# A Network Model of Interpersonal Alignment in Dialog

^{1}

^{2}

^{*}

## Abstract

**:**

## 1. Introduction

- Firstly, the linguistic form /a/ activates or primes its representation a in the mind of the recipient.
- Secondly, by the priming of the mental representation a by its manifestation /a/, items that are, for example, phonetically, syntactically or semantically related to a may be co-activated, that is, primed in the mind of the recipient, too. Take the word form /cat/ as an example for a prime. Evidently, this word form primes the form /mat/ phonetically, while it primes the concept dog semantically.

- A:
- both churches have those typical church windows, to the bottom angular, to the top just thus (pauses and performs a wedge-like gesture)
- B:
- gothically
- A:
- (slightly nodding) gothically tapering

- Firstly, alignment by the coupling or linkage of interlocutors due to the usage of paired primes, that is, by linguistic units which both are used to express certain meanings and which connect their mental representations interpersonally.
- Secondly, alignment by the structural similarity of the networks of representations that are possibly co-activated by these paired primes.

- We develop a framework in which the notion of alignment, that we take to be essential for the understanding of natural language dialog, is operationalized and made measurable. That is, we provide a formal, quantitative model for assessing alignment in dialog.
- This model, and thereby the notion of alignment, is exposed to falsifiability; it is applied to natural language data collected in studies on lexical alignment. Our evaluation indeed yields evidence for alignment in dialog.
- Our framework also implements a developmental model that captures the procedural character of alignment in dialog. Thus, it takes the time-bounded nature of alignment serious and, again, makes it expressible within a formal model and, as a result, measurable.

## 2. Related Work

## 3. An Experimental Setting of Alignment Measurement

## 4. A Network Model of Alignment in Communication

#### 4.1. Two-Layer Time-Aligned Network Series

- Variant I—unlimited memory span:
- −
- Interpersonal links: if at time t, agent $X\in \{A,B\}$ uses the item $l\in {V}_{t}$ to express the topic $T=\mathit{T}\left(t\right)$ that has been expressed by $Y\ne X$ in the same round of the game or in any preceding round on the same topic T by the same item, we span an interpersonal link $e=\{{v}_{A},{w}_{B}\}\in {E}_{t}$ between ${v}_{A}\in {V}_{A}$ and ${w}_{B}\in {V}_{B}$ for which ${\mathcal{L}}_{A}\left({v}_{A}\right)={\mathcal{L}}_{B}\left({w}_{B}\right)=l$, given that e does not already exist. Otherwise, if $e=\{{v}_{A},{w}_{B}\}\in {E}_{t-1}$, its weight ${\mu}_{t}\left(e\right)$ is increased by 1. The initial weight of any edge is 1.
- −
- Intrapersonal links: if at time t, agent $X\in \{A,B\}$ uses item l to express $T=\mathit{T}\left(t\right)$, we generate intrapersonal links between ${v}_{X}\in {V}_{X}$, ${\mathcal{L}}_{X}\left({v}_{X}\right)=l$, and all other vertices labeled by items that X has used in the same round of the game or in any preceding round on the same topic T. Once more, if the links already exist, their weights are augmented by 1.

Variant I models an unlimited memory where both interlocutors always remember, so to speak, every usage of any item in any preceding round of the game irrespective how long ago it occurred. Obviously, this is an unrealistic scenario that may serve as a borderline case of network induction. A more realistic scenario is given by the following variant that simulates a limited memory. - Variant II—limited memory span:
- −
- Interpersonal links: if at time t, agent $X\in \{A,B\}$ uses $l\in {V}_{t}$ to express topic $T=\mathit{T}\left(t\right)$ that has been expressed by $Y\ne X$ in the same or preceding round on the same topic by the same item, we span an interpersonal link $e=\{{v}_{A},{w}_{B}\}\in {E}_{t}$ between ${v}_{A}\in {V}_{A}$ and ${w}_{B}\in {V}_{B}$ for which ${\mathcal{L}}_{A}\left({v}_{A}\right)={\mathcal{L}}_{B}\left({w}_{B}\right)=l$, given that e does not already exist. Otherwise, e’s weight is adapted as before.
- −
- Intrapersonal links: if at time t, agent $X\in \{A,B\}$ uses item l to express $T=\mathit{T}\left(t\right)$, we generate intrapersonal links between ${v}_{X}\in {V}_{X}$, ${\mathcal{L}}_{X}\left({v}_{X}\right)=l$, and all other vertices labeled by items that X has used in the same round or in the preceding round on the same topic T. If the links already exist, their weights are augmented by 1.

#### 4.2. Mutual Information of Two-Layer Networks

**Definition 1.**Mutual information of primes and of two-layer graphs. Let $G=(V,E,\mathcal{L})$ be a labeled two-layer graph with the projections ${\pi}_{A}\left(G\right)={G}_{1}=({V}_{1},{E}_{1},{\mathcal{L}}_{1})$ and ${\pi}_{B}\left(G\right)={G}_{2}=({V}_{2},{E}_{2},{\mathcal{L}}_{2})$. Let further

