At the beginning of learning a new language, students go through a demanding process. They encounter the new linguistic system they need to master for the first time. This is when most students encounter the phonetic and alphabetic system of the language they are learning, and in teaching, attention should be paid to the alphabet of the language being learned, taking into account the graphemes and phonemes that this alphabet represents. Languages that use an alphabetic writing system represent their phonemes in various ways, so a grapheme corresponding to one phoneme can consist of a single letter, a derived letter, a string of letters, and the same grapheme can represent different phonemes, which are realized as distinct sounds depending on the context (e.g., Jelaska and Musulin 2011). It is necessary to associate spoken and written forms with the abstract concept of the corresponding linguistic unit to ensure that language reception and production are consistent with the concept of each phoneme. In doing so, students need to exert conscious effort, and the ability to segment words into the sounds that make them up and blend those sounds to form new words is called phonemic awareness (Yeong, Rickard Liow 2012). On the other hand, the ability to detect and manipulate the sound segments of spoken words, or understanding that spoken words can be broken down into smaller units such as syllables or phonemes, is called phonological awareness (Torgesen et al. 2007, Puffpuf 2009). Numerous studies have shown that phonological awareness is the most important individual phonological factor underlying effective reading acquisition, and that phonemic awareness skills in preschool children can successfully predict later reading achievements. Therefore, the aim of this research is to explore phonemic awareness in foreign language learners of the Croatian language. The goal is to provide an overview of the difficulties that speakers of Italian, Polish, and German encounter in mastering writing, listening, and pronunciation in the Croatian language. Such an overview can serve as a starting point in teaching Croatian spelling and pronunciation. Thus, this work aims to contribute to the field of mastering Croatian as a second language with an emphasis on applying insights to teaching. The study involved 27 participants, aiming to provide insight into phonemic awareness in terms of recognizing specific units isolated and in context at several different levels – phonemes in listening, phonemes in pronunciation, graphemes representing phonemes in reading and writing, as well as letters in writing, and phonemes in different positions in words – and at different levels of learning. To provide a detailed insight, this research is based on a case study. In second language research, a case study typically provides a detailed description of a specific group of students in their second language learning environment. The main goal of this research was to determine and explain the phonemic awareness of nonnative Croatian language learners based on the transfer of spoken language units into written text and vice versa. Mastery of phonemes and graphemes in the Croatian language was observed in three skills of individual language activities – writing based on dictation, listening to read sentences, and pronunciation based on reading. The research was based on theoretical knowledge about acquiring a foreign language, particularly related to the acquisition of phonemic awareness. It involved subjects whose native languages were Italian, Polish, and German at beginner, intermediate, and advanced levels of Croatian language proficiency. The results of this research showed a statistically significant result in writing phonemes based on dictation according to the level of Croatian language knowledge. It was found that beginners in learning Croatian achieved significantly lower accuracy in writing graphemes compared to advanced learners, while there was no statistically significant difference between other groups. No statistically significant difference was found in writing graphemes according to the participants' native languages. It was only shown that with increasing levels of Croatian language proficiency, the influence of the native language decreases, which is in line with the descriptors in the CEFR and existing knowledge about acquiring Croatian as a foreign language (e.g., Udier 2013, 2015, 2017). At the beginner level, participants correctly wrote the following graphemes: vowels (a, e, i, o, u), nasals (m, n; l, r, j), and from fricatives (b, d, g, p, k), two affricates (f, h) and one sibilant (c). Comparing graphemes in Italian, Polish, and German, it can be concluded that all correctly written graphemes are present in these languages and are part of the alphabet, i.e., they are alphabetic letters in these languages (Jelaska, Musulin 2011). At the beginner level, all participants did not correctly write twelve graphemes: three nasals (nj, lj, v), one plosive (t), four fricatives (s, š, z, ž), and four sibilants (č, ć, đ, dž). Minor difficulties at the beginner level were noted with writing two plosive graphemes (t, z) and two nasals (v, nj). The most difficulties were with writing all palatalized graphemes (š, ž, č, dž, ć, đ). In writing the phoneme v, Polish and German participants transferred the grapheme w from their native languages, which are used to write the same sounds, and similarly ń for nj by Polish participants. The grapheme lj caused such difficulties for some beginner-level participants that they did not write it at all. At the intermediate level, all participants correctly wrote all vowel graphemes (a, e, i, o, u), all fricatives (f, h, s, š, z, ž, č), most plosives (p, t, k, b, d), and only two nasals (l, v). This means that not all correctly wrote the nasal graphemes (m, n, nj), three liquids (lj, r, j), and from the fricatives, the voiced velar (g) and four sibilants (c, ć, đ, dž). At the advanced level, all participants correctly wrote all vowel graphemes (a, e, i, o, u), most nasals (m, n, l, lj, r, j), and from the fricatives all plosives (p, t, k, b, d, g), all fricatives (f, h, s, š, z, ž) and only one sibilant (c). This means that even at this level, not all participants correctly wrote all other sibilant graphemes (č, ć, đ, dž), the palatal nasal (nj), and the labio-dental nasal (v). Thus, participants at intermediate and advanced levels were generally successful in writing individual graphemes. At both levels, there were minor difficulties in writing four prepalatalized graphemes (j, lj, nj, ć), and more significant difficulties with voiced sibilants (dž, đ). At the intermediate level, there were difficulties with the other