Introduction: Studies consistently report that patients with schizophrenia exhibit qualitative abnormalities on language production tasks. These abnormalities are possibly associated with the severity of psychotic symptoms. Despite this, some studies have conflictingly suggested that patients with schizophrenia exhibit similar word frequency (WF) effects on lexical tasks compared to healthy subjects. Given that previous studies calculated WFs from language corpora, we aimed to investigate the relationship between WF and psychotic symptoms using a novel, simple method for calculating WF. Methods: Thirty-six patients with schizophrenia were included in the study. The severity of positive symptoms was measured using the Scale for the Assessment of Positive Symptoms (SAPS). One semantic and one letter fluency task were administered with the patients instructed to produce as many animal names and words beginning with the letter p in 60 s, respectively. Every response in the output was assigned (1) a corpus-based WF, extracted from the German-language lexical database dlexDB, and (2) a within-sample WF. The within-sample WF was calculated as the raw number of participants who produced the word. Spearman’s correlations were computed between the WF variables and symptoms. Results: Corpus-based WF exhibited skewed, kurtic, and/or non-normal distribution. Contrastingly, within-sample WF displayed normal, non-skewed, and non-kurtic distribution. There were no significant correlations between corpus-based WF and symptoms on both tasks. Conversely, within-sample WF on semantic fluency was significantly negatively and weakly correlated with the global SAPS score, as well as subscales measuring delusions and bizarre behavior. Further, within-sample WF on letter fluency was significantly positively and weakly correlated with the subscale measuring bizarre behavior of the SAPS scale. Conclusion: The differences in the data distribution patterns between corpus-based WF and within-sample WF indicate that different methodological frameworks may have better use of one or the other variable type. Further, significant correlations with positive symptoms were observed only for within-sample WF. It can be concluded that within-sample WF may be more appropriate for analyzing verbal fluency output in psychiatric research compared to corpus-based WF.