Making Our Democracy Work by Stephen Breyer

text-oriented system cannot work very well. A more realistic example will help me explain why.
    A federal statute permits citizens to sue the government and to recover damages for serious harm that federal officials wrongfully caused, including harm to their property. But the act contains a series of exceptions, including an exemption for harm to property caused by “any officer of customs or excise or
any other law enforcement officer.”
Whom exactly does the exempting phrase, “any other law enforcement officer,” cover? Does it cover only those law enforcement officers who carry out “customs or excise” duties? Or does it cover (and exempt from liability) other law enforcement officers as well, such as federal prison officials? 2
    Text-oriented judges will carefully examine the statute’s language. They may refer to a dictionary or look to surrounding language. They will see if the phrase has some special traditional or historical meaning. They will search the precedent, and, failing to find any strong reason for giving the words a specialized meaning, they will try to give them the meaning they carry in ordinary, non-statutory life. In this case, the key words “any other” are not technical, and dictionaries, history (other than legislative history), tradition, and precedent do not suggest any specialized meaning. So the judge may well conclude that the words “any other” mean what they say, namely, suit is barred against
any
other law enforcement officer, including prison officials.
    What is wrong with that? Let us break this question down into two parts: First, what is wrong with the assumption that language in a statute means what it means in ordinary life outside the statute? Unfortunately, such an assumption is rarely helpful. As those who study language have pointed out, we use words, strung together in sentences, uttered on particular occasions orally or in writing, to perform many different functions. We use them to ask questions, to make statements, to agree with others, to write contracts, to perform marriages, to pray, to vow, to inform, to estimate, to recommend. We also use them to write laws—a highly specialized activity. The assumption begs the further question: What part of “ordinary life”? 3
    The statute’s language may be vague, and the scope of its coverage may be uncertain. But it does not help us understand a vague statement to pretend that someone else “in ordinary life” made the statement. How often would it help us understand, say, a difficult point in a university lecture to pretend that the lecturer is not a lecturer but a journalist? Similarly, how often does it help us understand a statute’s vague or ambiguous language to pretend that its congressional authors were engaged in any activity other than the one they were engaged in, namely, writing a statute?
    It is also rarely helpful to rely on dictionaries. Nothing is wrong with turning to the dictionary when a court is trying to interpret a technical word, say the word “percentile” used in a special technical sense in a statute that incorporates the understanding of professional statisticians. In such a case, Congress might well have intended non-statisticiansto look to a dictionary to discover how statisticians use the word. 4
    Far more often, however, statutory uncertainty does not arise because the statute’s language has an unclear technical meaning or because ordinary readers fail to understand the general
kinds
of situations to which the statute’s language refers. We all know the meaning of the words “any other law enforcement officer.” We need not look them up in a dictionary. Rather, here the statute is ambiguous or uncertain in respect to the scope of its coverage. Its general language does not tell us precisely which situations fall outside Congress’s demonstrative intent.
    Suppose a statute uses the phrase “
any
court.” We have no trouble understanding the words, but we may have trouble