- ${S}_{0,0}(u,x)={S}_{0=0,0=0}(u,x)=\left\{{l}_{0}\right\}$; by definition, paired primes are both located in the zero sphere.
- ${S}_{1,1}(u,x)={S}_{1>0,1>0}(u,x)=\left\{{l}_{1}\right\}$; starting from u and x, respectively, ${l}_{1}$ is directly associated to ${l}_{0}$ in both interlocutor lexica.
- ${S}_{2,2}(u,x)={S}_{2>0,2>0}(u,x)=\left\{{l}_{2}\right\}$; ${l}_{2}$ exemplifies a word that is mediately associated to their respective primes in both interlocutor lexica the same way.
- ${S}_{1,0}(u,x)={S}_{1>0,0=0}(u,x)=\{{l}_{7},{l}_{9},{l}_{10}\}$ is the subset of words in ${S}_{1}^{{G}_{1}}\left(u\right)$ used by speaker A, but not by speaker B.
- ${S}_{2,0}(u,x)={S}_{2>0,0=0}(u,x)=\{{l}_{3},{l}_{6},{l}_{8},{l}_{11}\}$ is the subset of words in ${S}_{2}^{{G}_{1}}\left(u\right)$ used by speaker A, but not by speaker B.
- ${S}_{1,0}(x,u)={S}_{0,1}(u,x)={S}_{0=0,1>0}(u,x)=\{{l}_{14},{l}_{15},{l}_{17},{l}_{18}\}$ is the subset of words in ${S}_{1}^{{G}_{2}}\left(x\right)$ used by speaker B, but not by speaker A.
- ${S}_{2,0}(x,u)={S}_{0,2}(u,x)={S}_{0=0,2>0}(u,x)=\{{l}_{4},{l}_{12},{l}_{13},{l}_{16},{l}_{19}\}$ is the subset of words in ${S}_{2}^{{G}_{2}}\left(x\right)$ used by speaker B, but not by speaker A.
- ${S}_{1,2}(u,x)={S}_{2,1}(x,u)={S}_{1>0,2>0}(u,x)=\left\{{l}_{5}\right\}$; ${l}_{5}$ exemplifies a word that is used by both interlocutors but in a different way from the point of view of the paired primes u and x.
- $\forall 2<i\le 18\phantom{\rule{4pt}{0ex}}\forall 0\le j\le 18\phantom{\rule{-0.166667em}{0ex}}:{S}_{i,j}(u,x)=\varnothing $; note that 18 = | | − 1 where $=\{q,r,s,t,u,v,w,x,$$y,z,{v}_{6},{v}_{7},{v}_{8},{v}_{9},{v}_{10},{v}_{11},{v}_{12},{v}_{13},{v}_{14},{v}_{15},{v}_{16},{v}_{17},{v}_{18},{v}_{19}\}$.
- $\forall 0\le i\le 18\phantom{\rule{4pt}{0ex}}\forall 2<j\le 18\phantom{\rule{-0.166667em}{0ex}}:{S}_{i,j}(u,x)=\varnothing $.

**Theorem 1.**Equation 8 is a measure of mutual information (i.e., Equation 8 defines a measure for estimating MI by means of the $(i,j)$-sphere ${S}_{i,j}(v,w)$).

- On the one hand, both interlocutors are aligned in the sense that they activate a common sub-lexicon during their conversation out of which they recruit paired primes by expressing, for example, the same topic of conversation by the same word. Without such a larger sub-lexicon, the value of $D({G}_{1},{G}_{2})$ could be hardly near to zero as word usages of one interlocutor would tell us nothing about the word usages of the other. Thus, for higher values of $D({G}_{1},{G}_{2})$ it is necessary that the sub-lexicon shared among the interlocutors covers a bigger part of the overall dialog lexicon—note that this notion relates to the notion of graph distance of [59] as explained below. As seen by example of the JMG, there are hardly many paired primes in our study that are common to all pairs of interlocutors. Thus, in order to secure comparability among different pairs of interlocutors, it makes sense to focus on a single lexeme that is actually used by all pairs of interlocutors. In our present study this is exemplified by the lexeme button (see Figure 2 in Section 3).
- On the other hand, both interlocutors are additionally aligned in the sense that the focal primes v and w induce similar patterns of spreading activation [60] within their respective dialog lexica. That is, by commonly using the lexeme that equally labels the vertices v and w, their neighborhoods are activated in a way that is mutually predictable. In terms of geodesic distances, this means that both interlocutors have built dialog lexica that are similarly organized from the point of view of the paired primes v and w.

- First and foremost, we calculate classical indices of complex network theory separately for each of the layers of two-layer networks and aggregate them by a mean value to describe, for example, the average cluster value of interlocutor lexica in dyadic conversations. This is done for the cluster value C
_{1}[61], its weighted correspondents $\langle {C}_{w}\left(k\right)\rangle $ and $\langle {C}_{w}^{\mathit{ns}}\left(k\right)\rangle $ [62], the normalized average geodesic distance $\widehat{L}$ and the closeness centrality of graphs [30,63]. For an undirected graph $G=(V,E)$, $\widehat{L}$ is defined as follows:$$\begin{array}{c}\hfill \widehat{L}\left(G\right)=\left\{\begin{array}{cc}0& :\left|V\right|\le 1\hfill \\ \frac{2}{n(n-1)}\sum _{i=1}^{n}\sum _{j=i+1}^{n}{\delta}^{+}({v}_{i},{v}_{j})& :\left|V\right|1\hfill \end{array}\right.\in [0,1]\end{array}$$$$\begin{array}{c}\hfill {\delta}^{+}({v}_{i},{v}_{j})=\left\{\begin{array}{cc}\frac{\delta ({v}_{i},{v}_{j})-1}{n-1}& :{v}_{i}\ne {v}_{j}\phantom{\rule{4pt}{0ex}}\mathit{are}\phantom{\rule{4pt}{0ex}}\mathit{connected}\hfill \\ 1& :{v}_{i}\ne {v}_{j}\phantom{\rule{4pt}{0ex}}\mathit{are}\phantom{\rule{4pt}{0ex}}\mathit{unconnected}\hfill \end{array}\right.\end{array}$$ - The latter group of indices simply adapts classical notions of complex network theory to the area of two-layer networks. That is, they do not measure the dissimilarity of graphs as done, for example, by $D({G}_{1};{G}_{2})$. As an alternative to D, we utilize two graph distance measures [64,65,66]. More specifically, [59] consider the following quantity to measure the distance of two (labeled) graphs ${G}_{1},{G}_{2}$:$$\begin{array}{c}\hfill {d}_{B}({G}_{1},{G}_{2})=1-\frac{\left|\mathit{mcs}\right({G}_{1},{G}_{2}\left)\right|}{max\left(\right|{G}_{1}|,|{G}_{2}\left|\right)}\end{array}$$$$\begin{array}{c}\hfill {d}_{W}({G}_{1},{G}_{2})=1-\frac{\left|\mathit{mcs}\right({G}_{1},{G}_{2}\left)\right|}{|{G}_{1}|+|{G}_{2}|-|\mathit{mcs}({G}_{1},{G}_{2})|}\end{array}$$
- Last but not least, we consider an index of modularity that, for a given network, measures the independence of its candidate modules. As we consider networks with two modules, we use the following variant of the index of [68]:$$\begin{array}{c}\hfill Q\left(G\right)=\sum _{i=1}^{2}({e}_{ii}-{a}_{i}^{2})\end{array}$$

