Functional magnetic resonance imaging (fMRI) can reliably detect a network-level configuration of blood-oxygen-level-dependent (BOLD) referred to as the language network (LN). The BOLD signal that forms the LN is modulated by a range of cognitive tasks. Here we present the results of six task-based fMRI analyses which retrieved the LN, and report specific pattern-based (as opposed to coordinate-based) anatomical details essential for distinguishing the LN from other BOLD networks, and a preliminary cognitive mode interpretation based on how the task-induced BOLD signal changes associated with the LN was modulated by a range of task conditions. The cognitive tasks which elicited the LN were lexical decision, metrical stress, noun/verb discrimination, semantic association, and facial emotion discrimination. A review of the task-induced BOLD signal changes associated with the LN provided information that could lead to determination of a LN-associated cognitive mode; namely, greater activation for linguistic processing (e.g., distant > close semantic association), reduced activation when linguistic information must be ignored (e.g., when judging that a word-like letter string is not a real word), and possible extension to aspects of non-verbal communication (i.e., emotion discrimination). The LN has been demonstrated to be anatomically highly reliable, producing identifiable and highly specific network-level anatomical patterns. Based on task-induced BOLD signal changes over a wide range of cognitive tasks, the LN includes both activation and suppression depending on the linguistic information processing required to achieve the task goals, as well as possible extension to non-verbal communication, providing information to consider for determination of a LN-associated cognitive mode.