The framework of Hopper and Thompson
Due to the fact that transitivity appears to be a very important factor in the distribution of my data between participles and relative clauses, I am going to provide detailed analysis of it. Transitivity is a syntactic category which indicates whether a clause involves one or more “syntactically privileged arguments”.64 In the case of Ancient Greek these arguments are the subject in the nominative case and the direct object in the accusative case. Transitivity is related, but not identical to valency which refers to all kinds of arguments of a verb.65 In other words, it can also be defined as a property of the entire clause such that an activity is transferred from an agent to a patient.66 In their famous article Paul Hopper and Sandra Thompson proposed a model for analysing transitivity by identifying its components. In each of their criteria a clause can score high or low in transitivity. This model is especially attractive for this research, because it does not treat transitivity as a binary category. Treating transitivity as a scale can better explain why the distribution of participles and relative clauses on the basis of transitivity is not completely clear-cut.
This framework, however, is not completely free from flaws and criticism. The most significant point is that Hopper and Thompson treat all of the components of transitivity as equally important in all languages. This is not true, because words with the same meaning can have different degrees of transitivity in different languages, which means that some one parameter can have a greater importance in one language and be less important in another. Secondly, the categories they discuss refer to different elements of the clause – some concern the semantics of the verb, others the agent, and some others the object.67 However, if we think of transitivity as a property of the clause, not the verb, this should not be a problem. I am now going to discuss each of their categories and explain how I use them specifically for Ancient Greek.
The first, and most basic parameter is « participants ». No transfer of activity can be made if there is just one participant. From a typological perspective, it is an interesting question how to treat reflexive verbs, because the agent and the patient is the same person. However, a question that is more relevant is whether in a language with a case system, a direct object always has to be in the accusative case. Many languages, for example Polish or Hungarian, mark negation additionally with a genitive case on the object, which in an affirmative sentence would be in the accusative. Another problem is that some verbs normally take an object in a case other than the accusative with no semantic change, for example Greek ἀκούω normally takes the genitive and it can mean « hear, hear of or listen to ». One could say that since it does not take a direct object in the accusative, it is thus not transitive, which, considering the semantics of the verb, is rather counterintuitive. If we consider the definition of transitivity as a property of a clause where there is an interaction between two participants, one may be more likely to include verbs like ἀκούω as transitive. However, typologically, these changes of case from an accusative to something else, often a genitive, very often mark reduced transitivity. That is why I am rather reluctant to treat verbs not taking their objects in accusative as really transitive.
The second parameter of transitivity is « kinesis ». This is a defined as a distinction between action and non-action, or a state. Actions are naturally more transitive than states, although verbs denoting states can also take a direct object, for example: “I like you”. There is very little typological discussion about kinesis as a verbal category in the linguistics literature. In some cases the distinction is not very clear-cut. For example in English, in a sentence « I hear his voice » there really is no kinesis between the participants, but it would be difficult to argue that « to hear » is a stative verb. Hopper and Thompson explain that « actions can be transferred from one participants to another; states cannot. » Thus in my analysis in any controversial cases, if an action cannot be transferred, I treat is as scoring low in this category, even if it does not seem to be prototypically « stative ». An example of a participle with a high kinesis from my corpus is:
The third parameter is « aspect ». Hopper and Thompson consider it simply as a matter of telicity: telic verbs are more transitive than atelic verbs. This is not satisfactory in Ancient Greek, because aspect and telicity are quite a complicated matter and play a huge role in the verbal system. The grammatical aspect and the tense are bound together, especially in moods other than the indicative where the tense-based distinction is much less significant. This is highly relevant for our consideration of aspect and transitivity. Telicity, sometimes called « lexical aspect », which Hopper and Thompson equate with aspect, is a lexical property. As I have already explained, in Ancient Greek lexical aspect is overridden by a grammatical category of tense-aspect. In general, tenses built on the present stem have imperfective aspect, these are the imperfect, the present, and the future. By contrast, the tenses built on the aorist and perfect stems: aorist, perfect, pluperfect have perfective aspect. This combination of tense and aspect becomes almost entirely aspectual in participles: present participles do not have a clear time reference. If the participle is coverbal, the tense signifies that the action happens at the same time as the main verb, and if the participle is attributive, the present tense signifies the timeless and thus imperfective value of the participle. The aorist participles always refer to an action which is completed and so are always perfective. To be really perfective, the action must be completed and thus in the past, hence typologically languages are much more sensitive to aspectual differences in the past tense. For this reason, regarding Hopper and Thompson’s category of aspect, I judge it according to the tense-aspect of the verbs in my corpus, not their telicity, for example:
The participle οὐλομένην from ὄλλυμαι “to destroy”, even though it is highly adjectivized in this context, is of the aorist tense. For this reason, it scores “high” in the aspect category. An example of a participle classified as « low » in the aspect category is:
The next part of transitivity in Hopper and Thompson’s theory is « punctuality ». Punctual verbs are intrinsically more transitive than non-punctual verbs, which in English can be illustrated with the contrast between to pick – punctual, and to carry – non-punctual. Although this is again a lexical property, it also interacts with the grammatical aspect. It is simply difficult to imagine talking about a really punctual action in an imperfective aspect. It would then have to assume a meaning of a continuously repeated action. However, in order to stay close to the original framework, in my analysis I treat this category purely as a lexical feature and I disregard its tense. A participle of a punctual verb can be exemplified by:
In this example, we may wonder how to correctly interpret this participle. Should we read it as circumstantial with the meaning “when she was sitting” or attributively as an epithet “who was sitting”? In this case, I chose to treat it as an epithet describing some permanent characteristics of Thetis, since the same formula is repeated again in Iliad 18.36. Thus we can assume that this is not some ad hoc situation where Thetis was sitting in the depths of the sea at that moment, but that is a more or less permanent feature of Thetis as a Homeric character.
The fifth parameter is « volitionality ». When the agent is acting with high volitionality, the effects of the action are usually more apparent on the patient, which increases the transitivity of the clause. One can contrast “I wrote down his phone number” – volitional, and “I forgot his phone number” – non-volitional. The choice in assigning these values is sometimes quite arbitrary. It is not always clear whether an action is fully volitional, for example « to get angry ». It is impossible to determine whether an agent got angry against their active will, or did they actually decided to be angry and stayed angry until they decided to stop. I have decided to treat all verbs of emotion as non-volitional, since people generally do not really control their emotions. Examples of volitional verbs are abundant:68 “But the Achaian men went silently, breathing valor, Stubbornly minded each in his heart to stand by the others.”
The perfect participle μεμαῶτες from μέμαα “to desire” can certainly be described as volitional, especially that it refers here to an active desire to defend one’s companions in battle, rather than to some basic needs, which might not be volitional. There are also unambiguous examples of clearly non-volitional participles:
However we want to interpret the exact meaning of καμόντας, the aorist participle of κάμνω, “those who suffer” or simply “the dead”, it surely refers to a non-volitional action or state.
The next parameter is « affirmation ». Affirmative clauses are more transitive than negative ones. This is typologically confirmed by the fact, which I mention above, that some languages mark negation additionally with a different case and thus possibly removing the direct object.
Results of the transitivity analysis
In this section I am going to present the results of the analyses outlines above. All the percentage indicators refer to the percentage of use of individual examples in my corpus in a given criterion.
Starting again with the category of participants, we already can notice a clear difference between participles and finite verbs in relative clauses: only 33% of participles take a direct object, compared with 73% of finite verbs. Although there are no readily available statistics about transitivity of participles in Ancient Greek, it is not surprising. Participles, especially attributive ones, are syntactically quite close to adjectives. We have already seen that in some cases in the classical language, they might even be treated as nouns. For that reason, they are automatically less likely to behave like verbs and to take a direct object. Obviously, finite verbs do not have any limitations in this matter.
The next category on the list is kinesis. Here only 7% of participles in my corpus are characterised by high kinesis, in contrast to finite verbs of which 43% are high in kinesis. It is quite difficult to say why it should be the case that participles are likely to be low in kinesis. Maybe their adjectival nature is once more surfacing and limiting them, since it would be difficult to image an adjective that has high kinesis built in its meaning.