## 5. Random Models of Two-Layer Networks

Algorithm 1: Binomial case I: computing a set of randomized TiTAN series. |

Data: TiTAN series $\mathcal{X}=({X}_{1},\dots ,{X}_{k})$; number of iterations nResult: Set $\mathbb{Y}$ of n randomized TiTAN series ${\mathcal{Y}}_{i}=({Y}_{{i}_{1}},\dots ,{Y}_{{i}_{k}}),i=1..n$for $i=1..n$ dorewire ${X}_{k}=(V,{E}_{k})$ at random to get ${Y}_{{i}_{k}}=(V,{\widehat{E}}_{{i}_{k}})$; for $j=(k-1)..1$ dorandomly delete $|{\widehat{E}}_{{i}_{j+1}}|-|{E}_{j}|$ edges from ${\widehat{E}}_{{i}_{j+1}}$ to get ${\widehat{E}}_{{i}_{j}}$; ${Y}_{{i}_{j}}\leftarrow (V,{\widehat{E}}_{{i}_{j}})$; end${Y}_{{i}_{1}}\leftarrow (V,{E}_{1})$; $Y\leftarrow Y\cup \left\{({Y}_{{i}_{1}},\dots ,{Y}_{{i}_{k}})\right\}$; end |

Algorithm 2: Binomial case II: computing a set of layer-sensitive randomized TiTAN series. |

Data: TiTAN series $\mathcal{X}=({X}_{1},\dots ,{X}_{k})$; number of iterations nResult: Set $\mathbb{Y}$ of n randomized TiTAN series ${\mathcal{Y}}_{i}=({Y}_{{i}_{1}},\dots ,{Y}_{{i}_{k}}),i=1..n$for $i=1..n$ dorewire ${\pi}_{A}\left({X}_{k}\right)$, ${\pi}_{B}\left({X}_{k}\right)$ and ${\pi}_{AB}\left({X}_{k}\right)$ at random to build ${Y}_{{i}_{k}}=(V,{\widehat{E}}_{{i}_{k}})$ where ${X}_{k}=(V,{E}_{k})$ such that ${\pi}_{A}\left({Y}_{{i}_{k}}\right)\leftarrow \mathit{rand}\left({\pi}_{A}\left({X}_{k}\right)\right)$, ${\pi}_{B}\left({Y}_{{i}_{k}}\right)\leftarrow \mathit{rand}\left({\pi}_{B}\left({X}_{k}\right)\right)$ and ${\pi}_{AB}\left({Y}_{{i}_{k}}\right)\leftarrow \mathit{rand}\left({\pi}_{AB}\left({X}_{k}\right)\right)$; for $j=(k-1)..1$ dorandomly delete $|E\left({\pi}_{A}\left({Y}_{{i}_{j+1}}\right)\right)|-|E\left({\pi}_{A}\left({X}_{j}\right)\right)|$ edges from $E\left({\pi}_{A}\left({Y}_{{i}_{j+1}}\right)\right)$ to get ${A}_{{i}_{j}}$; randomly delete $|E\left({\pi}_{B}\left({Y}_{{i}_{j+1}}\right)\right)|-|E\left({\pi}_{B}\left({X}_{j}\right)\right)|$ edges from $E\left({\pi}_{B}\left({Y}_{{i}_{j+1}}\right)\right)$ to get ${B}_{{i}_{j}}$; randomly delete $|E\left({\pi}_{AB}\left({Y}_{{i}_{j+1}}\right)\right)|-|E\left({\pi}_{AB}\left({X}_{j}\right)\right)|$ edges from $E\left({\pi}_{AB}\left({Y}_{{i}_{j+1}}\right)\right)$ to get ${C}_{{i}_{j}}$; set ${Y}_{{i}_{j}}=(V,{\widehat{E}}_{{i}_{j}})$ such that ${\widehat{E}}_{{i}_{j}}={A}_{{i}_{j}}\cup {B}_{{i}_{j}}\cup {C}_{{i}_{j}}$ and ${\pi}_{A}\left({Y}_{{i}_{j}}\right)=(V\left({\pi}_{A}\left({X}_{k}\right)\right),{A}_{{i}_{j}})$, ${\pi}_{B}\left({Y}_{{i}_{j}}\right)=(V\left({\pi}_{B}\left({X}_{k}\right)\right),{B}_{{i}_{j}})$ and ${\pi}_{AB}\left({Y}_{{i}_{j}}\right)=(V\left({\pi}_{AB}\left({X}_{k}\right)\right),{C}_{{i}_{j}})$; end${Y}_{{i}_{1}}\leftarrow (V,{E}_{1})$; $\mathbb{Y}\leftarrow \mathbb{Y}\cup \left\{({Y}_{{i}_{1}},\dots ,{Y}_{{i}_{k}})\right\}$; end |

#### 5.1. The Binomial Model (BM-I)

#### 5.2. The Partition-Sensitive Binomial Model (BM-II)

#### 5.3. The Partition- and Edge-Identity-Sensitive Binomial Model (BM-III)

#### 5.4. The Event-Related Shuffle Model (SM-I)

#### 5.5. The Time Point-Related Shuffle Model (SM-II)

#### 5.6. Summary Attributes of Randomized Two-Layer Networks

## 6. Experimentation

#### 6.1. On the Temporal Dynamics of Lexical Alignment

#### Lexical Clustering

_{1}[61] as a function of time in Dialog 19${}^{*}$. In this example, clustering is very high (up to 80%) while in the corresponding randomized networks it is much lower: while the BM-II and BM-III converge in a cluster value of C

_{1}$\approx 35\%$, the BM-I rests below 20% of clustered items—both results coincide with general findings about random networks [61] that are known for small cluster values. This observation is at least not falsified by the SM-I and SM-II of Dialog 19${}^{*}$: both models produce rather stable cluster values that rapidly converge below the corresponding values of the BM-II and BM-III. Thus, by example of Dialog 19${}^{*}$, we get a first hint that clustering in dialog lexica is a likely phenomenon much beyond what can be expected by chance. In principle, the same relation holds in the case of the weighted variant of C

_{1}, that is, $\langle {C}_{w}\left(k\right)\rangle $ [62]. This is exemplified by Figure 9, which depicts a large gap between weighted clustering in natural in relation to randomized dialog lexica. Moreover, if we switch from the perspective of a dialog lexicon as a whole and compute separately for each of its sublexica before aggregating it by a mean value (see Section 4.2), we still see that this relation is retained as shown in Figure 10.

