In scriptura veritas: A methodologic proposal to compare writing and speaking for forensic purposes

Can language production processes in writing and speaking facilitate identification of fabrications? If so, are fabrications easier to detect in the written modality compared to the spoken modality? The aim of this study is to examine how language production processes manifest across truthful and fabricated narrative accounts of witnessed events. Specifically, the study asks if and how differences and similarities are manifested across written and spoken modalities, and if/how these differences/similarities can be analyzed within and across modalities. Linguistic processes in written and spoken language production are subject to working memory constraints (Goldman Eisler, 1970; McCutchen, 1996), and this can be seen in for example increased pause duration, frequency, and other disfluencies (Heldner & Edlund, 2010; Matsuhashi, 1981). It has also been found that lying leads to increased cognitive load (Williams, Bott, Patrick, & Lewis, 2013), leading to the hypothesis that fabrication would be visible in the production patterns in both modalities. Due to the synchronous nature of speaking, where the production process is overt, compared to the asynchronous nature of writing, where the process is covert, another hypothesis is that it will be easier to detect fabrication during writing.Two truthful and two fabricated narratives (whereof two spoken and two written) from a corpus of written and spoken narratives were used to qualitatively examine how pauses and disfluencies manifest in the two conditions across modalities. Written data consisted of a keystroke logging file and spoken as an audio file and transcription. Cognitive load was analyzed through text length, pause analyses and revisions (in writing) and reformulations (in speaking). The results show that, taking the difference in production speed across modalities into consideration, cognitive load can meaningfully be operationalized as pause frequency in relation to text length. Further, when measuring text length, the number of written/transcribed characters is proposed as a way of measuring this while taking repetitions (in speaking) and fragments (in writing) into consideration. In conclusion, the study suggests that comparable results in the modalities can be reached by making informed analytical choices.