway.
"If you eat your broccoli you will get your dessert" in standard American Parental English does not mean you might get your dessert if you don't eat your broccoli.

And then there is the phenomenon of Vacuous Implication, which leaves students gasping and writhing.

About "dialects"

Most Americans are not familiar with dialects in the sense I am using the word here, since the only really different dialects we have are Gullah and Hawaiian Pidgin, both of which are very hard to understand; although for example Appalachian English and African-American urban vernacular [1] are dialects of a milder sort. I grew up in Savannah and heard diluted Gullah sometimes on the street (didn't understand much). I am also rather familiar with Züritüütsch since we lived in Zürich for a year.

What the rest of the world call dialects have many distinctive properties:

They have nonstandard pronunciation to the point where they are difficult to understand.
They have differences in grammar. (Both Gullah and especially Hawaiian Creole have differences in grammar from Standard English.)
They have differences in vocabulary, enough sometimes to cause misunderstanding.

I grew up speaking an Atlanta dialect, which really did have differences in all those parameters. But what people today call a Southern accent is really just an accent (minor variations in pronunciation), not a dialect.

Hawaiian Creole, and possibly Gullah, but not the other dialects I mentioned, are singled out by linguists as creoles because they been modified heavy influence from another language. Züritüütsch is not a creole, but it is quite difficult for native German-speakers to understand. The Swiss situation particularly emphasizes the distinction between "dialect" and "accent". The typical native of Zürich speaks Züritüütsch and also speaks standard German with a Swiss accent.

Reference

[1] What Language Is (And What It Isn't and What It Could Be) by John H. McWhorter. Gotham, 2011.

Send to Kindle

Sue Van Hattum wrote in response to a recent post:

I’d like to know what you think of my ‘abuse of terminology’. I teach at a community college, and I sometimes use incorrect terms (and tell the students I’m doing so), because they feel more aligned with common sense.

To me, and to most students, the phrase “whole numbers” sounds like it refers to anything that doesn’t need fractions to represent it, and should include negative numbers. (It then, of course, would mean the same thing that the word integers does.) So I try to avoid the phrase, mostly. But I sometimes say we’ll use it with the common sense meaning, not the official math meaning.

Her comments brought up a couple of things I want to blather about.

Official meaning

There is no such thing as an "official math meaning". Mathematical notation has no governing authority and research mathematicians are too ornery to go along with one anyway. There is a good reason for that attitude: Mathematical research constantly causes us to rethink the relationship among different mathematical ideas, which can make us want to use names that show our new view of the ideas. An excellent example of that is the evolution of the concept of "function" over the past 150 years, traced in the Wikipedia article.

What some "authorities" say about "whole number":

MathWorld says that "whole number" is used to mean any of these: Any positive integer, any nonnegative integer or any integer.
Wikipedia also allows all three meanings.
Webster's New World dictionary (of which I have been a consultant, but they didn't ask me about whole numbers!) gives "any integer" as a second meaning.
American Heritage Dictionary give "any integer" as the only meaning.
Someone stole my copy of Merriam Webster.

Common Sense Meaning

Mathematicians think about and talk any particular kind of math object using images and metaphors. Sometimes (not very often) the name they give to a math object embodies a metaphor. Examples:

A complex number is usually notated using two real parameters, so it looks more complicated than a real number.
"Rings" were originally called that because the first examples were integers (mod n) for some positive integer, and you can think of them as going around a clock showing n hours.

Unfortunately, much of the time the name of a kind of object contains a suggestive metaphor that is bad, meaning that it suggests an erroneous picture or idea of what the object is like.

A "group" ought to be a bunch of things. In other words, the word ought to mean "set".
The word "line" suggests that it ought to be a row of points. That suggests that each point on a line ought to have one next to it. But that's not true on the "real line"!

Sue's idea that the "common sense" meaning of "whole number" is "integer" refers, I think, to the built-in metaphor of the phrase "whole number" (unbroken number).

I urge math teachers to do these things:

Explain to your students that the same math word or phrase can mean different things in different books.
Convince your students to avoid being fooled by the common-sense (metaphorical meaning) of a mathematical phrase.