The aspect, or rather tense-aspect, displays a slightly smaller but noticeable difference, with 27% of participles and 43% of transitive verbs in aorist or perfect tense-aspect. The preference for the present tense is due to the fact that it can have a timeless value, bringing it even closer to the adjective than when used with any different tense.
Proceeding to punctuality, it becomes very clear that there is a pattern here. In all categories of transitivity of Hopper and Thompson the participles score lower than finite verbs. The same happens here, as only 7% of participles have a punctual meaning, compared to 51% of finite verbs. Continuing the theme of attributive participles displaying highly adjectival characteristics, it also makes sense that punctual verbs are rather used in finite verbs in relative clauses, since adjectives are not really punctual in any way.
A very interesting result appears in the category of volitionality. Here, a very clear distribution can be noticed: 40% of participles have a volitional meaning, compared to 91% of finite verbs. This could also be explained in terms of the opposition between adjectival and verbal sides of the participles. Adjectives usually refer to attributes over which the subject has no control, for example colours, nationalities, state of emotions, features of character etc.
As I mentioned before, the categories of affirmation and mode are not relevant to this case study. All participles are in affirmative clauses and only three finite verbs are negated in relative clauses, which does not produce any significant conclusions. Similarly, mode is not something that can be a subject of comparison between participles and finite verbs, since participles do not have moods, thus any form of modality has to be expressed in the form of finite verbs. And indeed, 25% of verbs in relative clauses appear in moods other than indicative.
Table of contents :
1 Text mining for animal health epidemic intelligence: stakes and limits
1 Animal disease surveillance context
1.1 Stakes and limits of traditional surveillance
1.2 Novel sources of health surveillance information
1.3 Epidemic intelligence process
2 Indicator-based and event-based surveillance systems in animal health
2.1 International indicator-based surveillance systems
2.2 Event-based surveillance systems
2.3 Epidemiological information within the EI process
3 Online news sources for event-based surveillance
3.1 Animal health information newsworthiness
3.2 Structural features of online news
3.3 Challenges in describing and comparing EBS pipelines
4 Approaches, stakes and limit of text-mining methods
4.1 Document level
4.2 Entity level
2 Epidemic intelligence and information retrieval
1 Elaboration of a new framework for ne-grained epidemiological annotation
1.1 Motivation and context of framework elaboration
2 Retrieval of ne-grained epidemiological information
2.2 Supervised classication
2.3 Lexicosyntactic pattern-based approach
3 General discussion
3 Event extraction from news articles 79
1 Information extraction in PADI-web pipeline
2 Event extraction: a statistical approach
2.1 Statistical approach
2.2 Corpus and evaluation
3 Event extraction: lexicosyntactical approach
3.1 Lexicosemantic representation approach
3.3 Results and discussion
4 Using epidemiological features to improve information retrieval
1 Proposed approach
1.1 Morpho-syntactic features
1.2 Lexicosemantic features
2 Corpus and evaluation
2.2 Evaluation protocol
3.1 Morphosyntactic features
3.2 Lexicosemantic features
5 Integration of event-based surveillance systems into epidemic intelligence activities – Case studies
1 Dissemination of information in event-based surveillance
1.2 Preliminary results and discussion
2 Early detection of unknown diseases by event-based surveillance systems
2.1 Context of the study
2.2 Material and methods
2.3 Results and discussion
6 Major contributions and perspectives
1 Summary of the main contributions
2 Perspectives of text mining in epidemic intelligence
2.1 Retrieval of ne-grained epidemiological information
2.2 Event detection
2.3 Retrieval of related documents
2.4 Role of the expert
2.5 Detection of weak signals
3 Perspectives of the event-based surveillance systems
3.1 Information dissemination between sources
3.2 Enhancing One Health surveillance through cross-sectorial collaboration
3.3 Event-based surveillance systems in developing countries
3.4 Monitoring of online news through news aggregators