#### Lexical Closeness

- A low value of $\widehat{L}$ for a graph G indicates short paths between any pair of vertices in G that tends to be connected without any two disconnected components. In terms of dialog lexica this is tantamount to a high probability that any prime may activate any other item in the lexicon even though to a low degree. In other words, when receiving a word form /a/, the dialog lexicon of an interlocutor allows, in principle, for activating the complete sublexicon that he has generated till the moment of processing /a/. Both of these assessments are confirmed by our corpus of 11 experimental dialogs. The upper left part of Figure 12 reports the temporal dynamics of $\widehat{L}$ in Dialog 19${}^{*}$, while the upper right part of Figure 12 depicts the values of $\widehat{L}$ after being averaged over both interlocutor lexica. In both cases, we observe that short average geodesic distances emerge much more rapidly in the BM-I, the BM-II, and the BM-III. We also observe that all variants (whether natural or randomized) result in very low values of $\widehat{L}$ indicating the existence of the latter mechanism. However, we also observe that the SM-II (based on shuffling the time points of dialog events) perfectly approximates geodesic distances in natural dialogs. In any event, as dialog lexica finally produce short average geodesic distances, even though more slower than their binomial counterparts, they perfectly fit into the class of networks that have been called small-worlds ([61,73]) as they also have high cluster values. Note that the upper left and right part of Figure 12 hints at a constant drop of $\widehat{L}$ as the dialog evolves. In other words, nouns as considered here are not distinguished in their contribution to the decrease of $\widehat{L}$ as a function of the time of their utterance. This may explain why the SM-II approximates the dynamics of $\widehat{L}$ in natural dialogs.
- A high closeness centrality of a graph G indicates that its vertices coincide more or less regarding the sum of their geodesic distances to all other vertices of G [30]. In conjunction with a value of $\widehat{L}$ near to zero this means that all vertices operate on an equal footing as efficient entry points to the lexicon by which any other item is accessible with almost the same effort of spreading activation. This picture is only confirmed, if we average the standardized closeness centrality over both interlocutor lexica (depicted by the lower right part of Figure 12). Obviously, Dialog 19${}^{*}$ has a much higher closeness centrality than its random counterparts.

#### Lexical Modularity

#### Lexical Similarity

#### 6.2. A Classification Model of Alignment

- In spite of the fact that annotating dialogical data is very effortful so that even 24 dialogs can be seen to be a data set of medium size, any classification result based on such a small set is hardly expressive.
- On the other hand, alignment is a process variable that does not (necessarily) emerge suddenly at the endpoint of a conversation. Rather, alignment gradually evolves so that it is present at different stages of a conversation by different, possibly non-monotonic degrees. According to this view, we need to evaluate a larger interval of a conversation when measuring its alignment—ideally beginning with its end in a regressive manner.

- Straight lines parallel to the x-axis denote the F-scores of two baselines: (i) the known-partition-scenario (with an F-score of $.70591$) that has knowledge about the cardinality of each target class, and (ii) the equi-partition-scenario (with an F-score of $.62539$) that assumes equally sized target classes. Computing these baselines is done 1,000 times so that Figure 16 shows their average F-scores. Roughly speaking, an F-score of $.70591$ means that about 70% of dialogs are correctly classified by randomly assigning them onto both target classes subject to knowing their cardinality—a remarkably high value.
- On the opposite range of the curves in Figure 16 we get the values of the best performing classification (denoted by bullets). In each of the 50 classifications, reported by Figure 16, this variant integrates a genetic search for the best performing subset of 50 topological features. This includes indices of complex network theory (as, for example, the cluster coefficient and the average geodesic distance [73]), graph centrality measures [63], entropy measures ([52,75]) as well as the graph similarity measures described in Section 4.2. According to Figure 16, this variant produces a maximal F-score of $F=1$ in the case of three different classifications—this holds especially for the endpoint of the time series under consideration (with the x-coordinate 1). On average, this variant reaches an F-score of $F=0.9498$. That is, nearly $95\%$ percent of the dialogs are correctly classified by this nearly optimal approach. Thus, using a wide range of topological indices together with an optimization algorithm seems to perform very well, but to the price of an optimization that runs the risk of overfitting.
- In order to shed light on this risk, we compute two additional alternatives. Firstly, Figure 16 shows the values of a genetic search (denoted by circles) on the subset of those 12 features that result in an F-score of $F=1$ for the dialogs’ endpoints. We observe a loss in F-score down to an average score of $.8704$, which, at first glance, does not seem to be dramatic. But things are different if we apply the same subset of features in each of the fifty classifications without any additional optimization. In this case, the average F-score drops down to $.6626$ (i.e., below the upper baseline). Obviously, the optimized classifier performs very unreliably. That is, although we can classify two-layer networks according to (non-)alignment, the classifiers considered so far should be treated very carefully when processing heretofore unseen data.
- Figure 16 also shows that the latter assessment is preliminary. It presents the F-scores of a variant (denoted by squares) that has been produced by two features only, that is, $S({G}_{1},{G}_{2}){|}_{(v,w)}$ (using the lexeme button as the single paired prime) and ${d}_{W}({G}_{1},{G}_{2})$. On the one hand, we see that this variant produces an average F-score of only $.7322$ and, thus, outperforms the upper-bound of both baselines on a rather low level. Further, it also generates two outliers that fall below this upper-bound. However, we also observe that this variant produces very stable results by means of only two indices that according to Section 16, directly relate to alignment measurement. Moreover, the outlier on the right side of the corresponding curve may be due to a loss in the prominence of alignment at this earlier stage of conversation.

