The importance of meaning in ambient speech to eliciting cognitive performance effects

In open-plan offices employees have to perform demanding cognitive tasks in the presence of ambient speech due to conversations among colleagues or colleagues being on the phone. Ambient speech has been shown to impair cognitive performance at silent, concentrated work in laboratory studies. This holds true even if the background speech is irrelevant to the task at hand and is intended to be ignored by the listener. Many cognitive psychology experiments demonstrated this phenomenon, which is best explored in tests of verbal short-term memory, such as remembering a series of digits (see Hellbrück & Liebl 2008, for a summary). However, office work is far more than remembering single items in their correct order. A more typical example of office work is reading and text comprehension, as often written information needs to be processed and elaborated. Here, not only the serial order of the words needs to be stored, but the relation between the words must be elaborated to understand and derive information. Irrelevant background speech has been shown to impair performance in reading and text comprehension tasks (e.g. proofreading: Jones et al. 1990; Miles et al. 1988; reading comprehension, or text recall: Martin et al. 1988; Oswald et al. 2000). In these studies, however, the special importance of semantic content within ambient speech is striking; specifically, that irrelevant but semantically meaningful speech (e.g. mother tongue) reduced performance to a greater extent than semantically neutral speech (foreign language or reversed speech). In contrast to that, the semantic content of ambient speech is irrelevant to the extent of performance effects in tasks that solely rely on serial verbal short-term memory (e.g. Colle & Welsh 1976; Jones et al. 1990). This pattern of results is explained by the 'interference by process' principle (e.g. Macken et al. 1999). This principle is based on the assumption that performance reducing sound effects arise due to the similarity of processes involved in the voluntary processing of the task on the one hand and the automatic and involuntary processing of ambient speech on the other hand. Because proofreading, text recall and reading comprehension require semantic processing, the semantic content of the background speech contributes to its detrimental impact. Analogously, the semantic content of background speech plays no role in tasks with minimal load on semantic processing like, for example, verbal serial recall. Here, unrelated verbal items (e.g. digits, consonants, words) are presented successively and have to be recalled afterwards in exact presentation order. This strict serial reproduction criterion ensures that successful task performance is achieved without semantic processing of the memorized items. Accordingly, mother tongue reduces serial recall performance to the same extent than unknown foreign language (e.g. Colle & Welsh 1976; Jones et al. 1990). The present study consists of two experiments which tested the 'interference by process' principle (e.g. Macken et al. 1999) with respect to semantics by systematically varying both task characteristics and semantic content of background speech signals. The extent to which successful task performance relied on semantic processing was varied by using a reading comprehension task consisting of two subtasks. These subtasks differed regarding to the need to semantically process the presented information. The semantic content of background speech was varied differently in the two present experiments. In Experiment 1, the reading comprehension task had to be solved during semantically meaningful background speech (i.e., mother tongue) and during semantically neutral speech (i.e., foreign language). In Experiment 2, the semantic content of background speech was varied by step-wise reducing the coherence of mother tongue background speech. Here, coherent text, unrelated sentences, multiple word phrases (e.g. a rainy day, the girl's dress) or unrelated words were played-back