# Distributional properties

## Results

### Nouns

There were 203 novel nouns in total. Of the various reference constructions, these forms display full referentiality (subject or object position) in 140 cases (69.0%): 74 subjects (36.5%) and 66 objects (32.5%). An example of each is given in (1-2), below:

(1) The bleff isn't working, can you fix it please? (noun 2, participant 36)

(2) I want a froop next time I go to the supermarket. (noun 2, participant 57)

Co-occurrence with a determiner was more frequent (170 cases or 83.7%), which means there were 30 examples (or 14.8%) of determiner-noun combinations occurring in positions other than subject or object. Example (3), below, illustrates this:

(3) I went to the shop with my gleep. (noun 1, participant 1)

I argued in Section 3.3.2, above, that these should be analysed as being of intermediate referentiality. The chi-square test reveals that the difference between subject and object status is not statistically significant (p = 0.4990; df = 1, j2 = 0.457), while the difference between subject and object position taken together on the one hand, and presence of a determiner is nearly statistically significant (p = 0.0884; df = 1, j2 = 2.903). Finally, the difference between full referentiality (subject or object status) and merely partial or non-referentiality is very highly significant (p < 0.0001; df = 1, j2 = 29.207).

As for non-referential status, the remaining 33 novel nouns (16.2%) are used as predicate nominals or complements of prepositions and are not combined with a determiner. Each pattern is illustrated with an example below:

(4) Blond people are mooshes. (noun 2, participant 62)[1]

(5) It can be seen by the increased usage of rizification that the nutrients are not having the desired successful. (sic, noun 1, participant 4)[2]

The difference with partly or fully referential status is very highly significant (p < 0.001; df= 1, J2 = 92.458).

The results for full, intermediate, and non-referentiality are summarised in Table 1.

### Verbs

The participants produced 202 novel verb forms. Of these, 160 (79.2%) occurred in finite clauses. Some examples of finite main and subordinate clauses are given in (6-9), below:

(6) He rickled at the thought of a clown at his birthday party.

(verb 1, participant 58)

(7) I went clubbing last night and volhardied. (verb 1, participant 62)

(8) If we all linted all the time, the world would be a better place.

(verb 2, participant 32)

(9) The pain was so intense he almost wurzed. (verb 3, participant 13)

The remaining 42 examples featured participial forms or to-infinitival complements of verbs or other uses that involve predication much less clearly than the sequentially scanned events in (6-9) above. Some illustrative examples are (10-12):

(10) I began romcooing down the hallway. (verb 2, participant 22)

(11) She was bored so decided to sibbify her work. (verb 2, participant 3)

(12) How many people does it take to stificate an elephant? None, it's impossible without getting trampled on. (verb 3, participant 69)

The difference between the predication and non-predication uses of the novel verbs is statistically very highly significant (p < 0.0001; df= 1, x2 = 68.931).

• [1] Inflectional and derivational (see example 5) morphology was not excluded from my analysis, as it is included in the psycholinguistic literature on word classes as well.
• [2] The rather elaborate examples many participants came up with suggest that they really engaged with the task, and therefore that it is a useful instrument for investigating production.