The variable context: What counts and what doesn’t

In this section we will look at the variable contexts for past tense marking and plural marking. The variable context is the set of sentences where two or more variants are interchangeable or have the exact same meaning despite their different forms. For example, in Colloquial Singapore English, He bought a book yesterday is equivalent to He buy a book yesterday, where both the past tense form, bought, and the base form, buy, are interchangeable variants. It is necessary to precisely define the variable context for several reasons. First, it allows us to investigate the motivations behind the use of different variants in different situations. Second, it allows rough comparisons between different studies that define the variable context differently. Lastly, it allows the replication of similar future studies by other researchers.

The variable context for past tense marking

Previous studies of past tense marking (see Bickerton 1975; Patrick 1999; Rick- ford 1987; Singler 1990; Winford 1992 among others) have delineated the variable context in different ways. The variable context in the context of past tense marking refers to the absence or presence of past tense morphology. In our study, we follow the criteria set out in Poplack and Tagliamonte (2001), which gives the broadest possible definition of the variable context, thereby facilitating crossvariety comparisons. Their two criteria are reference to past time and instantiation of the past tense either as a regular -ed, a strong form, or a zero. Examples of zero form or base form usage in a past context are weak verbs drop and ask in (7) and come in (8).

• (7) Then the bus driver just drop us there even though we ask him, ‘Oh can you take us back to the bus interchange?’
• (Chinese Female, 22 years old)
• (8) Ah, waste matter collectors, where they come with a big lorry or van.

‘The waste matter collectors always came in a big lorry or van.’

(Chinese Male, 64 years old)

Past time reference includes punctual past (see Example (9a)), past habitual (see Example (9b)), present perfect (see Example (9c)), or past perfect (see Example (9d)). The following examples of possible variable contexts are from the interview data.

• (9a) Just yesterday I met with my primary school teacher
• (Chinese Female, 22 years)
• (9b) Quite often we played soccer in school.
• (Chinese Male, 25 years)
• (9c) I already travel a lot of places.

‘I had already travelled to a lot of places.’

• (Malay Male, 61 years)
• (9d) Then after he walked away, she told her son in Tamil, ‘Don’t bother, just continue eating.’

‘Then after he had walked away, she told her son in Tamil, ‘Don't bother, just continue eating.’

(Tamil Female, 33 years)

Only main finite verbs with a past reference are included in the data analysis of this quantitative study. In other words, any non-temporal usage of past tense morphology is not included in the data analysis. Examples of non-temporal usage from the interview data include conditional or modal uses like Example (10a), counterfactuals like Example (10b), and possible or counterfactual conditions like Example (10c).

• (10a) I wouldn’t say they are that bad, you know.
• (Malay Female, 39 years)
• (10b) I could have been military police or something.
• (Chinese Male, 25 years)
• (Юс) I hope you enjoyed my talk.
• (Tamil Male, 54 years)

Other non-temporal usages include fixed expressions like (lOd) or reported speech like (lOe). Fixed expressions are determined based on whether other forms of inflection are possible with a particular meaning (Shirai and Andersen 1995). For example, used in the sentence I think they’re just used to it cannot appear as other forms like uses or using, and is considered a fixed expression. Since fixed expressions and reported speech could be pre-fabricated chunks

44 Missing you - Past tense and plural marking

that are stored in an individual’s memory, they are excluded from the data analysis.

• (lOd) No, I used to study in the UK right.
• (Malay Male, 53 years old)
• (lOe) Then my grandma said, ‘Eh, they come back already.’
• (Malay Male, 35 years old)

Lastly, ellipsis like (1 Of) and ambiguous verbs which can be interpreted as past or non-past like (lOg), are also excluded from the analysis.

• (lOf) No I didn’t (ask for it).
• (Chinese Male, 25 years old)
• (lOg) So instead of English we take language arts which is a mixture of philosophy.
• (Chinese Male, 24 years old)

The take in (lOg) can be interpreted as either referring to the interviewee's particular cohort, or more generally, to past and current students who took or are taking language arts. Additionally, restricting the analysis to main finite verbs mean that auxiliary verbs like (lOh) are excluded from the analysis.

(1 Oh) I went to the same company as a recruit and then was posted back there. (Chinese Male, 25 years)

Lastly, as copula be can be omitted in Colloquial Singapore English, it is also not included in the analysis as there are three possible variants (past tense form, base form, and null form) for this feature and not two variants like the other verbs. Example (lOi) is an example of copula be in the interview data.

• (lOi) I remember, like the response was overwhelming.
• (Chinese Female, 18 years)

In terms of phonological environment, if a weak verb is followed by the consonants It/ or Id/ like (11), it will not be included in the data analysis because it is difficult to determine if the verb is truly marked for past tense or it is a phonological effect of the following consonant.

(11) I join dancing, and I play all sorts of games.

T joined the dance club and I played all sorts of games.’

(Chinese Female, 55 years old)

The verb join in Example (11) may sound like it is marked for past simply because the following consonant is a /d/.

If an interviewee repeats a verb multiple times like in Example (12), each repeated verb is considered a token in the analysis.

• (12) Yah, we travel, we traveled quite often together.
• (Malay Male, 35 years old)

In (12), both travel and traveled are included in the data analysis where travel is included as an unmarked token and traveled is included as a marked token.

To sum up, the variable contexts selected for quantitative analysis need to fulfill two criteria. First, reference to past time. Second, instantiation of the past tense form. In addition, it is necessary to make sure that all the contexts belonged to the same envelope of variation, where an unmarked form is the same way of saying what the marked form meant (Hackert 2008). When the past tense is instantiated as a regular -ed or strong form, it is coded as 1. When the past tense is instantiated as a zero, it is coded as 0. In other words, presence of the past tense form is coded as 1 and absence is coded as 0 in the data.

The variable context for plural marking

The variable context for plural marking is defined as the set of countable nouns that should be marked by the plural marker Is/, regardless of whether they cooccur with an overt plural marker (Tagliamonte et al. 1997). More specifically, the two criteria are plural reference of a countable noun and instantiation of the plural marker either as an Is/, including all its phonological variants, or a zero. Irregular plural nouns like women and fish are not included in the data analysis. An example of a zero form or base form usage of a noun with a plural reference is opportunity in (13).

• (13) A lot of opportunity.
• (Tamil Male, 37 years old)

Like other variation studies, nouns that are ambiguous between plural or singular reference like (14), and nouns that are categorically marked or unmarked for plurality like in terms of and humanities, are excluded from the statistical analysis.

• (14) So it can be like literary theory as applied to some other field.
• (Chinese Male, 25 years old)

In Example (14), field is ambiguous, as the speaker could either have a singular ox- plural reference in mind. In this example, he could be thinking about a particular field of study or several fields of study.

In terms of phonological environment, if a plural marker is followed by the consonant /s/ like (15), it will not be included in the data analysis because it will be difficult to determine if the noun is truly marked for plurality or it is simply a phonological effect of the following consonant.

• (15) Then it’ll be six generations so now it’s five.
• (Tamil Female, 65 years old)