## 7. General Discussion

## 8. Conclusions

## Acknowledgement

## References

- Pickering, M.J.; Garrod, S. Toward a mechanistic psychology of dialogue. Behav. Brain. Sci.
**2004**, 27, 169–226. [Google Scholar] [CrossRef] [PubMed] - Clark, H.H. Using Language; Cambridge University Press: Cambridge, UK, 1996. [Google Scholar]
- Maturana, H.R.; Varela, F.J. Autopoiesis and Cognition. The Realization of the Living; Reidel: Dordrecht, The Netherlands, 1980. [Google Scholar]
- Levelt, W.J.M. Speaking. From Intention to Articulation; MIT Press: Cambridge, MA, USA, 1989. [Google Scholar]
- Giles, H.; Powesland, P.F. Speech Styles and Social Evaluation; Academic Press: London, UK / New York, NY, USA, 1975. [Google Scholar]
- Clark, H.H.; Wilkes-Gibbs, D. Referring as a collaborative process. Cognition
**1986**, 22, 1–39. [Google Scholar] [CrossRef] - Branigan, H.P.; Pickering, M.J.; Cleland, A.A. Syntactic coordination in dialogue. Cognition
**2000**, 25, B13–B25. [Google Scholar] [CrossRef] - Garrod, S.; Anderson, A. Saying what you mean in dialogue: a study in conceptual and semantic co-ordination. Cognition
**1987**, 27, 181–218. [Google Scholar] [CrossRef] - Watson, M.E.; Pickering, M.J.; Branigan, H.P. An empirical investigation into spatial reference frame taxonomy using dialogue. In Proceedings of the 26th Annual Conference of the Cognitive Science Society, Chicago, IL, USA, 2004; pp. 2353–2358.
- Kamp, H.; Reyle, U. From Discourse to Logic. Introduction to Modelltheoretic Semantics of Natural Language, Formal Logic and Discourse Representation Theory; Kluwer: Dordrecht, The Netherlands, 1993. [Google Scholar]
- Lewis, D. Conventions. A Philosophical Study; Harvard University Press: Cambridge, MA, USA, 1969. [Google Scholar]
- Stalnaker, R. Common ground. Linguist. Phil.
**2002**, 25, 701–721. [Google Scholar] [CrossRef] - Collins, A.M.; Loftus, E.F. A spreading-activation theory of semantic processing. Psychol. Rev.
**1975**, 82, 407–428. [Google Scholar] [CrossRef] - Kopp, S.; Rieser, H.; Wachsmuth, I.; Bergmann, K.; Lücking, A. Speech-Gesture alignment. In Project Panel at the 3rd International Conference of the International Society for Gesture Studies, Evanston, IL, USA, June, 2007.
- Lücking, A.; Mehler, A.; Menke, P. Taking fingerprints of speech-and-gesture ensembles: approaching empirical evidence of intrapersonal alignment in multimodal communication. In Proceedings of the 12th Workshop on the Semantics and Pragmatics of Dialogue, King’s College, London, UK, June 2008; pp. 157–164.
- Church, K.W. Empirical estimates of adaptation: the chance of two noriegas is closer to p/2 than p
^{2}. In Proceedings of Coling 2000, Saarbrücken, Germany, July-August 2000; pp. 180–186. - Reitter, D.; Keller, F.; Moore, J.D. Computational modelling of structural priming in dialogue. In Proceedings of the Human Language Technology Conference of the North American Chapter of the Association for Computational Linguistics, New York, NY, USA, June 2006; pp. 121–124.
- Wheatley, B.; Doddington, G.; Hemphill, C.; Godfrey, J.; Holliman, E.; McDaniel, J.; Fisher, D. Robust automatic time alignment of orthographic transcriptions with unconstrained speech. In Proceedings of IEEE International Conference on Acoustics, Speechand Signal Processing (ICASSP-92), San Francisco, CA, USA, March 1992; pp. 533–536.
- Anderson, A.H.; Bader, M.; Gurman Bard, E.; Boyle, E.; Doherty, G.; Garrod, S. The HCRC map task corpus. Lang. Speech
**1991**, 34, 351–366. [Google Scholar] - Reitter, D.; Moore, J.K. Predicting success in dialogue. In Proceedings of the 45th Annual Meeting of the Association of Computational Linguistics (ACL), Praque, Czech Republic, June 2007; pp. 808–815.
- Branigan, H.P.; Pickering, M.J.; McLean, J.F.; Cleland, A.A. Syntactic alignment and participant role in dialogue. Cognition
**2007**, 104, 163–197. [Google Scholar] [CrossRef] [PubMed] - Garrod, S.; Fay, N.; Lee, J.; Oberlander, J.; MacLeod, T. Foundations of representations: where might graphical symbol systems come from? Cogn. Sci.
**2007**, 31, 961–987. [Google Scholar] [CrossRef] [PubMed] - Fay, N.; Garrod, S.; Roberts, L. The fitness and functionality of culturally evolved communication systems. Philos. Trans. R. Soc. Lond. B Biol. Sci.
**2008**, 363, 3553–3561. [Google Scholar] [CrossRef] [PubMed] - Schober, M.F.; Brennan, S.E. Processes of interactive spoken discourse: the role of the partner. In Handbook of Discourse Processes; Erlbaum: Mahwah, NJ, USA, 2003. [Google Scholar]
- Krauss, R.M.; Weinheimer, S. Concurrent feedback, confirmation, and the encoding of referents in verbal communication. J. Pers. Soc. Psychol.
**1966**, 4, 343–346. [Google Scholar] [CrossRef] [PubMed] - Weiß, P.; Pfeiffer, T.; Schaffranietz, G.; Rickheit, G. Coordination in dialog: alignment of object naming in the Jigsaw Map Game. In Proceedings of the 8th Annual Conference of the Cognitive Science Society of Germany, Saarbrücken, Germany, August 2007; pp. 4–20.
- Gleim, R.; Mehler, A.; Eikmeyer, H.J. Representing and maintaining large corpora. In Proceedings of the Corpus Linguistics 2007 Conference, Birmingham, UK, July 2007.
- Ferrer i Cancho, R.; Solé, R.V. The small-world of human language. Proc. R. Soc. Lond. B Biol. Sci.
**2001**, 268, 2261–2265. [Google Scholar] [CrossRef] [PubMed] - Ferrer i Cancho, R.; Solé, R.V.; Köhler, R. Patterns in syntactic dependency-networks. Phys. Rev. E
**2004**, 69, 051915. [Google Scholar] [CrossRef] [PubMed] - Mehler, A. Structural similarities of complex networks: a computational model by example of Wiki graphs. Appl. Artif. Intell.
**2008**, 22, 619–683. [Google Scholar] [CrossRef] - Mehler, A. On the impact of community structure on self-organizing lexical networks. In Proceedings of the 7th Evolution of Language Conference (Evolang7), Barcelona, Spain, March 2008; pp. 227–234.
- Motter, A.E.; de Moura, A.P.S.; Lai, Y.C.; Dasgupta, P. Topology of the conceptual network of language. Phys. Rev. E
**2002**, 65. [Google Scholar] [CrossRef] [PubMed] - Mukherjee, A.; Choudhury, M.; Basu, A.; Ganguly, N. Self-organization of the sound inventories: analysis and synthesis of the occurrence and co-occurrence networks of consonants. J. Quant. Linguist.
**2009**, 16, 157–184. [Google Scholar] [CrossRef] - Steyvers, M.; Tenenbaum, J. The large-scale structure of semantic networks: Statistical analyses and a model of semantic growth. Cogn. Sci.
**2005**, 29, 41–78. [Google Scholar] [CrossRef] [PubMed] - Church, K.W.; Hanks, P. Word association norms, mutual information, and lexicography. Comput. Linguist.
**1990**, 16, 22–29. [Google Scholar] - Miller, G.A.; Charles, W.G. Contextual correlates of semantic similarity. Lang. Cogn. Process
**1991**, 6, 1–28. [Google Scholar] [CrossRef] - Mehler, A. Large text networks as an object of corpus linguistic studies. In Corpus Linguistics. An International Handbook of the Science of Language and Society; De Gruyter: Berlin, Germany/New York, NY, USA, 2008; pp. 328–382. [Google Scholar]
- Diestel, R. Graph Theory; Springer: Heidelberg, Germany, 2005. [Google Scholar]
