The first experiment was conducted in the German Intensive Certificate Course 7 of the Savitribai Phule Pune University in an offline mode on 10th March 2025 from 4pm to 6pm. The 24 participants were total beginners, who were in the second half of A1 level i.e. A1.2 and used the textbook Menschen A1.1 and A1.2. The students had a basic understanding of the grammatical structures and vocabulary till lesson 15 of the book. They could write and understand basic structures and vocabulary related to the facilities and places in the cities, they had also learned the cases nominative, accusative and dative with the dative verbs and personal pronouns.

The aim of this experiment was to explore whether the AI generated song can help to revise the grammar topic: Dative Verbs and Dative personal Pronouns and strengthen the vocabulary learnt regarding the places and facilities in the city. A song was generated with the help of ChatGPT and the AI tool Suno.ai with a prompt in German, as the song was generated in German.

The German prompt that was given to CHATGPT was as follows: bitte gib mir einen Liedtext auf Niveau A1, der das Grammatikthema Dativ-Personalpronomen mit Dativ-Verben wie gefallen, gehören, helfen, danken und schmecken behandelt. Das Themenfeld ist die Stadt und ihre Einrichtungen, zum Beispiel: die Kirche, der Markt, das Kino, das Geschäft, die Schule, die Bibliothek, das Restaurant, der Spielplatz, die Post, der Park, die Bäckerei. Ich brauche einen Liedtext mit 4 Strophen, der sich nach Möglichkeit reimt. Style of Music- playful song on which one can dance and which is full of beats (its English translation*: please give me a song text at level A1 that deals with the grammar topic of dative personal pronouns with dative verbs such as like, belong, help, thank and taste. The topic area is the city and its facilities, for example: the church, the market, the cinema, the store, the school, the library, the restaurant, the playground, the post office, the park, the bakery. I need a song text with 4 verses that rhymes if possible).

After providing this prompt to ChatGPT, a song was generated with four stanzas incorporating the specified verbs and locations such as the park, library, post office, police station, theater, school, and museum. The generated lyrics were then inserted into the lyrics section of Suno.ai, which allows a maximum of 3000 words. The song's style was specified, and the title was assigned by the author of this paper, who served as the creator of the prompt. Suno.ai subsequently generated two versions of the song with different melodies, from which the following version was chosen for its engaging and memorable tune (A1, A2).

The experiment was designed to last approximately 100–120 minutes. It commenced with the creation of a sociogram depicting various places in the city (Fig. A3), followed by an interactive game where participants passed a ball while music played. When the music stopped, the participant holding the ball performed an action associated with one of the places or facilities in the city, and the other participants constructed sentences such as “Im park kann man spielen und Freunde treffen“ (one can play and meet friends in the park, Video of the activity-A4).

Subsequently, the students created smaller sociograms focusing on the locations mentioned in the AI-generated song, such as the bakery, park, museum, library, market, police station, theater, school, bus stop, and hospital. This activity enabled the teacher to introduce and reinforce new vocabulary from the song during the discussion and feedback. The first listening of the AI-generated song followed, during which the students were provided with a list of words, including some additional ones. Their task was to identify and mark the words they heard in the song. To enhance comprehension at the vocabulary level, the students then matched challenging verses with appropriate images that illustrated those verses.

During the second listening, the students were given the song lyrics along with a table of dative personal pronouns to follow along with the text. This was accompanied by an exercise requiring them to match nominative and dative personal pronouns to the relevant verses, as the prompt emphasized the use of dative personal pronouns. This stage was followed by grammar exercises from the workbook that reinforced the concept of dative personal pronouns.

Subsequently, the students practiced framing and answering Yes/No questions using five selected dative verbs. They engaged in peer interaction by asking and responding to these questions with four different classmates, ensuring a practical application of sentence structures. Following this, the students answered six fundamental questions related to the places mentioned in the AI-generated song, providing concise responses. In the final stage, the students were divided into groups of three to four participants. Each group discussed and answered a common question aimed at enhancing their speaking skills: “Was brauchst du in deinem Stadtteil?“ (what do you need in your city area?). Participants were expected to provide their responses along with appropriate reasoning for their choice of place or facility (Video of students-A5).

The results of the experiment were analyzed using a questionnaire administered to all 24 participants of the class through Google Forms (A6). The questionnaire, designed in English, consisted of objective questions aimed at gathering feedback on the effectiveness of the AI generated song.

