Each language has unique thought patterns which can cause a negative (or positive) interference between L1 and L2. Many authors focus on comparing the patterns in verb-framed (V-framed) and satellite-framed (S-framed) languages whereby studies often involve participants whose L1 and L2 are not in the same category. The influence of L1 has been researched in terms of how accurate (and/or native-like) learners' L2 is, the age at which L1 cognitive patterns are the most likely to become weaker, and the type of corrective feedback that helps learners develop L2 thought patterns. L1 thought patterns are formed in early childhood, and they influence our use of language. Native speakers of a certain language pay attention to particular details (e.g. manner or path), which influence the way they use language in order to describe reality. Such learned perception will transfer to their L2 which implies that acquiring L2 thought patterns takes time, effort and appropriate instruction which includes error correction. Recast has proven to be the most efficient type of corrective feedback, althought explicit instruction can be very beneficial as well. The present study examined how Croatian EFL learners encode path when describing boundary-crossing motion events in English in order to assess whether the choices were influenced by their L1 thought patterns. The study was conducted on three groups of participants – elementary, upper-intermediate and advanced – which were compared in terms of the average number of correct answers and the type of errors that had been made. The results showed the highest accuracy in the advanced group, and the lowest in the elementary group. The influence of L1 was noticeable in the answers of all three groups of participants who erroneously used prepositions with a locative meaning instead of a translocative one. Such answers were prevalent in the the elementary group while they were the least frequent in the advanced group.