Lexical analysis in schizophrenia: How emotion and social word use informs our understanding of clinical presentation

https://doi.org/10.1016/j.jpsychires.2015.02.024Get rights and content

Highlights

  • Lexical analysis tools are novel methods to assess speech in schizophrenia.

  • Metacognition in patients with schizophrenia was predicted by social word use.

  • Increased use of anger words predicted greater symptoms and lower functioning.

  • Findings suggest that lexical analysis can yield vital information about patients.

  • Social and emotion word use appears to be especially informative for schizophrenia.

Abstract

Background

The words people use convey important information about internal states, feelings, and views of the world around them. Lexical analysis is a fast, reliable method of assessing word use that has shown promise for linking speech content, particularly in emotion and social categories, with psychopathological symptoms. However, few studies have utilized lexical analysis instruments to assess speech in schizophrenia. In this exploratory study, we investigated whether positive emotion, negative emotion, and social word use was associated with schizophrenia symptoms, metacognition, and general functioning in a schizophrenia cohort.

Methods

Forty-six participants generated speech during a semi-structured interview, and word use categories were assessed using a validated lexical analysis measure. Trained research staff completed symptom, metacognition, and functioning ratings using semi-structured interviews.

Results

Word use categories significantly predicted all variables of interest, accounting for 28% of the variance in symptoms and 16% of the variance in metacognition and general functioning. Anger words, a subcategory of negative emotion, significantly predicted greater symptoms and lower functioning. Social words significantly predicted greater metacognition.

Conclusions

These findings indicate that lexical analysis instruments have the potential to play a vital role in psychosocial assessments of schizophrenia. Future research should replicate these findings and examine the relationship between word use and additional clinical variables across the schizophrenia-spectrum.

Introduction

The words that people choose offer windows into internal states, feelings, and views of the world around them. Computerized lexical analysis is a promising tool for examining word use, as it provides a fast, reliable method for scanning narratives by grouping words and word stems into thematic categories. Researchers have observed that analyzing word use, particularly in emotion and social categories, can yield information on a range of constructs, including cortical activity (Saxbe et al., 2013), childhood behavior problems (Slatcher and Trentacosta, 2012), and psychopathological symptoms (Rude et al., 2004). Employing lexical analysis tools, either as stand-alone or supplemental methods, offers the potential to increase accuracy in behavioral assessments, as these objective measures are free of some limitations inherent in other types of measures (e.g., social desirability bias). Lexical analysis would appear to be tailor-made for investigating word use in schizophrenia, which is characterized by disorganized speech that is highly reactive to phenomenological state (Burbridge and Barch, 2002, Docherty and Hebert, 1997). These tools are unique compared to most speech instruments used in schizophrenia research, as lexical analysis focuses solely on speech content rather than syntax or word order. However, few studies have utilized lexical analysis instruments in this population.

Previous studies implementing lexical analysis have focused on comparing word use in schizophrenia and control groups, and have observed clinically significant differences between groups (Buck et al., 2015, Cohen et al., 2009; Junghaenel et al., 2008; but also see St-Hilaire et al., 2008). There has been little research within patient groups to investigate whether word use is linked with schizophrenia symptoms. Cohen et al. (2009) published the lone study on this topic, observing that patients with schizophrenia high in anhedonia, a negative symptom characterized by reduced positive affect expression, used significantly more negative emotion words when discussing pleasant topics than those low in anhedonia or controls. Cohen and colleagues focused specifically on anhedonia; to our knowledge, no previous study has investigated whether word use is associated with the full spectrum of schizophrenia symptoms.

To enhance the clinical utility of lexical analysis instruments, it is imperative to examine whether word use can inform our understanding of commonly observed deficits in schizophrenia, such as metacognition and general functioning, where affective and social processes play critical roles. Metacognition, which is often defined as thinking about thinking (Flavel, 1979, Frith, 1992), involves a range of activities from discrete acts that require recognizing specific thoughts and feelings to synthetic acts that necessitate combining an array of intentions, thoughts, feelings, and connections between events into complex representations of others (see Gumley, 2011, Lysaker and Dimaggio, 2014, Lysaker et al., 2011). To demonstrate metacognition, one must exhibit an ability to integrate cognitive and emotional experiences in the moment and in memory, as well as recognize emotions in oneself and others. These abilities are often examined in schizophrenia by assessing the speech narratives of patients, as those with metacognitive deficits are likely to demonstrate greater impoverishment through language. Lexical analysis instruments offer the potential to examine speech narratives more deeply in order to identify moments when core psychological processes, such as emotional or social processes, are engaged. Despite the importance of emotional and social processes to metacognition and the common use of speech narratives during assessment, no previous study has investigated whether significant overlap exists between metacognition and emotion or social word use.

Regarding functioning, greater levels of negative emotion are significantly associated with lower functioning in schizophrenia (Blanchard et al., 1998). Gauging social relationships is also a key component of functional assessments. Examining associations between functioning and emotion and social word use, compared to other categories, could inform our understanding of how lexical analysis can be utilized to assess general functioning in schizophrenia. If significant relationships are observed, this would offer preliminary evidence for the utility of lexical analysis instruments as an objective screening measure of general functioning in patients.

