This paper provides an analysis of some argumentation in a biomedical genetics research article as a step towards developing a corpus of articles annotated to support research on argumentation. We present a specification of several argumentation schemes and inter-argument relationships to be annotated.
This paper provides an analysis of some
argumentation in a biomedical genetics research
article
The main claim of
The Knowledge Claim Discourse Model
(KCDM)
In the Introduction section of
The Results section of
This passage can be analyzed as an instance of a type of Practical Reasoning argument whose discourse goal is to justify the authors’ action (sequencing the DFNB93 region by using a custom DNA capture array) in order to achieve the authors’ goal (to identify the disease causing mutation in one affected individual). In addition, as will be discussed in the next section, the excerpt contains a causal argument. 2.3
The focus of our planned annotation efforts is on
argumentation for scientific claims and practice
implications. In this section we present our
analysis of several examples of this type of
argumentation, given mostly in the Results section of
Argument 1. a. Premise: Several genes in the DFNB93 region are expressed in the human inner ear. b. Premise (implicit generally accepted assumption): A mutation of a gene that is expressed in a human tissue or system may lead to an abnormality in that tissue or system. c. Premise: A certain individual (identified as V:14) has arNSHL. d. Conclusion: The mutations occurring in DFNB93 of V:14 are strong candidates for the cause of V:14’s arNSHL.
Argument 1 can be represented more
abstractly, for purposes of annotation of similar
arguments in the corpus, by the following
argumentation scheme. In addition to specifying the
premises and conclusion, we have added a critical
question. Critical questions associated with an
argumentation scheme provide a way to
challenge arguments instantiating the scheme
Premise: There is a causal pathway from G-type events to P-type events.
Premise: An individual has experienced P (a Ptype event).
Conclusion: Some G-type event experienced by that individual may be the cause of P.
Critical Question: What if the set of candidates G does not include the actual cause of the event?
Excerpt 3 contains the argument described in Argument 2.
“After the identified homozygous variants were filtered through the 1000 Genomes Project November 2010 release and dbSNP131, 47 previously unreported variants remained…” (p. 639) Argument 2. a. Premise (same as 1d): The mutations occurring in DFNB93 of V:14 are strong candidates for the cause of V:14’s arNSHL. b. Premise (implicit generally accepted assumption): If a variant is a frequent polymorphism then it is not likely to be the cause of a deleterious condition. c. Premise: All but 47 of the homozygous variants in DFNB93 of V:14 are frequent polymorphisms. d. Conclusion: One of the remaining 47 homozygous variants may be the genetic cause of V:14’s condition.
Excerpt 4 contains the argument described in Argument 3.
Excerpt 4: “… 47 previously unreported variants remained and included two exonic mutations, one splicing mutation, six nontranslated mutations, 16 intergenic (downstream or upstream) mutations, and 22 intronic mutations. The two exonic variants included one nonsynonymous variant … in PFIA1 and synonymous variant … in GAL3ST3 … The splice-site variant, c.637+1G>T … was located at … of CABP2 … The variants in PPFIA1 and CABP2 were subsequently validated by Sanger DNA sequencing, which only confirmed the splicing variant in CABP2. (p. 639). Argument 3. a. Premise (same as 2d): One of the remaining 47 homozygous variants may be the genetic cause of V:14’s condition. b. Premise (implicit generally accepted assumption): Only exonic or splice-site variants confirmed by Sanger DNA sequencing could be the cause of a genetic condition. c. Premise: Of the remaining 47 homozygous variants, only the c.637+1G>T splicing variant in CABP2 was confirmed. d. Conclusion: The c.637+1G>T variant in CABP2 may be the genetic cause of V:14’s condition.
Arguments 2 and 3 can be described as instances of the following argumentation scheme.
Premise: There exists a set of candidates C, one of which may be the cause of event E.
Premise: One or more members of C can be eliminated as candidates.
Conclusion: One of the remaining members of C may be the cause of E.
Excerpt 5 contains two arguments, described in Arguments 4 and 5.
Excerpt 5: “Next, we checked the inheritance of the CABP2 variant in the entire Sh10 family … and screened an additional 100 random Iranian controls to ensure that the variant is not a frequent polymorphism. The mutation was not detected in any of the controls, and inheritance was consistent with hearing loss in the family.” (p. 639).
