The word “eth” is the name of a letter used in earlier versions of English orthography (among other Germanic writing systems) for the so-called voiced (inter)dental fricative, this grapheme being pronounced with the same voiced sound, viz. [ɛð]. (The proper phonological designation is “lax,” not “voiced,” since English is a protensity language, not a voicing language.) The symbol inherited from Old English resembles a reversed numeral 6 with a stroke through the stem. While contemporary English orthography has dropped this item from its inventory, its phonetic/phonological counterpart, the voiceless (inter)dental fricative called “thorn” and represented in transcription by the Greek theta, i.e., [θ], survives as the digraph th.

The pronunciation of orthographic th in present-day English varies in large part with its position in the word (initial, medial/intervocalic, final), and secondarily with the word class to which a given item belongs. Taking the latter first, the deictics (demonstrative pronouns) this, that, there, thus, and thither, along with the personal pronoun they, all have initial eth, whereas non-pronominals have thorn, e. g., thistle, thatch, thorn, etc. Intervocalic th is exceptionlessly pronounced with eth, as in blather, hither, lather, etc. The directional deictic thither can be pronounced either with a medial eth or a thorn.

In the case of plural forms of items ending in th in the singular, there is a regular assimilation such that eth appears before the {-s} desinence realized phonetically as [z], e. g., path is sg. [paθ] but pl. [paðz], etc.

The distribution of eth and thorn in the immediate vicinity of a liquid (l and r) depends on which liquid it is and on their position in the word. In initial position before /r/, the pronunciation is regularly “voiceless” (throne, thrust, etc.), but medially it is “voiced” before r (e.g., brethren) and “voiceless” after r (e.g., arthritis), as it is after l (e. g., wealth).

An interesting case of distribution is that of the noun/verb pair with final th, viz. bath/bathe. Instead of the correlation expected from markedness theory of the marked sound (here, the tense thorn) obtaining in the marked category (the verb), we have an instance of complementation rather than replication, the markedness values being reversed (the marked sound appearing in the unmarked category and vice versa). Perhaps this distribution is to be explained as a garden-variety case of markedness DOMINANCE, since the two interdental sounds eth and thorn already constitute a marked (restricted) class in the phonology of English to begin with.

MICHAEL SHAPIRO