- Branigan, H.P.; Pickering, M.J.; Cleland, A.A. Syntactic priming in written production: evidence for rapid decay. Psychon. Bull. Rev.
**1999**, 6, 635–640. [Google Scholar] [CrossRef] [PubMed] - Bock, K.; Griffin, Z.M. The persistence of structural priming: transient activation or implicit learning? J. Exp. Psychol.
**2000**, 129, 177–192. [Google Scholar] [CrossRef] - Ferreira, V.S.; Bock, K. The functions of structural priming. Lang. Cogn. Process
**2006**, 21, 1011–1029. [Google Scholar] [CrossRef] [PubMed] - Mehler, A.; Weiß, P.; Menke, P.; Lücking, A. Towards a simulation model of dialogical alignment. In Proceedings of the 8th International Conference on the Evolution of Language (Evolang8), Utrecht, The Netherlands, April 2010.
- Halliday, M.A.K. Lexis as a linguistic level. In In Memory of J. R. Firth; Longman: London, UK, 1966. [Google Scholar]
- Rieger, B.B. Semiotic cognitive information processing: learning to understand discourse. A systemic model of meaning constitution. In Adaptivity and Learning. An Interdisciplinary Debate; Springer: Berlin, Germany, 2003; pp. 347–403. [Google Scholar]
- Barrat, A.; Barthélemy, M.; Vespignani, A. Dynamical Processes on Complex Networks; Cambridge University Press: Cambridge, UK, 2008. [Google Scholar]
- Caldarelli, G. Scale-Free Networks. Complex Webs in Nature and Technology; Oxford Uiversity Press: Oxford, UK, 2008. [Google Scholar]
- Caldarelli, G.; Vespignani, A. Large Scale Structure and Dynamics of Complex Networks; World Scientific: Hackensack, NJ, USA, 2007. [Google Scholar]
- Pastor-Satorras, R.; Vespignani, A. Evolution and Structure of the Internet; Cambridge University Press: Cambridge, UK, 2004. [Google Scholar]
- Bunke, H. What is the distance between graphs? Bull. EATCS
**1983**, 20, 35–39. [Google Scholar] - Garey, M.R.; Johnson, D.S. Computers and Intractability: A Guide to the Theory of NP-Completeness; W.H. Freeman: New York, NY, USA, 1979. [Google Scholar]
- Wilhelm, T.; Hollunder, J. Information theoretic description of networks. Phys. A
**2007**, 385. [Google Scholar] [CrossRef] - Dehmer, M. Information processing in complex networks: graph entropy and information functionals. Appl. Math. Comput.
**2008**, 201, 82–94. [Google Scholar] [CrossRef] - Bonchev, D. Information Theoretic Indices for Characterization of Chemical Structures; Research Studies Press: Chichester, UK, 1983. [Google Scholar]
- Kraskov, A.; Grassberger, P. MIC: mutual information based hierarchical clustering. In Information Theory and Statistical Learning; Springer: New York, NY, USA, 2008; pp. 101–123. [Google Scholar]
- Dehmer, M.; Varmuza, K.; Borgert, S.; Emmert-Streib, F. On entropy-based molecular descriptors: statistical analysis of real and synthetic chemical structures. J. Chem. Inf. Model.
**2009**, 49, 1655–1663. [Google Scholar] [CrossRef] [PubMed] - Dehmer, M.; Barbarini, N.; Varmuza, K.; Graber, A. A large scale analysis of information-theoretic network complexity measures using chemical structures. PLoS One
**2009**, 4, e8057. [Google Scholar] [CrossRef] [PubMed] - Tuldava, J. Methods in Quantitative Linguistics; Wissenschaftlicher Verlag: Trier, Germany, 1995. [Google Scholar]
- Cover, T.M.; Thomas, J.A. Elements of Information Theory; Wiley-Interscience: Hoboken, NY, USA, 2006. [Google Scholar]
- Bunke, H.; Shearer, K. A graph distance metric based on the maximal common subgraph. Pattern Recogn. Lett.
**1998**, 19, 255–259. [Google Scholar] [CrossRef] - Gärdenfors, P. Conceptual Spaces; MIT Press: Cambridge, MA, USA, 2000. [Google Scholar]
- Watts, D.J.; Strogatz, S.H. Collective dynamics of `small-world’ networks. Nature
**1998**, 393, 440–442. [Google Scholar] [CrossRef] [PubMed] - Serrano, M.Á.; Boguñá, M.; Pastor-Satorras, R. Correlations in weighted networks. Phys. Rev. E
**2006**, 74, 055101. [Google Scholar] [CrossRef] [PubMed] - Feldman, R.; Sanger, J. The Text Mining Handbook. Advanced Approaches in Analyzing Unstructured Data; Cambridge University Press: Cambridge, UK, 2007. [Google Scholar]
- Bunke, H.; Günter, S. Weighted mean of a pair of graphs. Computing
**2001**, 67, 209–224. [Google Scholar] [CrossRef] - Bunke, H.; Günter, S.; Jiang, X. Towards bridging the gap between statistical and structural pattern recognition: two new concepts in graph matching. In Proceedings of the Second International Conference on Advances in Pattern Recognition; Springer: Berlin, Germany/New York, NY, USA, 2001; pp. 1–11. [Google Scholar]
- Schenker, A.; Bunke, H.; Last, M.; Kandel, A. Graph-Theoretic Techniques for Web Content Mining; World Scientific: Hackensack, NJ, USA/London, UK, 2005. [Google Scholar]
- Wallis, W.D.; Shoubridge, P.; Kraetz, M.; Ray, D. Graph distances using graph union. Pattern Recogn. Lett.
**2001**, 22, 701–704. [Google Scholar] [CrossRef] - Newman, M.E.J.; Girvan, M. Finding and evaluating community structure in networks. Phys. Rev. E
**2004**, 69, 026113. [Google Scholar] [CrossRef] [PubMed] - Erdős, P.; Rényi, A. On random graphs. Publ. Math.
**1959**, 6, 290–297. [Google Scholar] - Cattuto, C.; Barrat, A.; Baldassarri, A.; Schehr, G.; Loreto, V. Collective dynamics of social annotation. PNAS
**2009**, 106, 10511–10515. [Google Scholar] [CrossRef] [PubMed] - Zlatic, V.; Bozicevic, M.; Stefancic, H.; Domazet, M. Wikipedias: Collaborative web-based encyclopedias as complex networks. Phys. Rev. E
**2006**, 74, 016115. [Google Scholar] [CrossRef] [PubMed] - Mehler, A. Text linkage in the Wiki medium — a comparative study. In Proceedings of the EACL Workshop on New Text — Wikis and blogs and other dynamic text sources, Trento, Italy, April 2006; pp. 1–8.
- Newman, M.E.J. The structure and function of complex networks. SIAM Rev.
**2003**, 45, 167–256. [Google Scholar] [CrossRef] - Altmann, G. Semantische diversifikation. Folia Linguist.
**1985**, 19, 177–200. [Google Scholar] [CrossRef] - Mehler, A. A quantitative graph model of social ontologies by example of Wikipedia. In Towards an Information Theory of Complex Networks: Statistical Methods and Applications; Birkhäuser: Boston, MA, USA/Basel, Switzerland, 2009. [Google Scholar]
- Manning, C.D.; Schütze, H. Foundations of Statistical Natural Language Processing; MIT Press: Cambridge, MA, USA, 1999. [Google Scholar]