Argument 4. a. Premise: The c.637+1G>T variant in CABP2 segregates with arNSHL in Sh10 (V:14’s pedigree). b. Premise (implicit generally accepted principle): A variant may be the cause of an autosomal recessive condition if it segregates with the condition in a pedigree, i.e., occurrence of the condition and the variant are consistent with an autosomal recessive inheritance pattern. c. Conclusion (implicit): The c.637+1G>T variant in CABP2 may be the cause of arNSHL in Sh10.
Although Argument 4 is in some respects similar to Mills’ Joint Method of Agreement and Difference (described in Jenicek and Hitchcock, 2004), its premise (4b) provides a causal explanation that is not part of that type of argument. Argument 4 can be described more precisely as an instance of the following argumentation scheme.
Premise: There exists a set of individuals Ipresent that have a feature F and property P. Premise: There exists a set of individuals Iabsent that do not have feature F and property P. Premise: There is a plausible causal link from F to P that could account for the presence of P in Ipresent.
Conclusion: F may be the cause of P in I-present. Critical Question: Is there some other feature G in I-present that could account for P, or is there some other factor G in I-absent that could account for the absence of P?
Argument 5 from Excerpt 4 is as follows.
a. Premise: The c.637+1G>T variant in CABP2 is present in the arNSHL affected members of Sh10. b. Premise: The variant does not occur in a control group. c. Conclusion (implicit): The c.637+1G>T variant in CABP2 may be the cause of arNSHL in Sh10.
Argument 5 can be described as an instance of the following argumentation scheme, based upon Mills’ joint method of agreement and difference. Note that its first critical question is shared with Causal Agreement and Difference, but its second critical question is not needed for that argumentation scheme, one of whose premises is that there is a causal mechanism that may account for the differences between I-present and I-absent.
Premise: A set of individuals I-present have a feature F and property P.
Premise: A set of individuals I-absent (distinct from I-present) do not have F and P.
Conclusion: F may be the cause of P in I-present. Critical questions: • Is there some other feature G in I-present that could account for P, or is there some other factor G in I-absent that could account for the absence of P? • Is there a plausible causal mechanism that explains how F leads to P?
Excerpt 6 contains a causal argument for how the c.637+1G>T variant of CABP2 could lead to hearing loss, as shown in Argument 6.
Excerpt 6: “… we evaluated the effect of the c.637+1G>T mutation on splicing … Analysis … revealed … indicating that the mutation of c.637+1G>T leads to a complete skipping of exon 6 … Skipping of exon 6 is expected to lead to a shifted reading frame and a premature truncation of the protein” (p. 639-0).
a. Premise: The c.637+1G>T mutation of CABP2 may have a deleterious effect on CaBP2 protein during synthesis by CABP2. b. Premise (implicit): CaBP2 protein plays a role in the auditory system. c. Premise (implicit generally accepted principle): Damage to a protein can lead to disease of the tissue or biological system in which that protein plays a role. d. Conclusion (implicit): A c.637+1G>T mutation of CABP2 may result in a disease of the auditory system.
Argument 6 can be described by the following argumentation scheme.
Premise: There is a partially known causal pathway from events of type G to events of type P. Conclusion: The occurrence of a G-type event may result in a P-type event.
Excerpt 7 contains Argument 7, which is similar to Argument 4 and can likewise be described as an instance of Causal Agreement and Difference. Excerpt 7: “We identified two families (Sh11 and He) with affected individuals who were homozygous in this region … Affected family members presented with an audiogram similar to the affected individuals in the Sh10 family… Sanger sequencing … revealed the same c.637+1G>T mutation in these families.” (p. 640)
a. Premise: Affected members of two families, Sh11 and He, have audiograms similar to those of affected family members of Sh10 and the c.637+1G>T variant in CABP2 segregates with hearing loss in those two families. b. Premise (implicit generally accepted principle): A variant may be the cause of an autosomal recessive condition if it segregates with the condition in a pedigree. c. Conclusion (implicit): The c.637+1G>T variant in CABP2 may be the cause of arNSHL in Sh11 and He.
Perhaps because they expect it to be obvious to the intended audience, the authors do not explicitly state Argument 8.
Argument 8. a. Premise (generalizing 4c, 5c, 7c): The c.637+1G>T variant in CABP2 may be the cause of arNSHL in several pedigrees. b. Premise (implicit generally accepted assumption): A homozygous mutation known to have a certain effect in some families will have a similar effect in anyone who inherits it. c. Conclusion (implicit): Anyone having homozygous c.637+1G>T variants of CABP2 may be affected by arNSHL.