The graphical analysis of the responses revealed that 100% of the participants expressed a preference for learning through songs in class. Furthermore, 95.8% of the students indicated that they found revising grammar and vocabulary through a song to be effective and enjoyable. All participants reported a positive experience with the lecture in which the experiment was conducted. According to the data, 100% of the participants confirmed that they were able to revise vocabulary effectively through the verses of the song. Additionally, 95.8% agreed that learning through a customized song was not only beneficial but also engaging and motivating. When asked whether AI-generated songs could facilitate learning German, 87.5% of the participants responded affirmatively, while 12.5% were uncertain and marked "Maybe."

Regarding the specific ways in which the AI-generated song contributed to their learning, 37.5% of the students highlighted that the catchy tune of the song helped them retain words and verses more effectively. Another 33.3% indicated that the grammatical structures, particularly the use of dative personal pronouns, became clearer through the song. Meanwhile, 29.2% reported that the song facilitated vocabulary revision, enhancing their ability to recall the words with greater ease.

This experiment was conducted in the same class on March 12, 2025, with a different objective: to investigate whether an AI-generated song could facilitate the pre-reading phase before introducing Lesson 16 from the textbook Menschen A1.2, which focuses on problems encountered in a hotel room. Since the topic was related to a hotel room, the AI-generated song emphasized relevant vocabulary, including air conditioner, bathroom, elevator, hairdryer, shower, TV, and heater. The song was generated by providing a prompt in German to ChatGPT.

The original prompt in German was: gib mir einen Liedtext mit 2 Strophen für Niveau A1. Ich möchte die 3Wörter “der Aufzug, die Klimaanlage, die Heizung, der Fernseher, das Radio, die Internetverbindung, das Licht, die Seife, das Handtuch, der Bademantel, der Föhn, der Wecker, das Telefon, die Dusche, das Wasser“ verwenden. Ich brauche den Liedtext im Zusammenhang mit dem Thema “Das Hotelzimmer“ (its English translation*: give me a song text with two verses for level A1. I want to use the following words: ‘elevator, air conditioning, heater, television, radio, internet connection, light, soap, towel, bathrobe, hairdryer, alarm clock, telephone, shower water.' I need the song lyrics in connection with the topic ‘hotel room’) As the AI-generated song was intended solely for vocabulary practice at the pre-reading stage, specific words related to the topic were included in the prompt. The textbook Menschen has a lexicon of photos in every lesson, which focuses on the key vocabulary essential for that particular lesson. So the words from this lexicon were chosen to be integrated into the AI-generated song. The song text generated by ChatGPT was pasted in the lyrics section in suno.ai without any instructions for the style of music. Out of the two versions created, these versions (A7, A8), were selected because they were long enough, featured repetitions of stanzas, and had a catchy tune.

The class began with the creation of a mind-map (Fig. A9) featuring items associated with a hotel room. Since students were already familiar with basic household objects, they could easily expand the list to include vocabulary related to hotels. This was followed by listening to the AI-generated song, during which students identified and circled words from the mind-map that appeared in the song. New vocabulary from the song was then matched with illustrations drawn by the teacher on the board.

During the second listening, students marked pairs of opposites found in the song, which led to a brief discussion on antonyms. This was followed by a round of dumb charades where the teacher mentioned a hotel-related word, and students acted it out immediately (Video- A10). Through these activities, the pre-reading phase introduced new vocabulary effectively using the AI-generated song. Subsequently, students completed vocabulary exercises from Menschen A1.2 (pages 91–92). They concluded with a productive task where they wrote dialogues simulating requests for assistance when something malfunctions in a hotel room, along with appropriate responses (Fig. A11).

The results of this experiment were not assessed through a Google Form. However, it was evident that the students had successfully learned the vocabulary, as they completed all subsequent exercises on the topic accurately and with ease, demonstrating a clear understanding. As a final productive task, the students wrote dialogues and illustrated corresponding images, which are included in the appendix.

The third experiment was conducted online on March 12, 2025, during the German Intensive Course 6 at Savitribai Phule Pune University, from 8:30 a.m. to 10:30 a.m. The participants were total beginners at the A1.2 level, using the textbook Menschen A1.1 and A1.2, with a basic understanding of grammatical structures and vocabulary up to Lesson 16. The objective was to investigate whether an AI-generated song could assist in revising the grammar topic Partizip 2/Perfekt (past participle) within the context of traveling. Since this topic had already been taught, the AI-generated song was intended as part of the revision phase. The song was created by providing a detailed prompt to ChatGPT.