two nasals (m, n), the trill (r), the voiced velar (g), and at the advanced level with the approximant (v). Although there were no difficulties with the approximant grapheme v at the intermediate level, the fact that not all advanced-level participants mastered it (one Polish participant transferred the grapheme w from their native language) classifies it among the graphemes that not all participants fully master even at the advanced level. At the intermediate level, no one correctly wrote the sibilant grapheme dž, and eight participants did not correctly write the grapheme đ. At the advanced level, four participants did not correctly write dž, and three did not correctly write đ. From this, it can be concluded that participants are more successful in writing graphemes at higher levels, but difficulties with writing graphemes specific to the Croatian language still occur. This research confirms the results of other studies, e.g., M. Musulin and Z. Jelaska (2015); S. Jukić and T. Diklić (2022); S. Jukić, T. Diklić, and B. Prosenjak (2022). The fact that writing palatal sibilants is challenging for non-native learners of Croatian is entirely expected (many native speakers also have difficulties with them), but it is surprising how challenging nasals are in writing, especially at the intermediate level, and that not all learners mastered them even at the advanced level, although this only applies to nj and v. In listening, or recognizing phonemes in a sentence, no statistically significant difference was found according to the level of knowledge of the Croatian language. However, there is a statistically significant difference in listening according to the native language of the participants. Specifically, students whose native language is Polish achieved significantly greater accuracy compared to participants whose native language is German (but not for other languages). Descriptive indicators for overall accuracy show a high average accuracy. The negatively skewed distribution showed a higher prevalence of higher scores. All results showed that participants could recognize individual phonemes with very high accuracy when they hear them. Accuracy below 60% was recorded for only four examples: mađ (59.3%), nadž, bać (55.6%), and đas (44.4%). In the examples of mađ and nadž, the cause of all incorrect responses was only the final phoneme, /ʒ́ / and /ǯ/, indicating that voiced sibilants are among the most challenging for non-native learners to acquire. Participants recognized the phonemes /ć/ and /ǯ/ with greater accuracy when they were at the beginning of the word (ćak and džag) than at the end of the word (bać and nadž). In the example of bać, most incorrect answers involved the failure to recognize the phoneme /ć/, but a few participants also failed to recognize the phoneme /b/. In the example of đas, an equal proportion of incorrect answers involved not recognizing /ʒ́ / or /s/. Such results may indicate a lack of mastery of the voicing contrast. Specifically, the results show that it is significantly more difficult to distinguish voiced b from its voiceless counterpart p, as well as voiced z from its voiceless counterpart s in sentences (Jukić 2022). In the pronunciation of Croatian phonemes, it was found that there is a statistically significant difference in accuracy in the task of pronouncing phonemes according to the level of knowledge of the Croatian language. Participants at the advanced level achieved significantly greater accuracy in pronouncing phonemes compared to participants at the beginner level, while no statistically significant difference was found between other groups. However, there is no statistically significant difference in pronunciation accuracy according to the participants' native language. Although participants whose native language is German achieved the lowest results in this task, no statistically significant difference was found compared to participants whose native language is Polish or Italian. All participants pronounced the 30 Croatian phonemes with very high accuracy, as determined by descriptive indicators: although the range was from 12 to 30, the median and mode were 27. Six evaluators agreed that all participants correctly pronounced the vowels i, u, and the consonants b, d, p, t, while there was one incorrect pronunciation for the vowels a, e, o, and the consonant g. The evaluators rated the pronunciation of the palatals nj, đ, lj the lowest. Interestingly, they found that only two participants incorrectly pronounced the postalveolar affricate č, but six participants incorrectly pronounced the palatal affricate ć. The pronunciation of dž was rated incorrectly by half as many participants as đ. Less than half (48%), or thirteen participants, incorrectly pronounced lj. This research confirmed that non-native learners follow a certain order in acquiring sounds (e.g., Jakobson 1969; Jelaska 2005), and thus in this research, vowels are acquired first, with minimal deviations in the pronunciation of high vowels, mid, and low vowels (a, e, o). After them, non-native speakers of Croatian acquire the pronunciation of stops (p, t, k, b, d, g), which was largely shown in this research. Namely, the evaluators rated the pronunciation of the four front stops p, t, b, d as correct for all participants, while the back stops were more challenging: the pronunciation of k was rated as incorrect for three participants, and g for only one participant. Although the number of participants who incorrectly pronounced certain phonemes varies, it is important to say that only three phonemes were incorrectly pronounced by more than 30% of the participants: nj (33%), đ (37%), lj (48%). The results of this research are consistent with theoretical insights into sound acquisition. For example, it was confirmed that participants find unmarked or more common sounds in human languages easier, while more marked or rarer sounds are more difficult (e.g., Eckman 1977, 1981). Besides the understandable positive transfer from the native language, participants in this research performed better in the pronunciation of vowels and stops that are generally more common in other languages and performed worse in the pronunciation of those that are more marked, specific to the Croatian language, or rarer in other languages. To examine the relationship between the writing of graphemes, the recognition of phonemes in a listening task, and the pronunciation of Croatian phonemes, Spearman's correlation coefficients (r) were calculated between the variables measuring accuracy in the listening tasks, writing tasks, and pronunciation tasks. The results showed significant moderate positive correlations between listening and writing (r=.51, p