**Figure 1.**Priming of representations within two networks of mental representations of an interlocutor A and B, respectively.

**Figure 2.**Critical (upper row) and uncritical (lower row) objects and their naming (/in diagonal slashes/) in the JMG.

**Figure 3.**Schematic depiction of an object arrangement in the JMG: agent A (left side) plays two instruction cards as does agent B (right side). Numbers indicate the order of the cards being played. The map in the middle shows the object arrangement after these for cards have been processed.

**Figure 4.**Schematic representation of a Two-layer Time-Aligned Network (TiTAN) series starting with two initially unlinked lexica of interlocutor A and B (upper left). Both interlocutor lexica are networked step by step (upper right) till, finally, a dialog lexicon emerges that is spanned by intra- and interpersonal links across the alignment channel (lower left). The lower right part of the figure highlights the role of turn-taking as the means by which dialog lexica (represented as TiTAN series) gradually evolve.

**Figure 5.**A graph-theoretical model of turning points of lexical alignment (cf. [42]). On the left side, a two-layer dialog lexicon is shown whose layers are completely separated as there are no links crossing the alignment channel. The right side depicts the opposite case where both interlocutor lexica are identical and where each item is linked across the alignment channel with its correspondent in the lexicon of the opposite interlocutor.

**Figure 6.**The final two-layer network of a TiTAN series that represents a gradually evolving dialog lexicon of two interlocutors. Initially, no items are interlinked. From turn to turn, more and more associations are established intra- and interpersonally so that the dialog lexicon is finally structured as depicted by the network. Edge weights are represented by the thickness of lines.

**Figure 7.**Two labeled graphs ${G}_{1}=({V}_{1},{E}_{1},{\mathcal{L}}_{1})$ and ${G}_{2}=({V}_{2},{E}_{2},{\mathcal{L}}_{2})$ as the projections ${\pi}_{A}\left(G\right)={G}_{1}$ and ${\pi}_{B}\left(G\right)={G}_{2}$ of a graph $G=({V}_{1}\cup {V}_{2},E,\mathcal{L})$ such that ${E}_{1}\subset E\supset {E}_{2}$ and $\mathcal{L}={\mathcal{L}}_{1}\cup {\mathcal{L}}_{2}$. ${G}_{1}$ and ${G}_{2}$ share, for example, an equally labeled vertex in the 1-sphere of $v\in {V}_{1}$ and $x\in {V}_{2}$, respectively.

**Figure 8.**The temporal dynamics of the cluster coefficient C

_{1}[61] (y-axis) as a function of time (x-axis) by example of the TiTAN series induced from Dialog 19${}^{*}$ (•) in relation to the corresponding BM-I (+), BM-II (×), BM-III (Δ), SM-I (□), and SM-II (∘) model.

**Figure 9.**The temporal dynamics of the weighted cluster coefficient $\langle {C}_{w}\left(k\right)\rangle $ [62] (y-axis) as a function of time (x-axis) by example of the TiTAN series induced from Dialog 19${}^{*}$ (•) in relation to the corresponding BM-I (+), BM-II (×), BM-III (Δ), SM-I (□), and SM-II (∘) model.

**Figure 10.**The temporal dynamics of the average cluster coefficient C

_{1}[61] (y-axis) as a function of time (x-axis) by example of both layers of the TiTAN series induced from Dialog 19${}^{*}$ (•) in relation to the corresponding BM-I (+), BM-II (×), BM-III (Δ), SM-I (□), and SM-II (∘) model.

**Figure 11.**The temporal dynamics of the cluster coefficient [61] (y-axis) as a function of time (x-axis) by example of 11 dialog lexica (green) in relation to the corresponding BM-I (red), BM-III (cyan), and SM-I (blue) models.

