Two experiments and a corpus-based (Hebrew language) analysis confirm the hypothesis, renouncing the Graded Salience Hypothesis, that addressees’ attention to meanings low on salience is actually drawn by speakers' explicit marking (the low-salience marking hypothesis).