# How to handle verbs on the NLG platform
Nouns (person, place, thing, or concept) and verbs (words that describe an action or state of being) are the hearts and souls of all sentences. Verbs in particular are among the essential building blocks of communication in any language. Apart from short interjections like "Caution!" there is no (complete) sentence without a verb: The verb controls the sentence structure and arranges the word order.
# Verbs are relevant for text generation in two areas:
- As a central element of sentences they determine the other parts of the text. Therefore, to get a grammatically correct text, you need to know how verbs work and what they affect.
- Verbs rarely come from the data when generating text, so you have the a lot of freedom for designing your text around them. You can use them to make texts lively and descriptive, and thus increase the text quality.
# What to expect in this guide
- You get an overview for the most important grammatical aspects of verbs, such as: what are verbs, how are they used and what are the general rules for their inflection.
- You learn how to manage verbs on the NLG platform (including automated annotation and verb containers)
- You will get style advice on building statements around meaningful verbs.
- This guide provides examples in several languages.
# Verbs - how they work in language
Verbs can take different forms depending on their grammatical function: This phenomenon is called verbal inflection. These inflections are language specific and usually depend on the subject of the phrase. Several properties of the subject and the sentence influence the inflection of the verb:
- The number (singular/plural), gender (feminine/masculine/neutral/...) and person (I/you/he, she or it) of the subject.
- Tense (present, past, future etc)
- Mood (indicative, subjunctive)
- Voice (active, passive).
Verbs also affect the sentence structure itself: they are the ones that matter when it comes to the objects in the sentence. How many objects a sentence contains, or whether there is an object at all, depends on the verb, as does the case of each object (accusative, dative, genitive).
To see multilingual examples on the platform start the lesson
# Compound Verbs
While simple verb forms, as shown above, need only one complete verb, compound verbs are formed by combining a verb with one or more auxiliary verbs.
Compound verb forms are found in different settings, for example:
- in compound tenses: I have seen (perfect tense/ EN), Ich hatte es gesehen. (Plusquamperfekt in DE), // Spanisches Beispiel// russisches Beispiel
- as constructions for different mood: I could go to Rome (conditional in EN), Das würde ich nicht tun (Conditional/DE)
- in passive constructions: The ball was hidden by the boy. (EN), Questa e-mail è scritta* in tedesco (I)
These auxiliary verbs help the verb, so to speak, to indicate the respective form. Almost always the auxiliary verb is the one that is inflected. As only those verbs that should inflect must be annotated on the NLG platform, usually annotating the auxliary verb is enough.
The main verb is often an infinitive or a participle. The participle may be adapted to the subject's number or gender, only if that is the case and if the subject is variable, it also has to be annotated.
In English the modal verbs (can, could, may, might, must, shall, should, will, would) will never be inflected so you don't have to annotate them.
# Verbs on the NLG Platform
# Annotation to ensure grammatical correctness
In a sentence, certain parts are dependent on others. If some parts of a sentence change, for example by changing data values, verbs may be required to change also.
NLG software needs to know these dependencies and we use a system of labels to mark (or annotate) that information in the text, it is used as a rule set for generating correct texts. If certain parts of the text change, for example, due to different data, all words referring to them can be adjusted correctly.
# What does automatic annotation do?
While you previously did the annotation work yourself, the automatic annotation feature now has the ability to apply these labels on its own on the statements you’ve written. The results of the grammar analysis by the software are presented as container suggestions, which suggest positions for the containers and their settings.
The automatic annotation analyzes and labels your statement from different points of view:
- It recognizes the variables in the written statement and links them to the according node.
- It recognizes the grammatical structure of the statement (or the parts you’ve marked) and tags the parts of speech.
- It identifies the words that build a unit, detects its grammatical role and annotates the grammatical cases, numbers or tense.
# Creating Verb Containers
An important aspect of annotation is to group together all the words that belong to a label. On the NLG platform the words are grouped on a functional level and then are managed together in a unit, which is called a container.
Two of the most important functions of containers are to store the reference to the variables or other nodes, and to hold information about the structure of a statement. Verb containers, unlike the other two container types, have the sole function of grammatical regulation because verbs cannot be derived from data or nodes.
The form that a verb takes is derived primarily from the subject. Depending on the properties of the subject, the verb changes. What other factors you need to determine so that the verb is output correctly may vary depending on the language.
# The components of Verb containers
Containers are specific to the kind of word they can hold and configure: noun containers, pronoun containers and verb containers. So you will store any grammar information about a verb in a verb container. A verb container cannot contain data from variables nor can it contain a branching.
To configure a verb container you have to enter the lemma of the verb. The software can propose a lemma via the automatic annotation feature or you enter it by hand.
What is a lemma?
In the NLG Cloud a lemma is the form of a verb that is used in a dictionary. For verbs, it is the infinitive, not inflected form of the verb that is considered as lemma. Examples:
- (EN): The lemma is the uninflected form, e.g. for runs, run, running, ran, the lemma is to run
- (DE)/(FR)/(ES): The lemma is the infinitive (singular present), e.g. for gesungen, singt, sang the lemma is singen, for va, alla, ira the lemma is aller.
