Pneumonia is an inflammation of one or both lungs, typically due to a bacterial, viral, or fungal infection. Pneumonia diagnosis involves highly skilled professionals to examine a chest radiograph and is prone to subjective variability. Also, computer-aided classification of pneumonia using deep learning models is constrained by the scarcity of annotated samples and the black-box nature. To eliminate both these challenges, we have devised an explainable pneumonia classification technique by utilizing a slot-attention based few-shot learning methodology. We designed an explainable slot-attention based classifier using a publicly available pneumonia dataset with a customized loss designed to control the size of explanatory patterns in the image. For better interpretation, heat maps are computed to spot regions perceived as important for the decision by the model. The proposed method for pneumonia classification attains an accuracy of 91.0% on covid- chestxray-dataset, which is at par with the state-of-the-art methods but by making use of much less training data. Interpretable visualizations ensure the application of these methods in clinical settings. The proposed method gives superior classification results, accurate explanations, effective generalization in comparison to state-of-the-art techniques in low data regimes. This work can assist medical practitioners in preliminary pneumonia screening and expedite the treatment process.