This study is divided into two parts. In the first part, I will consider the term of adjustment as the concept is used within the TOE. We will see that the activity of text production-recognition implies a process of adjustment in all cases, if this is defined as a process whereby a speaker or cospeaker seeks to match linguistic and cognitive representations. As long as the match is felt to be unproblematical, however, there are no explicit markers of adjustment and so, since the TOE deals with observables, the process of adjustment escapes analysis. In other cases, however, the match between linguistic and cognitive representations is not self-evident and, in these cases, the operation of adjustment – or readjustement – may appear in specific patterns of markers, as a speaker exercises a form of metalinguistic monitoring over his or her own production. In the second part I will consider two such types of linguistically-marked adjustment with their associated markers. In the first type a speaker signals explicitly that an initial representation is judged inappropriate in some respect and is being rectified, post hoc, accordingly. Here, the adjustment is initiated by the speaker. This case will be exemplified by examples from English of postposed WH– EVER clauses. In the second type a speaker, aware of a potentially inexact match between the linguistic representation and the projected cognitive representation, indicates this pre-emptively, telling the cospeaker that the utterance will require some adjustment. This case will be illustrated by certain approxi-mative uses of “like”, or cases where non-literal uses of language are linguistically signalled as such to the co-speaker, who is enjoined to operate the appropriate adjustment.