# Verb container: Set configuration or reference to subject (container)
On the NLG platform there are two ways to configure verb containers:
- You put all the information (number, person and tense) into the forms of the verb container itself. This means that the verb will be inflected automatically in a correct manner for this given parameters, but it will not be adapted automatically if changes occur.
- You choose copy grammar from role to refer to the subject (role). In this way, the system receives the information from the container that has been assigned the subject role. The changes in the settings of this container will affect the flection of the verb and so maintain the grammatical correctness of this statement if changes in the subject container occur. To use this referential way you have to create a container for the subject of the statement and give it a role name.
# When to create a verb container?
As a thumb of rule for annotating and creating containers is to check if changes in the subject are possible for different data sets or conditions. To keep the grammar structure correct for you must create a container for the case that changes are expected.
- This means you have to create a verb container, if the subject of the statement is derived from a variable.
- If your subject and verb are on different branches, then you need to set a container if the subject variants on the branches can differ in person, number, or gender.
- And when using compound verbs for this case you only have to create containers for the auxiliary verb.
But it is also important to remember that the more verb containers you set, the clearer the grammatical framework of the statement will be, which will protect you against any errors when you change words or other containers.
- Annotating verbs in english statements and creating verb containers
- Working with auxiliary verbs.
This lesson is also available in German with German example statements:
- Verben annotieren und verb container erstellen
- Die Besonderheiten von Hilfsverben.
# Using verbs for good text quality in NLG
“Write with nouns and verbs, not with adjectives and adverbs.” (William , Elements of Style)
On the NLG platform, the values from the data are very prominent because they form the basis for the statements and they also can appear directly as variables in the text. Nevertheless, when writing, it is worthwhile to start designing the sentences with the verb, because this can significantly increase the quality of the texts. The more precise and appropriate the verbs are, the more clear and understandable the text will be. And in addition to this the verb is the element you have the highest level of freedom for creating in a text generation project, because it is usually not directly derived from your data, like other elements of your text.
Use verbs as often as you can: Opting for verbs over nouns will not only make your sentences flow better but also reduce your word count because you will avoid articles (e.g., an, the) and prepositions (e.g., in, on) which are required to make nouns work.
# Different verb categories
To be able to use the power of verbs skillfully, it is helpful to have an overview of verbs. Depending on their stylistic and grammar functions, verbs can be divided into two main categories. Actually, the auxiliary verbs as found above under compound verbs belong here as a third category. But they are not as crucial for the style of a text as the verbs of the other two categories.
- action verbs are all verbs that show action. They make sentences vivid. These are the verbs you should use whenever possible.
- The doctor wrote the prescription.
- Emma buys a ticket.
- Anna prepara i biscotti.
- Emil überbrückt die Pause.
- linking verbs (state of being verbs): They connect (link) the subject to another word or words that describe or rename the subject. Linking verbs show no action but refer to states. Therefore, a text with many lacks the vividness of a text with more action verbs.
|is, are,||look, feel, seem, sound, taste, smell||appear, become, seem|
These three verbs are always linking verbs: to be, to become and to seem. There are verbs that can be linking verbs in some cases, but are action verbs in other cases. If you can substitute the verb with is it is probably a linking verb.
linking verbs examples
- Tom ist ein netter Junge.
- Er bleibt gesund.
- Mont Blanc bedeutet weißer Berg.
- De shampoo is verkrijgbaar in vijf geuren.
# How to choose the right Verbs
One rule of thumb is to choose the verb that carry specific meaning over more generic verbs. In many cases, it is the linking verbs that tend to be less precise and so weaker than action verbs. Correct weak verbs by omitting them and replacing them with a more meaningful verb. Notice that you will need to add information as you specify the nature of the action.
If you search your text on linking verbs or weak verbs, you will also discover other stylistic weaknesses, such as nominalization or passive voice, because they usually go hand in hand.
# Replace linking verbs with action verbs
A good way to make writing better is to check for linking verbs and then try to replace them with action verbs. This eliminates many stylistic weaknesses (such as nominal or passive constructions).
no : Anna is a lover of country living. yes: Anna treasures country living.
nein: Das Smartphone ist als Kameraersatz einsetzbar. ja: Das Smartphone macht exzellente Bilder.
no :She is of the opinion that... yes: She thinks that...
# Look for the most concise verbs
But it is not always linking verbs that are too generic to make a text colorful and precise. Other verbs can be weak as well. Try to find the most accurate verb this is in some cases the one that does not relay on an adverb or another addition to be precise:
no: The dog ran quickly through the bushes. yes: The dog dashed through the bushes.
nein: Der Pressesprecher macht eine Mitteilung. ja: Der Pressesprecher teilt mit.
nein: Setzen Sie sich mit uns in Verbindung. ja: Schreiben Sie uns.
# What's next?
Do you want to learn more about writing on the NLG platform? The guide Write first (opens new window) shows you how the software assists you as a writer in NLG projects.