This contribution considers the word error rate (WER) performance of a concatenated coding scheme in the presence of impulsive noise (IN), which is modeled as gated white Gaussian noise, with on-and off-times governed by a 2-state Markov model. The scheme consists of a Reed-Solomon (RS) outer code and a low-density parity-check (LDPC) inner code, which are separated by a block interleaver with finite depth. The Monte Carlo (MC) simulation of the WER of such communication systems is time-consuming, especially when targeting low error rates and examining several interleaver settings. We present a semi-analytical evaluation of the WER, which relies on a simple semi-analytical statistical model for the number of byte errors in a segment of the information word after LDPC decoding. To compute the error performance of the concatenated code corresponding to different parameters of the RS code and the interleaver, we require only the WER and byte error rate (ByteER) of the inner subsystem, determined by the LDPC code and the considered constellation, in the presence of stationary white noise. We show that the semi-analytical WER of the concatenated system closely matches the WER resulting from MC simulations and use the proposed model to investigate the effect of the interleaver depth on the WER performance.