This is the first study to implement lexical analysis to explore relationships between word use and overall symptoms, metacognition, and general functioning in individuals with schizophrenia. These relationships were examined across fifteen word use categories, spanning themes related to psychological processes and personal concerns. We expected word use to significantly predict all three clinical variables, and that emotion and social word categories would be the strongest individual predictors measured in this study.

Section snippets

Participants

Participants were outpatients from a Midwestern VA Medical Center (n = 17) and a community mental health clinic (n = 29) with confirmed DSM-IV diagnoses of schizophrenia (n = 17) or schizoaffective disorder (n = 29). Exclusion criteria included an age <18, presence of severe cognitive impairments as measured by a six-item cognitive screener (see Callahan et al., 2002), and incomplete baseline data for speech, symptoms, metacognition, or functioning. This data was part of a randomized controlled

Results

When word use categories were compared in schizophrenia and schizoaffective subgroups, we observed that biological words were used more frequently by participants with schizophrenia, t(45) = 2.34, p = 0.02. A trend level difference was observed in sadness words, t(45) = 1.96, p = 0.06, with the schizoaffective subgroup using these words at a higher rate. In subsequent analyses, controlling for these two results did not affect the interpretation of any finding.

As shown in Table 2, greater

Discussion

Our findings indicate that lexical analysis of speech can yield important information regarding schizophrenia symptoms, metacognitive abilities, and general functioning in individuals with schizophrenia. Each of these clinical variables was significantly predicted by word use categories, with negative emotion and social word categories demonstrating the strongest relationships across variables. The finding that word use categories explained over a quarter of the variance in overall symptoms is

Role of the funding source

Funding for this study was provided by the Department of Veterans Affairs, Health Services Research & Development Service (IIR 08-324, IAC 05-254, and CDA 10-034. The views in this paper express those of the authors. The Department of Veterans Affairs had no further role in design of this study, data collection, interpretation, or data analysis, in writing the manuscript or in the decision to submit the finished manuscript for publication.

Contributors

Kyle S. Minor conducted the literature search, performed data analysis, and wrote or co-wrote all drafts of the manuscript. Kelsey A. Bonfils, Lauren Luther, Ruth L. Firmin, Marina Kukla, Victoria R. MacLain, and Benjamin Buck co-authored all drafts of the manuscript and played key roles in data collection, analysis, and interpretation. Paul H. Lysaker served as an expert on interpreting the metacognition data and co-authored all drafts of the manuscript. Michelle P. Salyers was the Principal

Conflict of interest

All authors declare no conflicts of interest in the current study.

Acknowledgments

We wish to thank all of the patients who participated in this project. Additionally, we wish to thank staff and research assistants at both recruitment sites, in the Assertive Community Treatment (ACT) Center, and in the Cognition, Language, and Affect in Severe Psychopathology (CLASP) Laboratory at Indiana University – Purdue University Indianapolis. Special thanks to Ciara Spencer and Chase Brown for their help with data collection.

References (32)

  • B. Buck et al.

    Lexical characteristics of anticipatory and consummatory anhedonia in schizophrenia: a study of language in spontaneous narratives

    J Clin Psychol

    (2015)
  • J.A. Burbridge et al.

    Emotional valence and reference disturbance in schizophrenia

    J Abnorm Psychol

    (2002)
  • C.M. Callahan et al.

    Six-item screener to identify cognitive impairment among potential subjects for clinical research

    Med Care

    (2002)
  • A.S. Cohen et al.

    Understanding anhedonia in schizophrenia through lexical analysis of natural speech

    Cognition Emot

    (2009)
  • N.M. Docherty

    On identifying the processes underlying schizophrenic speech disorder

    Schizophr Bull

    (2012)
  • N.M. Docherty et al.

    Comparative affective reactivity of different types of communication disturbances in schizophrenia

    J Abnorm Psychol

    (1997)
  • Cited by (45)

    • Blinded Clinical Ratings of Social Media Data are Correlated with In-Person Clinical Ratings in Participants Diagnosed with Either Depression, Schizophrenia, or Healthy Controls

      2020, Psychiatry Research
      Citation Excerpt :

      Numerous studies have shown that written and spoken language could assist in monitoring and evaluating mental illness (Ozdas, Shiavi et al. 2004, Mundt, Snyder et al. 2007, Cohen and Elvevag 2014, Mowery, Smith et al. 2017, Arevian, Bone et al. 2020, de Boer, Brederoo et al. 2020). Evidence of schizophrenia manifested in language use includes aphasias, loose associations, derailment, tangentiality, neologisms, poverty of speech, decreased fluency, use of unorganized word lists, use of ambiguous pronouns, poor word associations in sentences, increased repetition of words, and deviances in grammar (Kuperberg and Caplan 2003, Strous, Koppel et al. 2009, Stephane, Kuskowski et al. 2014, Buck and Penn 2015, Minor, Bonfils et al. 2015, Moro, Bambini et al. 2015, Iter, Yoon et al. 2018), and language-related evidence may help to indicate not only illness but illness severity (Hui, Longenecker et al. 2012). Language processing areas have been associated with psychiatric symptoms (Jung, Lee et al. 2019).

    View all citing articles on Scopus
    View full text