Normalization in floating-point representation refers to:

• A. Setting exponent to zero.
• B. Converting to integer.
• C. Rounding to nearest.
• D. Adjusting the significand and exponent so that the leading digit is non-zero.

Answer: (D)

Normalization in floating-point representation means adjusting the significand and exponent so the leading digit of the significand is non-zero. In binary floating-point form, this yields a representation like 1.xxxx times 2^E for every nonzero number, which makes the value unique and allows the available mantissa bits to carry as much precision as possible. If the leading digit could be zero, the same number might be represented in multiple ways, wasting precision and complicating arithmetic. So the essence of normalization is shifting the significand to have a non-zero leading bit and adjusting the exponent accordingly. Rounding to nearest, converting to integer, or simply setting the exponent to zero pertain to other aspects like precision or encoding choices, not the normalized form. Zero is handled separately because it has no nonzero leading digit.