This is one of those questions where the textbook says "yes" in one chapter and "no" in another, and both chapters are right. The chapter on cube roots cheerfully writes (-8)^{1/3} = -2 and moves on. The chapter on square roots sternly warns that (-4)^{1/2} is not a real number and you must never try to compute it. Ten pages apart, two opposite policies, no explanation for the gap.

The gap has a very clean explanation. Whether a negative base can carry a fractional exponent depends entirely on the denominator of that fraction. Odd denominators are safe. Even denominators are not. And for exponents that are not fractions at all — \pi, \sqrt{2}, any irrational — negative bases leave the real numbers entirely.

Here is the map.

Case 1 — odd denominator (cube roots, fifth roots, ...)

Every real number, positive or negative, has a unique real n-th root when n is odd. Cube roots are the cleanest example. Ask "what number cubed gives -8?" and there is exactly one answer: -2, because (-2)^3 = -2 \cdot -2 \cdot -2 = -8. The positive candidate +2 cubes to +8, not -8, so it is eliminated. Only -2 works.

So (-8)^{1/3} = -2 is unambiguous. The cube root of a negative is a negative, and nothing breaks. The same story plays out for any odd-denominator fractional exponent: (-32)^{1/5} = -2, because (-2)^5 = -32; (-1)^{1/7} = -1; (-27)^{1/3} = -3.

The structural reason is that raising to an odd power preserves sign. x^3 sends negatives to negatives and positives to positives; it is a one-to-one function over all real numbers. Inverting it — taking the cube root — therefore returns a unique real answer for every real input, including negative ones. All odd-root-denominator fractional exponents of negative bases are safe.

Case 2 — even denominator (square roots, fourth roots, ...)

Now things break. Consider (-4)^{1/2}. This asks for a real number whose square is -4. Any real number squared is non-negative — 2^2 = 4, (-2)^2 = 4, 0^2 = 0. There is no real number whose square is -4.

You have two options.

Option 1. Declare (-4)^{1/2} undefined in the real numbers. This is what pre-calculus and board-level textbooks do. The domain of f(x) = x^{1/2} is restricted to x \geq 0, full stop.

Option 2. Work in the complex numbers, where (-4)^{1/2} = 2i. The imaginary unit i is defined by i^2 = -1, so (2i)^2 = 4i^2 = -4. Engineering and physics courses (and JEE Advanced, eventually) switch to this option as soon as complex numbers arrive.

Both options are consistent; they just disagree about where the answer lives. Pre-calculus stays in the reals and calls the expression meaningless. Complex analysis extends the universe and finds an answer. If you are in a class where i has not yet been introduced, you are in the first world and (-4)^{1/2} is off-limits.

Case 3 — general non-integer exponent (e.g. \pi)

What about (-2)^{\pi}? Here the exponent is irrational — not a fraction at all — so "even vs odd denominator" has no meaning.

In the real numbers, (-2)^{\pi} is not defined under any usual convention. You might hope to compute it as a limit of rational approximations, the same trick that defines 2^{\pi} as the limit of 2^{3.1}, 2^{3.14}, 2^{3.141}, \ldots. But the approximation fails for negative bases. Consecutive rational approximations to \pi have denominators that are sometimes odd, sometimes even — and when the denominator is even, the expression is undefined; when odd, it has a sign that flips depending on the numerator's parity. The sequence of candidate values oscillates between real and undefined, with no limit to converge to.

In complex analysis you can assign (-2)^{\pi} a value by choosing a branch of the complex logarithm, but different branches give different answers. There is no canonical choice. The expression is messy even for professional mathematicians.

The summary: for general irrational exponents, the base must be positive. Otherwise you are outside the real numbers and the answer depends on choices that a school algebra class will not make.

The safe working rule

In a school algebra class, assume all bases are positive unless the problem explicitly uses a negative base and the exponent is either an integer or an odd-denominator fraction. This one rule avoids every edge case at once. Negative bases with integer exponents are always fine — (-3)^4 = 81. Negative bases with odd-denominator fractions are fine. Negative bases with anything else: avoid.

Why (-8)^{1/3} works but (-8)^{2/6} looks like it shouldn't

A real gotcha. You have just agreed that (-8)^{1/3} = -2. Algebraically, \tfrac{1}{3} = \tfrac{2}{6}. So naively,

(-8)^{2/6} \stackrel{?}{=} (-8)^{1/3} = -2.

But apply a^{p/q} = (a^{1/q})^p with p = 2, q = 6: you would take the sixth root of -8 first, which is not real (even denominator!), then square. The "first step" fails. Contradiction.

Resolution: the rule \tfrac{1}{3} = \tfrac{2}{6} holds for fractions as abstract quantities, but not automatically as labels for exponents when the base is negative. When you apply the power-of-a-power law to a fractional exponent over a negative base, you must keep the fraction in lowest terms. Otherwise, two "equivalent" fractions as exponents can give different verdicts — one real (\tfrac{1}{3} works), one complex or undefined (\tfrac{2}{6} routed through a sixth root).

The standard convention is: for a negative base, evaluate a^{p/q} only when \gcd(p, q) = 1 and q is odd. That keeps the answer real and unambiguous.

Worked numeric examples

What NOT to try

The rule (ab)^n = a^n \cdot b^n is reliable when a and b are non-negative. With negatives and fractional exponents it is dangerous. Watch this:

(-4)^{1/2} \cdot (-9)^{1/2} \stackrel{?}{=} ((-4)(-9))^{1/2} = 36^{1/2} = 6.

Plausible, but wrong. In the reals, (-4)^{1/2} does not exist at all, so the product on the left is meaningless. If you insist on complex numbers, (-4)^{1/2} = 2i and (-9)^{1/2} = 3i, so the product is 2i \cdot 3i = 6i^2 = -6. Either way, the naive "combine first" step gives the wrong answer.

The rule (ab)^{1/2} = a^{1/2} \cdot b^{1/2} only holds when both a and b are non-negative — or when you are very careful about which branch of the complex square root you are using. In a pre-calculus context, do not combine fractional exponents across negative bases; handle each case in its own right.

Practical upshot for exam problems

In an Indian board or entrance exam, you will almost never be asked to evaluate something like (-2)^{\pi} or to multiply (-4)^{1/2} by anything. When a negative base with a fractional exponent appears, it is nearly always of the (-8)^{1/3} type — odd denominator, unique real answer.

If you see (-8)^{1/3}, it is fair game: the answer is -2. If you see (-4)^{1/2} on an objective paper, it is probably a domain-restriction question: the expression is undefined over the reals, and the answer to "for which x is \sqrt{x-5} real?" is x \geq 5. If the problem mentions complex numbers or i, then (-4)^{1/2} = 2i is expected. Read the context before computing.

Closing

The pattern is short. Odd denominator of a fraction is safe for any real base — positive or negative — because odd-power functions are one-to-one over the reals. Even denominator requires a positive base, because even powers destroy sign information and the inverse has no real answer for negatives. For general irrational exponents, only positive bases are safe, because the rational-approximation trick needs a consistent sign to converge. When in doubt, stick to positive bases; you will never go wrong.