Such an argument could be described by the following argumentation scheme.
Premise: P is true of some members S of a class C.
Conclusion: P is true for all members of C. Critical question: What if the individuals in S are exceptional with respect to P? The conclusion of Argument 8 is needed as a premise of Argument 9 for the practice implications of the article given in Excerpt 8 (which, unlike the other excerpts in this paper, comes from the article’s Discussion section).
“In conclusion, we identified mutations in CABP2 in individuals with moderate-to-severe hearing loss. Mutations in CABP2 cause an audiometric phenotype that is seen in most families segregating arNSHL. Our results suggest the importance of screening for mutations in CABP2, as well as in TECTA, in families with this milder audiometric phenotype.” (p. 644) Argument 9. a. Premise (implicit): The reader’s goal is to prevent or mitigate the occurrence of arNSHL. b. Premise (implicit, same as 8c): Someone having homozygous c.637+1G>T variants of CABP2 may be affected by arNSHL. c. Premise (implicit): Screening may determine if someone has homozygous c.637+1G>T variants. d. Premise: (implicit) Knowing if someone has homozygous c.637+1G>T variants is necessary to prevent or mitigate the occurrence of arNSHL. e. Conclusion: It is desirable to screen for c.637+1G>T variants in CABP2.
Argument 9 can be described as a form of Practical Reasoning.
Premise: Agent’s goal is to prevent or mitigate the occurrence of D.
Premise: The occurrence of G may result in D. Premise: Doing Act may result in Agent’s knowing if G.
Premise: Knowing if G is necessary to prevent or mitigate D.
Conclusion: It is desirable for Agent to do Act. 3
The previous section illustrates a chained relationship in Arguments 1-3, i.e., the conclusion of Argument i is a premise of Argument i+1. Arguments 4 and 5 share the same conclusion: The c.637+1G>T variant in CABP2 may be the cause of arNSHL in Sh10. The conclusions of Arguments 4, 5, and 7 (The c.637+1G>T variant in CABP2 may be the cause of arNSHL in Sh11 and He) in combination support the premise of Argument 8, whose conclusion is: Anyone having homozygous c.637+1G>T variants of CABP2 may be affected by arNSHL. The conclusion of Argument 8 is further supported by the conclusion of Argument 6: A c.637+1G>T mutation of CABP2 may result in a disease of the auditory system.
To provide an explanation for why the authors chose to provide various arguments, rather than merely observing their presence in the text, we must consider how the authors expect their arguments to be challenged or evaluated by the intended audience. Note that the chain of Arguments 1-3 could be challenged by posing the instantiated critical question of Argument 1: What if the cause of V:14’s genetic condition was not in the set of candidates that were tested? Rather than directly responding to that critical question, the authors continue with Argument 4 whose claim is that the c.637+1G>T variant is the cause of arNSHL in V:14’s family, Sh10. In other words, Argument 4 makes a broader claim, a claim that subsumes the claim of Argument 3.
Argument 4 can itself be challenged by posing its critical question: Is there some other feature G in I-present that could account for P…? Then one could explain why the authors include Argument 5, in which the Sh10 family is compared to a control group.
Argument 8 can be challenged by posing its critical question: What if the individuals in S are exceptional with respect to P? The biochemical argument 6 that a c.637+1G>T mutation of CABP2 may result in a disease of the auditory system provides a response to that challenge.
Dialogue games have been used to model
argumentation between intelligent agents
The Dialogue Rules of SDG reflect weaker ordering constraints in text than in dialogue and the fact that the reader is imaginary: The permissible replies of the reader to Argue are: implicit Challenge, implicit Pose, or silence. The permissible reply of the author to Challenge or Pose is Argue.
1 4 5
This paper described our analysis of some argumentation schemes and inter-argument relationships in a research article as part of our initial effort to create an annotation scheme. We are continuing to analyze representative articles as preparation for developing and evaluating the annotation scheme. Our longer term goal is to create a freely available corpus of open-access, full-text scientific articles from the biomedical genetics research literature, annotated to support research on argumentation.
We thank Dr. Malcolm Schug of the Biology Department of the University of North Carolina Greensboro for verifying our interpretation of the arguments in the genetics article.