The language of alcohol: Similarities and differences in how drinkers and policymakers frame alcohol consumption

Introduction
The primary objective of the paper was to compare semantic domains reported by drinkers and policymakers in their alcohol consumption narratives. The research question was ‘How do drinkers and policymakers use semantic domains to construct alcohol consumption narratives?’.

Design
Secondary analysis of open-ended survey responses (The Drinker Corpus: TDC) and three English alcohol policies (The Policy Corpus: TPC).

Methods
Wmatrix software was used to identify semantic domains that appeared more frequently in our corpora compared to general usage. Wmatrix outputs a log-likelihood (LL) score; a score of 6.63 corresponds to a p value of 0.05, indicating frequently used domains.

Results
Five domains appeared more frequently in both corpora than general usage: ‘Cause & Effect/Connection’; ‘Disease’; ‘Drinks and Alcohol’; ‘Excessive drinking’; ‘Knowledge’ (LL >226.68 for all). Domains were represented differently in the two corpora; the TPC focused on long-term health consequences, like liver disease, whereas the TDC talked about short-term consequences like hangovers. The ‘Emotional actions’ domain appeared in the TDC more than expected (LL = 231.26). Drinkers reported experiencing positive and negative emotions following drinking. The ‘Social actions/states/processes’ domain was used more frequently in the TPC than expected (LL = 408.17). Policymakers talked about changing ‘behaviour’ in partnership with organisations rather than working with drinkers.

Discussion and Conclusions
This study shows that while drinker and policymaker alcohol consumption narratives draw on the same semantic domains, how these domains are used to construct these narratives differs. To improve the effectiveness of policy initiatives, we recommend greater dialogue between policymakers and drinkers.