In the current TEI Guidelines, <usg> is defined as an element which marks up "usage information in a dictionary entry."

Prototypically, usage information is a label which can be attached at various points in the entry hierarchy in order to signal e. g. restrictions in terms of geographic regions, domains of specialized language or stylistic properties for the particular lexical item that it is attached to.

Consider for instance the following senses taken from a German entry for Pflaume “plum" where usage information is provided by labels taken from fixed sets of values for stylistic and diatopic properties:

<sense n="1">
        <def xml:lang="de">Frucht des Pflaumenbaums</def>
        <def xml:lang="en">fruit of the plum tree</def>
</sense>
<sense n="2">
        <usg type="register" norm="colloquial">ugs.</usg>
<def xml:lang="de">Pflaumenbaum</def>
<def xml:lang="en">plum tree</def>
</sense>
<sense n="3">
        <usg type="style" norm="casual">salopp</usg>
        <usg type="style" norm"expletive">Schimpfwort</usg>
<def xml:lang="de">ungeschickter, untauglicher Mensch</def>
<def xml:lang="en">awkward, ineligible  person</def>
</sense>
<sense n="4">
        <usg type="geo" norm="regional">landsch.</usg>
<usg type="style" norm="casual">salopp</usg>
<def xml:lang="de">anzügliche, leicht boshafte Bemerkung</def>
<def xml:lang="en">offensive, slightly mischievous remark</def>
</sense>

In contrast to the example above, the following sample features an occurrence of a more verbose usage description that does not rely on a fixed vocabulary. The sample is taken from a Serbian dialect dictionary. The quote in the dialect is further qualified by a usage hint: “(said by a peasant woman in the field in hot weather)" which provides a particular context in which the quote was recorded.

<cit type="example">
<quote>„Ду́ни, ве́тре, се́јче леб да пе́че"</quote>
<usg type="hint">(рекла сељанка на њиви за време врућине)</usg>
<bibl>(<placeName>Дубница</placeName>).</bibl>
</cit>

Types of Usage

1. You should always use @type with <usg>. Usage labels describe a wide range of linguistic phenomena. Classifying them should be considered a good practice.
2. Currently, the TEI Guidelines contain an overuse of <usg> for describing phenomena that could be covered by alternative, more narrowly defined TEI elements. It should be considered a good practice to use the most specific TEI element available.
3. It is good practice to normalize the values of the <usg> elements because dictionaries are not always consistent in the way they use their usage labels. For instance, we could have abbreviated and unabbreviated labels in the same dictionary.

We recommend the following restrictions on how you use <usg>:

• Do not use <usg type="lang"/> to mark up the name of a language in an etymological or other discussion. The recommended way to encode this information is using <lang> element within <etym>.

INCORRECT

<entry xml:id="MZ.RGJS.сајдисльк">
<form type="lemma"><orth>сајдисль́к</orth></form>
<gramGrp><pos>м</pos></gramGrp>
<usg type="lang">тур.</usg>
<sense>
<def>уважавање.</def>
</sense>
</entry>

CORRECT

<entry xml:id="MZ.RGJS.сајдисльк">
<form type="lemma"><orth>сајдисль́к</orth></form>
<gramGrp><pos>м</pos></gramGrp>
<etym>
<lang value="tr">тур.</lang>
</etym>
<sense>
<def>уважавање.</def>
</sense>
</entry>

…

• Do not use <usg type="hyper"> or <usg type="syn"> to mark lexical relations such as hyperonymy or synonymy. The recommended way to encode lexical relations is using the reference mechanism provided by <xr>.
• do not use <usg type="colloc"/> to mark up collocations. Instaed, opt for <colloc type="object"/> or <colloc type="preposition">.

In the following example from a Serbian-German-Latin dictionary, the sense of the adjective дангубан (futile, wasteful) is labeled with a collocate: н. п. посао (e.g. job).

INCORRECT

<entry xml:id="VSK.SR.дангубан" sortKey="дангубан">
    <form type="lemma"><orth>да̀нгубан</orth></form>
    <!-- etc.-->
    <gramGrp source="#digitalEd"><pos>adj.</pos></gramGrp>
    <sense xml:id="d1e162786">
        <usg type="colloc">н. п. посао</usg>
        <!-- etc.-->
    </sense>
</entry>

CORRECT

<entry xml:id="VSK.SR.дангубан" sortKey="дангубан">
    <form type="lemma"><orth>да̀нгубан</orth></form>
    <!-- etc.-->
    <gramGrp source="#digitalEd"><pos>adj.</pos></gramGrp>
    <sense xml:id="d1e162786">
        <colloc type="noun"><lbl>н. п.</lbl> посао</colloc>
        <!-- etc.-->
    </sense>
</entry>

• <usg type="hint"> should be used as fallback for cases where the usage information do not fall into one of the recognized cases discussed above.
• <usg type="dom">(domain) is used to specify the (e.g. scientific, professional etc.) domain in which the annotated lexical information (e.g. a sense, a cit etc.) is used.