As far as I am concerned L. casei and L. paracasei can be differentiated by the amplification of the 16S rRNA gene of the aforementioned species using PCR - please see the work done by Ward and Timmins (1999). An alignment of the V1 region of the 16S rRNA genes of Lactobacillus casei and Lactobacillus paracasei shows the bases comprising the species-specific PCR primers - screenshot taken from Ward and Timmins (1999):
So, perhaps in your study you have not amplified the correct region of the 16S rRNA gene of L. casei and L. paracasei?
With regards to the classification of these organisms. Just as @skymningen has already mentioned in the comment section, the definition of a bacterial species (not even to mention strains) is somewhat debatable and tricky.
However, here is what Mori et al (1997) suggest:
The primary structures of the 16S rRNA genes of the type strains of Lactobacillus casei and related taxa were determined by PCR DNA-sequencing methods. The sequences of Lactobacillus casei, Lactobacillus zeae, Lactobacillus paracasei, and Lactobacillus rhamnosus were different. The K nuc values ranged from 0.0040 to 0.0126. On the basis of the K nuc values and the levels of DNA-DNA relatedness among the strains of these species, the L. casei-related taxa should be classified in the following three species: L. zeae, which includes the type strains of L. zeae and L. casei; a species that includes the strains of L. paracasei and L. casei ATCC 334; and L. rhamnosus.