Abstract
The observation that the standard solution of the paradox of the Liar is not satisfactory as a pragmatic solution of a semantic problem restores its former status as a semantic antinomy. Since the antinomy originates from Tarski's T scheme, a conservative modification of the standard semantics is looked for, which would prevent applying the scheme T to anomalous statements. Two such modifications are considered. The first is simpler and implies Kleene's weak tables for three-valued logic. The second, more complex but also more intuitive, agrees with Kleene's strong tables. This undermines the seemingly natural intuition that a sentence and the sentence attributing truth to it have the same truth conditions, which is compatible only with the weak tables