JSON supports various types of values. At a high level: Objects, Arrays, and various kinds of Scalars. A Spanner column of type JSON can support any of these:

spanner> create table t_json (i int64, j json) primary key (i);
spanner> insert into t_json (i,j) values (1,JSON '[1,2,3]');
spanner> insert into t_json (i,j) values (2,JSON '"foo"');
spanner> insert into t_json (i,j) values (3,JSON '{"a": "b"}');
spanner> select * from t_json;
+---+-----------+
| i | j |
+---+-----------+
| 1 | [1,2,3] |
| 2 | "foo" |
| 3 | {"a":"b"} |
+---+-----------+
Spanner's Python Client assumes that all JSON values are Objects. If you read back a value that is not an Object, you will get an error that looks like:

ValueError: dictionary update sequence element #0 has length 3; 2 is required
The root cause of this issue is this line here: https://github.com/googleapis/python-spanner/blob/main/google/cloud/spanner_v1/data_types.py#L47

        super(JsonObject, self).__init__(*args, **kwargs)