Date Published: May 31, 2019
Publisher: Public Library of Science
Author(s): Derya Çokal, Vitor Zimmerer, Douglas Turkington, Nicol Ferrier, Rosemary Varley, Stuart Watson, Wolfram Hinzen, Dante R. Chialvo.
Everyday speech is produced with an intricate timing pattern and rhythm. Speech units follow each other with short interleaving pauses, which can be either bridged by fillers (erm, ah) or empty. Through their syntactic positions, pauses connect to the thoughts expressed. We investigated whether disturbances of thought in schizophrenia are manifest in patterns at this level of linguistic organization, whether these are seen in first degree relatives (FDR) and how specific they are to formal thought disorder (FTD). Spontaneous speech from 15 participants without FTD (SZ-FTD), 15 with FTD (SZ+FTD), 15 FDRs and 15 neurotypical controls (NC) was obtained from a comic strip retelling task and rated for pauses subclassified by syntactic position and duration. SZ-FTD produced significantly more unfilled pauses than NC in utterance-initial positions and before embedded clauses. Unfilled pauses occurring within clausal units did not distinguish any groups. SZ-FTD also differed from SZ+FTD in producing significantly more pauses before embedded clauses. SZ+FTD differed from NC and FDR only in producing longer utterance-initial pauses. FDRs produced significantly fewer fillers than NC. Results reveal that the temporal organization of speech is an important window on disturbances of the thought process and how these relate to language.
Spontaneous speech in people with schizophrenia has long been shown to manifest altered patterns of linguistic organization, including reduced syntactic complexity and increased syntactic errors [1–4]. This is unsurprising given close links between thought and language: speech production is nothing other than the process of converting a thought into a temporal sequence of speech units. Disturbances of the thought process are therefore likely to be reflected in speech disturbances. This is obvious in the case of the symptom of formal thought disorder (FTD), which is manifest and diagnosed as disorganized speech  and hence closely related to language. Clinically, however, such speech is rated through terms such as derailment, illogicality, tangentiality, or poverty of content, and it is unclear what distinguishes it in linguistic terms. Recent work suggests that language in FTD can be differentiated from that of people with schizophrenia without FTD and controls by linguistically fine-grained variables, e.g. their anomalous use or production of definite noun phrases (e.g. the man on the left) but not indefinite ones (e.g. a man who likes toast) . The same linguistic variables can distinguish FDRs from neurotypical controls . People with FTD also differ from people with schizophrenia without FTD through a lower proportion of subordinated clauses (e.g. Paul saidthat he liked it), but not of syntactic errors . Here we hypothesized that thinking disturbances may be reflected in the temporal organization of speech as well, i.e. the normal rhythm of structural linguistic units following each other as separated by intermittent pauses.
These results confirm our hypothesis that dysfluency patterns distinguish participants with SZ from non-clinical controls, and both clinical and non-clinical groups from each other. However, differences between participants with and without SZ only started to appear when unfilled pauses and fillers were looked at separately and were indexed to specific syntactic positions. That is, unlike in earlier work, [21, 35] participants with SZ did not produce more unfilled (silent) pauses after co-varying for IQ, age, and education. However, those earlier results were obtained from reading tasks rather than spontaneous speech; nor did they discriminate between SZ with and without FTD. In the latter task, speakers must continuously generate new thoughts of their own. This may help to explain why an increase in empty pauses was found here only relative to specific syntactic positions, namely utterance-initial ones and those before embedded clauses. Clauses index units of information or thoughts of particular kinds. Strikingly, no group differences at all were found in within-clause pauses, contrasting with the alternative pattern seen in aphasia . Within-clause pauses have been widely regarded as reflecting speech planning difficulties at the level of word-finding or lexical retrieval. This pattern suggests that in the speech generation process, the problem seen in SZ does not lie at a lexical but at a grammatical level: it arises at the boundaries of larger syntactic units where complete thoughts are encoded. Our findings specifically point to a difficulty in the organization of thought in speech at positions involving either the utterance or clausal boundaries in embedded positions. This finding contrasts with an earlier study , where no differences between a group of six people with FTD and neurotypical controls in unfilled pauses in clausal boundary positions was seen. However, these authors made no syntactic distinction between utterances and embedded clauses, no group with SZ but without FTD was included, and durations were not considered.