A word-frequency counter needs to split text, normalise case, count occurrences and report the top results. Why is choosing a dictionary for the counting step the decision that shapes the whole program?

C. Because every word needs a lookup, and a dictionary makes that O(1).

The counting step performs one lookup per word, so its cost is multiplied by the length of the text. O(n) with a dictionary and O(n²) with a list of pairs, fine on a paragraph and hopeless on a book. Splitting and normalising have no such leverage. Dictionaries are not sorted either, so reporting still needs a sort, cheap, because it runs once rather than n times.