The answer to the question will likely vary from grammarian to grammarian. My short answer is that distribution is a necessary condition for identifying nominals and morphological criteria, e.g. plural -s, are sufficient.
To be classified as a nominal, a given token of a lexical item must be able to appear in a position that is associated with nominals. In other words, the distribution of the lexical item is a necessary condition. So, for instance, if an expression can appear as a subject, it is a candidate for nominal status, e.g.
Tom stopped.
It stopped.
The laughing stopped.
The first stopped.
Since Tom, it, the laughing, and the first can appear as subjects, one might classify each as (containing) a nominal. In other words, since these expressions can appear in a position that is associated with stereotypical nominals, they are all candidates for nominal status.
Note, however, that satisfying a necessary condition does not mean that the candidate is definitely a nominal. It just means that it is potentially viewed as a nominal. Thus one could argue that first in the fourth example is actually not a nominal, but rather it is an adjective that immediately introduces an elided noun (the first whatever).
A sufficient condition for identifying nominals is the ability to take plural -s. Any item that forms its plural with -s is, I believe, going to be accepted by most as a nominal, e.g.
boy vs. boy-s
time vs. time-s
discussion vs. discussion-s
Since these lexical items form their plural with -s, they are easily classified as nominals. The -s condition is sufficient.
Another criterion that one might employ and view as sufficient is the ability to be introduced by the definite article the or the indefinite article a. If one takes this criterion as sufficient then laughing, and first above are definitely nominals.
But there are of course numerous expressions that have the distribution of typical nouns (necessary condition), but that one has difficulty classifying as a nominal based upon morphological criteria (sufficient condition). As Adam points out, gerunds are a good example. Gerunds have the distribution of nominals and they can be introduced by an article, but they do not take plural -s, e.g. **the laughings*, **the discussings*, etc. Thus whether or not one classifies gerunds as nominals is going to depend on which criteria one takes to be more important. My stance takes gerunds to be a separate category that straddles two other categories. Gerunds are nominals with respect to their head but verbs with respect to their dependents.
In the big picture, each grammarian will decide for him- or herself which criteria are most pertinent for classifying lexical items. I think there is therefore variability in how exactly the term nominal is employed.