**Figure 12.**The temporal dynamics of the normalized average geodesic distance (upper left), its average over both interlocutor lexica (upper right), the standardized closeness centrality [63] (lower left) and its corresponding average (y-axis) by example of the TiTAN series induced from Dialog 19${}^{*}$ (•) in relation to the corresponding BM-I (+), BM-II (×), BM-III (Δ), SM-I (□), and SM-II (∘) models.

**Figure 13.**The temporal dynamics of the normalized average geodesic distance and the standardized closeness centrality (y-axis) both averaged over the layers of two-layer networks by example of 11 dialog lexica (green) in relation to the corresponding BM-I (red), BM-III (cyan), and SM-I (blue) models.

**Figure 14.**The temporal dynamics of the index of modularity [68] (y-axis) averaged over the layers of two-layer networks by example of 11 dialog lexica (green) in relation to the corresponding BM-I (red), and SM-I (blue) models.

**Figure 15.**The temporal dynamics of $S({G}_{1},{G}_{2}){|}_{(v,w)}$ (left) and ${d}_{W}$ (right) (y-axis) by example of 11 dialog lexica (green) in relation to the corresponding BM-I (red), BM-III (cyan), and SM-I (blue) models.

**Figure 16.**Distribution of F-scores (y-axis) of 50 classifications (x-axis) each of which operates on 24 dialogs as represented by two-layer networks. The classification starts with the endpoints of the 24 dialogs (i.e., from the left on the x-axis) and then regresses event by event. The ordinate plots the corresponding F-scores. Bullets (•) denote the values of the best performing classification as a result of a genetic search for the best performing subset of 50 topological features. Circles (∘) denote the values of a genetic search for the best performing subset of 12 features that give an F-score of $F=1$ for the first classification (i.e., the endpoints of the dialogs). Diamonds are the corresponding F-scores that are produced by using all these 12 features without any additional optimization. Squares (□) denote the F-scores of only two features, $S({G}_{1},{G}_{2}){|}_{(v,w)}$ and ${d}_{W}({G}_{1},{G}_{2})$. Finally, the straight lines denote the F-scores of two baselines.

**Figure 17.**Three scenarios of dialog network formation: (1) schematic depiction of a time point of a TiTAN series. (2) Formation of clusters at a certain time point of a TiTAN series. The vertices labeled by A denote a general noun that is used by both interlocutors in different thematic contexts. (3) Cluster formation with interlocutor lexica as a result of sudden topic changes.

**Table 1.**Overview of the corpus of 24 dialogs based on the JMG played by 24 naïve participants and 13 confederates. Asterisks code experimental dialogs in which only naïve interlocutors participated (in contrast to confederate dialogs). Column 2 codes whether the dialog manifests alignment or not according to a manual annotation as explained in Section 6.2. The corpus belongs to a larger corpus of 32 dialogs (in preparation). Data has been partly annotated using the Ariadne system [27]. $\left|V\right|$ is the lexicon size, $|{L}_{A}|$ and $|{L}_{B}|$ are the sizes of the interlocutors’ sublexica. $\left|E\right|$ is the number of associations (edges) that have been induced by the algorithm of Section 4. #events is the number of nominal word forms in referential function and #turns is the number of turns.

Network | Alignment | $\left|V\right|$ | $\left|E\right|$ | $|{\mathit{L}}_{A}|$ | $|{\mathit{L}}_{B}|$ | #events | #turns |

1 | no | 39 | 197 | 15 | 24 | 123 | 30 |

1${}^{*}$ | yes | 44 | 267 | 25 | 19 | 164 | 36 |

4 | yes | 34 | 145 | 15 | 19 | 100 | 30 |

5 | no | 39 | 232 | 17 | 22 | 118 | 34 |

5${}^{*}$ | yes | 42 | 254 | 23 | 19 | 143 | 31 |

6 | no | 43 | 227 | 21 | 22 | 113 | 30 |

6${}^{*}$ | yes | 46 | 343 | 23 | 23 | 151 | 34 |

7 | yes | 37 | 145 | 17 | 20 | 111 | 36 |

7${}^{*}$ | no | 43 | 227 | 20 | 23 | 165 | 30 |

8 | yes | 39 | 163 | 18 | 21 | 112 | 30 |

8${}^{*}$ | yes | 46 | 237 | 18 | 28 | 161 | 52 |

19 | yes | 40 | 169 | 18 | 22 | 110 | 32 |

19${}^{*}$ | yes | 63 | 367 | 28 | 35 | 201 | 48 |

33 | yes | 27 | 104 | 12 | 15 | 96 | 32 |

33${}^{*}$ | yes | 44 | 225 | 18 | 26 | 156 | 50 |

34 | no | 34 | 137 | 13 | 21 | 117 | 34 |

34${}^{*}$ | yes | 32 | 146 | 17 | 15 | 118 | 28 |

35 | yes | 20 | 65 | 9 | 11 | 100 | 30 |

36 | yes | 37 | 133 | 13 | 24 | 113 | 41 |

36${}^{*}$ | yes | 50 | 213 | 30 | 20 | 134 | 42 |

37 | yes | 24 | 111 | 11 | 13 | 98 | 34 |

37${}^{*}$ | yes | 40 | 216 | 20 | 20 | 133 | 30 |

38 | yes | 30 | 140 | 12 | 18 | 118 | 28 |

38${}^{*}$ | yes | 50 | 244 | 31 | 19 | 157 | 41 |

© 2010 by the authors; licensee MDPI, Basel, Switzerland. This article is an Open Access article distributed under the terms and conditions of the Creative Commons Attribution license http://creativecommons.org/licenses/by/3.0/.

## Share and Cite

**MDPI and ACS Style**

Mehler, A.; Lücking, A.; Weiß, P.
A Network Model of Interpersonal Alignment in Dialog. *Entropy* **2010**, *12*, 1440-1483.
https://doi.org/10.3390/e12061440

**AMA Style**

Mehler A, Lücking A, Weiß P.
A Network Model of Interpersonal Alignment in Dialog. *Entropy*. 2010; 12(6):1440-1483.
https://doi.org/10.3390/e12061440

**Chicago/Turabian Style**

Mehler, Alexander, Andy Lücking, and Petra Weiß.
2010. "A Network Model of Interpersonal Alignment in Dialog" *Entropy* 12, no. 6: 1440-1483.
https://doi.org/10.3390/e12061440