Right-Wing Rock or Just Rock? A Computational Linguistic Analysis of Frei.Wild

Rechtsrock is a subgenre of rock music that spreads right-wing ideology, and this study uses computational linguistic methods to determine whether the band Frei.Wild should be classified as politically right-leaning or as part of the general German rock genre.

The research utilizes lexical analyses and classification experiments to analyze the similarities between right-wing extremist rock music and the band Frei.Wild, finding that while they maintain ambiguity, their tendency leans towards the right-wing spectrum, with roughly half of their songs classified as right-wing extremist by computational models.

Lexical Analyses

The initial phase of the study involved examining word level patterns across three distinct corpora: a corpus of songs by Frei.Wild, a reference corpus for right-wing extremist rock (Rechtsrock), and a corpus for general German rock. The goal was to obtain an overview of dominant themes in the lyrics by extracting the ten most frequent lemmata from each set.

The frequency patterns revealed thematic differences between the genres:

In Rechtsrock, ‘Heimat’ is found to cooccur with words such as blood, honor, and pride.

In contrast, ‘Heimat’ in the German rock corpus only appears three times in the context of ‘Haus’, while ‘Deutschland’ appears once.

Further investigation focused on concordance analyses for key terms. For example, frequent collocates of ‘Vaterland’ (fatherland) in Rechtsrock include ‘stehen’, ‘blut’, and ‘schweiß’. In contrast, the Frei.Wild corpus favors the compound word ‘Heimatland’ and the regional identifier ‘Südtirol’. The analysis suggests that while both genres view their homeland through a right-wing extremist lens featuring violent imagery, nationalistic ideas, and an 'us against them' narrative, they do not use the exact same co-referential expressions.

Classification Experiments

The study employed several computational approaches to classify Frei.Wild songs as either Rechtsrock or general German rock. The methodology involved:

1. Computing inter-band similarity using a TF–IDF representation of the lyrics and cosine similarity to identify where similarities arise between bands.

2. Training random forest and Support Vector Machine (SVM) classifiers on both reference corpora to predict the category of Frei.Wild songs, using TF-IDF vectors initially, followed by sentence embedding models for incorporating contextual meaning.

3. Testing combinations of two genres (Frei.Wild vs. Rechtsrock, and Frei.Wild vs. German Rock) to see which genre Frei.Wild would be the hardest to separate from using a binary classifier trained on the reference corpora with TF-IDF representations, achieving an ROC-AUC score of 98% for general German rock vs Frei.Wild and 89% for Rechtsrock vs Frei.Wild.

Results of Classification

The classification experiments demonstrated that Frei.Wild positions itself between the two reference corpora, acting as a border case.

In the TF-IDF classification experiment on the complete Frei.Wild corpus, 51.06% are classified as Rechtsrock (24 out of 47).

When using sentence embeddings, Frei.Wild lyrics are classified as Rechtsrock in 59.57% of cases.

The density plots visualizing classification probabilities showed a unimodal distribution centered around 0.5 for Frei.Wild data, indicating high uncertainty in the classification. However, when directly trained to separate Frei.Wild from the two other classes, Frei.Wild and Rechtsrock were harder to separate which indicates they are more similar. The results confirm that when contextual information is included via sentence embeddings, Frei.Wild tends to be associated more strongly with right-wing rock.

Temporal Analysis

A temporal dimension was added to the final analysis to investigate how classification results change over time, considering the skinhead background of the lead singer. The temporal plots revealed that Frei.Wild started out as less right-wing extremist, grew very extremist in the 2010s and has now settled to the right of the center, according to classification scores. This trend contradicts Frei.Wild’s self-portrayal regarding leaving their skinhead past behind.

