Copula absence has been extensively investigated in English-based
creoles (EBCs) and African American Vernacular English (AAVE) (Rickford
1991).  As Walker and Meyerhoff (2004) pointed out, however, a handful
of EBCs (e.g., Barbadian, Jamaican, Trinidadian, Guyanese) have
received most of the attention, and there is the need for more
variationist analyses of copula absence in other EBCs.  In this paper,
I demonstrate a multivariate analysis of copula absence in Hawai‘i
Creole (HC), an English-based creole spoken in the islands of Hawai‘i. 
The implication of the results involves a pan-creole tendency as well
as a current decreolization process in HC.
 
Amongst the research in copula absence in EBCs and AAVE, the
hierarchical ordering of the following grammatical environment has been
reported as most significant.  Rickford and Blake (1990) argue that if
a prior creole origin were to leave its vestiges in a decreolizing or
decreolized variety, we would expect to find the following patterns of
copula absence.

_ gonna > _ V+ing > _ Adj > _ Loc > _ NP

Previously, copula absence in HC was quantitatively
investigated with distributional analyses some thirty years ago (Day
1973; Pearlman 1973).  Day’s (1973) study of HC provided one of the
first substantive quantitative studies in a creole continuum.  Day
(1973) observed an internal implicational pattern within the use of
null copula in 4 environments. 

           _ V+ing > _ Adj > _ Loc > _ NP

Perlman (1973) also found a similar hierarchy of the variability of
copula in 5 environments. 

         _ gon > _V+ing > _ Adj > _ Loc > _ NP

The present study analyzed sociolinguistic interviews with rural O‘ahu
residents collected by the project External Influences and Internal
Variation in Current Hawai‘i Creole (PI Jeff Siegel) in 2001-2005. 
Compared with the outer islands, the island of O‘ahu is reportedly
where the most “decreolized” and therefore the most acrolectal
varieties are found (Romaine 1994).  Data from the twenty speakers
stratified by age and gender were quantified and analyzed using
Goldvarb 2001 (Robinson, Tagliamonte & Lawrence 2001).  Tokens were
coded for the following linguistic factors: grammatical person and type
of subject, following grammatical category, and preceding as well as
following phonological context.

Results indicate that the current HC speech clearly demonstrates an
implicational pattern that implies a prior creole origin argued by
Rickford and Blake (1990).  However, evidence from social constraints
shows that the creole pattern is observed similarly across age groups. 
Also, comparison with the previous studies (Day 1973; Perlman 1973)
reveals that the probability of the occurrence of copula absence in 5
environments has not been changed largely in past thirty years.