Studies in language variation and change normally conduct quantitative analyses of (relatively) large samples. In this article we demonstrate how even meagre tokens can provide key information about the characteristics of change in conditions of loss. As a case study, we study double demonstrative constructions (DDEMO_NPs) in Ontario, Canada. We show that DDEMO_NPs are obsolescent, restricted to older male speakers doing blue collar jobs. Building on theories of dialect recession (Schilling-Estes and Wolfram, 1999), we argue that on their way to moribund status, DDEMO_NPs undergo a process of ‘concentration’ of the kind postulated by Jankowski and Tagliamonte (2017). Specifically, we demonstrate that from among the potential range of forms and functions of demonstratives, speakers relatively frequently use DDEMO_NPs as ‘indefinite this’ (Prince, 1981a), a fairly recent usage of this to introduce a new referent into discourse. This finding suggests that loss of a form is not a simple matter of decreasing frequency. Rather, qualitative scrutiny of the data is essential to expose more nuanced trajectories of decline. Additionally, studying obsolescent dialect features may reveal the origins of new uses; we will advance the idea that DDEMO_NPs provide an alternative perspective on the emergence of indefinite this via discourse-pragmatic innovation.

语言变异和变化的研究通常对（相对）大样本进行定量分析。在本文中，我们展示了即使是微薄的代币也可以提供有关损失条件变化特征的关键信息。作为案例研究，我们研究了加拿大安大略省的双重示范结构（DDEMO_NPs）。我们表明 DDEMO_NP 已过时，仅限于从事蓝领工作的年长男性演讲者。基于方言衰退理论（Schilling-Estes 和 Wolfram，1999 年），我们认为在走向垂死状态的过程中，DDEMO_NP 经历了 Jankowski 和 Tagliamonte（2017 年）假设的那种“集中”过程。具体来说，我们证明了在指示词的形式和功能的潜在范围中，说话者相对频繁地使用 DDEMO_NPs 作为“不定this ' (Prince, 1981a)，这是一个相当近的用法，用于在话语中引入一个新的所指对象。这一发现表明，形式的丢失并不是降低频率的简单问题。相反，对数据进行定性审查对于揭示更细微的下降轨迹至关重要。此外，研究过时的方言特征可能会揭示新用途的起源；我们将推进这样的想法，即 DDEMO_NPs通过话语-语用创新为不确定的this的出现提供了另一种观点。