General Discussion

The study confirms that Frei.Wild exhibits enemy/‘us against them’ constructions and an appeal to resistance tied to nationalistic narratives such as preserving certain values and the region/country that they call home. The quantitative approach backs qualitative studies by showing that at least half of the songs would be classified as right-wing extremist in a random sample, suggesting that while they are situated in the middle with a tendency towards Rechtsrock, any affinity to music judged as ‘youth-endangering’ and unconstitutional should be seen as negative. The analysis also provided insight into how computational methods can improve the process of identifying right-wing extremist tendencies in music, especially in "border cases like Frei.Wild.

Based on the computational linguistic analysis presented in this paper, here are specific improvements that can be implemented in AI systems, along with their potential capabilities:

1. Use a multi-stage classification pipeline combining Lexical Feature Extraction (TF-IDF/Leave-One-Out) and Contextual Semantic Embeddings (Sentence Transformers).

2. Implement a Random Forest or SVM classifier trained on these combined features to predict the political leaning of music.

3. Develop genre classification models that compare a target song against reference corpora (e.g., Rechtsrock vs. General German Rock) using cosine similarity and machine learning classifiers (like Random Forest, as shown in Section 6).

4. Incorporate Temporal Analysis into the classification system by feeding the year of release as an additional feature to predict genre drift or ideological shifts over time for a specific artist.

5. Utilize Leave-One-Out analysis to identify the most discriminative lexical features (e.g., 'frei', 'vaterland', 'tod') that drive classification decisions, allowing for interpretable model development in borderline cases like Frei.Wild.

6. Employ fine-tuned language models (like German Semantic STS V2) to generate contextual embeddings for lyrics, significantly improving the AI's ability to capture nuanced semantic meaning beyond simple word frequency (as demonstrated by the shift from TF-IDF to Sentence Embeddings in Section 6.3).

7. Create a system capable of quantifying border case ambiguity by measuring the uncertainty (e.g., probability distribution unimodality around 0.5) of the classification model, rather than just providing a binary label, allowing for nuanced reporting of political affiliation risk.

The improved AI system can perform the following specific tasks:

1. Identify and flag music that resides in a border case between two defined categories (e.g., Right-Wing Rock vs. General German Rock) with a quantifiable measure of uncertainty, rather than forcing a rigid binary classification where the model is highly uncertain (probability near 0.5).

2. Predict the political leaning of an unknown song by leveraging both structural lexical cues and deep contextual semantic understanding (embeddings), leading to higher accuracy than models relying on single features alone.

3. Track ideological drift in an artist's catalog over time, automatically detecting when a band's lyrical content shifts from extremist toward mainstream or vice versa, providing real-time monitoring for media analysis institutions.

4. Provide transparent and interpretable justifications for classification decisions by highlighting the specific high-impact words (e.g., 'vaterland', 'blut') that most heavily influenced the model's prediction, which is crucial for mitigating false positives in content moderation systems.

5. Develop a robust system capable of distinguishing between ideological commitment and mere aesthetic incorporation (e.g., differentiating the rhetorical use of Heimat in a nationalist context versus its use in general rock), thereby reducing the false classification rate for bands like Rammstein or Böhse Onkelz which have complex histories.

Rechtsrock is a subgenre of rock music that spreads right-wing ideology, often instrumentalized to recruit adolescents into the radical scene. Monitoring institutions counteract this by manually examining and, in some cases, banning extremist content; however, there are border cases that evade regulation. We present a study aimed at determining whether such a case, the band Frei. Wild, should be classified as politically right-leaning or as part of the general German rock genre. We sampled a German rock dataset and created a corpus for right-wing rock to use as reference in this analysis and found that we can confirm the intuitions from previous investigations that Frei. Wild successfully maintains an ambiguity with regard to their political affiliation. However, the tendency is towards the right-wing spectrum. Lexical analyses reveal nationalistic narratives and two high-performing classifiers (up to 97% ROC-AUC score) label more than half of their songs as right-wing extremist. Our analysis provides insight into how computational methods can improve the process of identifying right-wing extremist tendencies in music, especially in borderline cases like Frei. Wild. The code and data are made available for future research.