We present a new framework for compositional distributional semantics in which the distributional contexts of lexemes are expressed in terms of anchored packed dependency trees. We show that these structures have the potential to capture the full sentential contexts of a lexeme and provide a uniform basis for the composition of distributional knowledge in a way that captures both mutual disambiguation and generalization. Aligning Packed Dependency Trees: A Theory of Composition for Distributional Semantics David Weir∗ University of Sussex Julie Weeds∗ University of Sussex Jeremy Reffin∗ University of Sussex Thomas Kober∗ University of Sussex We present a new framework for compositional distributional semantics in which the distribu- tional contexts of lexemes are expressed in terms of anchored packed dependency trees. We show that these structures have the potential to capture the full sentential contexts of a lexeme and provide a uniform basis for the composition of distributional knowledge in a way that captures both mutual disambiguation and generalization. 1. However, in contrast to previ- ous proposals, the higher-order dependency relations provides structure to the space that is crucial to our definition of composition. Each co-occurrence associated with a lexeme such as wooden is typed by the path in the dependency tree that connects the lexeme wooden with the co-occurring lexeme (e.g., fired). This allows us to en- code a lexeme’s distributional knowledge with a hierarchical structure that we call an Anchored Packed Dependency Tree (A PT). we denote the weight of the distributional feature ⟨τ, w′⟩ of the lexeme w with the expression W(w,⟨τ, w′⟩). 2.2 Anchored Packed Trees Given a dependency tree corpus C⊂ TV,R and a lexeme w∈ V, we are interested in cap- turing the aggregation of all distributional contexts of w in C within a single structure. We achieve this with what we call an Anchored Packed Dependency Tree (APT). APTs 732 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/COLI_a_00265 by guest on 08 August 2026 Weir et al. Aligning Packed Dependency Trees are central to the proposals in this article: Not only can they be used to encode the aggregate of all distributional features of a lexeme over a corpus of dependency trees, but they can also be used to express the distributional features of a lexeme that has been contextualized within some dependency tree (see One possibility is fully incremental left- to-right composition, where, working left-to-right through the string of lexemes, the elementary A PTs for the first two lexemes are composed, with the resulting A PT then being composed with the elementary A PT for the third lexeme, and so on. It is always 742 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/COLI_a_00265 by guest on 08 August 2026 Weir et al. Aligning Packed Dependency Trees possible to compose A PTs in this fully incremental way, whatever the structure in the dependency tree. The tree structure, however, is critical in determining how the adjacent APTs need to be aligned. Aligning Packed Dependency Trees Table 3 Neighbors of uncontextualized shoot/N compared with shoot/N in the contexts of green/J and six-week/J, using⨆ INT with aligned and unaligned representations. Aligning Packed Dependency Trees that those internal (and external) contexts that are not supported by a majority of the lexemes in the sentence will tend to be considered insignificant and therefore will be ignored in similarity calculations. By using shifted PPMI, it should be possible to further reduce the number of non-zero dimensions in a representation constructed using⨆ UNI , and this should also allow us to control the specificity/generality of the neighbors observed. 5.3 Phrase-Based Composition Tasks Here we look at the performance of one instantiation of the A PT framework on two benchmark tasks for phrase-based composition. 5.3.1 Experiment 1: The M&L2010 Data Set. Conversely, in the context of the subject face one might expect glowed to be close to beamed and not close to burned. 750 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/COLI_a_00265 by guest on 08 August 2026 Weir et al. Aligning Packed Dependency Trees This data set was used in the evaluations carried out by Grefenstette et al. (2013) and Dinu, Pham, and Baroni (2013). These evaluations clearly follow the experimental procedure of Mitchell and Lapata and do not evaluate against mean scores. Instead, separate points are created for each human annotator, as discussed in Section 5.3.1. The multi-step regression algorithm of Grefenstette et al. The direction of the NN relationship between the words leads to a different alignment of the A PTs and consequently a different representation for the phrases. There are other approaches that incorporate theoretical ideas from formal semantics and machine learning, use syntactic information, and specialize the data structures to the task in hand. For adjective–noun phrase composition, Baroni and Zamparelli (2010) 752 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/COLI_a_00265 by guest on 08 August 2026 Weir et al. Aligning Packed Dependency Trees and Guevara (2010) borrowed from formal semantics the notion that an adjective acts as a modifying function on the noun. The A PT generated by a dependency tree, however, provides contextualized lexeme representa- tions where the weights have been adjusted by the influence of the contextual lexemes so that the co-occurrences relating to the correct usage have been appropriately up- weighted, and the co-occurrences found in other circumstances down-weighted. In 756 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/COLI_a_00265 by guest on 08 August 2026 Weir et al. Aligning Packed Dependency Trees other words, A PT structures automatically perform word sense induction on lexeme- level representations and this is demonstrable through the lexeme similarity measure.