Work on summarization has explored both reinforcement learning (RL) optimization using ROUGE as a reward and syntax-aware models, such as models whose input is enriched with part-of-speech (POS)-tags and/or dependency information. However, it is not clear what is the respective impact of these approaches beyond the standard ROUGE evaluation metric, which arguably fails to capture several important qualitative aspects of texts. Especially, RL-based for summa-rization is becoming more and more popular. In this paper, we provide a detailed comparison of these two approaches and of their combination along several dimensions that relate to the perceived quality of the generated summaries: how many words are repeated in the output ? How close to the ground truth is the generated distribution of part-of-speech tags ? What is the impact of sentence length ? How good are relevance and grammaticality ? Using the standard Gigaword sentence summarization task, we compare an RL self-critical sequence training (SCST) method with syntax-aware models that leverage POS tags and/or Dependency information. We show that on all qualitative evaluations, the combined model gives the best results, but also that only training with RL and without any syntactic information already gives nearly as good results as syntax-aware models with less parameters and faster training convergence.