The original prompt in German was: generiere bitte einen Liedtext, um das Grammatikthema Partizip 2 für das Niveau A1 zu wiederholen. Nimm dabei Verben, die haben und sein als Hilfsverb haben, und grundlegende A1-Verben (zum Beispiel: aufstehen, fernsehen, fahren, bleiben, wohnen, frühstücken, gehen, schwimmen, essen, spielen, kommen, anfangen, machen, lesen, aufräumen, lernen, kaufen, fliegen – diese Verben sind nur Beispiele und müssen nicht alle verwendet werden). Das Lied soll über das Thema “Mein Urlaub“ sein. Der Liedtext ist für A1-Kinder und soll 2 Strophen beinhalten (its English translation:* please generate a song text to revise the grammar topic of participle 2 for level A1. Use verbs that ‘to have’ and ‘to be’ as auxiliary verbs and basic A1 verbs (for example: to get up, to watch TV, to drive, to stay, to live, to have breakfast, to go, to swim, to eat, to play, to come, to start, to make, to read, to tidy up, to learn, to buy, to fly) these verbs are only examples and all may not have to be used. The song should be about the topic “My vacation”. The lyrics are for A1 level and should contain two stanzas).

After adding this prompt, a song text was generated with two stanzas and a refrain. But there were three complex phrases that the A1 students would not have understood. So another prompt was added to edit the song. Prompt in German: Ich brauche die Verben: ‘werden’ und ‘hat...genossen’ nicht. Gib mir ein anderes Version (its English translation*: I don't need the verbs: 'become' and 'enjoyed'. Give me another version). After this instruction for editing the song text, ChatGPT removed the verses that contained these two verbs but added another difficult verse, which was in Präteritum, i.e., the simple past tense, which was not taught to the students. With the help of another small instruction, the song text was again edited. Prompt in German: Ich brauche ‘freute mich sehr’ nicht. Das ist ein Lied für A1 (its English translation*: I don't need 'freute mich sehr'. This is a song for A1 level).

The version created after this prompt was the final version, which was submitted to suno.ai for tuning. In the section on the style of music, the following instruction was given in German: das Lied soll ein Reiselied sein (its English translation*: the song is supposed to be a travel song).

Out of the two versions created a 2:42 version was selected and used in the class (A12, A13).

The experiment, planned for 90–100 minutes, began with a brief discussion about travel, during which the teacher shared her recent travel experience, and students added one or two sentences about their most recent trips. During the first listening of the AI-generated song, students identified and marked the verbs, noting that most were in the Partizip 2 form, with two in the past tense and one in the present tense.

Following this, the verbs were dramatized through a dumb-charades activity, where students acted out the verbs, followed by matching song verses to relevant images to reinforce comprehension. To review grammar, students completed an online exercise on Learning Apps that involved selecting the correct Partizip 2 form, linked to the theme “Travel to Ayvalik.” This was followed by a Wordwall game, where students spun a virtual wheel, asked Yes/No or Wh-questions using the past participle, and their peers responded accordingly. Subsequently, students rewrote the entire song, originally written in the Partizip 2 tense, in the present tense (Fig. A14). Their versions were cross-checked using AI, while the teacher simultaneously generated the present-tense version of the song on Suno.ai. The prompt in German was: Das Lied soll ein Reiselied sein. Bitte die frühere Melodie (its English translation: the song is supposed to be a travel song. Please use the previous melody).

Despite this instruction, the song was not created by AI using the same earlier tune. However, it was now transformed into the present tense, retaining the same lyrics (A15). For checking the students’ song, this new version in the present tense was played, and the students were amazed that the text they transformed is a real song. As a productive writing exercise, the students wrote about their travel experience (A16).

The results were analyzed using a Google Form filled by all 20 participants (Fig. A17). The analysis revealed that 100% of the students enjoyed learning and revising grammar and vocabulary through songs, with 95% agreeing that the song’s verses helped them reinforce vocabulary. Similarly, 95% found that a customized song was beneficial, engaging, and motivating.

Regarding AI-generated songs, 70% believed they could aid in learning German, and 90% expressed interest in incorporating such songs in future lessons. Additionally, 40% indicated that the song clarified grammatical structures, particularly the Partizip 2 forms, while another 40% mentioned that the catchy tune helped them retain the words and verses. Meanwhile, 20% reported that the song effectively supported vocabulary revision. Overall, the results highlight that students appreciated the lecture and showed enthusiasm for integrating AI-generated songs as an enjoyable, motivating, and effective method for learning a foreign language.

Fig . 2. Student perceptions of an AI-generated song used to teach Partizip 2 and travel-related vocabulary in German.