Bug with lemmatisation: lemma of "cameras" is "cameras"?

[Saibo-creator]

How to reproduce the behaviour

import spacy
nlp = spacy.load('en_core_web_sm')
doc = nlp("cameras, fungi")
for token in doc:
    print(token._lemma)
# return  cameras, fungi, expected: camera, fungus

Info about spaCy

• spaCy version: 2.3.8
• Platform: macOS-10.16-x86_64-i386-64bit
• Python version: 3.8.13
• Models: en

[adrianeboyd]

The underlying cause is the same reason as described here: #5343 (comment)

If your texts are longer with more context ("The photographer always carries two cameras with her when she photographs fungi in the woods.") then the POS is more likely to be correct.

[adrianeboyd]

Let me convert this to a discussion...