The article presents a new theory of stress and linguistic rhythm, based on Liberman's (1975) proposals. It argues that stress subordination in English is not primarily determined by individual segments or syllables, but by a hierarchical rhythmic structure that organizes syllables, words, and syntactic phrases. The theory uses two key ideas: relative prominence is defined by a relation on constituent structure, and linguistic rhythm is defined by the alignment of linguistic material with a "metrical grid." The perceived stress of an utterance reflects the combined influence of constituent structure and grid alignment, reminiscent of traditional verse scansion. The theory is called "metrical" and focuses on assigning relative prominence based on constituent structure. Section 1 applies the metrical theory to English phrasal stress, arguing its value in explaining arbitrary stress features and rules. Section 2 extends this to word stress, using a traditional view of [+stress] assignment but explaining word-level stress patterns through the metrical theory. Section 3 introduces alignment with a metrical grid, fundamental to understanding the "rhythm rule" and relative stress at the syllabic level. The article discusses the traditional description of English phrasal stress, noting that it has been challenged by some scholars who argue that stress should not extend beyond the word level. However, the authors argue that English is a stress language, not a tone or pitch-accent language, and that stress patterns have independent rhythmic properties. They use reiterant speech experiments to show that stress patterns are independent of intonation contours. The article then presents a relational theory of English phrasal stress, where relative prominence is defined on constituents rather than segments. This theory uses a binary relation on pairs of nodes in a tree to represent relative prominence. The theory is compared to the stress-number theory, which uses a multivalued feature. The relational theory is argued to be more explanatory, as it rationalizes the arbitrary properties of stress patterns and rules. The article also discusses the advantages of the metrical theory of phrasal stress, including the n-ary nature of the stress feature, the syntagmatic definition of nonprimary stress values, the lack of local phonetic import for stress levels, the cyclic application of phonological rules, the widespread pattern of change caused by stress rules, the use of crucial variables, and the convention of disjunctive ordering. The article then turns to the treatment of word stress, arguing that words have an internal metrical structure. It presents a relational theory of word stress, where relative prominence is defined on constituent structure. The theory is compared to the stress-number theory, and the article discusses the advantages of the relational theory, including the binary nature of the stress feature, the syntagmatic definition of stress values, the use of larger patterns for phonetic implementation, the preservation of relative prominence under embedding, the local nature of the phonological effect, and the local conditioning of the rules. The article concludes byThe article presents a new theory of stress and linguistic rhythm, based on Liberman's (1975) proposals. It argues that stress subordination in English is not primarily determined by individual segments or syllables, but by a hierarchical rhythmic structure that organizes syllables, words, and syntactic phrases. The theory uses two key ideas: relative prominence is defined by a relation on constituent structure, and linguistic rhythm is defined by the alignment of linguistic material with a "metrical grid." The perceived stress of an utterance reflects the combined influence of constituent structure and grid alignment, reminiscent of traditional verse scansion. The theory is called "metrical" and focuses on assigning relative prominence based on constituent structure. Section 1 applies the metrical theory to English phrasal stress, arguing its value in explaining arbitrary stress features and rules. Section 2 extends this to word stress, using a traditional view of [+stress] assignment but explaining word-level stress patterns through the metrical theory. Section 3 introduces alignment with a metrical grid, fundamental to understanding the "rhythm rule" and relative stress at the syllabic level. The article discusses the traditional description of English phrasal stress, noting that it has been challenged by some scholars who argue that stress should not extend beyond the word level. However, the authors argue that English is a stress language, not a tone or pitch-accent language, and that stress patterns have independent rhythmic properties. They use reiterant speech experiments to show that stress patterns are independent of intonation contours. The article then presents a relational theory of English phrasal stress, where relative prominence is defined on constituents rather than segments. This theory uses a binary relation on pairs of nodes in a tree to represent relative prominence. The theory is compared to the stress-number theory, which uses a multivalued feature. The relational theory is argued to be more explanatory, as it rationalizes the arbitrary properties of stress patterns and rules. The article also discusses the advantages of the metrical theory of phrasal stress, including the n-ary nature of the stress feature, the syntagmatic definition of nonprimary stress values, the lack of local phonetic import for stress levels, the cyclic application of phonological rules, the widespread pattern of change caused by stress rules, the use of crucial variables, and the convention of disjunctive ordering. The article then turns to the treatment of word stress, arguing that words have an internal metrical structure. It presents a relational theory of word stress, where relative prominence is defined on constituent structure. The theory is compared to the stress-number theory, and the article discusses the advantages of the relational theory, including the binary nature of the stress feature, the syntagmatic definition of stress values, the use of larger patterns for phonetic implementation, the preservation of relative prominence under embedding, the local nature of the phonological effect, and the local conditioning of the rules. The article concludes by